Abstract
Mammalian cytochrome b5 (cyt b5) and cytochrome b5 reductase (b5R) are electron carrier proteins required for many membrane-embedded oxidoreductases. Both cyt b5 and b5R have a cytosolic domain anchored to the membrane by a single transmembrane helix (TM). It is not clear if b5R, cyt b5 and their partner oxidoreductases assemble as binary or ternary complexes. Here we show that b5R and cyt b5 form a stable binary complex, and that b5R, cyt b5 and a membrane-embedded oxidoreductase, stearoyl-CoA desaturase 1 (SCD1) form a stable ternary complex. The formation of the complexes significantly enhances electron transfer rates, and that the single TM of cyt b5 and b5R mediated assembly of the complexes. These results reveal a novel functional role of TMs in cyt b5 and b5R and suggest that an electron transport chain composed of a stable ternary complex may be a general feature in oxidoreductases that require the participation of cyt b5 and b5R.
Introduction
Cytochrome b5 (cyt b5) and cytochrome b5 reductase (b5R) are obligatory partners for a number of oxidoreductases such as fatty acid desaturases and elongases, oxygenases, and cytochrome P450s (cyt P450)(Schenkman and Jansson 2003, Elahian, Sepehrizadeh et al. 2014). They form part of an electron transport chain that transfers electrons from a reductant, nicotinamide dinucleotide (NADH) or nicotinamide dinucleotide phosphate (NADPH), to the flavin adenine dinucleotide (FAD) cofactor on b5R(Yamada, Tamada et al. 2013), the heme moiety of cyt b5(Vergeres and Waskell 1995) and finally the metal ions in the catalytic center of oxidoreductases.
Mammalian stearoyl CoA desaturase-1 (SCD1) is a membrane-embedded oxidoreductase that catalyzes the rate-limiting step in the formation of the first double-bond in saturated fatty acids. SCD1 has a major role in the regulation of fatty acid metabolism and membrane synthesis and is a validated drug target(Paton and Ntambi 2009). The enzymatic cycle of SCD1 involves redox transition of a diiron center, which requires participation of b5R and cyt b5 to deliver reducing equivalent from NADH (Scheme 1).
Extensive structural and functional studies have been conducted on the soluble forms of b5R and cyt b5 that lack their transmembrane (TM) segments. These studies show that the soluble domains of b5R and cyt b5 are sufficient to support electron transfer, and that charged residues on the surface of the soluble domains of b5R and cyt b5 likely mediate their interactions(Dailey and Strittmatter 1979, Strittmatter, Kittler et al. 1992, Nishida and Miki 1996, Kawano, Shirabe et al. 1998, Shirabe, Nagai et al. 1998, Samhan-Arias, Almeida et al. 2018). However, less attention has been paid to the role of their TM domains, and a stable b5R/cyt b5 binary complex has never been isolated. Studies of electron transfer from cyt b5 to SCD1 have been limited to demonstration that cyt b5 is required for the activity of SCD1(Paton and Ntambi 2009, Nagao, Murakami et al. 2019). Whether cyt b5 and SCD1 form a stable complex and whether a stable complex enhances the activity of SCD1 have not been explored. These questions are significant in terms of understanding how each redox component of the electron transport chain may function in the native environment and the mechanism of electron transfer. Knowledge of stable binary or ternary complexes of electron transfer partners is also relevant in developing novel strategies to inhibit membrane-bound oxidoreductases, for example, SCD1, which is a validated drug target for many types of cancers(Ackerman and Simon 2014, Theodoropoulos, Gonzales et al. 2016, Savino, Fernandes et al. 2020, Oatman, Dasgupta et al. 2021), neurodegenerative diseases(Vincent, Tardiff et al. 2018, Fanning, Haque et al. 2019, Nuber, Nam et al. 2021), and metabolic diseases(Ntambi, Miyazaki et al. 2002, Gutiérrez-Juárez, Pocai et al. 2006, Aljohani, Syed et al. 2017).
