Fine‐tuning the yeast GAL10 promoter and growth conditions for efficient recombinant membrane protein production and purification

Abstract

One of the most common issues in producing membrane proteins in heterologous expression systems is the low yield of purified protein. The solubilization efficiency of the recombinant membrane protein from biological membranes is often the limiting step. Here, we study the effects of titration of the GAL10‐CYC promoter of Saccharomyces cerevisiae, induction time, and culture media, on the rat mitochondrial uncoupling protein (UCP1) production and solubilization levels. We found that a maximum threshold of solubilized UCP1 (70%) is reached at 0.003% galactose concentration, independently of time, temperature, and detergent‐to‐protein ratio during solubilization. Supplementation with 0.1% amino acids of the S‐lactate medium at induction resumes cell growth and recombinant protein production. The purified UCP1 protein (0.2 mg/L) is homogenous in DDM detergent and active after reconstitution in proteoliposomes. To extend the impact of our findings, we applied the same promoter titration to produce the GFP‐AT7B human transporter and found an optimal galactose concentration of 0.0015%. The protein data bank analysis revealed that these galactose concentrations are 300 times lower than usual. We propose a novel strategy for the recombinant production of membrane proteins in the yeast S. cerevisiae, which unlocks the use of this inexpensive eukaryotic host for membrane protein production.

1. INTRODUCTION

The yeast expression system is widely used to produce endogenous or heterogenous eukaryotic proteins (Phan et al. 2001; Ton and Rao ²⁰⁰⁴). Saccharomyces cerevisiae host is well‐characterized, cost‐effective, easy to manipulate, and conserves basic cellular machinery and signal transducing pathways with higher eukaryotes. It is a preferred host system for the expression of eukaryotic proteins containing post‐translational modifications (glycosylation, phosphorylation) that cannot be performed in bacterial expression systems. It also provides the possibility of functional in vivo or in vitro experiments on native organelles after their isolation. Yeast expression systems have been used for the structural determination of 180 membrane proteins (see Dilworth et al. ²⁰¹⁸, and this study). The mammalian mitochondrial uncoupling protein 1 (UCP1) is an integral membrane protein found in the inner mitochondrial membranes of brown adipose tissue. It is responsible for non‐shivering thermogenesis by its ability to burn off cellular nutrients as heat. UCP1 uncouples respiration from ATP synthesis by dissipating the electrochemical proton gradient (Nicholls 2021). It is a potential therapeutic target for treating obesity and diabetes by increasing energy expenditure. Functional and structural analyses of UCP1 have been achieved using S. cerevisiae as an expression host (Gagelin et al. 2023; Jones et al. ²⁰²³). However, many questions about the mechanism of activation by free fatty acids are still unresolved. There is a need for a robust expression host able to support strong uncoupling of mitochondrial respiration induced by UCP1 variants that failed to be produced in HEK cells (Kang and Chen 2023). Heterologous protein expression in yeast is possible owing to the availability of a large set of promoters, selectable markers, and engineered strains (Dilworth et al. 2018). However, solubilization of the recombinant protein from yeast membranes remains critical and often a limiting step (Stuart et al. 2001; Winkler et al. ²⁰⁰¹) for further purification. GFP‐based optimization scheme, selection of appropriate hosts, and growth temperature have been successfully applied to S. cerevisiae GAL1/10 promoter system for tuning membrane protein expression and purification (Dieudonné et al. 2023; Drew et al. ²⁰⁰⁸). Here we propose a novel method of GAL1/10 promoter titration and culture medium optimization that resolves metabolic burden and toxicity associated with membrane protein overproduction, suppresses the formation of recombinant membrane protein aggregates, and significantly increases the efficiency of solubilization of the recombinant membrane protein from yeast membrane. We applied this approach to the purification and functional characterization of UCP1 and the effective solubilization of the human GFP‐ATP7B transporter.

