Divalent Cations and Lipid Composition Modulate Membrane Insertion and Cancer-Targeting Action of pHLIP

Abstract

The pH-Low Insertion Peptide (pHLIP) has emerged as an important tool for targeting cancer cells; it has been assumed that its targeting mechanism depends solely on the mild acidic environment surrounding tumors. Here, we examine the role of Ca²⁺ and Mg²⁺ on pHLIP’s insertion, cellular targeting and drug delivery. We demonstrate that physiologically-relevant concentrations of either cation can shift the protonation-dependent transition by up to several pH units towards basic pH and induce substantial protonation-independent transmembrane insertion of pHLIP at pH as high as 10. Consistent with these results, the ability of pHLIP to deliver the cytotoxic compound monomethyl-auristatin-F to HeLa cells is increased several-fold in presence of Ca²⁺. Complementary measurements with model membranes confirmed this Ca²⁺/Mg²⁺-dependent membrane-insertion mechanism. The magnitude of this alternative Ca²⁺/Mg²⁺-dependent effect is also modulated by lipid composition—specifically by the presence of phosphatidylserine—providing new clues to pHLIP’s unique tumor targeting ability in vivo. These results exemplify the complex coupling between protonation of anionic residues and lipid-selective targeting by divalent cations, which is relevant to the general signaling on membrane interfaces.

Graphical Abstract

INTRODUCTION

The selective targeting of tumors is an important aspect of developing and optimizing anti-cancer therapies. In the past decade, the pH-Low Insertion Peptide (pHLIP) has emerged as a promising tool in tumor imaging and targeted drug delivery ^1–5. pHLIP remains stable in solution, yet it can also insert into membranes under mildly acidic (pH ~ 6) conditions and translocate cargo molecules (including anti-cancer drugs) conjugated to its C-terminus across the lipid bilayer ^5–8. Model in vitro studies suggest that the interaction of pHLIP with lipid membranes involves an initial interfacial binding of unfolded peptide at neutral pH, with subsequent insertion as a transmembrane helix upon acidification ^9,10. The latter is believed to be the molecular mechanism responsible for pHLIP’s selective targeting of tumors, which are known to produce a slightly more acidic extracellular microenvironment (pH ~ 7) than healthy tissues (pH 7.4) ^11,12. The advent of pHLIP has led to the development of other promising protonation-driven cancer targeting peptides like ATRAM ¹³ and TYPE7 ¹⁴, with a supposedly similar mode of action. Whether such a small difference in pH can solely explain the selectivity of pHLIP towards tumors in vivo, or whether other mechanisms are involved remains unknown.

One of the biggest challenges in deciphering the molecular mechanisms driving pHLIP tumor targeting arises from the often-overlooked discrepancy in experimental conditions between the studies aimed at delivering compounds into cancer cells and studies with model lipid vesicles. While delivery studies into cells are normally performed in the presence of physiological concentrations of divalent ions (present in growth media or extracellular fluid) ^6,15,16, experiments with vesicles have been, so far, been conducted in the absence of Ca²⁺ or Mg^{2+ 10,17–21}. Furthermore, most mechanistic studies have been performed on vesicles containing only phosphatidylcholine, thus neglecting the complex nature of the plasma membrane. For instance, we have previously demonstrated that variation in lipid composition leads to a significant variation in the apparent pK_a of the insertion of pHLIP ^19,20. Here, we examine the role of lipid composition and divalent cations in the pHLIP targeting of both model membranes and cancer cells. Using a combination of spectroscopic and cellular techniques, we show that the membrane insertion of pHLIP is strongly promoted by physiological concentrations of Ca²⁺ and Mg²⁺. Our results also suggest a strong regulatory link between membrane lipid composition and extracellular concentrations of Ca²⁺/Mg²⁺ on the membrane insertion of pHLIP, which may have general implications for signaling on membrane interfaces.

RESULTS

Ca²⁺ promotes the cellular interaction of pHLIP

The cellular interaction between pHLIP, labeled at its N-terminus with the fluorophore Alexa488 (A488-pHLIP), and human MDA-MB-231 breast cancer cells was inspected by fluorescence microscopy under two different conditions: (1) in the absence of divalent cations to reflect conditions most commonly used to characterize pHLIP interactions with vesicles, and (2) in the presence of 1.8 mM Ca²⁺, as a simplified mimetic of extracellular divalent cation concentration. In both cases, cells are incubated with A488-pHLIP at the appropriate pH, washed and imaged. No significant fluorescence was observed at pH 7.4 in the absence of divalent cations (Fig. 1b), pointing to a weak cellular interaction of pHLIP-A488 under these conditions. It should be noted that the formation of the interfacial form of the peptide is still expected, but likely removed by the washes performed before microscopy imaging. As expected, acidification of the solution to pH 5.0 resulted in a large increase in A488 fluorescence intensity (Fig. 1a), attributed to the pH-dependent membrane insertion character of pHLIP. At a pH of 7.4, the addition of 1.8 mM Ca²⁺ also resulted in a significant increase in A488 fluorescence intensity (Fig. 1c), suggesting that the presence of Ca²⁺ induces the interaction of A488-pHLIP with cellular membranes without the need for acidification.

FIGURE 1. Comparison of pH-dependent and Ca²⁺-dependent cellular targeting by pHLIP.

The cellular interaction of pHLIP N-terminally conjugated to an Alexa488 fluorophore (A488-pHLIP) was determined by fluorescence microscopy (a-f) and flowcytometry (g) using MDA-MB-231 cells. (a-c) An increase in A488 intensity (green), indicative of pHLIP cellular interaction, was observed either by acidification (a) or the addition of 1.8 mM Ca²⁺ at pH 7.4 (c) to a sample at pH 7.4 lacking Ca²⁺ (b). (d, f) Hoechst staining of MDA-MB-231 cells used in microscopy images. (g) Flow-cytometry was used to quantify the effects of [Ca²⁺] on the interaction of A488-pHLIP with MDA-MB-231 cells at pH 7.4 to mimic the extracellular pH of healthy cells or at mildly acidic pH 6.0. At both pH the presence of Ca²⁺ led to higher fluorescence, due to larger populations of membrane bound pHLIP. Normalized fluorescence represents the fold increase over cells incubated without Ca²⁺ at pH 7.4. Data are represented as mean values from which the error bars represent the standard error of the mean (n = 3). Representative microscopy images of conditions used in flowcytometry measurements are shown in Fig. S1.

The effect of Ca²⁺ on pHLIP cellular interactions was quantified by flow cytometry. At pH 7.4, the presence of Ca²⁺ led to a concentration-dependent increase in A488 fluorescence, with a 1.8 and 2.8-fold increase at 1.2 and 1.8 mM Ca²⁺, respectively (Fig. 1g). These results are consistent with the fluorescence increase observed by microscopy (Fig. 1a–f) and confirm the ability of Ca²⁺ to promote the interaction of pHLIP with cells. A similar trend was observed when cells were incubated with A488-pHLIP at pH 6.0, albeit with higher A488 intensity, likely due to the pH-dependent insertion of pHLIP. Representative microscopy images of the conditions used in the flowcytometry measurements are shown in Fig. S1. These results indicate that acidic conditions and Ca²⁺ work in tandem to induce the interaction of pHLIP with cells and its possible membrane insertion. Understanding the role of divalent cations as possible regulators of pHLIP membrane interaction is, therefore, critical to establishing the molecular mechanism of pHLIP under physiological conditions.