Results
Existence of binary and ternary complexes in cells
We first examined whether SCD1, cyt b5, and b5R form binary or ternary complexes in cells. By fusing SCD1 with a green fluorescent protein (GFP), cyt b5 with a Myc tag, and b5R with a hemagglutinin (HA) tag, we monitored cellular localization of the three proteins by immunofluorescence confocal microscopy. When SCD1 was expressed, meshwork-like distribution of GFP fluorescence was observed (Extended Data Fig. 1), consistent with its localization to the endoplasmic reticulum (ER) membranes(Man, Miyazaki et al. 2006). Cells co-expressed with SCD1 and cyt b5 exhibited overlapping fluorescence, and although the fluorescence from cyt b5 clustered with most of that from SCD1 as shown in yellow color in the merged image (Fig 1a), cyt b5 seemed to have a wider distribution than SCD1 likely due to different expression levels of these proteins and participation of cyt b5 in multiple redox pathways. Colocalization was also evident in cells co-expressing cyt b5 and b5R (Fig. 1b), consistent with their roles in mediating electron transfer to redox enzymes. As expected, co-localization of all three proteins was also observed when all three were co-expressed in the same cells (Fig. 1c). These observations suggest that cyt b5 may form binary complexes with b5R or SCD1 and that the three may form a ternary complex.
To test whether the proximity in their expression pattern leads to the formation of stable binary or ternary complexes, we next examined their interactions by co-immunoprecipitation (co-IP). We found co-IP of cyt b5 and SCD1, cyt b5 and b5R, and all three when either two or three proteins were co-expressed. (Fig. 1f and 1g). We also generated binary fusions by connecting SCD1 and cyt b5, and cyt b5 and b5R to test their assembly with b5R and SCD1, respectively. The SCD1-cyt b5 had a C-terminal GFP tag, and cyt b5-b5R an N-terminal Myc tag. Colocalization analysis showed that SCD1 and cyt b5-b5R (Fig. 1d), SCD1-cyt b5 and b5R (Fig. 1e) were in close proximity in cells. Co-IP results (Fig. 1h and 1i) indicate that the binary fusions afford stronger complex formation compared to individual SCD1, cyt b5, and b5R.
Stable binary complex between b5R and cyt b5
We proceeded to large-scale production of cyt b5 and b5R complex for further biochemical characterizations. However, simply co-expressing full-length cyt b5 and b5R did not produce sufficient amount of complex. To increase the yield of the complex, and encouraged by previous reports of production of stable dimeric membrane proteins after fusion of two monomers (Steiner-Mordoch, Soskine et al. 2008, Nasie, Steiner-Mordoch et al. 2010, Stockbridge, Robertson et al. 2013), we adopted the strategy of expressing a fusion protein of full-length cyt b5 and b5R as a concatenated chimera with a linker connecting the C-terminus of cyt b5 with the N-terminus of b5R (Fig. 2a). The linker contained a tobacco etch virus (TEV) protease recognition site and can be cleaved after purification. The fusion protein was expressed and purified, and the amount was sufficient for further biochemical studies (Fig. 2b left). The fusion protein contained both FAD and heme, as indicated by the UV/Vis absorption spectrum (Fig. 2b right), and eluted as a single peak on a size-exclusion column (SEC). When the linker was cleaved by TEV protease, cyt b5 and b5R stayed together as a stable complex with the same elution volume as the fusion protein (Fig. 2b left). However, when the soluble domains of cyt b5 and b5R were expressed as a fusion protein, the two soluble domains did not stay as a complex after cleavage of the linker (Extended Data Fig. 3a and 3d). These results indicate that interactions between the TM segments are required to maintain the binary complex.
We next measured the rate of electron transfer in the stable binary complex of cyt b5 and b5R. We measured the reduction of cyt b5 in the context of a stable complex with b5R or as an individual protein mixed with b5R. We found that the time course of cyt b5 reduction can be fit with a biphasic exponential function (k1 and k2) when cyt b5 and b5R were in a binary complex (Extended Data Fig. 2a). The biphasic kinetics of electron transfer was also observed in soluble fusion of cyt b5 and b5R with similar rates with those of full-length fusion. In contrast, the time course can be fit with a single exponential function when the soluble forms of cyt b5 and b5R were added (Extended Data Fig. 2b). The observed electron transfer rate (k1) was ∼34-fold faster in the binary complex than that in the mixture of individual proteins in 150 mM NaCl. Thus, the formation of a stable cyt b5 and b5R complex enhances spatial proximity and the precise alignment of the two soluble domains to facilitate electron transport.
