Surfactant protein C peptides with salt-bridges (“ion-locks”) promote high surfactant activities by mimicking the α-helix and membrane topography of the native protein

Ethics statement

The animal study was reviewed and approved by the Institutional Animal Care and Use Committee of the Los Angeles Biomedical Research Institute at Harbor-UCLA Medical Center (protocol # 12958). All procedures and anesthesia were performed in accordance with the American Veterinary Medical Association guidelines.

Materials

All organic solvents for sample synthesis were HPLC grade or better. Phospholipids were obtained from Avanti Polar Lipids, Inc. (Alabaster, AL). Deuterated water was NMR quality from Aldrich Chemical Co (St. Louis, MO).

Synthesis of SP-C peptides

A suite of SP-C peptides, composed of the 34-residue SP-Cff, SP-Cff ion-lock 1, SP-Css and SP-Css ion-lock 1 and the 33-residue SP-C33 UCLA, SP-C33 ion-lock 1 and SP-C33 ion-lock 2 (Fig. 1), was synthesized with FastMoc™ (Fields et al., 1991) or standard Fmoc procedures on a Leu-HMPB NOVA resin using ABI 431A (Applied Biosystems, Foster City, CA), Symphony Multiple Peptide (Protein Technologies, Tucson, AZ) or Liberty Microwave Peptide (CEM Corp, Matthews, NC) synthesizers. The SP-C mimic peptides were each carboxylated at the C-terminus and amidated at the N-terminus, and all residues were double-coupled to insure optimal yield (Waring et al., 2005). C-terminal carboxyl SP-C peptides were cleaved from the resin using the standard phenol:thioanisole:ethanedithiol water:trifluoracetic acid (0.75:0.25:0.5:0.5: 10, v:v) cleavage-deprotection mixture (Walther et al., 2010). Crude peptides were then purified (better than 95%) by preparative HPLC using a VYDAC diphenyl or C8 (1^′′ by 12^′′ width by length) column at 20 ml/min. SP-C peptides were each eluted from the column with a 0–100% (water to acetonitrile with 0.1% TFA as an ion pairing agent added to both aqueous and organic phases) linear gradient in one hour. The purified product was freeze-dried directly and the mass was confirmed by Maldi TOF mass spectrometry.

Attenuated total reflectance-fourier transform infrared (ATR-FTIR) spectroscopy

Infrared spectra were recorded at 37 °C for selected SP-C peptides in Fig. 1 using a Bruker Vector 22 FTIR spectrometer (Pike Technologies, Madison, WI) with a deuterium triglyceride sulfate (DTGS) detector, averaged over 256 scans at a gain of 4 and a resolution of 2 cm⁻¹ (Waring et al., 2005; Walther et al., 2007b; Walther et al., 2010; Gordon et al., 2008). For spectral recordings of SP-C peptides in hexafluoroisopropanol (HFIP) or aqueous buffer solutions, self-films were first prepared by air-drying peptide originally in 100% HFIP onto a 50 × 20 × 2 mm, 45° germanium attenuated total reflectance (ATR) crystal for the Bruker spectrometer. The dried peptide self-films were then overlaid with solutions containing 35% HFIP/65% deuterated-10 mM sodium phosphate buffer (pD 7.4) (Glasoe & Long, 1960) or 100% deuterated-10 mM sodium phosphate buffer (pD 7.4) before spectral acquisition; control solvent samples were similarly prepared, but without peptide. Spectra of peptides in solvent were obtained by subtraction of the solvent spectrum from that of the peptide-solvent. For FTIR spectra of SP-C peptides in synthetic surfactant lipid multilayers (DPPC:POPC:POPG; weight ratio, 5:3:2), the peptide was co-solvated with lipid systems in HFIP (lipid:peptide mole ratio of 10:1) and transferred onto an ATR crystal (Pike Technologies). The organic solvent was then removed by flowing nitrogen gas over the sample to produce a lipid-peptide film. The lipid-peptide films were next hydrated (≥35%) (Yamaguchi et al., 2001) with 10 mM deuterated sodium phosphate buffer (pD 7.4) vapor in nitrogen for 1 h prior to acquiring spectra (Gordon et al., 2000; Yamaguchi et al., 2002). The spectra for peptide in lipid mixtures were obtained by subtracting the lipid spectrum hydrated with 10 mM deuterated phosphate buffer (pD 7.4) from that of peptide in lipid with 10 mM phosphate buffer (pD 7.4). Infrared spectra were subsequently recorded for the lipid-peptide films at 37 °C, similar to those described for peptide self-films (see above). Relative amounts of α-helix, β-turn, β-sheet, or random (disordered) structures in either peptide self-films or lipid-peptide films were estimated using Fourier deconvolution (GRAMS/AI8, version 8.0, Themo Electron Corporation, Waltham, MA) and area of component peaks calculated using curve-fitting software (Igor Pro, version 1.6, Wavemetrics, Lake Oswego, OR) (Kauppinen et al., 1981). FTIR frequency limits were: α-helix (1662–1645 cm⁻¹), β-sheet (1637–1613 cm⁻¹ and 1710–1682 cm⁻¹), turn/bend (1682–1662 cm⁻¹), and disordered or random (1650–1637 cm⁻¹) (Byler & Susi, 1986).