Ca²⁺ promotes pHLIP-mediated translocation of cargo molecules and cytotoxicity

We evaluated the effect that Ca²⁺ may have on the ability of membrane-inserted pHLIP to translocate cargo molecules into cancer cells using a construct, in which the cytotoxic agent MMAF (monomethyl auristatin F) is conjugated to the C-terminus of pHLIP via a disulfide bond (pHLIP-MMAF). We know from our previous reports, that pHLIP can translocate, release MMAF into the cytoplasm, and induce cancer cell death in a pH-dependent manner ^22,23. The presence of 1.2 mM Ca²⁺ greatly increased cytotoxicity at pH 7.4, indicated by an at least 10-fold decrease in the IC₅₀ from ≥ 8.5 μM in the absence of Ca²⁺ (Fig. 2a, black) to 0.9 ± 0.1 μM (Fig. 2a, red). The decrease in cellular viability observed in absence of Ca²⁺ is attributed to non-specific killing due to background levels of endocytosis upon association of pHLIP-MMAF to cell membranes, as we previously reported ^22,23. Nevertheless, these results are consistent with the increase in A488 intensity observed by microscopy and flow cytometry (Fig. 1) and indicates that Ca²⁺ promotes the interaction of pHLIP with cells, and possibly its transmembrane insertion.

FIGURE 2. Effect of pHLIP-MMAF treatment on the viability of cultured of HeLa cells.

The directional insertion of pHLIP results in the intracellular delivery of C-terminally conjugated MMAF, which is then released into the cytosol through cleavage of the connecting disulfide bond by the reducing cytosolic environment.. Measurements were performed at pH 7.4 and 6.0 in the presence or absence of 1.2 mM Ca²⁺ (the extracellular [Ca²⁺] at pH 7.4). At both pH tested the presence of Ca²⁺ led to higher cytotoxicity due to the more favorable insertion of pHLIP into cells, leading to increased translocation of compounds across the plasma membrane. Horizontal dotted lines and color-coded arrows indicate the concentration of pHLIP-MMAF required to kill 50% and 80% cells under each condition tested. (a) At pH 7.4, the presence of Ca²⁺ at pH 7.4 resulted in a 10-fold decrease of the IC₅₀ from ≥ 8.5 to 0.9 ± 0.1 μM (b) At pH 6.0, addition of Ca²⁺ led to a 5-fold decrease in IC₅₀ from 1.6 ± 0.1 to 0.3 ± 0.1 μM. The increased steepness of the transition led to a 10-fold difference in the [pHLIP-MMMAF] required to kill 80% of cells. All measurements were normalized to the media control (0 μM, pH 7.4), as 100% cell viability, in which the error bars represent standard error of the mean (n = 3).

Treating cells with pHLIP-MMAF at pH 6.0 in the absence Ca²⁺ led to a 5-fold decrease in IC₅₀ (1.6 ± 0.1 μM; Fig. 2b, black) as compared to pH 7.4 (≥ 8.5 μM; Fig. 2a, black). Inclusion of Ca²⁺ in the treatment solution, resulted in another 5-fold decrease in IC₅₀ (0.3 ± 0.1 μM; Fig. 2b, red). The addition of Ca²⁺ not only resulted in lower IC₅₀ values, but also in a steeper cytotoxicity response. We attribute this change to a more favorable transmembrane insertion of pHLIP and subsequent drug delivery in the presence of Ca²⁺. The increased cooperativity led to a 10-fold difference in the concentration of pHLIP-MMAF required to kill 80% of cells between both conditions tested. Where 0.53 μM pHLIP-MMAF was sufficient to kill 80% of cells in the presence of Ca²⁺, while an equimolar concertation of pHLIP-MMAF only resulted in 23% cytotoxicity in the absence of Ca²⁺.

Our cytotoxicity results, together with the fluorescence microscopy (Fig. 1a–c) and flow cytometry measurements (Fig. 1g), show that the presence of physiological Ca²⁺ concentrations promote the ability of pHLIP to translocate compounds across cellular membranes. Physiological conditions must therefore be considered when characterizing and optimizing the interaction and mechanism of pHLIP-like cancer targeting systems.

Ca²⁺-dependent transmembrane insertion into model POPC membranes

The common way of characterizing pH-dependent membrane interactions of pHLIP is through a combination of changes in its tryptophan (Trp) emission maximum and circular dichroism (CD) spectrum ^10,17–21. Here, we used these techniques to compare the spectral responses of pHLIP observed at neutral pH in the presence of Ca²⁺ to those observed upon transmembrane insertion at acidic pH.

The spectral hallmarks of the interfacial (State II) and transmembrane (State III) states of pHLIP and have been firmly established by previous studies ^10,19,24. Our data collected in the absence of Ca²⁺ are consistent with these hallmarks (Fig. 3). Specifically, Trp fluorescence maximum of the interfacial State II of pHLIP (populated in the presence of large unilamellar vesicles (LUV) composed of POPC at pH 8.0) is 354 nm (Fig. 3a, black), indicating that the Trps are in a relatively polar environment. In State III (populated at pH 4), the maximum is blue-shifted to 342 nm (Fig. 3a, blue), consistent with a more nonpolar environment. Our data collected at pH 8, but in the presence of 2.0 mM Ca²⁺ (Fig. 3a, red), clearly indicate that, under these conditions, pHLIP exhibits fluorescence hallmarks of the transmembrane State III.

FIGURE 3. Comparison of pH-dependent and Ca²⁺-dependent insertion of pHLIP into model POPC membranes.

The transmembrane insertion of pHLIP was characterized by (a) Trp fluorescence and (b) circular dichroism (CD) in LUV and by oriented circular dichroism (OCD) in flat multilayers. (a) At pH 8.0, the interfacial form pHLIP presents a Trp position of maximum of 354 nm (black). Acidification of the sample to pH 4.0 leads to a 12 nm blue shift to 342 nm, characteristic of pHLIP’s transmembrane State III (blue) ¹⁰. Introducing 2.0 mM Ca²⁺ (a proxy for extracellular divalent cation concentration) while maintaining the pH constant at 8.0 results in a similar 9 nm blue shift to 345 nm (red). (b) CD measurements in the absence of Ca²⁺ at pH 8.0 (black) and 4.0 (blue) show characteristic signals for unstructured peptides and α-helices respectively, consistent with its previously reported pH-dependent transmembrane insertion ¹⁰. The addition of 1.0 mM Ca²⁺ at pH 8.0 resulted in a typical α-helical spectrum (red) with a similar ellipticity as the one observed at pH 4.0 in the absence of Ca²⁺. (c) OCD measurements were performed in the presence of Ca²⁺ at pH 8.0 (red); and in the absence of Ca²⁺ at pH 4.0 (blue) and 8.0 (black). The latter two conditions correspond to those of the transmembrane State III and interfacial State II of pHLIP respectively. The Ca²⁺-inserted and pH-inserted samples exhibited similar spectra with a single minimum ~ 228 nm, characteristic of the OCD spectra of transmembrane α-helices ²⁵.