The rate of electron transfer between soluble forms of b5R and cyt b5 is known to be sensitive to ionic strength(Meyer, Shirabe et al. 1995). We next tested how the ionic strength affects the rate of electron transfer from b5R to cyt b5 in the stable complex. The time courses of cyt b5 reduction in the binary complex and the mixture of individual proteins were followed in buffer with different NaCl concentrations, and the rates were calculated from either double exponential fitting for the fusion proteins or single exponential fitting for the individual proteins. The full-length and soluble fusions displayed similar trend of ionic strength dependence where the electron transfer rates (k1 and k2) peaked at ∼50 mM NaCl and decreased as [NaCl] increased (Fig. 2c). The difference between no NaCl and 150 mM NaCl is only 1.04-fold. However, when the two proteins were not assembled as a complex, ionic strength has a more significant effect on the rate of electron transfer (Fig. 2d): an ∼8-fold decrease was observed from no NaCl to 150 mM NaCl. These results suggest that formation of a stable binary complex aligns the soluble domains in position for electron transfer so that electrostatic interactions have a much smaller role in guiding and facilitating the proper interactions of the soluble domains.
Stable binary complex between cyt b5 and SCD1
We next investigated whether the full-length cyt b5 can form a stable complex with SCD1. We found that simply co-expression of the two proteins do not produce high level of SCD1-cyt b5 binary complex. We then applied the fusion protein strategy and linked the C-terminus of SCD1 to the N-terminus of the full-length cyt b5 with a TEV recognition site in the linker (Fig. 3a). The SCD1-cyt b5 fusion protein had sufficient yield and eluted as a single peak on a size-exclusion column (Fig. 3b and Methods). After cleavage of the linker, SCD1 and cyt b5 stayed together in a stable complex as indicated by the single peak on the size exclusion column; the UV-Vis spectrum of the fusion protein did not show any change after TEV protease cleavage (Fig. 3b). When the soluble domain of cyt b5 was fused to SCD1, the two did not stay together as a complex after the linker was cleaved (Extended Data Fig. 3b and 3e), indicating that the TM segment of cyt b5 is required for the formation of the binary complex. The stable binary complex of SCD1-cyt b5 is capable of receiving electron from b5R, as indicated by the decrease of 340 nm absorbance from NADH (Fig. 3c).
We then measured electron transfer rate between the cyt b5 and SCD1 (Method and Fig. 3d) by following the optical change of cyt b5. The reduced form of cyt b5 was followed by its Soret peak at 423 nm. Under anaerobic condition, the first fast-rising phases (t < 20 s) correspond to the reduction of cyt b5 by b5R after equimolar of NADH was added. After the exhaustion of NADH, the phases of re-oxidation (t > 20 s) of cyt b5 by SCD1 appeared. The faster electron transfer rate between cyt b5 and SCD1 in the binary complex as opposed to the mixture of individual cyt b5 and SCD1 indicates that the binary complex is fully functional and that the formation of the binary complex likely facilitates the alignment and interactions of the two proteins inducive to electron transfer.
Stable ternary complex between b5R, cyt b5, and SCD1
Encouraged by the biochemical isolation of the two binary complexes, we examined the production of a ternary complex of SCD1, cyt b5 and b5R. We took a similar strategy of connecting all three full-length proteins with TEV protease cleavable linkers as shown in Fig. 4a. The fusion protein can be expressed and purified, and eluted as a single peak from a size-exclusion column (Fig. 4b). When the linkers were cleaved, all three proteins stayed together as a ternary complex (Fig. 4b).
The ternary complex was fully functional, as indicated by the production of oleoyl-CoA when supplied with NADH and stearoyl-CoA (Fig. 4c). The rate of oleoyl-CoA production was significantly faster than those in individual proteins or SCD1-cyt b5 binary complex (Fig. 4d), indicating that the ternary complex forms an electron transport chain and enhances the alignment of electron donors and acceptors.
Models of the binary and ternary complexes
To gain insight into the TM association patterns of SCD1, cyt b5 and b5R, we generated docking models of their TM regions and did mutational studies to validate key residues involved in the interactions. We first made models of the TM regions of cyt b5 and b5R as predicted by TMHMM server(Krogh, Larsson et al. 2001). The TM helix of cyt b5 was then docked to SCD1, and similarly to that of b5R. Best models with fewest clashes were selected and energy minimized (Methods). The selected models of SCD1-cyt b5 (TM) and cyt b5 (TM) -b5R (TM) were aligned with respect to the cyt b5 (TM). The aligned model yielded a ternary complex where SCD1 and b5R (TM) mounted on different sides of the cyt b5 (TM) (Extended Data Fig. 5a). This indicates that the cyt b5 (TM) can interact with both b5R (TM) and SCD1 simultaneously to facilitate the formation of the ternary complex, and the formation of binary or ternary complex is independent. We then performed all-atom MD simulation with the ternary complex model embedded in a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) bilayer (Extended Data Fig. 5a). Three independent 200 ns simulations showed small root-mean-square deviations (RMSD) of 6.6 ± 1.3 Å for cyt b5 (TM) and 8.3 ± 1.6 Å for b5R (TM) (Extended Data Fig. 5b). The largest root-mean-square fluctuations (RMSF) of cyt b5 (TM) and b5R (TM) were on residues near surfaces of the membrane, while those buried in the lipidic environment had much less fluctuation (Extended Data Fig. 5c). Collectively, our model was sufficiently stable during MD simulation to be recognized as a ternary complex.