To estimate the orientation of the α-helix of various SP-C peptides in surfactant lipid multilayers deposited on the germanium crystal surface, gold wire polarizers (Perkin Elmer, Waltham, MA) were rotated from 0° to 90° to obtain polarized IR spectra of each lipid-peptide film (Gordon et al., 1996). The insertion (or tilt) angle Θ for the SP-C helical axis with respect to the normal of the germanium surface was calculated based on the dichroic ratio R (Beevers & Kukol, 2006), which is equal to the ratio of A_parallel to A_{perpendicular}. Here, A_parallel and A_{perpendicular} are defined as the respective areas of absorbance at 0° (parallel) and 90° (perpendicular) polarizations for the amide I band centered at ∼1656 cm⁻¹. These measures assumed a thick film approximation with the values for the electric field components of the evanescent wave to be E_x = 1.398, E_y = 1.516, E_z = 1.625, and an angle α = 39° for the vibrational dipole relative to the molecular axis of the helix to derive an order parameter S (Beevers & Kukol, 2006; Chia et al., 2002). Here, S is equal to $\left[\left(E_x^2-R{E}_y^2+E_z^2\right)/\left(E_x^2-R{E}_y^2+E_z^2\right)\right]÷\left[\left(3{cos}^2\alpha -1\right)/2\right]$. The experimental tilt angle, Θ, was then calculated from this order parameter, by noting that S = (3cos²Θ−1)/2. However, it should be noted that the multilayer peptide-lipid samples stacked on the germanium crystal are unlikely to be perfectly ordered (Arkin, MacKenzie & Brunger, 1997; Kukol et al., 1999), and this introduces an isotropic contribution that will artificially reduce the dichroic ratio R. Accordingly, the experimental Θ angles obtained here with respect to the germanium crystal should be viewed as maximum tilt angles, with the actual tilt angles with respect to the normal of the multilayer lipid surface likely to be lower (Beevers & Kukol, 2006).

For selected SP-C peptides in surfactant lipids, the stability of the α-helical component was also analyzed by monitoring the hydrogen/deuterium (H/D) exchange for amide protons with the FTIR spectrometer. The peptide-lipid sample on the ATR crystal was flushed with 10 mM deuterated phosphate buffer (pD 7.4) vapor in nitrogen at 37 °C, and the decay of the amide II band area (1525–1565 cm⁻¹) was determined as a function of time (0–6 h) to assess the time-dependent H/D exchange of the amide group (Vandenbussche et al., 1992; Beevers & Kukol, 2006; Almlen et al., 2011).

Prediction of aggregation-forming regions in SP-C peptides

SP-C peptides were analyzed with PASTA (Trovato et al., 2006; Trovato, Seno & Tosatto, 2007) to characterize those domains most likely to form β-sheet, particularly in polar environments such as aqueous buffer or the lipid–water interface of membranes (Gordon et al., 2008). The PASTA algorithm systematically calculates the relative energies of the various pairing arrangements by assessing a pair-wise energy function for residues facing one another within a β-sheet. Using a database of known 3D-native structures, PASTA computes two different propensity sets depending on the directionality (i.e., parallel or antiparallel β-sheets) of the neighboring strands. PASTA assigns relative energies to specific β-pairings of two sequence stretches of the same length, and assumes that the lower relative energies enhance aggregation by further stabilizing the cross-β core. Theoretical PASTA predictions of SP-C and SP-C mimic aggregation were performed by submitting the primary sequences of these peptides (Fig. 1) to the PASTA Version 1.1 (http://protein.cribi.unipd.it/pasta) website (Trovato, Seno & Tosatto, 2007). PASTA calculates the relative energies of the most stable pairings (i.e., more negative energies denote more stable conformations), and outputs them as an HTML file (maximum pairings: 100). The PASTA energies for the top-ranked pairings may predict whether the β-sheet domains of one peptide will be more stable than those of other peptides. The energies for these pairings may also be used to compute an aggregation profile for those regions within a peptide most likely to aggregate as β-sheets. One cautionary note in using the PASTA algorithm is that it does not correct for the actual environmental polarity of a given protein domain, and also does not explicitly account for salt-bridges forming between charged ion-pairs.

Hydropathy predictions of SP-C peptides inserting into lipid bilayers as transmembrane helices

Membrane Protein Explorer (MPEx; Version 3.2.9) is a Java program that analyses hydrophobic lipid–protein interactions in membranes (http://blanco.biomol.uci.edu/mpex). With the hydropathy analysis mode (Snider et al., 2009), hydropathy plots are produced using the augmented Wimley-White (WW) whole-residue hydrophobicity scale that predicts transmembrane helices for protein sequences with high accuracy (Jayasinghe, Hristova & White, 2001; Liang et al., 2012). This augmented scale is experimentally based on the partitioning of hydrophobic pentapeptides (Wimley, Creamer & White, 1996) and salt-bridge pairs (Wimley et al., 1996) into n-octanol (i.e., solvent mimic of the hydrocarbon (non-polar) region of the bilayer), and accounts for whole-residue energy contributions due to the peptide backbone, side-chains and salt-bridge pairs (Jayasinghe, Hristova & White, 2001; Snider et al., 2009). Protein sequences may be submitted to the MPEx program, and the resulting plots are presented as hydropathy (kcal/mol) versus the sequence residue number, averaged over a sliding window of 19 amino-acid residues. Higher positive hydropathy values reflect enhanced lipid bilayer partitioning for any putative membrane helices. The default scale for the hydropathy analysis uses n-octanol, and there is an option for engaging potential salt-bridges using an “on–off” switch on the MPEx control panel.

In vitro adsorption surface pressure measurements

Adsorption experiments were done at 37 ± 0.5 °C in a Teflon^® dish with a 35 ml subphase (0.15 M NaCl + 1.5 mM CaCl₂) stirred to minimize diffusion resistance as described previously (Notter, Finkelstein & Taubold, 1983). Surfactants were prepared by co-solvating the lipid components (i.e., DPPC:POPC:POPG, 5:3:2, weight ratio) with the SP-C peptide in organic solvent HFIP (100%), dried under nitrogen, exposed to house vacuum to remove residual solvent, resuspended by hand vortexing in 0.15 M NaCl (normal saline) adjusted to pH 7.0 with 0.1 N sodium bicarbonate, heated to 65 °C intermittently for 30 min, and refrigerated for 12 h prior to use. At time zero, a bolus of these surfactant preparations containing 2.5 mg lipid in 5 ml of 0.15 M NaCl + 1.5 mM CaCl₂ was injected into the stirred subphase, and adsorption surface pressure (surface tension lowering below that of the pure subphase) was measured as a function of time by the force on a partially submerged, sandblasted platinum Wilhelmy slide (Notter, Finkelstein & Taubold, 1983). The final surfactant concentration for adsorption studies was uniform at 0.0625 mg phospholipid/ml (2.5 mg surfactant phospholipid/40 ml of final subphase).