A similar pattern is observed with conformational changes measured by CD spectroscopy for solution samples containing LUV (Fig. 3b) or samples in oriented multilayers (Fig. 3c). In the absence of Ca²⁺ at pH 8.0, the CD spectrum of pHLIP showed a single minimum ~ 200 nm, typical of unstructured peptides and consistent with the largely unfolded interfacial State II (Fig. 3b, black) ¹⁰. In contrast, the CD spectra of pHLIP collected at pH 4.0 in the absence of Ca²⁺ (transmembrane State III) and at pH 8.0 in the presence of Ca²⁺ were equivalent (Fig. 3b, blue and red), with both spectra presenting a double minimum at ~ 209 and 222 nm, characteristic of α-helices.

The pH- and Ca²⁺-induced helices also have similar orientations with respect to membrane normal, as revealed by the oriented circular dichroism (OCD) measurements performed in lipid multilayers (Fig. 3c). Both OCD spectra (red and blue curves) have a single sharp minimum at 228 nm, characteristic of transmembrane α-helices ^28,29,25. The spectra are also consistent with the previously reported OCD spectrum of the transmembrane State III of pHLIP ^26,27. As expected, these spectra are quite different from that measured at pH 8.0 in the absence of Ca²⁺ (black line), i.e., under conditions corresponding to the interfacial State II. Together, all fluorescence and CD measurements indicate that the Ca²⁺-inserted state resides in the lipid bilayer in the same conformation as the conventional pH-inserted transmembrane State III of pHLIP.

Effects of lipid composition on Ca²⁺-mediated insertion of pHLIP

One of the hallmarks of cancer cells, is the change in lipid composition of the outer leaflet of the plasma membrane, specifically the increase in phosphatidylethanolamine (PE) and phosphatidylserine (PS) ^28–32. Previously we have established that additions of many non-PC lipids, including PE and PS, have a substantial effect on the pH-dependent insertion of pHLIP ^19,20. Here, we compared how the membrane insertion of pHLIP into pure POPC bilayers and bilayers containing 25% of either POPE or POPS, is affected by the presence of Ca²⁺.

Trp emission spectra were collected in the presence of POPC LUV at pH 10.0 and increasing Ca²⁺ concentrations to characterize the Ca²⁺-mediated membrane insertion of pHLIP. The positions of their maxima were then compared to the known maxima of the interfacial State II (Fig. 4a red dashed line) and transmembrane State III (Fig. 4a blue dashed line) forms of pHLIP at pH 10.0 and 4.0, respectively. Incremental addition of Ca²⁺ at pH 10.0 led to progressive decreases in pHLIP Trp maxima until they approached saturation at 343 nm in 5.0 mM Ca²⁺ (Fig. 4a, filled circles). The positions of maxima observed at 4.0–5.0 mM Ca²⁺ match those observed for the transmembrane State III of pHLIP. Together with the α-helical conformation determined by CD (Fig. 3b) and transmembrane orientation determined by OCD (Fig. 3c), these results confirm that Ca²⁺ induces the transition of pHLIP into its transmembrane State III. These results also imply that 2.0 mM Ca²⁺, a proxy for extracellular divalent cation concentration, is sufficient to induce the transmembrane insertion of ~ 70% of pHLIP present in the sample at pH 10.0. The Ca²⁺-dependent blue shifts were reversible in the presence of EDTA (Fig. 4a, open circles).

FIGURE 4. Lipid modulation of Ca²⁺-dependent membrane interactions of pHLIP.

Ca²⁺ titrations at constant pH 10.0 and pH-titrations in the presence of 2.0 mM Ca²⁺ were performed in the presence of POPC (a, b), 25POPE:75POPC (c, d), or 25POPS:75POPC (e, f) LUV. (a, c, and e) The intrinsic fluorescence of pHLIP was measured at pH 10.0 and increasing [Ca²⁺]. Positions maximum for the two known interfacial forms of pHLIP (State II in POPC and II^S in non-POPC LUV) and transmembrane State III are indicated by the horizontal dashed lines ²⁰. Incremental addition of Ca²⁺ at pH 10.0 resulted in large concentration dependent blue shifts regardless of lipid composition (filled circles). In all cases saturation was achieved at the Trp positions of maxima expected for the transmembrane State III of pHLIP. The large spectral changes observed show that Ca²⁺ induces the transmembrane insertion of pHLIP without any acidification, even at pH 10.0 (red arrows). Addition of EDTA reversed the Ca²⁺-dependent spectral changes (open circles). (b, d, and f) The presence of 2.0 mM Ca²⁺ promoted the pH-dependent insertion of pHLIP into all LUV compositions tested, indicated as an increase in its transition pK_a (blue arrows). A 1.5 pH unit increase was observed POPC from 6.1 ± 0.1 to 7.6 ± 0.1 and a 1.5 pH shift from 5.7 ± 0.1 to 7.2 ± 0.1 was calculated for the 25POPE:75POPC LUV. While a larger 2.4 pH unit shift from 5.4 ± 0.1 to 7.8 ± 0.1 was detected in 25POPS:75POPC LUV. The transition pK_a of each curve is graphically indicated by color-coded vertical dotted lines.

Control measurements in the absence of membranes were performed to confirm that the Ca²⁺-dependent Trp blue shifts were not caused by aggregation-induced Trp burial. No significant changes in the Trp emission spectrum of pHLIP were observed after the addition of Ca²⁺ (356 nm, Fig. S2a, red) as compared to samples lacking Ca²⁺ (357 nm, Fig. S2b, black). This indicates that the results observed in the presence of LUV are membrane-dependent and not due to peptide aggregation. The small 1 nm blue shift from 357 nm after the addition of Ca²⁺ suggests, however, that Ca²⁺ likely affects the unstructured “conformation” of pHLIP in solution. Moreover, the effect of Ca²⁺ concentration on the aggregation of POPC LUV was also tested by light scattering measurements. No Ca²⁺-dependent membrane aggregation of POPC LUV was detected under the concentrations used in this study (Fig. S3a).

We next characterized the effect of Ca²⁺ on the protonation-dependent insertion of pHLIP into POPC LUV using intrinsic Trp fluorescence as a function of pH. In the absence of Ca²⁺, the pH-dependent insertion of pHLIP into POPC bilayers yielded a protonation-dependent membrane insertion pK_a of 6.1 ± 0.1 (Fig. 4b, black), consistent with previous reports ^21,33. The presence of 2.0 mM Ca²⁺ (Fig. 4b, brown) led to two prominent changes in the pH-dependent insertion of pHLIP: 1) A pronounced decrease in the Trp fluorescence maximum at pH 10.0 (Fig. 4b, red arrow) and 2) a significant shift of the titration curve towards basic pH (Fig. 4b, blue arrow). The 7 nm blue shift at pH 10.0 from 354 nm to 347 likely reflects the Ca²⁺-mediated membrane insertion of pHLIP independent of pH. The presence of 2.0 mM Ca²⁺ also led to a 1.5 pH-unit shift in the pK_a of the titration curve to 7.6 ± 0.1 (Fig. 4b, brown) compared to 6.1 ± 0.1 in the absence of Ca²⁺ (Fig. 4b, black). These results, together with our measurements at constant pH 10.0 (Fig. 4a), indicate a dual effect for Ca²⁺ on the interaction of pHLIP with POPC membranes: 1) Ca²⁺ induces the transmembrane insertion of a significant fraction of the pHLIP population without need of acidification. 2) Ca²⁺ promotes the pH-dependent transmembrane insertion of the remaining interfacial population.

