Grazing-angle Neutron Diffraction Study of Water Distribution in Membrane Hemifusion: From Lamellar to Rhombohedral Phase Shuo Qian 1,2, Durgesh K. Rai 1 1
Neutron Scattering Division and 2Center for Structural Molecular Biology, Oak Ridge National Laboratory, Oak Ridge, TN 37831, U.S.A.
Corresponding Author * Shuo Qian; Oak Ridge National Laboratory; PO Box 2008; MS-6393; Oak Ridge, TN 37831; email:
[email protected]; phone: 865-241-1934
Materials and Methods Materials and Sample Preparation: The substrate-supported sample was prepared by following the established organic solventdeposit-incubation method for multilayer membrane films1. DOPC and DOPE were mixed in a 1:1 molar ratio in a solvent composed of trifluoroethanol: chloroform = 1:1 (v/v). The mixture then was deposited on a quartz substrate and the organic solvent was evaporated in air to allow the formation of a visually uniform multilayer. Afterward, the sample was placed in vacuum for about an hour to eliminate any traces of organic solvent. Then the multilayer sample was rehydrated with saturated D2O vapor in a sealed jar at 40 °C overnight. The film looked uniform upon visual inspection and was used within 3 days of preparation for neutron diffraction measurement. A total of three substrates with about 5 mg of lipid on each were stacked together for the experiment. This approach significantly increased the sample amount in the neutron beam and therefore reduced the required experiment time. Aluminum foil strips were placed between
the quartz substrates to allow quick changes and equilibration of relative humidity (RH) during the experiment. The grazing-angle neutron diffraction data were collected at the Bio-SANS instrument at the High Flux Isotope Reactor and the EQ-SANS instrument at the Spallation Neutron Source (SNS) at Oak Ridge National Laboratory2. At the Bio-SANS, the neutron wavelength was set to 5.3 Å 43
with a wavelength spread of FWHM Δλ/λ ~0.10 by tilting a neutron velocity selector
(Mirrotron Inc., Budapest, Hungary). No further monochromator was used so we could use the much-needed neutron flux with a compromised wavelength resolution. The sample-to-detector distance (SDD) was fixed at 1.13 m. At the EQ-SANS, an SDD of 1.3 m with a wavelength band from 2.6 to 6.2 Å was used. The time-of-flight instrument EQ-SANS provides much superior wavelength resolution over the entire neutron wavelength band compared with the Bio-SANS. The difference in the wavelength resolution dominated the corresponding instrument resolution differences between the Bio-SANS and the EQ-SANS. Therefore, the Bio-SANS data were used only for preliminary sample screening; all the data used in the structural reconstruction were measured at the EQ-SANS. For both instruments, configurations were used to provide an effective q-range of ~0.03 Å-1 to 0.7 Å-1. The area detectors used at both instrument consisted of linear tubes manufactured by GE (GE Measurement and Control; Twinsburg, OH, USA), and developed and assembled by Oak Ridge National Laboratory staff3. The detectors were about 1×1 m in total size with a pixel size of 5.5×4 mm. A beam stop with boron nitride was placed in front of the area detectors to block the direct beam. Sample Relative Humidity Control The lipid membrane sample was housed in a sample cell about 10×10×8 cm with controlled relative humidity (RH) and temperature throughout the experiment. The sample was held by a
sample holder housed within the cell. The sample holder was in tight contact with the cold plate to ensure effective temperature control. Once closed, the cell was gastight to ensure no exchanges of water vapor between the ambient air and the generated humidified air that contained a desirable D2O/H2O ratio. The desired RH was provided by flowing humidified air generated by an RH generator (Model RH-200, L&C Science, Hialeah, FL) in the range of 3–95% RH. Briefly, saturated water vapor was generated by an evaporator inside the RH generator, and a separate dry air flow was fed into the generator. Two mass flow controllers, controlled by a PID controller, were used to regulate the separate flows of dry and saturated air that were mixed to obtain the specified RH air flow. The humidified air was transferred by a temperature-controlled tube to the sample position. The controlling RH sensor used for feedback control was positioned near the sample position. Another RH sensor (SHT75, Sensirion, Switzerland) was installed at another location in the cell as a separate measure to check the consistency of the RH in the sample cell. The air flow rate was adjustable to up 500 cc/min, providing ample air mass to humidify the entire sample cell uniformly. The evaporator of the RH generator was fed by a water reservoir in which the water content could be changed to the desired D2O/H2O ratio. Both RH and temperature were controlled remotely by a computer. The accuracy of the RH was within ±1.8%, dictated by RH sensors. Membrane Diffraction Data Collection and Reduction: The data collection effort used the existing SANS instruments data acquisition systems without any special implementations. Under different RH conditions, the sample changed from onedimensional (1D) lamellar to multi-dimensional non-lamellar structures. Lamellar diffraction produces only reflection along the qz, which is perpendicular to the plane of the sample substrate.
