Rachael L.
Grime‡
ab,
Richard T.
Logan‡
a,
Stephanie A.
Nestorow
a,
Pooja
Sridhar
a,
Patricia C.
Edwards
c,
Christopher G.
Tate
c,
Bert
Klumperman
d,
Tim R.
Dafforn
a,
David R.
Poyner
e,
Philip J.
Reeves
*f and
Mark
Wheatley
*gb
aSchool of Biosciences, University of Birmingham, Birmingham, B15 2TT, UK
bCentre of Membrane Proteins and Receptors (COMPARE), University of Birmingham and University of Nottingham, Midlands, UK
cMRC Laboratory of Molecular Biology, Francis Crick Avenue, Cambridge, CB2 0QH, UK
dDepartment of Chemistry and Polymer Science, Division of Polymer Science, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa
eLife and Health Sciences, Aston University, Birmingham B4 7ET, UK
fSchool of Life Sciences, University of Essex, Wivenhoe Park, Essex CO4 3SQ, UK. E-mail: preeves@essex.ac.uk
gCentre for Sport, Exercise and Life Sciences, Institute for Health & Wellbeing, Alison Gingell Building, Coventry University, Coventry, CV1 2DS, UK. E-mail: mark.wheatley@coventry.ac.uk
First published on 2nd August 2021
Membrane proteins are of fundamental importance to cellular processes and nano-encapsulation strategies that preserve their native lipid bilayer environment are particularly attractive for studying and exploiting these proteins. Poly(styrene-co-maleic acid) (SMA) and related polymers poly(styrene-co-(N-(3-N′,N′-dimethylaminopropyl)maleimide)) (SMI) and poly(diisobutylene-alt-maleic acid) (DIBMA) have revolutionised the study of membrane proteins by spontaneously solubilising membrane proteins direct from cell membranes within nanoscale discs of native bilayer called SMA lipid particles (SMALPs), SMILPs and DIBMALPs respectively. This systematic study shows for the first time, that conformational changes of the encapsulated protein are dictated by the solubilising polymer. The photoactivation pathway of rhodopsin (Rho), a G-protein-coupled receptor (GPCR), comprises structurally-defined intermediates with characteristic absorbance spectra that revealed conformational restrictions with styrene-containing SMA and SMI, so that photoactivation proceeded only as far as metarhodopsin-I, absorbing at 478 nm, in a SMALP or SMILP. In contrast, full attainment of metarhodopsin-II, absorbing at 382 nm, was observed in a DIBMALP. Consequently, different intermediate states of Rho could be generated readily by simply employing different SMA-like polymers. Dynamic light-scattering and analytical ultracentrifugation revealed differences in size and thermostability between SMALP, SMILP and DIBMALP. Moreover, encapsulated Rho exhibited different stability in a SMALP, SMILP or DIBMALP. Overall, we establish that SMA, SMI and DIBMA constitute a ‘toolkit’ of solubilising polymers, so that selection of the appropriate solubilising polymer provides a spectrum of useful attributes for studying membrane proteins.
In this study, we report the first systematic investigation into the effect of the polymer structure on the range of conformational changes exhibited by a dynamic protein (rhodopsin, the dim light receptor in the eye activated by light) when encapsulated in a SMALP, SMILP or DIBMALP. Rhodopsin is a G-protein-coupled receptor (GPCR). GPCRs are central to cell signalling throughout the evolutionary tree from humans to viruses (but not bacteria) and form the largest class of ‘chemical switches’ in biology. They transduce signals from chemical messengers acting on a cell, such as hormones and neurotransmitters, into biochemical changes within the cell via activation of intracellular signalling cascades.20 As a result, they regulate almost every physiological process. Consequently, GPCRs form the largest family of membrane proteins in the human genome (with >800 receptors) and are very important to the pharmaceutical industry as they are the therapeutic target of 30–40% of clinically-prescribed drugs.21,22 GPCRs share a common protein architecture of a bundle of seven transmembrane helices (TMs) and adopt a wide spectrum of conformational states in executing their cell signalling role.23 Rhodopsin is an archetypical GPCR for which the intermediate structures between the inactive and active states have been defined in great detail.24 This makes rhodopsin an excellent ‘probe’ to investigate the effect of the chemical composition of the stabilising polymer belt of structurally-related ‘SMALP-like’ lipid particles on the conformational changes of the encapsulated membrane protein.