Measurements were replicated with LUV containing 25% of the zwitterionic lipid POPE or the anionic lipid POPS. The addition of POPE or POPS-containing LUV did not result in the characteristic 4 nm Trp blue shift associated with the interfacial State II of pHLIP in the absence of Ca²⁺. This effect has been previously described and relates to the formation of the spectroscopically silent interfacial State II^S (Fig. 4c and 4e, black dashed line) ²⁶. Regardless of the starting interfacial form, the incremental addition of Ca²⁺ resulted in concentration-dependent Trp blue shifts. The observed endpoint between the different LUV used were similar with a 12 nm decrease from 356 nm in the absence of Ca²⁺ to 344 nm at 5.0 mM Ca²⁺ observed for 25POPE:75POPC (Fig. 4c). While a 14 nm decrease from 356 nm to 342 nm was observed in the case of 25POPS:75POPC LUV under the same conditions (Fig. 4e). Compared to zwitterionic POPC and POPE membranes, the Ca²⁺-dependent transition of pHLIP into its transmembrane State III was significantly steeper in the presence of the anionic POPS lipids. This suggests that the POPS lipids exposed in the membranes of cancer cells promote the Ca²⁺-dependent membrane insertion of pHLIP.

The insertion of pHLIP was also measured for 25POPE:75POPC (Fig. 4 c, d) and 25POPS:75POPC LUV (Fig. 4 e, f). Overall the effects of Ca²⁺ in these lipid compositions are qualitatively similar to those in pure POPC (Fig. 4 a, b), and are characterized by the induction of insertion at neutral and basic pH (arrow 1, Figs. 4 b, d, f) and changes in pK_a for acid-induced insertion (arrow 2). The quantitative effects, however, are lipid dependent. In the absence of Ca²⁺, the lipid-dependent changes in pK_a are consistent with those reported previously ^17,19,20. Namely, the insertion in mixed lipid compositions requires stronger acidic environment (e.g., pK_a = 5.8 ± 0.1 in 25POPE:75POPC LUV and a pK_a = 5.4 ± 0.1 in 25POPS:75POPC LUV) as compared to pure POPC (pK_a = 6.1 ± 0.1 for POPC). The presence of 2.0 mM Ca²⁺ leads to a blue shift in the Trp fluorescence maxima at pH 10.0 as well as a move of the titration curves towards alkali pH, regardless of lipid composition. The magnitude of the associated changes, however, is modulated by lipid composition and were particularly prominent for the protonation-dependent insertion of pHLIP. In the case of the zwitterionic 25POPE:75POPC LUV the presence of Ca²⁺ led to a 1.5 pH unit increase in pK_a from 5.7 to 7.2 (Fig. 4d). Meanwhile, the anionic 25POPS:75POPC LUV showed a larger 2.4 pH unit increase from 5.4 to 7.8 between samples lacking and containing Ca²⁺, respectively (Fig. 4d). Together with the Ca²⁺ titration measurements, these results reveal a strong lipid dependence for the Ca²⁺-dependent modulation of pHLIP membrane insertion with a strong preference towards anionic POPS bilayers.

Effects of Ca²⁺on the thermodynamics of pHLIP membrane insertion

Our measurements of pHLIP insertion in the presence of Ca²⁺ demonstrated the simultaneous presence of two distinct effects: (i) induction of protonation-independent membrane insertion at neutral and basic pH (Fig. 4 a, c, and e), and (ii) shift of the protonation-dependent insertion toward more neutral pHs (Fig. 4b, d, and f). To quantify the thermodynamics of both processes, we repeated the same experiments done at 2.0 mM Ca²⁺ (presented in Fig. 4) at intermediate [Ca²⁺] (Fig. S4).

The thermodynamics of the protonation-independent membrane insertion of pHLIP was characterized with the free energy term ΔG_TM^[Ca2+]. This parameter was derived from the fractional insertion, f_TM, which was calculated using the changes in fluorescence emission maximum (Eq. 4) at pH 10.0 for each [Ca²⁺] tested in our pH-titrations (Fig. S4). This thermodynamic approach is essentially identical to that used in the derivation of the so-called biological hydrophobicity scale ³⁴ and in the quantitation of insertion by molecular dynamics simulations ³⁵. The results presented in Fig. 5a demonstrate that all lipid compositions present a similar susceptibility to this process, particularly at [Ca²⁺] ≥ 1.0 mM. At lower [Ca²⁺], however, membranes composed purely of POPC lipids show a lower susceptibility than the mixed lipid compositions. Nevertheless, at least half of the population of pHLIP is already inserted at the physiologically relevant concentrations of extracellular Ca²⁺ at neutral and acidic pH.

FIGURE 5. Effects of Ca²⁺ on thermodynamics of membrane insertion of pHLIP.

The membrane insertion of pHLIP into POPC (orange), 25POPE:75POPC (purple), and 25POPS:75POPC (green) LUV was measured as a function of pH at [Ca²⁺] ranging from 0 – 2.0 mM (for complete set of raw data see Fig. S5). (a) The effect of Ca²⁺ on the free energy of pH-independent insertion of pHLIP ΔG_™^[Ca2+], which were calculated using Eq. 4. This free energy is based on the fractional insertion f_™ measured at pH 10.0 and equals ΔG_™^[Ca2+] = 0 kcal/mol when half of pHLIP molecules are inserted (dashed line). (b) The effect of Ca²⁺ on the free energy of protonation-dependent insertion of pHLIP, calculated from the pK_a’s according to Eq. 3. For each lipid composition, the data are plotted as a difference between the free energy in the presence of Ca²⁺ (ΔG_™^{H+, Ca2+}) and the corresponding free energy in the absence of Ca²⁺ (ΔG_™^H+). By definition, all curves originate from the common origin point (i.e., ΔG_™^{H+, Ca2+} − ΔG_™^H+ = 0 kcal/mol at 0 mM Ca²⁺ for each lipid composition). (c) The cross-correlation of the free energies determined for the protonation-dependent and protonation-independent insertion of pHLIP. For each concentration of Ca²⁺, the pH-independent mode (obtained from panel a) is plotted against the free energy difference for pH-dependent mode (obtained from panel b). The lines represent the linear approximations for each lipid composition with the following parameters: Slopes: POPC = 0.4 ± 0.1, 25POPE:75POPC = 1.0 ± 0.2, and 25POPS:75POPC = 2.7 ± 0.2. Intercepts (Y value at X = 0): POPC = −1.8 kcal/mol, 25POPE:75POPC = −1.8 kcal/mol, and 25POPS:75POPC = −2.6 kcal/mol.

To examine the effects of Ca²⁺ on the thermodynamics of protonation-dependent insertion, we used the same procedure than in our previous publications on the lipid-dependent modulation of pHLIP insertion ^19,20. Specifically, we used pH-titration plots to determine the apparent pK_a for the insertion transition (Eq. 2), from which the free energy was calculated using Eq. 3. The results presented in Fig. 5b show that the effect in pure POPC and 25POPE:75POPC mixtures is almost indistinguishable and saturates at about 2 kcal/mol after 1.0 mM Ca²⁺ is present. In contrast, the gain in free energy of protonation-dependent insertion in 25POPS:75POPC is more pronounced and reaches 3.0 kcal/mole at 2.0 mM Ca²⁺.Interesting observations come from a cross-correlation analysis, in which the protonation-dependent gain in free energy is plotted against the protonation-independent (calcium-induced) free energy of insertion (Fig. 5c). Remarkably, the correlations are linear, yet the slope varies dramatically with lipid composition, suggesting that the lipid composition modulates the interplay of various modes of pHLIP membrane insertion on a thermodynamic level (see discussion).