In vitro surfactant activity measured with captive bubble surfactometry

Adsorption and dynamic surface tension lowering ability of all above described surfactant preparations were measured with a captive bubble surfactometer at physiological cycling rate, area compression, temperature, and humidity (Walther et al., 2010). We routinely analyze surfactant samples of 1 μl (35 mg phospholipids/ml) in the captive bubble surfactometer and perform all measurements in quadruplicate.

Ventilated, lung-lavaged, surfactant-deficient rabbit model

Young adult New Zealand white rabbits (weight 1.0–1.3 kg) received anesthesia with 50 mg/kg of ketamine and 5 mg/kg of acepromazine intramuscularly prior to placement of a venous line via a marginal ear vein. After intravenous administration of 1 mg/kg of diazepam and 0.2 mg/kg of propofol, a small incision was made in the skin of the anterior neck for placement of an endotracheal tube and a carotid arterial line. After placement of the endotracheal tube muscle paralysis was induced with intravenous pancuronium (0.1 mg/kg). During the ensuing duration of mechanical ventilation anesthesia was maintained by continuous intravenous administration of 3 mg/kg/h of propofol and intravenous dosages of 1 mg/kg of diazepam as needed; muscle paralysis was maintained by hourly intravenous administration of 0.1 mg/kg of pancuronium. Heart rate, arterial blood pressures and rectal temperature were monitored continuously. Respiratory function was followed during mechanical ventilation by measurements of serial arterial blood gases and pulmonary compliance over the first 2 h after a standardized saline lavage protocol. For this timeframe of study, the model reflected a relatively pure state of surfactant insufficiency in animals with mature lungs.

After stabilization on the ventilator, saline lavage was performed with repeated intratracheal instillation and removal of 30 mL of normal saline until the PaO₂ dropped below 100 mm Hg (average 3 lavages). Maintenance fluid was provided by a continuous infusion of Lactated Ringer’s solution at a rate of 10 ml/kg/h. Edema fluid appearing in the trachea was removed by suctioning. When the PaO₂ was stable at less than 100 mm Hg in saline lavaged animals, an experimental or control surfactant mixture was instilled into the trachea at a dose of 100 mg/kg body weight and a concentration of 35 mg/ml. The rabbits were ventilated using a Harvard volume controlled animal ventilator (tidal volume 7.5 ml/kg, a positive end-expiratory pressure of 3 cm H₂O, an inspiratory/expiratory ratio of 1:2, 100% oxygen, and a respiratory rate to maintain a PaCO₂ at ∼40 mm Hg). Airway flow and pressures and tidal volume were monitored continuously with a pneumotachograph connected to the endotracheal tube and a pneumotach system. Arterial pH and blood gases were done every 15 min. Dynamic lung compliance was calculated by dividing tidal volume/kg body weight by changes in airway pressure (peak inspiratory pressure minus positive end-expiratory pressure) (ml/kg/cm H₂O). Animals were sacrificed 2 h after surfactant administration with an overdose of pentobarbital. End-points were pulmonary gas exchange (arterial pH, PaCO₂ and PaO₂) and pulmonary mechanics (dynamic compliance).

FTIR spectroscopic analysis of the secondary conformations for SP-C mimics with and without ion-locks

To test the structural effects of inserting ion-locks into SP-C peptides, a suite of SP-C mimics was prepared based on native SP-C sequences (i.e., SP-Css and SP-Cff) and these native SP-C sequences with a Glu⁻–Lys⁺ pairing in the hydrophobic midsection (i.e., SP-Css ion-lock 1 and SP-Cff ion-lock 1; see Fig. 1). Representative FTIR spectra of the amide I band for SP-Css and SP-Cff in surfactant lipids (DPPC:POPC:POPG 5:3:2, weight ratio) in Fig. 2A were similar, each showing a major β-sheet component with a predominant peak at ∼1630–1631 cm⁻¹ and a miniscule companion peak at ∼1691 cm⁻¹, and a lower α-helical component with a mid-field shoulder at ∼1657 cm⁻¹. Subsequent deconvolutions confirmed that SP-Css and SP-Cff in surfactant lipids principally adopt β-sheet conformations (≥ ∼ 58%) with minor contributions from α-helix, loop-turn and disordered structures (Table 1). In additional studies (Table 2), FTIR spectra and deconvolutions for SP-Css and SP-Cff in a membrane-interfacial lipid mimic (i.e., 35% HFIP/65% deuterated-sodium phosphate buffer, pD 7.4) or 100% aqueous buffer (i.e., 10 mM deuterated phosphate buffer, pD 7.4) indicated comparably high β-sheet (≥36%–50%), with smaller proportions from other conformations. Consistent with these findings, prior FTIR spectroscopy of 34-residue SP-Cff showed reduced α-helix on short-term storage in surfactant lipids (Walther et al., 2000), while earlier CD analyses of synthetic, 35-residue SP-Css or SP-Cff in detergent micelles also indicated enhanced β-sheet and reduced α-helix (<20%) (Johansson et al., 1995). In our hands, SP-Css and SP-Cff exhibited only high β-sheet and low α-helix in surfactant lipids (Fig. 2; Table 1) or environments of differing polarity (Table 2).