Non-lamellar diffraction, such as the rhombohedral phase, however, produces both reflection along qz and off-specular reflections in the qr-qz plane that need to be recorded separately. We describe the data collection and reduction separately for specular and off-specular peaks. Specular peaks are reflections along qz. They were collected by oscillating ω scan, also called “rocking scan.” The sample was rotated around the x axis, vertically from ω = 0° to ~10° in steps of Δω=0.1° or 0.2°, for about 5 minutes at each step. After normalization to the proton charge (on the EQ-SANS), the data at each step were corrected for detector dark current, pixel sensitivity, and empty background by the facility-supplied reduction software package Mantid (http://mantidproject.org/). Then all the data sets in the 2D qr-qz map were summed up and analyzed using routines implemented in MATLAB (MathWorks Inc., Natick, MA) to give the complete intensity of the peaks along qz. The integrated intensity I was obtained by integrating the peak from the 2D qr-qz map. The diffraction amplitudes |F| obtained from the integrated diffraction peak intensity needed to be corrected for the Lorentz factor CL, the absorption correction factor Cabs, and the geometry correction factor Cgeo following Eq. (1)4: 𝐹 =
𝐼 𝐼! 𝐶! 𝐶!"# 𝐶!"# (2𝜋/𝛺) ,
(1)
where I0 is the incident neutron flux, 𝛺 is the angular velocity of the scan. The Lorentz factor is given by 𝐶! = 1/sin2θ ,
(2)
where θ is the scattering angle. The absorption factor is given by 𝐶!"# =
!!!"# (!!!"/!"#$) !!"
,
(3)
where 𝜇 is the absorption coefficient of the sample, d is the thickness of the sample. Cgeo is a constant, as the amount of sample immersed in the neutron beam remained approximately unchanged for the angular range in the scans. The off-specular peaks in non-lamellar diffraction are collected by orientating the sample substrate at an angle of ~1° with respect to the incident neutron beam. After similar normalization to the proton charge, dark current, pixel sensitivity, and background correction, the diffraction amplitude is given by 𝐹 =
𝐼 𝐼! 𝐶! 𝐶!"# 𝐶!"# 𝑡 ,
(4)
where t is the scan time. The Lorentz factor is given by !
𝐶! = !"#! !"#! !"#$ ,
(5)
where Δ is the incident angle, α is the angle between diffracted beam and sample surface (X-Y plane); β is the angle between the Y and the X-Y plane projection of the diffracted beam (Fig. S1). The absorption factor is given by
𝐶!"# =
! ! ! )) !"#$ !"#$ ! ! !"( ! ) !"#$ !"#$
!!!"# (!!!"(
.
(6)
The geometry correction factor Cgeo remains a constant in this case as well. where 𝜇 is the absorption coefficient of the sample, d is the thickness of the sample.
Figure S1: The sample and diffraction geometry. The sample is held vertically and can be rotated around axis X(ω); the sample normal direction is defined as Z; Y direction is perpendicular to the X-Z plane and parallel to the sample surface. In reciprocal space, qz is defined to be perpendicular to the sample substrate while qr is in the plane of the substrate. The specular diffraction peaks fall along qr with diffraction angle 2θ. The off-specular peaks distribute in the qr-qz plane. ɑ is the angle between the diffracted beam and the X-Y plane (sample surface); β is the angle between the Y and the X-Y plane projection of the diffracted beam.
Figure S2: The hexagonal phase of the sample below RH 40%.