Comparison of Ca²⁺ and Mg²⁺ dependent effects

To determine whether other divalent cations exhibited an effect similar to that of Ca²⁺, we repeated our experiments with LUV in the presence of 2.0 mM Mg²⁺. Similar to Ca²⁺, incremental increases in [Mg²⁺] at constant pH 10.0 in the presence of POPC LUV showed concentration-dependent Trp blue shifts (Fig. S5a). The endpoint at saturating Mg²⁺ concentrations, however, did not reach the expected Trp fluorescence maximum for the transmembrane State III. Instead, the signal saturated at ~ 348 nm, 6 nm higher than the 342 nm expected when the entire population is present in its transmembrane conformation (Fig. 4a). Similarly, Mg²⁺ titrations in the presence of zwitterionic 25POPE:75POPC LUV showed a reduced effect for Mg²⁺ compared to Ca²⁺, resulting in a 9 nm blue shift to 347 nm from 356 nm (Fig. S5c), compared to the 12 nm blue shift in the presence of Ca²⁺ (Fig. 4c). Measurements with 25POPS:75POPC also showed a reduced effect when compared to Ca²⁺. In the case of these anionic bilayers, the presence of Mg²⁺ only resulted in a 4 nm maximal blue shift to 352 nm from 356 nm in the absence of divalent cations (Fig. S5e). The inability of Mg²⁺ to yield the same blue shifts as equimolar concentrations of Ca²⁺ indicates that Mg²⁺ is less efficient in promoting the transition of pHLIP into its transmembrane state. Despite these weaker effects, the ability of Mg²⁺ to induce the transmembrane insertion of pHLIP without lowering the pH was confirmed by OCD measurements (Fig. S6). The intermediate Trp fluorescence maxima were therefore interpreted as a combination of transmembrane and interfacial pHLIP populations.

We also performed pH-titrations in the presence of 2.0 mM Mg²⁺ and vesicles with different composition (POPC, 25POPE:75POPC, and 25POPS:75POPC). As in the case of Ca²⁺, the presence of Mg²⁺ led to a decrease in the Trp fluorescence maximum at pH 10.0 compared to measurements in the absence of divalent cations (Fig. S5b, d, and f). This is consistent with the Mg²⁺-dependent insertion of pHLIP into membranes without the need for protonation. This effect appears to be more moderate with Mg²⁺ than with Ca²⁺ (Fig. 4b, d, and f). The decrease in Trp position of maximum at pH 10.0 was accompanied by a shift of the curves towards alkali pH (Fig. S5, blue arrows). The magnitude of this shift, however, was reduced compared to equimolar concentrations of Ca²⁺. In the case of POPC, the presence of 2.0 mM Mg²⁺ led to a 1.0 pH unit increase in its pK_a from 6.1 ± 0.1 to 7.1 ± 0.1 (Fig. S5b), compared to the 1.5 pH units gain in the case of Ca²⁺ (Fig. 4b). Similarly, the presence of 2.0 mM Mg²⁺ increased the pK_a of the pH-dependent insertion of pHLIP into 25POPE:75POPC LUV by 0.3 pH unit shift from 5.7 ± 0.1 to 6.1 ± 0.1 (Fig. S5d), compared to 1.0 pH unit in the presence of Ca²⁺ (Fig. 4d). A shift of the pH-titration curve towards a more basic pH range was also observed for 25POPS:75POPC bilayers when 2.0 mM Mg2+ was present (Fig. S6f). The shorter buffering range of HEPES buffer compared to the traditional phosphate buffer (see methods for details), however, limited our measurements to pH ≥ 6.0. This prevented us from accurately determining the pKa under these conditions, but it was estimated to be ≤ 6.0 compared to 5.4 in the absence of Mg²⁺.

Our measurements with Mg2+ show that the modulation of pHLIP membrane insertion is not specific to Ca2+, suggesting a more general role of divalent cations in this process. Nevertheless, in all lipid compositions, the addition of Ca2+ had a more prominent effect compared to that of Mg2+. Preferential effects of specific divalent cations, particularly for calcium, on protein membrane interactions have been previously observed for other systems such as α-synuclein 36, highlighting the importance of proper cation conditions when characterizing membrane active proteins.

DISCUSSION

The results presented here provide new insights into our understanding of the physicochemical mechanisms underlying the ability of pHLIP to target tumors and other disease tissues. The standard pH-dependent targeting mechanism involves protonation of key acidic residues (e.g., D14 and D25) ^21,33,37 and subsequent membrane insertion. At mildly acidic pH (pH < 6.0), pHLIP adopts a transmembrane conformation, with characteristic spectroscopic signatures that have been well-established for model membrane systems ^10,24,26. Here, we demonstrated that addition of either Ca²⁺ or Mg²⁺ at neutral or even basic pH, results in the same fluorescence and CD changes as those usually observed at low pH (Fig. 3). The addition of divalent cations produces two independent effects: (1) pH-independent insertion at neutral and basic pH (arrows 1 in Fig. 4b, d, f and Fig. S6b, d, and f), and (2) modulation of pKa for the acid-driven insertion (arrows 2 in Fig. 4b, d, f and Fig. S6b, d, and f). Both effects have been found to be modulated by the lipid composition (Figs. 4, 5), which might be a contributing factor in tumor targeting ^28–32.

We illustrate the effects of divalent cations on the bilayer insertion of pHLIP using the schemes in Fig. 6a. The top panel contains a standard pH-dependent insertion scheme ^19,20,24, which depicts the conversion of the interfacial unfolded state, populated at high and neutral pH, and the transmembrane helical state, populated al low pH. Insertion is characterized by the transition pK_a, which is influenced by both mutations of pHLIP ^18–20,38 and variations in lipid composition ^{17,19,20,27,39}. The free energy for this protonation-dependent transition, ΔG_TM^H+, is calculated from the pK_a’s using Eq. 3. The presence of divalent cations investigated here requires revising this standard pH-dependent insertion scheme (Figure 6a, lower panel). Indeed, the presence of Ca²⁺ introduces two major effects: induction of the protonation independent insertion at neutral and acidic pH (red arrow) and the modulation of the acid-induced insertion (blue arrow). The first transition is characterized by a free energy ΔG_TM^[Ca2+], calculated using Eq. 4; and the results are presented in Fig. 5a. The second transition is best described by the difference in free energies of protonation in the presence and absence of Ca²⁺ ΔG_TM^{H+, Ca2+} - ΔG_TM^H+, calculated separately for each lipid composition (Fig. 5b). The cross-correlation analysis of the free energies of pH-dependent and pH-independent effects of Ca²⁺ is shown in Fig. 5c and discussed below.

FIGURE 6. Thermodynamic scheme of pHLIP membrane interactions.