FTIR spectroscopy determinations for either SP-Css ion-lock 1 or SP-Cff ion-lock 1 in surfactant lipids indicated enhanced α-helix and reduced β-sheet, when compared to the corresponding measurements of the native SP-Css or SP-Cff sequences. For example, FTIR spectra of SP-Css ion-lock 1 in surfactant lipids (Fig. 2A) indicated high α-helix (at ∼1657 cm⁻¹), and very low β-sheet (at ∼1630 cm⁻¹). Importantly, deconvolution showed predominant α-helix (i.e., ∼63%) and minor β-sheet (∼6%) for SP-Css ion-lock 1 in surfactant lipids, as opposed to the reduced α-helix (∼25%) and elevated β-sheet (∼58%) determined from FTIR spectra of its parent SP-Css in this environment (Fig. 2A; Table 1). Consequently, insertion of the ion-lock (i.e., Glu⁻-20–Lys⁺-24) in the mid-section of the poly-Val sequence produced a remarkable conformation shift in the host SP-Css, from an ‘amyloid-like’ β-sheet profile to one that is primarily α-helix (Fig. 2A). Additional FTIR spectra of SP-Css ion-lock 1 in aqueous and membrane-interfacial HFIP milieu also demonstrated elevated α-helix values (≥ ∼ 41%) over those obtained for the parent SP-Css (Table 2), suggesting that an intrapeptide salt-bridge (at Glu⁻-20–Lys⁺-24) stabilizes the α-helical conformation in environments of differing polarity. In further studies, FTIR spectra of SP-Cff ion-lock 1 in surfactant lipids showed enhanced α-helix (at ∼1653 cm⁻¹) and low β-sheet (at ∼1630 cm⁻¹) (Fig. 2A). Deconvolution of these SP-Cff ion-lock 1 spectra verified that the major structural component in lipids was α-helix (∼44.6%), produced by a conformational change from the very high β-sheet (∼74.3%) of its parent SP-Cff (Fig. 2A; Table 1). Incorporation of the ion-pair (i.e., Glu⁻-20–Lys⁺-24) induced a similar conformational shift for SP-Cff in the lipid-mimic HFIP solution from a β-sheet conformation (∼53.2%) to one that is predominately α-helix (∼53.0%) (Table 2). Although these FTIR spectral results suggest at least a partial salt-bridge for SP-Cff ion-lock 1 in lipid and lipid-mimic environments, additional FTIR experiments indicate that this salt-bridge may not be as stable as that seen with SP-Css ion-lock 1. Unlike SP-Css ion-lock 1, SP-Cff ion-lock 1 folds primarily as loop-turns in aqueous buffer (Table 2) and even shows diminished α-helix content in surfactant lipids relative to that of SP-Css ion-lock 1 (i.e., ∼44.6 vs. ∼63.1% in Fig. 2A and Table 1). Such α-helical reductions may be partially due to the Phe groups of SP-Cff ion-lock 1 (Fig. 1) weakening the putative salt-bridge between Glu⁻-20–Lys⁺-24, either through direct or indirect mechanisms (see below).

It is also worthwhile to contrast the helical properties determined for FTIR spectra of SP-Css ion-lock 1 and SP-Cff ion-lock 1 with those of SP-C33 UCLA and several SP-C33 ion-locks. SP-C33 UCLA has the same sequence as SP-C33 (Almlen et al., 2010; Almlen et al., 2011) with poly-Leu substituting for the poly-Val in the C-terminal region (Fig. 1). Previous CD and FTIR studies have reported that SP-C mimics of the poly-Leu class, including SP-C33 (Almlen et al., 2011) and its homologous precursor SP-C(Leu) (Nilsson et al., 1998), each show elevated α-helix when incorporated into lipids or lipid-mimics. In support of these experimental results, earlier Molecular Dynamics simulations of SP-C33 in methanol for 10 ns indicated that the hydrophobic poly-Leu region adopted an α-helical conformation in this membrane-interfacial mimic (Almlen et al., 2010). The FTIR spectrum of SP-C33 UCLA in surfactant lipids overlapped that of SP-Css ion-lock 1 (Figs. 2A and 2B), and spectral deconvolution indicated similarly high α-helix (Table 1). Moreover, FTIR spectra of SP-C33 UCLA in either aqueous or HFIP solutions showed α-helical components (at ∼ 1653–1655 cm⁻¹) nearly identical to those of SP-Css ion-lock 1 (not shown), as well as similar α-helix levels (i.e., ∼40%–51%) using spectral deconvolution (Table 2). Although SP-Css ion-lock 1 and SP-C33 UCLA indicate comparably high α-helix in aqueous buffer, aqueous HFIP buffer or surfactant lipids (Fig. 2: Tables 1 and 2), SP-C33 UCLA requires replacement of the poly-Val sequence with ten Leu residues to achieve this stabilization. Contrarily, SP-Css ion-lock 1 needs only a single ion-pair at Glu⁻-20–Lys⁺-24 to maintain high α-helix. In further research, the SP-C33 ion-lock 1 and SP-C33 ion-lock 2 mimics were prepared by inserting a single (i.e., Glu⁻-20–Lys⁺-24) or double (i.e., Lys⁺-15–Glu⁻-19 and Glu⁻-20–Lys⁺-24) ion-lock into the SP-C33 sequence (Fig. 1). FTIR spectra and deconvolution indicated α-helix for these SP-C33 ion-locks in aqueous, HFIP-water and surfactant lipids similar to those of SP-C33 (Fig. 2; Tables 1 and 2). For example, the FTIR spectra of SP-C33 UCLA and the SP-C33 ion-locks in surfactant lipids nearly overlap with α-helix maxima at 1655 cm⁻¹ (Fig. 2B), and spectral deconvolutions of each show elevated α-helix of ∼65%–67% (Table 1). Because the insertions of either one or two charged ion-pairs were unable to raise the α-helical content beyond that of the parent SP-C33 UCLA (Tables 1 and 2), the SP-C33 primary sequence may already be maximized for α-helix. Interestingly, the α-helicity of SP-Css ion-lock 1 in surfactant lipids (i.e., ∼63%) is only slightly below those of the SP-C33 and the SP-C33 ion-locks (i.e., ∼65%–67%), suggesting that the SP-Css ion-lock 1 in lipids is also almost optimized for α-helix.