2. RESULTS

2.1. Titration of UCP1 production in the yeast GAL10 promoter expression system

In this study, UCP1 is expressed using the pYeDP60 plasmid, where heterologous protein production is driven by the strong GAL10‐CYC1 inducible fusion promoter, regulated by the Gal4 transcriptional activator. The W303.1b/Gal4 strain, which carries both the wild‐type GAL4 gene and an integrated GAL10p‐GAL4 cassette, ensures controlled Gal4 overproduction in response to galactose. In glucose‐containing medium, GAL4 is repressed by constitutive inhibitor binding. Upon glucose depletion and galactose supplementation, GAL4 is released, activating the GAL10 promoter. This sequential induction leads to high GAL4 transactivator synthesis, followed by expression of the target protein from the plasmid (Figure S1, Supporting Information).

Two protocols were designed mainly for practical purposes, as shown in Figure 1. The respiration protocol, with a media change, enables yeast spheroplast preparation on the same day for activity assays the following morning. This culture protocol for Respiration assay (R‐culture) has been successfully applied to measure UCP1‐dependent uncoupling of respiration on yeast spheroplasts (Gagelin et al. 2023; Piel et al. ²⁰²¹). The S‐lactate medium is changed before induction at OD_600nm = 4, and galactose is added together with fresh medium and incubated for 4 h. The purification protocol allows cost‐effective upscaling without additional media changes. In the culture protocol for UCP1 Purification (P‐culture), UCP1 expression is induced by galactose addition at OD_600nm = 1 for 16 h without changing the media. To compare both protocols, yeast proteins are extracted and precipitated with TCA, and UCP1 levels are quantified by immunodetection. Figure 2a shows UCP1 and VDAC protein accumulation levels as a function of galactose concentration. Both protocols yield maximal amounts of UCP1 at 0.2% galactose and UCP1 protein levels start decreasing at 0.003% galactose. In the P‐protocol, galactose concentration can be decreased to 0.025% without affecting the maximal levels of UCP1 production per cell. In the absence of the galactose inducer, UCP1 protein is not detected. Next, we investigated the time course of UCP1 production (Figure 2b). At 1% galactose concentration, UCP1 protein appears 30 min after the addition of galactose, whereas, at 0.2%, 0.05%, and 0.01% galactose concentration, UCP1 is first detected at 1, 2, and 4 h, respectively. In all four conditions, UCP1 protein reached its maximum level of accumulation after 4 h without further increase after 16 h induction, suggesting a limiting factor in the medium.

FIGURE 1.

Culture protocols for UCP1 expression in yeast S. cerevisiae. In both protocols, yeast cells are cultured in S‐lactate minimal medium to maintain respiratory metabolism while maintaining selective pressure and preventing plasmid loss to ensure tighter regulation of expression. Respiration protocol culture (R‐culture): UCP1 is induced at OD_600nm = 4 in fresh medium for 4 h. Large‐scale culture for purification (P‐culture): induction is done at OD_600nm = 1 for 16 h.

FIGURE 2.

Kinetics of UCP1 production in yeast depend on galactose concentrations. Cultures volumes (mL) corresponding to a constant OD_600nm = 5 (V = 5/OD_600nm) are harvested and proteins are precipitated by TCA. Equivalent amounts of total protein extracts (1 μg) are loaded in each well for valid comparison. UCP1 (33 kDa band) and mitochondrial porin VDAC (29 kDa) are immunodetected. (a) Effect of the galactose concentration (indicated in %) on UCP1 levels in R‐ and P‐cultures. (b) Time‐course of UCP1 accumulation in yeast. UCP1 inclusion bodies produced in E. coli (50 and 500 ng) are loaded as positive controls in panels (a) and (b), respectively.