(a) Thermodynamic scheme of pHLIP membrane insertion in the presence and absence of Ca²⁺. In the absence of divalent cations, the only factor driving the membrane insertion of pHLIP is its protonation-dependent insertion with a ΔG_™^H+. The presence of Ca²⁺ leads to a more favorable protonation-dependent insertion (lower acidity levels required for insertion) with a free energy of ΔG_™^{H+, Ca2+} − ΔG_™^H+. Additionally, Ca²⁺ introduces a new pH-independent membrane insertion free energy with a ΔG_™^[Ca2+]. While the scheme only shows the free energies in the presence of Ca²⁺, other divalent cations (e.g., Mg²⁺) with corresponding free energies are also expected to modulate pHLIP insertion (see Supplementary Fig. S5). (b) Schematic representation of pHLIP fragments in a putative unstructured and helical conformation (green) bound to POPS lipids in the presence of Ca²⁺ (purple). The coordination of Ca²⁺ by Asp 12 or Asp 25 in pHLIP and the known possible coordination sites of divalent cations ^40,42,43 in POPS lipids is indicated by dashed red lines. In case of POPC, the interactions coordinating to serine group on the lipid will be absent (e.g., top two dashed lines), resulting in weaker coupling of lipid-Ca²⁺-protein interactions.

The interplay between lipid composition and calcium concentrations on pHLIP membrane insertion is best illustrated by the plots of the corresponding free energies for pH-dependent and calcium-induced insertion (Fig. 5c). For pure POPC, the dependence is rather shallow, suggesting a limited ability of pH-dependent modulation by Ca²⁺ on the insertion of pHLIP. For a 25POPE:75POPC lipid composition, the slope equals one, suggesting that both effects are equally important. The presence of 25% POPS results in a dramatic increase of the slope, indicating that POPS when coupled with Ca²⁺ strongly promotes pH-dependent insertion. We speculate that the reasons for this lipid-dependent modulation of insertion thermodynamics may be related to the ability of different lipids to co-coordinate divalent cations with the acidic sidechains of pHLIP (Fig. 6b).

It has been long established that interactions of cations with lipid bilayers depend on the nature of the cation and the lipid headgroups ³⁹. However, the complex nature of these interactions is still an active area of experimental and computational research. For example, Melcrova and co-authors recently demonstrated that lipid membranes have substantial calcium-binding capacity, with several types of binding sites present ⁴⁰. It is worth noting that the mechanism of action of Ca²⁺ and Mg²⁺ on the insertion of pHLIP is not likely to involve changes in surface potential. This is evident from the fact that the effects of divalent cations led to pK_a shifts in the opposite direction as those caused by the decrease in surface potential due to the additions of high concentrations of monovalent cations ⁴¹ or the decrease of anionic lipid content ^19,20. While the exact nature of the coupling between lipid headgroups and divalent cations in the insertion of pHLIP requires further investigation, it is reasonable to assume that the ability of Ca²⁺ and Mg²⁺ to coordinate bonding with lipid headgroups and acidic residues, plays an important role. This effect has been illustrated in Fig. 6b, which shows the putative interaction between POPS lipids and the unstructured and helical conformations of pHLIP (Fig. 6b, green) in the presence of Ca²⁺. We hypothesize that divalent cations (Fig. 6, purple spheres) mediate this interaction through anionic residues in pHLIP (e.g., D14 and D25, indicated in red), which are involved in its protonation-mediated folding and insertion ^9,21,33,37. It should be noted that in the case of POPS we expect Ca²⁺ and Mg²⁺ to be coordinated not just by the phosphates present in the lipids and anionic groups in pHLIP (dashed lines), but also by the POPS anionic head group or possibly its ester oxygens ^40,42,43. Zwitterionic lipids on the other hand lack a coordination group in their head group, which would lead to lower binding of divalent cations and a lessening of the Ca²⁺/Mg²⁺ - mediated effects on pHLIP insertion. Similar interactions have been suggested for other membrane-binding proteins in the literature ^36,44–46.

How does this new knowledge on the role of Ca²⁺ and Mg²⁺ change our understanding of the mechanism of selective targeting of tumor cells by pHLIP? The conventional explanation relies on the mild acidification produced by tumors, which lowers the outside pH to ~7.0, supposedly providing the necessary selectivity ¹⁵. While plausible for zwitterionic lipids, this explanation does not hold in the presence of anionic POPS. As demonstrated in Fig. 4f (black) and in our previous publications ^19,20, no pHLIP insertion into 25POPS:75POPC bilayer is observed at pH >5.8. Since the transfer of phosphatidylserine to the outer leaflet of the plasma membrane is known to occur in cancerous cells, other mechanisms besides acidification are expected to contribute to targeting selectivity. Here, we demonstrate that divalent cations constitute an important factor, modulating interactions of pHLIP with model membranes and cells alike. The presence of physiological concentrations of Ca²⁺ improves both cellular labeling (Fig. 1) and drug delivery by pHLIP (Fig. 2). This has important implications for the optimization of cancer imaging and targeted drug delivery strategies that rely on pHLIP variants and similar membrane-active peptides. Future strategies should account for both pH-dependent and divalent cation-dependent mechanisms as well as lipid variations in targeted cells. Moreover, the modulation of the protonation-dependent insertion pathways with Ca²⁺ and Mg²⁺ may be important for conformational switching in other physiological processes (e.g., entry of bacterial toxins ^47,48 and insertion of apoptotic regulators ^49,50).

MATERIALS AND METHODS

pHLIP solid-phase synthesis

pHLIP with a cysteine residue at its N-terminus, (H₂N-GCEQNPIYWARYADWLF-TTPLLLLDLALLVDADEGTG-CONH₂) or its C-terminus (H²N-GGEQNPIYWARYADWLF-TTPLLLLDLALLVDADEGTCG-CONH₂) were prepared by Fmoc solid-phase synthesis on an automated microwave peptide synthesizer (CEM Liberty Blue) using rink amide resin (CEM, 0.19 mmol/g loading capacity). The peptides were purified via RP-HPLC (Phenomenex Luna prep 5 μm Omega Polar C18 250 × 21.20 mm; flow rate 5 mL/min; phase A: water 0.1% TFA; phase B: acetonitrile 0.1% TFA; gradient 60 min from 95/5 A/B to 0/100 A/B. The purity of the peptides was determined by RP-HPLC (Phenomenex Luna 5 μm Omega Polar C18 250 × 100 mm; flow rate 5 mL/min; phase A: water 0.01% TFA; phase B: acetonitrile 0.01% TFA; gradient 60 minutes from 95/5 A/B to 0/100 A/B), and their identity was confirmed via MALDI-TOF MS (Shimadzu 8020).