Effects of ion-locks on the aggregation domains of SP-C peptides detected with PASTA analysis

PASTA calculations were next performed to assess whether the enhanced α-helicity induced by ion-lock insertion into either SP-Css or SP-Cff (Fig. 2; Tables 1 and 2) is related to the disruption of β-sheet regions in the host peptides. The PASTA algorithm systematically determines the relative energies for various paired sequences by assessing an energy function for facing residues that are H-bonded within a β-sheet (Trovato et al., 2006). Here, PASTA predicted that the native SP-Cff and SP-Css peptides will each form β-sheet at their respective C-terminal domains. Specifically, PASTA energy calculations showed that the most likely pairing for SP-Cff or SP-Css will be an interpeptide, parallel β-sheet (i.e., parallel in-register arrangement (PIRA)) for the C-terminal residues 7–27 (Fig. 1) with a relative PASTA energy of −29.2 (Trovato et al., 2006; Trovato, Seno & Tosatto, 2007). Using a database of 179 peptides as a benchmark (Trovato, Seno & Tosatto, 2007), an energy threshold of < −29.0 indicates a probability > 98% (true positive rate) that SP-Cff and SP-Css are each “amyloid-like” (i.e., have a high β-sheet tendency) for the corresponding C-terminal segments (i.e., residues 7–27). This prediction of elevated β-sheet for SP-Css and SP-Cff was confirmed by FTIR spectra of these peptides in surfactant lipids with high β-sheet and low α-helix (Figs. 2 and 3; Table 1). Furthermore, Fig. 3 shows that incorporation of Glu⁻-20–Lys⁺-24 in either SP-Cff or SP-Css increased the relative PASTA energy from −29.2 to −22.2 for the 7–27 sequence, and also raised the α-helix levels of the particular SP-C ion-locks in surfactant lipids. Because FTIR deconvolutions (Fig. 2; Table 1) demonstrated that the enhanced α-helix in SP-Css ion-lock 1 or SP-Cff ion-lock 1 in lipids occurred concomitantly with reduced β-sheet, the charged ion-pair may partially augment α-helicity by disrupting β-sheet in the C-terminal region.

Interestingly, Fig. 3 shows that an analogous insertion of ion-locks in SP-C33 does not increase the α-helix levels of SP-C33 ion-lock 1 and SP-C33 ion-lock 2 in surfactant lipids, despite the relative PASTA energies for these daughter ion-locks being substantially higher than that of the parent SP-C33. The inability of ion-locks to further raise the α-helicity of SP-C33, unlike the large α-helix increases seen for ion-lock incorporation into SP-Css and SP-Cff, may be due to the already high α-helix and low β-sheet present in the SP-C33 peptide (Fig. 3). FTIR spectral deconvolutions for SP-C33, SP-C33 ion-lock 1 and SP-C33 ion-lock 2 in surfactant lipids each indicated < ∼ 8%β-sheet (Table 1), and additional recruitment of α-helix from β-sheet residues may not be possible. In this context, SP-Css ion-lock 1 in surfactant lipids may be similarly optimized for high α-helix, as FTIR deconvolutions showed elevated α-helix and diminished β-sheet matching those of SP-C33 and SP-C33 ion-locks (Table 1).

Prediction of transmembrane α-helices for SP-C mimics using MPEx hydropathy calculations

Because secondary structure analysis using FTIR spectroscopy or PASTA do not directly measure membrane binding of helical peptides, the tendency of SP-C peptides to form transmembrane α-helices was next evaluated with hydropathy calculations using the Membrane Protein Explorer (MPEx) program. MPEx determines the hydropathy (kcal/mol) for protein sequences using the augmented Wimley-White (WW) hydrophobicity scale and is based on measurements of the partitioning of hydrophobic pentapeptides and salt-bridge pairs into n-octanol (Jayasinghe, Hristova & White, 2001). Previous MPEx calculations have predicted transmembrane helices for protein sequences with high accuracy and are especially useful in preliminary characterizations of novel proteins or peptides (Snider et al., 2009; Liang et al., 2012). One important feature of MPEx is its option for engaging potential salt-bridges using an “enter-remove” icon on the control panel (Snider et al., 2009). However, note that dipalmitoylated SP-C is excluded from the below comparisons, because the MPEx and PASTA algorithms do not correct for how covalently attached, non-peptide moieties (e.g., the palmitoyl groups in native SP-C; Fig. 1) may modify membrane incorporation or peptide aggregation, or both.

As a surrogate for native SP-C, the rSP-C sequence was submitted to MPEx, which predicted that this SP-C mimic would insert into lipid bilayers as a transmembrane α-helix with its most favorable hydropathy (i.e., 11.3 kcal/mol) through the C-terminal segment 12–30. The rSP-C sequence was chosen as a control for our MPEx analyses of SP-C ion-locks, because the recombinant protein is highly homologous with the dipalmitoylated SP-C (Fig. 1), lacks the two palmitoylcysteine groups of the native protein and has been previously subjected to structural and functional experiments. To simultaneously visualize the respective propensities of SP-C peptides to insert into lipids as transmembrane helix and to form β-sheet, the “Hydropathy from MPEx” (ordinate) is plotted versus the “Relative PASTA Energy” (abscissa) for various SP-C mimics (Fig. 4). The more positive hydropathy values (i.e., bottom-to-top) reflect greater incorporation as a transmembrane helix, while higher relative PASTA energy values (i.e., left-to-right) represent a lower tendency to form β-sheet. If rSP-C binds to lipid bilayers with the same helical conformation as in chloroform-methanol (i.e., α-helix for residues 9–34) (Luy et al., 2004), then Fig. 4 predicts that rSP-C readily incorporates as a transmembrane α-helix through its C-terminus. The very low PASTA energy indicated for rSP-C (Fig. 4) may also be responsible for the metastable helical properties observed in earlier 2D-NMR studies (Luy et al., 2004), as β-sheet multimers or fibrils should here be thermodynamically favored over α-helix (Fig. 3) (Trovato et al., 2006; Trovato, Seno & Tosatto, 2007).