The simulation also revealed the flexible nature of the N-terminal residues (40-60) of SCD1, whose average RMSF = 2.5 Å was significantly larger compared to the average RMSF = 0.3 Å of the rest of protein (Extended Data Fig. 5c). This region could dissociate from a groove on the surface of the soluble domain of SCD1 and expose a positively charged interface for potential interactions with the soluble domain of cyt b5. This observation is corroborated by the crystal structures of mouse and human SCD1(Bai, McCoy et al. 2015, Wang, Klein et al. 2015, Shen, Wu et al. 2020), whose N-terminal residues assume different conformations.
To examine if the model is accurate, we introduced single point mutations on the TM helices to the SCD1-cyt b5-TEV and cyt b5-b5R-TEV fusion constructs (Fig 5). Small hydrophobic residues were replaced with a bulky Trp, and polar and large hydrophobic residues were replaced with Ala. Tryptophan substitution on TM domains was shown to weaken or disrupt dimerization of homodimeric ClC transporters while maintaining function of the monomer (Robertson, Kolmakova-Partensky et al. 2010, Chadda, Krishnamani et al. 2016). The mutants were purified and digested with TEV protease and examined on a size-exclusion column. By monitoring the absorbance at 413 nm, where the binary complexes and monomeric cyt b5 have the same molar extinction coefficient, we were able to determine the ratio of dimer to monomer (Extended Data Fig. 4).
For the SCD1-cyt b5 complex, the M126A on the TM of cyt b5 achieved ∼50% of dissociation of cyt b5 from SCD1 (Fig. 5a). This is consistent with the docking model in which the M126 interacts with Y84 and F243 on the TMs of SCD1 (Fig. 5d). Introduction of the bulky Trp in the middle of TM helix of cyt b5 also destabilized the complex. The strongest effect of Trp mutant was observed on the L121, A122, and V123 of cyt b5 (Fig. 5a), which were predicted to be on the side facing SCD1 and involved in the hydrophobic interactions. When the corresponding residues on the SCD1 were mutated, partial dissociation of the complex was also observed (Fig. 5b).
We also mutated residues predicted to be involved in the TM interactions in the cyt b5- b5R complex. An Arg (R128) on the TM helix of cyt b5, which is not energetically favorable to be in a lipidic environment was predicted to interact with a Ser and His on the TM of b5R (Fig. 5e), partially neutralizing its charge. The SEC result on the R128A mutant showed that ∼40% of cyt b5 fell apart from b5R (Fig. 5c). Similar to the results in SCD1-cyt b5, Trp mutations to hydrophobic residues (L121, V123, and A124) on TM helix of cyt b5 reduced the amount of stable complex (Fig. 5c) by disruption of potential hydrophobic stacking patterns with b5R.
Soluble domains of SCD1, cyt b5, and b5R have very clear complementary electrostatic charge distributions (Extended Data Fig. 6a). Both SCD1 and b5R have positively charged surfaces while cyt b5 has a negatively charged surface, suggesting that cyt b5 may interact one at a time with either b5R or SCD1.
We docked the cyt b5 separately with SCD1 and b5R. Top-ranked docking models brought the heme in cyt b5 close to the FAD in b5R (Extended Data Fig. 6b) and the diiron center in SCD1 (Extended Data Fig. 6c), which implies direct electron tunneling(Winkler and Gray 2014, Gilbert Gatty, Kahnt et al. 2015) between these cofactors. Based on the predictions, cyt b5 engages b5R and SCD1 with a similar surface region. To further test this interaction model, we mutated two charged residues on the interface of cyt b5 (Extended Data Fig. 6b and 6c) and measured their binding affinities with either b5R or SCD1. The affinities to both b5R and SCD1 deceased modestly in these mutants (Extended Data Fig. 7), suggesting that these residues are involved in binding with both b5R and SCD1.