Preparation of conjugates

Conjugating Alexa488 C5 maleimide (Invitrogen #A10254) to the N-terminus cysteine of pHLIP was achieved by dissolving pHLIP in DMF with 50 mM HEPES, pH 7.2, followed by the addition of 1 molar equivalent of Alexa488 C5 maleimide. The solution was flushed with nitrogen and mixed at room temperature for 4 hours. pHLIP-MMAF was prepared by conjugating Py-ds-Prp-MMAF (Levena Biopharma) to pHLIP with a C-terminus cysteine residue utilizing the same procedure. The desired pHLIP conjugates were isolated using the same techniques described for the pHLIP peptides. The purity of the pHLIP conjugates were determined by RP-HPLC, and their identity was confirmed by MALDI-TOF MS. Alexa488-pHLIP: purity >99%; calculated (M+H+) = 4874, found (M+H+) = 4874. pHLIP-MMAF: purity >98%; calculated (M+H+) = 5030, found (M+H+) = 5030. The Alexa488-pHLIP conjugate was quantified at 493 nm by UV/Vis absorbance spectroscopy using the molar absorption coefficient of Alexa488 C5 maleimide (72,000 M⁻¹⋅cm⁻¹). The pHLIP-MMAF conjugate was quantified at 280 nm using the molar absorption coefficient of pHLIP (13940 M⁻¹⋅cm⁻¹). The conjugates were then lyophilized in 10⁻⁸ mole aliquots.

Cell culture

Human cervical adenocarcinoma HeLa and human breast adenocarcinoma MDA-MB-231 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) high glucose supplemented with 10% fetal bovine serum (FBS), 100 units/mL penicillin, and 0.1 mg/ mL of streptomycin. The cells were cultured in a humidified atmosphere of 5% CO2 at 37 °C.

Cell viability assay

HeLa cells were plated in 96-well plates at a cell density of 5,000 cells/well and incubated overnight. pHLIP-MMAF was solubilized in an appropriate volume of Dulbecco’s phosphate buffered saline (DPBS) containing 1.2 mM calcium (Sigma #D8662) or lacking calcium (Sigma #D8537) pH 7.4, so that upon pH adjustment the desired treatment concentration (10 μM) was obtained. The samples were then gently sonicated for 30–60 s using a bath sonicator (Branson Ultrasonics). After removal of cell media, the cells were washed twice with phosphate buffered saline (PBS), and then pHLIP-MMAF was added to the appropriate wells and incubated for 5 minutes at 37 °C. Then, the media was adjusted to the desired pH (final volume = 50 μL) using a pre-established volume of DPBS buffered with acetic acid, pH 4.0, and incubated for 2 hours. Acetic acid was used for sample acidification due to the high affinity and/or chelating properties of other acids (i.e. citric acid) towards Ca²⁺. Following the treatment, the plate was washed once with 100 μL of complete DMEM and then recovered for 72 hours at 37 °C in 100 μL of complete DMEM. Cell viability was assessed with the MTT colorimetric assay. Briefly, MTT was solubilized in PBS (10 mg/mL) with brief sonication, and 10 μL was added to each well. After incubation for 2 hours at 37 °C, the formazan crystals were solubilized in 200 μL of dimethyl sulfoxide (DMSO), and the absorbance at 580 nm was measured using an Infinite F200 PRO microplate reader (Tecan). Cell viability was normalized to control cells treated with media at pH 7.4.

Cell binding experiments

MDA-MB-231 human breast cancer cells were harvested and washed twice with PBS, pH 7.4. Alexa488-pHLIP was solubilized in an appropriate volume of 10 mM HEPES, 19.5 mM NaCl, pH 7.4 without calcium so that upon a 2-fold dilution, and after pH adjustment the desired treatment concentration (1 μM) was obtained. Next, 220,000 cells were incubated in suspension with Alexa488-pHLIP at two times the desired concentration without calcium for 5 minutes at 37 °C. After the incubation, an equal volume of 10 mM HEPES, 19.5 mM NaCl, pH 7.4 containing either 0, 2.4, or 3.6 mM calcium was added to bring the final calcium concentration to 0, 1.2, 1.8 mM, and incubated for an additional 5 minutes at 37 °C. Then, the pH was adjusted (final volume = 300 μL) using a pre-established volume of 10 mM HEPES, 19.5 mM NaCl, containing the appropriate calcium concentration, buffered with acetic acid, pH 4.0, and incubated for 10 minutes at 37 °C. The cells were then washed at the same pH and calcium concentration as the treatment, and fixed with 4% paraformaldehyde (PFA) for 10 minutes at 4 °C. The cells were then resuspended in PBS and analyzed by flow cytometry using a BDFacs Canto II flow cytometer equipped with a 488 nm argon laser and a 530/30 bandpass filter. The data was analyzed using FACSDiva version 6.1.1 software. The fluorescence data are expressed as mean arbitrary fluorescence units and were gated to include all healthy mammalian cells. Fluorescence was normalized to cells treated at pH 7.4 with 0 mM calcium.

For microscopy, MDA-MB-231 cells were seeded on glass coverslips pretreated with polylysine and allowed to reach ~70% confluency. The cells were then treated with Alexa488-pHLIP (1 μM) at pH 7.4 or 6.0 for 10 minutes at 37 °C as described above. Following the treatment, the cells were washed once at the same pH and calcium concentration as the treatment, and immediately fixed with ice cold methanol for 10 minutes. The fixed cells were then washed twice with PBS and incubated with 1 μg/mL Hoechst (Invitrogen #H3570) in PBS for 10 minutes at room temperature and washed twice again. The coverslips were mounted onto a slide with fluoromount (SouthernBiotech #0100–01) and stored at 4 °C until images were taken with a Nikon Eclipse Ti microscope with a 20× objective.

Vesicle preparation

The appropriate volume of lipid stocks dissolved in chloroform were dried under a nitrogen stream and dried overnight using high vacuum. The dried lipids were re-suspended in 50 mM phosphate buffer, pH 8.0 to a final concentration of 20 mM and large unilamellar vesicles (LUV) were formed by extrusion using a Mini-Extruder (Avanti Polar Lipids, Alabaster, AL). Extrusion was performed using 0.1 μm nucleopore polycarbonate membranes (Whatman, Philadelphia, PA) and the prepared stocks were stored at −4 °C. Lipids used in this study: Palmitoyl-oleoyl-phosphatidylcholine (POPC), palmitoyl-oleoyl-phosphatidylserine (POPS), and 1-palmitoyol-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) were purchased from Avanti Polar Lipids (Alabaster, AL)

Trp fluorescence measurements

The steady-state Trp fluorescence emission measurements of pHLIP were performed on a SPEX Fluorolog FL3–22 steady-state fluorescence spectrometer (Jobin Yvon, Edison, NJ) equipped with double-grating excitation and emission monochromators. Experiments were carried out on 2×10 mm cuvettes oriented perpendicular to the excitation beam. Sample temperature was maintained constant at 25 °C using a Peltier device from Quantum Northwest (Spokane, WA.). Measurements were performed using 2 μM pHLIP and 1.0 mM LUV after 20 min sample equilibration. Spectra were collected between 300–450 nm with an excitation wavelength of 285 nm at 1.0 nm steps, using 3.0 and 4.0 nm slits on the excitation and emission monochromators, respectively and averaged over 3 scans. The positions of maximum of the averaged spectra were determined by fitting them to a log-normal distribution using the following formula ⁵¹:

$$\begin{gathered} For\lambda >{\lambda }_{max}-\frac{\rho \Gamma }{{\rho }^2-1}, \\ I\left(\lambda \right)=I_{0}exp\left[\frac{ln2}{l{n}^2\rho }l{n}^2\left(1+\frac{\left(\lambda -{\lambda }_{max}\right)\left({\rho }^2-1\right)}{\rho \Gamma }\right)\right] \end{gathered}$$ (Eq. 1a)

$$\mathit{While}\mathit{for}\lambda <{\lambda }_{max}-\frac{\rho \Gamma }{{\rho }^2-1},I(\lambda )=0$$ (Eq. 1b)

Where I₀ is the maximal intensity of the analyzed spectrum at the fluorescence maximum λ_max, Γ is the width of the spectrum at the half maximum intensity, and ρ represents the asymmetry of the distribution.