The relative tendencies of SP-Css ion-lock 1 and SP-Cff ion-lock 1 to penetrate into lipid bilayers as transmembrane α-helices were also predicted with the joint MPEx and PASTA plot in Fig. 4. Earlier MPEx research showed that hydrophobic helices bearing salt-bridges may penetrate deeply into lipid bilayers, because there is only a slight thermodynamic penalty for replacing the hydrophobic residues of a transmembrane sequence with a neutral, charged ion-pair (e.g., Glu⁻ and Lys⁺) (Snider et al., 2009; Jayasinghe, Hristova & White, 2001). Moreover, previous electrostatic calculations indicated that salt-bridges with close-range distances (≤ 4.0 Å) between the cationic residue “N” and the nearest anionic residue “O” are mostly α-helix stabilizing, while disengaged salt-bridges with long-range distances (>5 Å) may be helix destabilizing (Kumar & Nussinov, 1999; Kumar & Nussinov, 2002). As indicated above, MPEx allows hydropathic analysis of SP-C ion-lock sequences in membranes assuming the salt-bridge is either engaged (i.e., close-range salt-bridge) or disengaged (i.e., long-range salt-bridge). With a close-range salt-bridge between the Glu⁻-20–Lys⁺-24 pair, MPEx forecasts that a highly α-helical SP-Css ion-lock 1 would deeply embed into the lipid bilayer due to the high hydropathy (i.e., ∼10.3 kcal/mol) of its segment 12–30. Alternatively, with a long-range salt-bridge between Glu⁻-20–Lys⁺-24, the corresponding MPEx hydropathy would be much lower (i.e., ∼4.0 kcal/mol) and predicts a more superficial association between a less helical SP-Css ion-lock 1 and the bilayer. Our FTIR results indicating high α-helicity for SP-Css ion-lock 1 in surfactant lipids are consistent with a close-range salt-bridge forming in this environment, permitting deep insertion of this SP-C mimic as a transmembrane α-helix. Specifically, FTIR deconvolution in Table 1 showed that an SP-Css ion-lock 1 in surfactant lipids is highly α-helical (∼63.1%). Thus, the SP-Css ion-lock 1 α-helix (∼21 residues) may span the width of the bilayer, as it lies within the 17–25 residue window reported for other transmembrane α-helices (Liang et al., 2012). The PASTA prediction that the Glu⁻-20–Lys⁺-24 pair in SP-Css ion-lock 1 converts segment 7–27 from β-sheet to α-helix (Figs. 3 and 4) also supports the MPEx assignment that α-helix is confined to the C-terminal region. Consequently, helical SP-Css ion-lock 1 with a close-range salt-bridge shows a high hydropathy, which is only slightly less than that of rSP-C (i.e., 10.3 vs. 11.3 kcal/mol in Fig. 4), and should mimic the transmembrane incorporation of native SP-C peptides. On the other hand, the reduced α-helicity observed for SP-Cff ion-lock 1 in surfactant lipids is compatible with a disruptive, long-range salt-bridge that lowers α-helical levels to ∼44.6% (Table 1). The resulting SP-Cff ion-lock 1 α-helix is too short (∼15 residues) to span the width of the bilayer (Liang et al., 2012) and instead may be limited to a single monolayer. These studies suggest that the replacement of the two serines in SP-Css ion-lock 1 with vicinal phenylalanines disorders the close-range, salt-bridge between Glu⁻-20–Lys⁺-24, through either direct or indirect mechanisms.

FTIR secondary structure and theoretical analyses similarly examined the insertion of SP-C33 mimics into the lipid bilayer as transmembrane α-helices. Figure 4 indicates higher MPEx and PASTA energy values for SP-C33 relative to those of rSP-C, predicting that the swap of poly-Leu with the poly-Val sequence of native SP-C produces an SP-C33 mimic that associates with lipids as a transmembrane α-helix. In support of this assignment, FTIR showed that the principal spectral component for SP-C33 UCLA in surfactant lipids was α-helix and that the α-helicity (67.2% from Table 1) for SP-C33 UCLA is sufficiently high for this SP-C mimic to incorporate as a transmembrane helix (∼22 residues long). In this context, Fig. 4 also predicts that SP-C33 ion-lock 1 or SP-C33 ion-lock 2 (i.e., SP-C33 mimics with one or two Glu⁻-20–Lys⁺-24 pairs; Fig. 1) will incorporate into lipid bilayers either as transmembrane α-helices with close-range salt-bridges or disordered α-helices with long-range salt-bridges and reduced membrane affinity. Because our FTIR results show only high α-helix levels for these SP-C33 ion-locks matching those of the parent SP-C33 in surfactant lipids (Table 1), both SP-C33 ion-lock 1 and SP-C33 ion-lock 2 will likely form engaged salt-bridges, and incorporate into surfactant lipids as transmembrane α-helices.

Orientation-dependent FTIR analysis of the helical axes of SP-C mimics in surfactant lipids

The topography of helical SP-C mimics in surfactant lipid multilayers was experimentally assessed by recording polarized infrared spectra of each lipid-peptide film that was rotated from 0° to 90°. Orientation-dependent FTIR spectroscopy of native SP-C (Vandenbussche et al., 1992; Pastrana, Maulone & Mendelsohn, 1991) and unrelated membrane proteins (Chia et al., 2002; Beevers & Kukol, 2006) was previously used to determine the Θ angles between their respective helical axes and the normal to the lipid bilayer plane. Early polarized FTIR studies indicated that native bovine and porcine SP-C proteins (Vandenbussche et al., 1992; Pastrana, Maulone & Mendelsohn, 1991) each insert into surfactant lipids as a transmembrane helix whose long molecular axis parallels the phospholipid acyl chains and a maximum angle of ∼20–24° with respect to the bilayer normal. In the present research, the polarized FTIR spectra of the amide I band for SP-Css ion-lock 1 in surfactant multilayers were highly anisotropic, with high and low absorbance at 0° and 90° (Fig. 5A). The dichroic ratio R for SP-Css ion-lock 1 is computed as the ratio of the integrated amide I absorption at parallel and perpendicular with polarized infrared light. The use of this dichroic R indicated a maximum tilt angle Θ = 22.6° for the helical axis of SP-Css ion-lock 1 with respect to the bilayer normal (Table 1), similar to those determined previously for native SP-C (Vandenbussche et al., 1992; Pastrana, Maulone & Mendelsohn, 1991). Based on its high α-helicity (63.1%) (Fig. 2A; Table 1), the SP-Css ion-lock 1 probably incorporates in surfactant lipids as a transmembrane α-helix (∼21 residues long) spanning the bilayer with a maximum 22.6° tilt, particularly if this SP-C mimic is highly hydropathic due to a salt-bridge between Glu⁻-20–Lys⁺-24 (Fig. 4). When compared with FTIR results from SP-Css ion-lock 1 or native SP-C, however, the polarized FTIR spectra for SP-Cff ion-lock 1 in surfactant lipids were less anisotropic with a lower dichroic R and demonstrated a correspondingly higher maximum tilt angle (Θ = 34.2°) (Table 1). Because of its reduced α-helix levels (∼44.6%) calculated from FTIR spectral deconvolutions and higher maximum tilt angle (Figs. 2A and 5B; Table 1), the relatively short (∼15 residues) α-helix of SP-Cff ion-lock 1 in surfactant lipids is probably limited to one bilayer leaflet. One hypothesis worth exploring is that the low α-helicity and more oblique Θ observed for SP-Cff ion-lock is due to a failure of its Glu⁻-20–Lys⁺-24 to form an engaged salt-bridge (i.e., close-range) in surfactant lipids.