Discussion
In summary, we demonstrated that SCD1, cyt b5, and b5R form a stable ternary complex and that cyt b5 forms a stable binary complex with either b5R or SCD1. The stable complexes are mediated by TM domains of the proteins, and that the rates of electron transfer greatly increase in these complexes. The formation of stable complexes also suggests that the redox pair of b5R and cyt b5 may exist as a complex to interact with other downstream proteins.
These results led us to propose a working model of an electron transport chain shown in Fig. 6. Cyt b5 and b5R form a stable binary complex which further interacts with their target membrane-bound oxidoreductases to form a ternary complex. While the assembly of the binary or ternary complexes is mainly mediated by the interactions between their transmembrane domains, electrostatic interactions between the soluble domains help to steer and further align the donor-acceptor redox pairs. Cyt b5, while its single TM is sandwiched between b5R and SCD1 to stabilize the ternary complex, its soluble domain is mobile and can interact alternatively between b5R and SCD1 to relay electrons. A recent study revealed that lipid solvation energies drive the association and dissociation of a dimeric ClC transporter (Chadda, Bernhardt et al. 2021), and we surmise that the formation of the ternary SCD1-cyt b5-b5R complex maybe influenced by the lipid content of the ER membrane.
The association of the soluble domains of b5R and cyt b5, which harbor the cofactors for electron transport, is weak (KD ∼ µM, Extended Data Fig. 7) and likely transient. Anchoring them to the membrane would constrain their diffusion to the 2-dimensional lipid bilayer removing the third dimension of free diffusion in cytosol and thus enhances their ability to form the binary complex for electron transfer. Our results indicate that the interactions between the b5R, cyt b5, and SCD1 transmembrane domains greatly enhance the stability of the binary and ternary complexes. Such complexes bring the soluble domains of SCD1, cyt b5, and b5R in proximity and may rigidify each component for the optimal alignment required for efficient electron transfer.
Soluble domains b5R and cyt b5 are connected to their TM helix with a long linker predicted to be unstructured. This may afford certain degrees of conformational freedom allowing the soluble domains to contact with one another for efficient electron transfer. Such flexibility may also allow cyt b5 to shuttle electrons between b5R and SCD1 in the ternary complex.
Recent studies(Ahuja, Jahr et al. 2013, Yamamoto, Dürr et al. 2013, Zhang, Huang et al. 2015, Jeřábek, Florián et al. 2016, Zhang, Huang et al. 2016, Yamamoto, Caporini et al. 2017, Yamamoto, Caporini et al. 2017) using nuclear magnetic resonance and molecular dynamic (MD) simulation have shown interactions between TM helix of cyt b5 and that of cyt P450, an oxidoreductase with a single TM helix. The authors found that when both full-length cyt b5 and cyt P450 were incorporated into a single lipid nanodisc(Zhang, Huang et al. 2016), their TMs could interact and that the interactions facilitate electron transport between the soluble domains of cyt b5 and cyt P450(Zhang, Huang et al. 2015). Interestingly, a conserved motif on cyt b5 (L121 – L125) that was thought to interact with cyt P450 in nanodiscs(Yamamoto, Caporini et al. 2017) is also identified in the current study to mediate interactions with SCD1 and b5R.
Understanding the interactions of SCD1, cyt b5, and b5R and ultimately obtaining structures of the ternary complex in different redox states will help our understanding of SCD1 and other oxidoreductases that rely on cyt b5 and b5R. The knowledge also provides insights into designing small molecules that target the interactions between the transmembrane domains which have not been the target regions of the proteins.
Methods
DNA constructs
The cDNA of mouse SCD1, full-length cyt b5, and full-length b5R were codon optimized and synthesized. Fusions of SCD1-cyt b5, cyt b5-b5R, SCD1-cyt b5-b5R were generated by PCR. The linkers between each domain were either a TEV protease site (ENLYFQ/G) for the cleavable fusion or a flexible linker (GGSGGGSG) for the non-cleavable fusion. The SCD1 and SCD1-cyt b5 fusions were cloned into a pEG BacMam vector with a TEV protease site prepended to a C-terminal GFP tag. Because the C-terminus of b5R ends on the interface of the FAD-binding domain and NADH-binding domain, no extra residue should be introduced after the C-terminus of b5R domain to preserve its functional integrity. Therefore, SCD1-cyt b5-b5R was cloned into a pEG BacMam vector with an N-terminal GFP tag appended with a TEV protease site. For immunofluorescence imaging and coimmunoprecipitation assays, cyt b5 and cyt b5-b5R with a N-terminal Myc tag, and b5R with a N-terminal HA tag were cloned into a pEG BacMam vector. For large-scale protein expression and purification, cyt b5, b5R, and cyt b5-b5R were cloned into a pFastBac Dual vector with an octa-histidine tag and TEV protease site. For Octet binding assays, the soluble domains of cyt b5 (4 - 89) and b5R (24 - 301) were cloned into a pET vector.