Our study establishes for the first time that the choice of polymer used to solubilise a membrane protein dictates the changes in conformation observed for the encapsulated protein. Moreover, these different polymers effectively provide a ‘toolkit’ that can be exploited by researchers to study different conformational states and provide a range of stabilities that expand the range of down-stream applications and allow the investigator to tailor the properties of SMALP-like lipid particle to their research needs.
(1) |
Mass = c × Mr | (2) |
Fig. 1 Structurally-related polymers used to solubilise membrane proteins. (a) SMA, (b) SMI and (c) DIBMA. |
In order to investigate the effect of the chemical composition of the polymer belt on the dynamic properties of the encapsulated protein, we characterised photoactivation of rhodopsin encapsulated in a SMALP, SMILP or DIBMALP. Activation of rhodopsin has been defined in great detail and occurs via a series of specific intermediates.24 The photoreactive chromophore of rhodopsin is 11-cis retinal covalently bound to Lys296 by a protonated Schiff base. Absorption of a photon of light induces rapid isomerisation of 11-cis retinal to the all-trans configuration, a potent agonist. This drives the conversion of dark-adapted inactive rhodopsin through a series of structurally-defined intermediates, each with its characteristic absorption maximum (λmax), to the fully-activated conformation metarhodopsin II (Meta II) which absorbs at 382 nm (Fig. S1†).24 In addition to crystal structures of the dark inactive rhodopsin and fully active Meta II, there are crystal structures of the intermediate states bathorhodopsin, lumirhodopsin and Meta I28–31 making rhodopsin the most structurally-defined GPCR. These structures reveal that rhodopsin activation is not a gradual build-up of conformational change, as there is relatively little change in the overall structure of light-activated rhodopsin as it transitions between the photointermediates from the dark ground-state as far as Meta I. In contrast however, conversion of Meta I to Meta II is accompanied by a large conformational change at the cytoplasmic end of transmembrane helix-6 (TM6), which moves out from the helical bundle, opening a binding crevice to allow docking of the G-protein, transducin (Gt).32,33 This large outward movement of the cytoplasmic end of TM6 (by 7–14 Å) is a characteristic feature of full activation of family A GPCRs in general.23,34 This detailed knowledge of the structural and spectroscopic features of the conformational states of rhodopsin during its photoactivation means that we can track the sequential conformational changes from the inactive dark state to the fully active Meta II conformation. Rhodopsin is therefore an excellent ‘probe’ to investigate the effect of the chemical composition of the stabilising polymer belt of a ‘SMALP-like’ lipid particle on the conformational changes of the encapsulated membrane protein. Furthermore, using bovine retina rhodopsin as the encapsulated membrane protein allowed us to use the wild-type protein resident in rod cell disc membranes and avoided the use of an engineered recombinant construct with introduced reporter groups, or structural changes, that might affect conformational changes.
Dark-adapted rhodopsin possesses a characteristic absorption peak (λmax) at 500 nm,24 and this peak was clearly evident when rhodopsin was solubilised by either the widely-used detergent n-dodecyl-β-D-maltopyranoside (DDM) or by encapsulation in one of the polymer-stabilised lipid particles (SMALP, SMILP or DIBMALP) (Fig. 2a–d). The efficiency of solubilisation of rhodopsin was similar for DDM (1%), SMA (2.5%), SMI (2.5%) and DIBMA (5%), where the number in the brackets is the concentration used in each case. The extraction efficiency values relative to DDM were 99 ± 5.5%, 97 ± 2.6% and 98 ± 4.8% (mean ± s.e.m., n = 3) for SMA, SMI and DIBMA respectively. Optimisation of rhodopsin solubilisation by DIBMA (Table S1†) revealed that a slightly higher concentration of DIBMA (5%) was required to generate the same extraction efficiency as SMA (2.5%) and SMI (2.5%). Photoactivation (30 s, hν > 495 nm) of the DDM-solubilised rhodopsin generated the characteristic changes in the rhodopsin spectrum, with loss of the peak at 500 nm corresponding to ground-state rhodopsin and appearance of a new absorption peak (λmax) at 382 nm (Fig. 2a), consistent with the formation of the fully-active rhodopsin conformation Meta II (Fig. S1†).24 Likewise, photoactivation of rhodopsin encapsulated in a SMALP or SMILP caused a loss of the peak at 500 nm but in contrast to DDM, generated a new absorption peak with λmax = 478 nm rather than 382 nm (Fig. 2b and c). This revealed that photoconversion of rhodopsin in a SMALP or SMILP proceeded only as far as Meta I (λmax = 478 nm; Fig. S1†) but did not progress further to the active Meta II conformation (λmax = 382 nm), even after an extended photobleaching period of 150 s (Fig. 2b and c). Both SMA and SMI contain styrene rings which give absorption in the UV region (<300 nm) of the spectrum.16,17 In marked contrast to SMALP or SMILP, photoactivation of rhodopsin encapsulated in a DIBMALP progressed beyond Meta I to the active Meta II conformation (λmax = 382 nm; Fig. 2d). The slight absorption observed >430 nm may indicate the presence of low levels of residual photointermediates. The slightly higher absorbance <400 nm in the dark-state Rho-DIBMALP spectrum compared to Rho-SMALP and Rho-SMILP, is probably due to greater light-scattering by the larger DIBMALP.