Analogous oriented FTIR experiments were performed to characterize the membrane topography of SP-C33 and SP-C33 ion-locks in surfactant lipids. As noted above, conventional FTIR analysis of SP-C33 UCLA, SP-C33 ion-lock 1 or SP-C33 ion-lock 2 (Fig. 1) in surfactant lipids each showed spectra with a predominant α-helix band centered at 1655 cm⁻¹ (Fig. 2B) and elevated levels (65%–67%) of α-helix (Table 1). The polarized FTIR spectra of the amide I bands for SP-C33 UCLA (Fig. 5D), SP-C33 ion-lock 1 (Fig. 5C) and SP-C33 ion-lock 2 (not shown) were all highly anisotropic, indicating maximum Θ angles of ∼23° (Table 1). Consequently, SP-C33 UCLA and the SP-C33 ion-locks incorporate into surfactant lipid as transmembrane α-helices (∼21–22 residues long) spanning the bilayer with a maximum tilt angle of ∼23°, confirming that they accurately reproduce the membrane topography of native SP-C proteins (Vandenbussche et al., 1992; Pastrana, Maulone & Mendelsohn, 1991). Moreover, SP-C33 ion-lock 1 and SP-C33 ion-lock 2 each probably contain stable salt-bridges as predicted from MPEx-hydropathy analysis (Fig. 4; top right quadrant), which would facilitate the deep penetration of these SP-C mimics into surfactant bilayers. Importantly, our orientation-dependent FTIR results with SP-C33 UCLA (Fig. 5D; Table 1) are in good agreement with those earlier reported for SP-C33 incorporated into surfactant lipid bilayers (Almlen et al., 2011), despite these studies using different experimental conditions for peptide synthesis and FTIR spectral measurements.

FTIR analysis of hydrogen/deuterium (H/D) exchange for amide protons of SP-C mimics in surfactant lipids

For selected SP-C mimics with surfactant lipids, the extent of peptide helix insertion into the lipid bilayer was characterized by measuring the hydrogen/deuterium (H/D) exchange of amide protons. Previous studies of native SP-C (Vandenbussche et al., 1992; Pastrana, Maulone & Mendelsohn, 1991) and unrelated membrane proteins (Chia et al., 2002; Grimard et al., 2001) indicated that protons participating in the H-bonds of highly ordered structures such as the transmembrane α-helix may show a lack of deuterium exchange for up to 6–28 h. The H/D exchange for amide protons that maintain the transmembrane α-helix was assessed as follows. SP-C mimic-lipid samples on the ATR crystal were hydrated by bubbling nitrogen through a 10 mM deuterated phosphate buffer (pD 7.4) at 37 °C, and the amide I and amide II peaks were recorded with FTIR spectra over the range 1500–1700 cm⁻¹ as a function of time (0–6 h). The peak between 1600 and 1700 cm⁻¹ reflects the amide I band (i.e., ν (C=O) of the peptide bond) that is sensitive to the secondary conformations of the protein, while the second band between 1525 and 1565 cm⁻¹ is primarily due to amide II (i.e., δ (N–H) of the peptide bond) (Vandenbussche et al., 1992; Jackson & Mantsch, 1995). The ratio of the amide II to amide I peak areas is proportional to the H/D exchange, as such exchange would shift the amide II band (at ∼1543 cm⁻¹) to that of amide II’ (at ∼1450 cm⁻¹) (Pastrana, Maulone & Mendelsohn, 1991). Contrarily, the absence of H/D exchange would suggest that the H-bonds are isolated from aqueous media (e.g., transmembrane α-helices), or are in aqueous environments in ordered conformations where exchange is slow.

In Fig. 6 the percentage (%) unexchanged hydrogens for the SP-Css ion-lock 1 in surfactant lipids is shown as a function of time (0–6). Analogous H/D curves for membrane proteins were previously analyzed as the sum of multiple linear exponential (first-order) decays (Grimard et al., 2001). Besides fast- and intermediate-H/D exchange compartments (i.e., with t_1/2’s between 10 and 120 min involving ∼49% of the helical residues), there is a final compartment for helical amide hydrogens (∼51%) that resists H/D exchange for t ≤ 6 h (Fig. 6). The inaccessible fraction at long-times is probably due to segments buried deeply as transmembrane α-helix, as the residual amide II band remains centered at ∼1543 cm⁻¹ (not shown). Because spectral deconvolutions indicated ∼ 21 α-helical residues for SP-Css ion-lock 1 in surfactant lipids (Table 1), ∼ 11 α-helical residues are readily calculated as unexchanged for this SP-C mimic at long-times (Almlen et al., 2011; Beevers & Kukol, 2006). Importantly, similar H/D exchange kinetics was earlier reported for bovine and porcine SP-C (Vandenbussche et al., 1992; Pastrana, Maulone & Mendelsohn, 1991), indicating that SP-Css ion-lock 1 is accurately mimicking the ability of native SP-C proteins to embed into surfactant lipids as transmembrane α-helices. In additional studies, the kinetic H/D exchange profile for SP-Css ion-lock 1 was nearly identical to that of SP-C33 UCLA (Fig. 6), and very similar to that reported earlier for the conventional SP-C33 (Almlen et al., 2011). The comparable H/D exchange and maximum tilt angles observed for SP-Css ion-lock 1, SP-C33 UCLA and SP-C33 (Figs. 5 and 6; Table 1) (Almlen et al., 2011) indicate that these three SP-C mimics similarly incorporate into surfactant lipids as transmembrane α-helices. Interestingly, the H/D curve for SP-Cff ion-lock 1 in Fig. 6 demonstrates both similarities and differences with those of SP-Css ion-lock 1 and SP-C33 UCLA. Despite SP-Cff ion-lock 1 showing fast-, intermediate- and very slow-exchange compartments analogous to those of the other two SP-C mimics, there was overall a more rapid H/D exchange for SP-Cff ion-lock 1 in Fig. 6. These discrepancies were particularly noticeable at 6 h, when the unexchanged hydrogen component for SP-Cff ion-lock 1 was ∼23.2%, while those for SP-Css ion-lock 1 and SP-C33 UCLA were each ∼51% (Fig. 6). Consequently, only ∼3.5 helical residues of SP-Cff ion-lock 1 showed unexchanged hydrogen at long-times in Fig. 6, compared to the ∼11 helical residues that remain unexchanged in either SP-Css ion-lock 1 or SP-C33 UCLA. The sharply lower unexchanged helical protons for SP-Cff ion-lock 1 in surfactant lipids (Fig. 6) may be due to a more flexible membrane α-helical component or increased accessibility of helical amide hydrogens to the aqueous medium.