Immunofluorescence imaging
The HEK 293S cells in FreeStyle 293 media (Invitrogen/Thermo Fisher) supplemented with 2% fetal bovine serum (FBS; Sigma) were plated one day before transfection onto glass coverslips coated with poly-lysine in a 24-well plate. The cDNAs in pEG BacMam vectors were transfected into cells with Lipofectamine 2000 (Invitrogen/Thermo Fisher) per manufacturer’s instructions. About 24 h after transfection, cells were fixed with 2% paraformaldehyde for 10 min and then washed three times with phosphate buffered saline (PBS). Cells were blocked and permeabilized with PBSAT (PBS + 1% bovine serum albumin, BSA + 0.1% Triton X-100) for 10 min. For cells expressing only GFP-tagged proteins, coverslips were washed three times with PBS and mounted onto glass slides with ProLong Diamond (Invitrogen/Thermo Fisher). For cells expressing Myc and/or HA tagged proteins, primary antibodies against Myc and/or HA tag diluted in PBSAT were added and incubated for 45 min at room temperature (RT). Coverslips were washed three time with PBS before the incubation with Alexa Fluor 555 (for Myc tag) and/or Alexa Fluor 647 (for HA tag) conjugated secondary antibodies (Invitrogen/Thermo Fisher) diluted in PBSAT for 30 min at RT. Finally, coverslips were washed and mounted as mentioned before.
Confocal images were acquired with a Zeiss LSM-710 confocal microscope using a 63× oil immersion objective (Zeiss, Plan-Apochromat 63×/1.4 Oil DIC M27) with Immersol 518F immersion oil (Zeiss). Alexa Fluor 647, Alexa Fluor 555, and GFP were detected sequentially with 633 nm HeNe laser, 561 nm diode-pumped solid-state laser, and 488 nm Argon laser. Crosstalk between the channels was avoided by adjusting emission regions. Single optical sections at a resolution of 1024×1024 pixels were acquired at two different zoom levels (1.5× and 4×).
Coimmunoprecipitation
The HEK 293S cells were plated one day before transfection in a 6-well plate. The pEG BacMam vectors containing target cDNAs were transfected with Lipofectamine 2000 (Invitrogen/Thermo Fisher) and incubated for 2 days. Cells on the plate were washed in PBS before scraping. Cell membranes were solubilized in lysis buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol) plus 0.2% Triton X-100 and Protease Inhibitor Cocktail (Roche) for 1 h at 4°C. Cell debris were pelleted by centrifugation. The supernatants of cell lysate were incubated with either pre-equilibrated GFP nanobody-conjugated NHS-Activated Sepharose 4 Fast Flow Agarose (GE Healthcare) or Pierce Protein A Agarose (Invitrogen/Thermo Fisher) with rabbit anti-Myc antibodies for 30 min at 4°C. The resins were extensively washed in lysis buffer plus 0.1% Triton X-100 within 5 min at 4°C. 4× Laemmli Sample Buffer (Bio-Rad) was added and samples were run in SDS-PAGE without extra elution steps. Bands of target proteins were visualized by western blotting with mouse anti-GFP (Invitrogen/Thermo Fisher), anti-Myc, and anti-HA antibodies as primary antibodies and IRDye-800CW anti-mouse IgG (Licor) as secondary antibody. Images were taken on an Odyssey infrared scanner (Licor).