A
B
Figure S3: The contour plots around peaks in the diffraction patterns (A) RH 57% and (B) RH 50%. The index numbers follow the ones used in Table 2.
Unit cell, Patterson map, Phase Determination, and Normalization:
The D-spacings of the lamellar phase at RH 85% are determined by the peak position: 𝐷=
!! !
!!
= !.!"!Å!! = 48.0 ± 1.6Å ; at RH 82%, by the peak position: 𝐷 =
!! !
!!
= !.!"#Å!! =
47.6 ± 1.6Å. The uncertainties were estimated from the instrument configuration, especially the pixel spatial resolution. -
The rhombohedral phase is of symmetry R3. The crystal lattice of a unit cell can be defined by !
crystal axes (𝑎,0,0), (− !, !
!
(0,0,0), (!, ! !,𝑐), (0,
!
!
!! !
,0), (0, 0,3𝑐). It consists of three primitive cells positioned at
,2𝑐). With the crystal axes, the peaks are indexed by (h,k,l) based on a !
set of reciprocal vectors B1=(!,
!
,0),B2= (0, !!
!
!
,0) and B3= (0, 0,!!). The lattice constants !!
were c=41.6 ±1.2Å and a=57.4 ±2.1 Å for RH 57%; c=40.8 ±1.2Å and a=55.4 ±2.0 Å for RH 50%. The Patterson map was constructed using amplitudes |F(h,k,l)|5,6: 𝑃 𝐫 =
!,!,!
𝐹 ℎ, 𝑘, 𝑙
!
cos (2𝜋 ℎ𝑩𝟏 + 𝑘𝑩𝟐 + 𝑙𝑩𝟑 ∙ 𝒓) .
(7)
The lipidic structure is centrosymmetric; therefore, the phases of the amplitudes are either positive or negative. The swelling method was applied for both the lamellar phase (Fig. S4a) and the rhombohedral phase (Fig. S4b, c) to obtain the phases. For the rhombohedral phase, it was applied to the series of peaks along (h,k)= (0,0) and (h,k)=(1,0) separately 7,8. The relative phases between them needed to be determined. There are four possibilities among two series. We inspected the four possibilities and ruled out ones that didn’t make physical sense according to the continuation of the lipid bilayer structure9.
(a)
(b)
(c)
Figure S4: (a) Swelling method for phase determination applied to lamellar phase for diffraction amplitudes along qz; RH 85% data are shown in blue, RH 82% data are shown in red. (b) Swelling method for phase determination applied to the rhombohedra phase for different (h,k)
series: (h,k)=(0,0), (c) (h,k)=(1,0); RH 57% data are shown in blue, RH 50% data are shown in red. The vertical axis represents amplitudes in relative scale. The structure was reconstructed with phases determined from the swelling method and amplitude structure: 𝜌 𝐫 =
!,!,! 𝐹(ℎ, 𝑘, 𝑙)cos
(2𝜋 ℎ𝑩𝟏 + 𝑘𝑩𝟐 + 𝑙𝑩𝟑 ∙ 𝒓) .
(8)
The density was normalized according to the actual NSLD of the D2O/H2O and lipid chains. The experimental NSLD was scaled to normalized NSLD by 𝜌!"#$%&'()* = 𝑎 ∙ 𝜌!"#!$%&!'( + 𝑏, where a and b are constant scaling factors. Based on the fact that the maximum density in the system was that of the D2O/H2O mixture (𝜌!"#$% = 0.4992×1015m-2), and the minimum density was that of the lipid hydrocarbon chain (𝜌!!!!!"# =-0.037×1015m-2), a and b can be determined.
Figure S5: Neutron scattering length density distribution consisting of two adjacent bilayers at RH 50%. (A) Contoured density distribution along X. (B) Contoured density distribution along Y. Blue is indicative of water distribution while white is devoid of water. The color map unit for is 1015 m-2. Water Change in the Lamellar Phase: It is assumed that the changes in the shoulder peak were caused solely by the water change associated with the headgroup. The areal difference under the shoulder peaks in a unit area (𝐴!"#$ ) provides a total scattering length difference that is used to estimate the number of water molecules, Nwater: 𝑁!"#$% =
!!"#$ ∗( !!!.!"×!"!" !!"# ∗!!"#$ ) !!"#$% ∗!!"#
%$.