In vitro surface activity determined with adsorption surface pressure measurements

Synthetic surfactant preparations were formulated by mixing synthetic lipids (Lipids) consisting of 5:3:2 (weight ratio) DPPC:POPC:POPG, with 3.0% by weight SP-Css ion-lock 1, SP-Cff ion-lock 1, SP-Cff, SP-C33 ion-lock 1, SP-C33 UCLA (positive control) and Lipids (negative control). Here, adsorption was determined as the Adsorption Surface Pressure (mN/m) as a function of time (0–15 min), with higher surface pressures reflecting lower surface tensions. The time-course plots in Fig. 7 indicate that each of these SP-C mimics attained maximal adsorption surface pressure within ∼5–10 min. The relative order for the surfactant activities of the various preparations was as follows: SP-C33 ion-lock 2 ≥ SP-Css ion-lock 1, SP-C33 UCLA ≫ SP-Cff ion-lock 1 ∼ SP-Cff > Lipids. Previous in vitro studies on SP-C(Leu), a homologous precursor to our SP-C33 UCLA, with a lipid mixture of DPPC/POPG (7:3 ratio by mass) showed similarly enhanced adsorption surface pressure curves (Nilsson et al., 1998).

In vitro dynamic surface activity determined with captive bubble surfactometry

Synthetic surfactant formulations were also prepared for in vitro experiments with captive bubble surfactometry and in vivo studies using ventilated, lung-lavage rabbits by mixing synthetic lipids (Lipids), consisting of 5:3:2 (weight ratio) DPPC:POPC:POPG, with 3.0% by weight SP-Css ion-lock 1, SP-Cff ion-lock 1, SP-C33 ion-lock 1, SP-Cff, SP-Css, SP-C33 UCLA (positive control) and Lipids (negative control). The SP-Css ion-lock 1 surfactant had very high surface activity in captive bubble experiments and reached minimum surface tension values <1 mN/m during each of ten consecutive cycles of dynamic cycling (rate of 20 cycles/min, Fig. 8). Importantly, our SP-C33 UCLA (positive control) attained low surface tension values <1 mN/m that not only were identical to those of SP-Css ion-lock 1 (Fig. 8), but also matched those of SP-C33 in earlier captive bubble experiments with DPPC/POPG lipid mixtures. The SP-C33 ion-lock 1 surfactant exerted similar low surface tensions as SP-C33 UCLA. Contrarily, SP-Cff ion-lock 1, Lipids (negative control), SP-Cff and SP-Css all reached significantly higher minimum surface tension values of 19, 16, 11 and 14 mN/m (p < 0.001) versus SP-Css ion-lock 1 and SP-C33 UCLA after ten cycles on the captive bubble surfactometer (Fig. 8). In summary, the relative order for the surfactant activities determined with captive bubble surfactometry for the various preparations was as follows: SP-Css ion-lock 1 ∼ SP-C33 UCLA ∼ SP-C33 ion-lock 1 ≫ SP-Cff ∼ SP-Css ≫ Lipids ∼ SP-Cff ion-lock 1.

In vivo activity of synthetic surfactants in ventilated, lung-lavaged, surfactant-deficient rabbits

The pulmonary activity of SP-C mimic surfactants (described above for captive bubble experiments) was investigated in comparison to SP-C33 UCLA (positive control) and Lipids alone (negative control) during a 120 min period following intratracheal instillation of these surfactants into ventilated rabbits with ARDS induced by in vivo lavage (Fig. 9). Oxygenation and lung compliance increased quickly to a plateau phase (within ∼30–60 min) after instillation of SP-Css ion-lock 1 and SP-C33 UCLA, with SP-C33 ion-lock 1 reaching similarly high functional plateaus after somewhat longer times (within ∼60 min) (Fig. 9). On the other hand, the SP-C mimics SP-Cff and SP-Cff ion-lock 1 restored more slowly (within ∼90 min) and only partially the oxygenation and dynamic compliance observed with SP-Css ion-lock 1 and SP-C33 UCLA. The corresponding oxygenation and dynamic compliance curves obtained for SP-Css (not shown for clarity) overlapped those of SP-Cff in Fig. 9. Last, instillation of the negative control of Lipids alone had minimal effects on arterial oxygenation or compliance. The relative order of pulmonary activity in terms of both oxygenation and compliance (≥90 min) was given as: SP-Css ion-lock 1 ∼ SP-C33 UCLA (positive control) ∼ SP-C33 ion-lock 1 ≫ SP-Cff ∼ SP-Css > SP-Cff ion-lock 1 ≫ Lipids (negative control). Although the differences in oxygenation and compliance between SP-Css ion-lock 1, SP-C33 UCLA and SP-C33 ion-lock 1 were not statistically significant starting at 90 min after surfactant instillation, each of these SP-C mimics consistently exceeded the corresponding performance of SP-Cff, SP-Css, SP-Cff ion-lock 1 and lipids only surfactants (p < 0.01) (Fig. 9).