Large-scale expression and purification of proteins
Expression of SCD1-containing proteins (SCD1, SCD1-cyt b5, and SCD1-cyt b5-b5R) were conducted in HEK 293S cells using the BacMam system(Goehring, Lee et al. 2014). Baculovirus were generated from pEGBacMam vectors with target cDNAs and amplified in Sf9 (Spodoptera frugiperda) cells. HEK 293 cells were maintained in FreeStyle 293 media (Invitrogen/Thermo Fisher) supplemented with 2% FBS (Sigma) in a 37°C incubator with 8% CO2 atmosphere at 100 rpm. Baculovirus after three passages (P3) were added to HEK 293S cells at a density of 3 × 106 mL-1 at a 7.5% v/v ratio and incubated overnight before adding 10 mM sodium butyrate and lowering temperature to 30°C. Media were supplemented with transferrin and ferric chloride as described previously(Shen, Wu et al. 2020). 0.5 mM δ-aminolevulinic acid and 100 µM riboflavin were added in media to enhance the biosynthesis of heme and FAD group, respectively. Three days after infection, cells were harvested and resuspended in lysis buffer plus Protease Inhibitor Cocktail (Roche), 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 mM MgCl2 and DNase I. Solubilization of cell membranes was achieved by incubating with 30 mM n-Dodecyl-β-D-Maltopyranoside (DDM, Anatrace) for 2 h at 4°C under gentle agitation. Target proteins in supernatants were captured by GFP nanobody resins. After washing the resins with 20 CV washing buffer (lysis buffer plus 1 mM DDM), proteins were released by TEV protease digestion during which TEV protease site linkers in the cleavable fusions were also cleaved. Eluents were concentrated (Amicon 50-kDa cutoff, Millipore) and loaded onto a size-exclusion column (Superdex 200 10/300 GL, GE Health Sciences) equilibrated with FPLC buffer (20 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DDM).
Expression of full-length cyt b5, full-length b5R, and cyt b5-b5R was conducted in High Five (Trichoplusia ni) cells using the Bac-to-Bac system. Baculovirus were generated from pFastBac Dual vectors with target cDNAs. 1.5% v/v of P3 virus was added to cells at a density of 3 × 106 mL-1. The δ-aminolevulinic acid (Santa Cruz) and/or riboflavin (Sigma) were supplemented in media as mentioned before. Cells were harvested three days after infection. Purification procedure was similar to that for HEK 293 cells except for the usage of cobalt-based affinity resin (Talon, Clontech) to capture His-tagged proteins.
UV-Vis spectroscopy and enzymatic assays
UV-Vis spectra were recorded using a Hewlett-Packard 8453 diode-array spectrophotometer (Palo Alto, CA). The time courses of the NADH consumption at 340 nm and the spectral change of cyt b5 heme at 423 nm were obtained with an Applied Photophysics (Leatherhead, UK) model SX-18MV stopped-flow instrument. The observed rates, kobs, were obtained by fitting the time courses to either 1- or 2-exponential functions.
Continuous turnover reactions of SCD1, the binary complex of SCD1-cyt b5, and the ternary complex of SCD1-cyt b5-b5R were performed similarly as previously described. Briefly, 3 µM of SCD1 plus equimolar of cyt b5 and b5R, SCD1-cyt b5 plus equimolar of b5R, and SCD1-cyt b5-b5R in FPLC buffer were incubated with substrate stearoyl-CoA (Sigma). NADH was added to start reaction. Aliquots of reaction mixtures were retrieved and quenched at different time points, and analyzed in high-performance liquid chromatography (HPLC). The initial rates were calculated by linear fitting of time courses within 1 min after reaction started.
Modeling of binary and ternary complex
The TM regions of cyt b5 and b5R were prediction by TMHMM server(Krogh, Larsson et al. 2001) and residue 108 - 134 of cyt b5 and residue 1 - 28 of b5R were used to model a TM helix a in I-TASSER server(Yang and Zhang 2015). The helical models with highest C-score were chosen for docking in Memdock(Hurwitz and Wolfson 2021). The diiron-containing mouse SCD1 structure (PDB ID: 6WF2) was used for docking. The resulting top 5 models with smallest Memscore were manually assessed based on topology and orientation in membrane predicted by PPM server(Lomize, Pogozheva et al. 2012). The selected models of SCD1-cyt b5 (TM) and cyt b5 (TM)-b5R (TM) were placed in a membrane bilayer of POPC and energy-minimized and equilibrated in the CHARMM36 force field(Huang, Rauscher et al. 2017). The two binary models were aligned to cyt b5 (TM) to generate a ternary model.
We used SWISS-MODEL server(Waterhouse, Bertoni et al. 2018) to construct homology models of mouse soluble cyt b5 (1 - 89) and b5R (24 - 301). The surface electrostatic potential of SCD1, cyt b5, and b5R were calculated in APBS(Jurrus, Engel et al. 2018). Dockings of soluble cyt b5 and b5R, and SCD1 and soluble cyt b5 were performed in HADDOCK server(van Zundert, Rodrigues et al. 2016) using a flexible docking protocol. Charged residues on the presumed binding interfaces were defined as active residues to restrain their proximity during docking process. The FAD in b5R and heme in cyt b5 were also included as part of the interfaces. All structure figures were prepared in Pymol (Schrödinger LLC.) or ChimeraX(Pettersen, Goddard et al. 2021).