(9)
The areas under the 1D profile were obtained by summing up the NSLD that was larger than the average headgroup density of ~0.25×1015 m-2, which was also half of the D2O/H2O mixture NSLD (𝜌!"#$% = 0.5×1015 m-2) used in the experiment. Lunit ~ 1 Å is the unit length used to generate the profile; vwater is ~29.7 Å3, the volume of a single water molecule10. The Number of Water Molecules between the Merging Bilayers in the Rhombohedral Phase: The number of water molecules between the merging bilayers in the rhombohedral phase was calculated by integration of the NSLD that was larger than 0.375×1015m-2 (medium point between average headgroup density of ~0.25×1015 m-2 and water density of ~0.5×1015 m-2), over the volume between Z= ~ −20 Å and ~20 Å in the unit cell. This total scattering length provides a total volume of water in between the bilayers; therefore, the number of the water is 𝑁!"#$% =
!!!.!"#×!"!" !!"# ∗!!"#
%$!!"#$% ∗!!"#
%$,
(10)
where ρexp is the experiment NSLD distribution reconstructed; Vvoxel is the unit voxel used in the reconstruction, typically 1Å3; ρwater is the NSLD of water, ~0.5×1015 m-2; vwater is ~29.7 Å3 the volume of a single water molecule10. Because the cutoff value 0.375 ×1015m-2 is higher than any of the lipid NSLDs, the results were not sensitive to the actual Z range as long as the range covered the central water pocket region. References (1) Pan, D.; Wang, W.; Liu, W.; Yang, L.; Huang, H. W. Chain Packing in the Inverted Hexagonal Phase of Phospholipids: A Study by X-Ray Anomalous Diffraction on Bromine-Labeled Chains. J. Am. Chem. Soc. 2006, 128 (11), 3800–3807. (2) Heller, W. T.; Cuneo, M.; Debeer-Schmitt, L.; Do, C.; He, L.; Heroux, L.; Littrell, K.; Pingali, S. V.; Qian, S.; Stanley, C.; et al. The Suite of Small-Angle Neutron Scattering Instruments at Oak Ridge National Laboratory. J. Appl. Crystallogr. 2018, 51 (2). (3) Berry, K. D.; Bailey, K. M.; Beal, J.; Diawara, Y.; Funk, L.; Steve Hicks, J.; Jones, A. B.; Littrell, K. C.; Pingali, S. V.; Summers, P. R.; et al. Characterization of the Neutron Detector Upgrade to the GP-SANS and Bio-SANS Instruments at HFIR. Nucl. Instrum. Methods Phys. Res. Sect. Accel. Spectrometers Detect. Assoc. Equip. 2012, 693, 179–185. (4) Ding, L.; Weiss, T. M.; Fragneto, G.; Liu, W.; Yang, L.; Huang, H. W. Distorted Hexagonal Phase Studied by Neutron Diffraction: Lipid Components Demixed in a Bent Monolayer. Langmuir 2005, 21 (1), 203–210. (5) Warren, B. E. X-Ray Diffraction; Addison-Wesley Pub. Co.: Reading, Mass., 1969. (6) Qian, S.; Wang, W.; Yang, L.; Huang, H. W. Structure of the Alamethicin Pore Reconstructed by X-Ray Diffraction Analysis. Biophys. J. 2008, 94 (9), 3512–3522. (7) Bragg, S. L.; S, F. R.; Perutz, M. F. The Structure of Haemoglobin. Proc R Soc Lond A 1952, 213 (1115), 425–435. (8) Qian, S.; Wang, W.; Yang, L.; Huang, H. W. Structure of Transmembrane Pore Induced by Bax-Derived Peptide: Evidence for Lipidic Pores. Proc. Natl. Acad. Sci. 2008, 105 (45), 17379–17383.
(9) Qian, S.; Huang, H. W. A Novel Phase of Compressed Bilayers That Models the Prestalk Transition State of Membrane Fusion. Biophys. J. 2012, 102 (1), 48–55. (10) Gerstein, M.; Chothia, C. Packing at the Protein-Water Interface. Proc. Natl. Acad. Sci. U. S. A. 1996, 93 (19), 10167–10172.