Fig. 2 Spectra of dark-state and photoactivated rhodopsin in (a) DDM, (b) SMALP, (c) SMILP and (d) DIBMALP. |
The active conformation of rhodopsin can be stabilised by occupation of the G-protein binding crevice on the cytoplasmic side of the receptor. This can be achieved by incubation with a small 11-mer peptide corresponding to the C-terminus sequence of the α5 helix of the α-subunit of transducin, referred to as G(t)-peptide35,36 or by formation of a rhodopsin:mini-Go complex.34 Mini-G proteins comprise the guanosine triphosphatase (GTPase) domain of the α-subunit of heterotrimeric G-proteins and are fully capable of stabilising GPCRs in their active state.37 Although full activation of rhodopsin to Meta II was not observed in SMALP or SMILP, it was feasible that rhodopsin in a SMALP or SMILP could transition inefficiently to Meta II but at levels too low to detect. To address this, photoactivation was performed in the presence of Gt peptide or mini-Go to trap any Meta II formed. However, neither G(t)-peptide (500 μM) nor mini-Go (present in 500-fold molar-excess) increased Meta II formation (Fig. S2†). Therefore, it is highly unlikely that rhodopsin encapsulated in a SMALP or SMILP can transition to the Meta II conformation. It is noteworthy that the two polymers that prevented Meta II formation (SMA and SMI) both possess a styrene aromatic ring but when this is replaced by aliphatic diisobutylene in DIBMA (Fig. 1) rhodopsin can transition to the active Meta II conformation. It is known that the phenyl rings of the styrene moieties in the polymer belt intercalate between the lipid acyl chains perpendicular to the plane of the lipids,5 which can affect the lipid packing order and the transition temperature for gel to liquid phase transitions.17 This could have important functional ramifications as membrane fluidity could impact on the conformational flexibility of dynamic membrane proteins like GPCRs, including rhodopsin. Interestingly, rhodopsin solubilised by the detergent digitonin also gets trapped at Meta I upon activation possibly due to the digitonin micelle forming a more rigid environment.38 In contrast, solubilisation by DIBMA has been shown to cause very little perturbation to the packing and phase transition temperature of the lipids.17 Furthermore, electron paramagnetic resonance spectroscopy and course-grained molecular dynamics simulations revealed that the dynamics of lipids in a DIBMALP are less constrained than in a SMALP and the affinity of spin-labelled lipids for the polymer belt was more pronounced in a SMALP.39 There are also differences in lipid transfer, which is slower among DIBMALPs than among SMALPs.40 The current study shows that although light-induced formation of Meta I occurs within a SMALP or SMILP environment, there is some degree of conformational restriction endowed by the nature of the copolymer belt that prevents the full activation of rhodopsin to Meta II observed within a DIBMALP. The suggestion that the copolymer component of a SMALP or SMILP can restrict the range of conformational transitions adopted by the encapsulated protein, either directly or via changes in lipid packing and dynamics, is supported by recent studies on the human adenosine A2A receptor (A2AR) and the photoreceptor/transducer complex (NpSRII2/NpHtrII2) from Natronomonas pharaonisfrom.11,41 Stimulation of A2AR in a SMALP by the agonist NECA generated only small conformational changes11 and the conformational freedom of NpSRII2/NpHtrII2 was also restricted in a SMALP.41 Interestingly, NpSRII2 could be fully activated when encapsulated in a DIBMALP.42
Fig. 3 Size of the Rho-LPs determined by DLS. A representative experiment is shown for (a) Rho-SMALP, (b) Rho-SMILP and (c) Rho-DIBMALP. |
Fig. 4 Temperature-dependent changes in Rho-LP diameter determined by DLS, for Rho-SMALP (), Rho-SMILP () and Rho-DIBMALP (). Data are mean ± s.e.m. (n = 3). |