Size-exclusion chromatography of mutants
Mutations on the TM domains of SCD1, cyt b5, and b5R were introduced to the constructs of linker-cleavable fusions of SCD1-cyt b and cyt b5-b5R by QuikChange site-directed mutagenesis. All mutations were confirmed by sequencing. Expression and purification were done similarly to the wild-type (WT) fusions. Purified proteins were loaded onto a SEC column (SRE-10C SEC-300, Sepax) in an HPLC system with a diode-array detector (SPD-M20A, Shimadzu). Samples were run in FPLC buffer at a flow rate of 0.75 mL/min and monitored at 423 nm. Elution profiles were normalized to the peak corresponding to either SCD1-cyt b5 or cyt b5-b5R complex for comparison.
Molecular dynamic simulation
The simulation system was prepared in CHARMM-GUI Membrane Builder(Jo, Kim et al. 2007, Jo, Kim et al. 2008, Jo, Lim et al. 2009, Wu, Cheng et al. 2014, Lee, Patel et al. 2019) with the ternary model of SCD1-cyt b5(TM)-b5R(TM). The structure was protonated at neutral pH and was embedded in a POPC bilayer. The system was solvated with TIP3P water and 150 mM NaCl (including neutralization ions). The final system had a size of 100 Å × 100 Å × 107 Å.
The simulations were performed in Gromacs v.2021.2(Abraham, Murtola et al. 2015) with the CHARMM36 force field(Lee, Cheng et al. 2016) in the isothermal-isobaric (NPT) ensemble. The system temperature was maintained at 303.15 K using the Langevin temperature coupling method with a friction coefficient of 1 ps-1. The semi-isotropic Nosé-Hoover Langevin-piston method was used to maintain the pressure at 1 atm. The 10–12 Å force-based switching was used for the Lennard-Jones interaction. The particle mesh Ewald method was used for the long-range electrostatic interactions. Three 200 ns unrestrained production simulations were performed using a timestep of 2 fs. Trajectory data were analyzed with the utilities in Gromacs(Abraham, Murtola et al. 2015).
Octet Biolayer Interferometry
Soluble cyt b5 and b5R were expressed in BL21(DE3) cells with pET vectors containing target cDNA. Mutations in cyt b5 were introduced by QuikChange site-directed mutagenesis and were confirmed by sequencing. Protocols for expression of these proteins were adapted from previous reports(Mulrooney and Waskell 2000, Bando, Takano et al. 2004). Purification procedure was similar to that for full-length cyt b5 and b5R as mentioned before.
Biolayer Interferometry (BLI) assays were performed at 30 °C under constant shaking at 1000 rpm using an Octet system (FortéBio). The immobilization of ligand proteins on amine reactive second-generation (AR2G) biosensors (Sartorius) was done following manufacturer’s instruction. Briefly, biosensor tips were activated in 20 mM 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) and 10 mM N-hydroxysulfosuccinimide (Sulfo-NHS) for 300 s. Then the tips were loaded with soluble cyt b5 at a concentration of 5 µg/mL in FPLC buffer for 600 s. The tips were quenched in FPLC buffer plus 1 M ethanolamine for 300 s. The tips with immobilized ligands were equilibrated in FPLC buffer plus 0.1% BSA to reduce non-specific binding. Then, they were transferred to wells of a concentration gradient (5, 2.5, and 1.25 µM) of analysts (soluble b5R or SCD1) in buffer B for association and returned to the equilibration wells for dissociation. Binding curves were aligned and corrected with the channel of no analyst protein. The association and disassociation phases were fitted with 1-exponential function to extract kon and koff of the binding, which were used to calculate dissociation constant KD.
Author Contributions
M.Z., A.T., J.S., and G.W. conceived the project. J.S. and G.W. conducted experiments. J.S. and M.Z. wrote the initial draft and all authors participated in revising the manuscript.
Competing interests
The authors declare no competing financial interests.
Corresponding authors
Ming Zhou (mzhou{at}bcm.edu), Ah-Lim Tsai (Ah-Lim.Tsai{at}uth.tmc.edu)
Extended Data
Acknowledgments
We thank Dr. Theodore G Wensel for the access of Zeiss confocal microscope and western blot-related materials; Dr. Melina A Agosto for the help with confocal microscopy and co-IP experiments; and Joshua I Rosario Sepulveda for the help with western blot. This work was supported by grants from NIH (DK122784 to M.Z. and A.T., HL086392 and GM098878 to M.Z.), and Cancer Prevention and Research Institute of Texas (R1223 to M.Z.).