Experiments performed between pH 8.0 and 4.0 in the absence of Ca²⁺ or Mg²⁺ were carried out in 10 mM phosphate buffer. The high binding propensity of phosphates for divalent cations which then precipitate as a salt makes this buffer unsuitable for measurements in the presence of Ca²⁺ or Mg². For this reason, we substituted 10 mM phosphate buffer for either 10 mM HEPES + 19.5 mM NaCl (experiments between pH 8.0–6.0) or 10 mM borate buffer + 19.5 mM NaCl (experiments at pH > 8.0) due to their lack of affinity for Ca²⁺ or Mg²⁺. The shorter buffering range of HEPES buffer, however, limited the experimental pH range to pH ≥ 6.0. Sample acidification was achieved by the addition of small aliquots of acetic/acetate buffer. As in the case of several buffers, many acids commonly used to induce acidification (i.e. citrate) chelate Ca²⁺ and Mg²⁺, eliminating any divalent cation-mediated effect. Acetate, however, is compatible with assays performed in the presence of Ca²⁺ or Mg²⁺. Divalent cation titrations were performed at pH 10.0 to minimize possible contribution form protonation events in our measurements.

The pK_a of the protonation-dependent insertion of pHLIP into LUV was calculated by fitting the data to the following equation using nonlinear least-square analysis ^19,38:

$$\mathrm{\lambda }=\frac{{\mathrm{\lambda }}_N+{\mathrm{\lambda }}_L({10}^{m\left(p{K}_a-pH\right)})}{1+{10}^{\mathrm{m}\left(\mathrm{p}{\mathrm{K}}_a-\mathrm{pH}\right)}}$$ (Eq. 2)

where λ is the Trp position of maximum measured as a function of pH, the λ_n and λ_L parameters correspond to the saturating Trp positions of maximum at high and low pH, m is the slope of the transition, and pK_a denotes the negative logarithm of the dissociation constant.

Membrane insertion calculations

The protonation-dependent membrane insertion free energy (ΔG_TM^H+) was calculated using the following equation:

$$\mathrm{\Delta }{G}_{TM}^{H+}=-2.3RT\cdot p{K}_a$$ (Eq. 3)

Where R is the gas constant (1.985×10⁻³ kcal K⁻¹ mol⁻¹) and T is the experimental temperature in Kelvin (298 K). The effect of Ca²⁺ on this process is given by the term ΔG_™^{H+, Ca2+}− ΔG_™^H+ which is calculated by subtracting the protonation-dependent membrane insertion free energy in the presence of Ca2+ (ΔG_™^{H+, Ca2+}) from the one obtained in the absence of Ca2+ (ΔG_™^H+). This term is by definition equal to zero in the absence of divalent cations and is lower in more favorable interactions.

The Ca²⁺-induced protonation-independent membrane insertion free energy (ΔG_™^[Ca2+]) was estimated using the decreases in Trp fluorescence maximum at pH 10.0 as proxies for the transmembrane populations of pHLIP at each [Ca²⁺] using the following formula ³⁵:

$$\Delta G_{TM}^{\left[Ca2+\right]}=-RTln\left(\frac{f_{TM}}{f_{IF}}\right)$$ (Eq. 4)

Where ΔG_™^[Ca2+] represents the free energy of the Ca²⁺-induced insertion of pHLIP measured at pH 10.0 for a particular [Ca²⁺], R and T denote the gas constant (1.985 × 10⁻³ kcal K⁻¹ mol⁻¹) and experimental temperature (298 K) respectively and f_™ and f_IF correspond to the fractional transmembrane and interfacial populations. For reference, a ΔG_™^[Ca2+] = 0 kcal/mol represents a f_™ = 0.5, a ΔG_™^[Ca2+] < 0 kcal/mol represents a f_™ > 0.5, and a ΔG_™^[Ca2+] > 0 kcal/mol represents a f_™ < 0.5.

Circular dichroism and oriented circular dichroism

Circular dichroism (CD) and oriented CD (OCD) measurements were performed using an upgraded JASCO-720 spectropolarimeter (JASCO, Easton, MD). CD spectra were collected using 15.0 μM pHLIP and 1.0 mM POPC LUV on a 1.0 mm optical path cuvette and corrected to their appropriate backgrounds. At least 30 scans were collected for each spectrum and averaged for each sample. As for fluorescence, measurements in the presence of divalent cations were performed using 10 mM HEPES buffer + 19.5 NaCl, while 10 mM phosphate buffer was employed for experiments in the absence of divalent cations and acidification induced by acetic/acetate buffer.

OCD spectra were obtained by creating a stack of oriented multilayers on a quartz disc as previously described ⁵². This technique allows the differentiation of interfacial α-helices, which present with a double minimum at ~ 205 and 222 nm, from transmembrane α-helices, which have a single minimum at ~ 228 nm ^52,53. Briefly, pHLIP and POPC lipids were codissolved at a 1:100 ratio in methanol (10 mM lipids and approximately 0.1 mM pHLIP) and 2.5 μL of the mixture was layered carefully on a ~ 1.0 cm circle at the center of a 2.5 cm disc. The solvent was then air dried and hydrated using warm air at ~ 100% relative humidity. The prepared disc was mounted on a sealed tube with the sample side pointing inwards. Samples were kept hydrated by placing a drop of water in the tube before closing. At least 50 scans were collected at four different orientations increasing by 90°-angles along the central axis and averaged. For measurements in the presence of divalent cations a 2 μL drop of 2.0 mM Ca²⁺ or Mg²⁺ dissolved in 5.0 mM HEPES buffer was added in-between each bilayer of the multilayer stack. Similarly, a drop of buffer at pH 4.0 was added in-between each layer for the collected spectrum at low pH. Buffer drops were completely dried before continuing with the multilayer stack protocol. The background signal was determined by collecting the spectra of a multilayer stack in the absence of pHLIP. The data is presented normalized to the ellipticity of their minima due to difficulties calculating the peptide concentration present.

LUV aggregation measurements

The aggregation of LUV containing POPC, 25POPS:75POPC, 25POPE:75POPC, and 75POPS:25POPC (as a control) in the presence of increasing concentrations of Ca²⁺ or Mg²⁺ were determined by measuring light scattering changes at 400 nm as previously described ^54,55. Samples were performed on a 4×10 mm quartz cuvette containing 0.1% LUV in 10 mM HEPES buffer + 19.5 mM NaCl. Cuvette orientation did not affect the results obtained. Similar measurements have been previously performed to characterize the aggregation of 100POPS vesicles and shown to reproduce results obtained by small-angle-scattering and dynamic light scattering ^54,55.

HIGHLIGHTS.

1. Ca²⁺ and Mg²⁺ promote the membrane insertion of pHLIP at neutral to basic pH

2. Divalent cations modulate the insertion pK_a of pHLIP

3. Divalent cations promote pHLIP cellular insertion and drug delivery

4. Effect of divalent cations is modulated by lipid composition