In the current study, encapsulation in a DIBMALP provided rhodopsin with a greater degree of thermostability than in a SMALP (Fig. 6a and b). However, it appears that there are protein-specific differences with respect to the comparative stability of proteins encapsulated in a SMALP versus a DIBMALP. For example, the thermostability of the human serotonin transporter (hSERT) is comparable in a SMALP or a DIBMALP but the A2AR in a DIBMALP was reported to be less stable to long-term storage (6 days) at 4 °C than in a SMALP.44,46
Rho-LP were characterised further using sedimentation velocity analytical ultracentrifugation (AUC; Fig. 5). The data were analysed using the continuous c(s) analysis method to determine sedimentation coefficients and molar masses, using the SEDFIT software and the method of Schuck.27 The large peak in each trace, with sedimentation coefficients of ∼1.5S (Rho-SMALP; Fig. 5a), ∼2S (Rho-SMILP; Fig. 5b) and 3S (Rho-DIBMALP; Fig. 5c) respectively, is likely to be free polymer and polymer aggregates. The second peak in each trace is compatible with the known sizes of the Rho-LPs. The sedimentation coefficient of 4.6S for Rho-SMALP is consistent with a molar mass of ∼89 kDa. It has been determined previously that the SMA polymer and lipid of a SMALP contribute ∼35 kDa of a ∼70 kDa SMALP encapsulating the protein ZipA.7 This would imply that the Rho component of Rho-SMALP is ∼54 kDa. Given that the mass of Rho, including the post-translational modifications (two glycosylation sites plus two palmitoylation sites) is ∼43 kDa,47 this would indicate that the encapsulated Rho was a monomer and not a dimer (∼86 kDa) or higher oligomer. This is relevant as although it is known that a single light-activated Rho is sufficient to activate its G-protein transducin,48 in native rod outer segment membranes Rho can organise into rows of dimers.49,50 The fact that monomeric Rho and not Rho dimers were encapsulated in the SMALP might be a ramification of the precise molecular mechanisms by which SMA monomers are incorporated into the membrane and polymerise to excise the SMALPs. What is clear, is that encapsulation of Rho monomers was not due to there being insufficient space within a SMALP to accommodate the 14-transmembrane helices (TMs) of a rhodopsin dimer, as Alternative Complex III in a functional super-complex with cytochrome oxidase was isolated encapsulated within a SMALP despite having a total of 48-TMs.12 The DLS experiments presented earlier in this study highlighted that Rho-SMALP and Rho-SMILP were of similar size. AUC revealed that Rho-SMILP had a sedimentation coefficient 4.1S consistent with a molar mass of ∼87 kDa. Lipid-only ‘empty’ SMILP containing only 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC-SMILP) without encapsulated protein gave a single major peak (2.35 S, Fig. S3†) consistent with a molar mass of 56 kDa, with a minor component at ∼5S. This is 31 kDa less than the Rho-SMILP, suggesting encapsulation of a single Rho (∼43 kDa), like Rho-SMALP, and not a dimer (∼86 kDa). DLS revealed that Rho- DIBMALP is larger than either Rho-SMALP or Rho-SMILP and this was apparent in the AUC as Rho-DIBMALP had a sedimentation coefficient of 5.2 S and a predicted molar mass of 117 kDa. Lipid-only ‘empty’ DIBMALP (DMPC-DIBMALP) gave a single major AUC peak (3.35 S, Fig. S3†) consistent with a molar mass of 65 kDa, with a minor component at ∼5S. This is 52 kDa less than the Rho-DIBMALP, again suggesting encapsulation of a Rho monomer (∼43 kDa) and not a dimer (∼86 kDa). In addition to investigating the effect of temperature on the lipid particles (Fig. 4), the thermostability of the encapsulated protein within each Rho-LP was also compared. Each Rho-LP was incubated at a range of temperatures between 4 °C–60 °C, using 5 °C increments, and the thermal stability of Rho was determined by the decrease in absorbance at 500 nm which represents the release of the 11-cis-retinal chromophore (Fig. 6a). The Rho within a Rho-SMALP was the least thermostable with a T50 = 45.7 ± 0.5 °C (mean ± s.e.m., n = 3). Encapsulation of Rho in DIBMALP conferred an increase in Rho thermostability of 6 °C (T50 = 51.4 ± 0.8 °C mean ± s.e.m., n = 3) compared to Rho-SMALP (Fig. 6a). The lower thermostability of Rho in a SMALP compared to a DIBMALP may reflect the lower thermostability of the SMALP particle compared to the DIBMALP particle (Fig. 4). Rho was also more thermostable in a SMILP than a SMALP (Fig. 6a), although T50 = >50 °C, a precise T50 value could not be determined for Rho-SMILP as above 55 °C a slight precipitate formed which prevented data collection at higher temperatures. The rate of thermal decay of Rho was determined at 37 °C for each Rho-LP using a pre-heated spectrophotometer and monitoring absorbance at 500 nm for 16 h (Fig. 6b). Consistent with data presented above, the Rho encapsulated in a SMALP was the least thermostable at 37 °C (Rho-SMALP; t1/2 = 183 ± 5 min; mean ± s.e.m. (n = 3)). In contrast, Rho within a SMILP or a DIBMALP showed no thermal decay over the 16 h period. It is noteworthy that the rate of Rho decay in the Rho-SMALP at 37 °C was very similar to the loss of ligand binding capability of another GPCR, the human adenosine A2A receptor (A2AR) encapsulated in a SMALP incubated at 37 °C (A2AR-SMALP; t1/2 = 148 ± 13 min; mean ± s.e.m. (n = 3)).6 As an extreme test of stability, the effect of lyophilisation on Rho-LPs was quantified using absorbance at 500 nm to assess Rho decay following lyophilisation and re-hydration with buffer (Fig. 6c). Despite such harsh treatment, ∼50% of the Rho encapsulated in a SMALP or DIBMALP was recovered with the chromophore intact (54 ± 3% and 48 ± 7% respectively, mean ± s.e.m. (n = 3)). However, Rho encapsulated in SMILP was even more stable and was unaffected by lyophilisation (recovery 106 ± 2%, mean ± s.e.m., n = 3; Fig. 6c). The availability of such a stable storage form of a membrane protein as a lyophilised SMILP could be a useful resource not only for storage but as a reagent for downstream applications.
Fig. 5 Sedimentation velocity AUC analysis of Rho-LPs. A representative AUC experiment is shown for (a) Rho-SMALP, (b) Rho-SMILP and (c) Rho-DIBMALP. |
Defining the characteristics of Rho, a native GPCR, encapsulated by three different but structurally-related polymers (Rho-SMALP, Rho-SMILP and Rho-DIBMALP) enabled direct comparisons to be made. SMALP and SMILP allowed conformational change in light-activated Rho as far as Meta I but the effect of the styrene ring in these polymers, or constraints due to their smaller size, prevented complete photoconversion of Rho to the fully-active Meta II observed in Rho-DIBMALP. Consequently, SMALP and SMILP, both of which prevented full activation of Rho, might facilitate structural studies on conformational intermediates or drug discovery programmes where stabilisation of an inactive or partially-active intermediate receptor conformation is required. In contrast, GPCR-DIBMALP may provide a better platform for a screen in a discovery programme for a drug that fully activates the receptor. It would be important nevertheless, to confirm the conformational sub-state of individual GPCRs solubilised by SMA, SMI or DIBMA and not assume that all GPCRs behave exactly like rhodopsin. Furthermore, not all important GPCR conformational states will be accessible to investigation simply through choice of solubilising polymer alone.
This study shows for the first time, that the dynamic range of conformational changes exhibited by a membrane protein encapsulated in a nanoscale lipid-particle is dictated by the chemical composition of the stabilising polymer employed to generate the nanoparticle. Using Rho, a member of the large GPCR family of membrane proteins that constitute the primary therapeutic target in drug discovery,21 we establish that SMA, SMI and DIBMA provide a ‘toolkit’ of solubilising polymers. Selection of the appropriate solubilising polymer provides a spectrum of useful attributes for studying membrane proteins, including generation of specific intermediate states and differences in thermostability of the encapsulated protein. This new insight into the expanded utility of the SMA-like polymers will have practical benefits for researchers studying membrane proteins using a wide range of approaches.
Footnotes |
† Electronic supplementary information (ESI) available: Supplementary table (Table S1) and figures (Fig. S1–S3). See DOI: 10.1039/d1nr02419a |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2021 |