2.2. Effective solubilization of UCP1 is inversely correlated with galactose concentration

First, we tested three mild detergents and the n‐Dodecyl phosphocholine (DPC) detergent, which was shown to inactivate UCP1 and other mitochondrial carriers (Piel et al. 2021), on crude mitochondrial fractions obtained from the P‐protocol with 1% galactose (Figure 3a). Mitochondrial membrane solubilization with n‐dodecyl‐β‐D‐maltopyranoside (DDM), lauryl maltose neopentyl glycol (LMNG), or 7‐cyclohexyl‐1‐heptyl‐β‐D‐maltoside (Cymal‐7) detergents at a detergent/protein ratio (w:w) of 10 results in poor extraction of UCP1 (around 3%, the upper 33 kDa band in the pellet fraction) but in full solubilization of VDAC, which appears as a lower 29 kDa band in the soluble fraction. In contrast, DPC partially solubilizes UCP1 (Figure 3a). This pattern of differential solubilization between DPC and other mild detergents suggests that the recombinant UCP1 is not correctly folded into the host membranes. We hypothesized that decreasing the UCP1 levels in yeast mitochondria may improve its extraction by mild detergents. At 0.05% galactose concentration, UCP1 solubilization is enhanced with DDM, LMNG, Cymal‐7, and 1,2‐dihexanoyl‐sn‐glycero‐3‐phosphocholine (DHPC) but not with 3‐[(3‐cholamidopropyl)‐dimethylammonio]‐1‐propane sulfonate (CHAPS) and glyco‐diosgenin (GDN) (Figure 3b). Adding cholesteryl hemisuccinate (CHS) to stabilize membrane proteins during solubilization (Dieudonné et al. 2023) does not improve solubilization levels except for DHPC. In contrast, at 0.003% galactose concentration the solubilization of UCP1 by DDM is strongly improved (Figure 3c, lane 60 min). A time course of UCP1 solubilization at 4°C with a 10:1 DDM to protein ratio shows that the maximum level (35%) of UCP1 solubilization is reached in the first 15 min with no further improvement in the next 2 h incubation (Figure 3c). Increasing the temperature to 21°C does not increase the solubilization efficiency (data not shown). Decreasing the DDM to protein ratio to 1:1 at 10 mg/mL protein concentration shows similar results (Figure 3d) and is the most suitable for the next purification step.

FIGURE 3.

Solubilization of UCP1 from crude mitochondria is inversely correlated to galactose concentrations. UCP1 is induced at OD_600nm = 1 for 16 h using the P‐protocol with (a) 1%, (b) 0.05%, (c, d) 0.003% galactose. In (b), the presence of CHS in DDM micelles (1:4 ratio) is tested (−/+). Solubilization is held at 4°C for 1 h (a, b) or monitored over time (c, d). Detergent concentration is 1% and protein concentrations are 1 mg/mL except for (d) at 10 mg/mL. CM = crude mitochondria. Soluble (S) and pellet (P) fractions are separated by ultracentrifugation at 100,000g and equivalent volumes are analyzed by Western blot. UCP1 levels are estimated by immunodetection as described in section 4. In panel (a) both UCP1 (33 kDa band) and mitochondrial porin VDAC (29 kDa band) are immunodetected. UCP1 inclusion bodies produced in E. coli (50 ng) are loaded as positive controls in panels (a), (c), and (d).

2.3. Amino acids supplementation increases UCP1 production and yeast exponential phase length

Both R‐ and P‐culture protocols yield similar amounts of UCP1 after 4 and 16 h induction, respectively (Figure 2a). However, in both cases, the final OD_600nm is below 5, suggesting a limiting nutrient in the medium. In the R‐culture protocol, not changing the medium indeed suppresses UCP1 production (Figure 4a). It has been shown that protein production in yeast under the ADH2 promoter triggers a metabolic burden that adding a selected number of amino acids could partially resolve (Görgens et al. 2005). We hypothesized that supplementing the culture media with all amino acids during induction may help restore cell growth and protein production. To test this hypothesis in the R‐culture protocol, amino acids (mix of 0.1% casamino‐acids and 20 mg/L tryptophan) are added at induction time (OD_600nm = 4), and cells are collected 4 h later. Figure 4a shows that adding amino acids fully restores UCP1 production regardless of galactose concentrations and slightly exceeds the effect of fresh medium. Next, we investigated the optimal time and amounts of amino acids addition (Figure 4b). Doubling amino acids concentration in S‐lactate in the pre‐induction culture at OD_600nm = 0.03 had no or limited effect on UCP1 production. Adding 5 times more amino acids at the time of induction does not further increase UCP1 protein levels. Next, we tested the addition of amino acids in the P‐culture protocol. At galactose concentrations of 1 and 0.01%, the addition of amino acids does not affect UCP1 expression. However, at the lowest concentration of galactose (0.003%; Figures 4c and S2a), amino acid addition increases 2.6 times UCP1 protein levels per cell (p ≤ 0.001) as well as stimulates cell growth with a final OD_600nm = 7.5, twice higher than without amino acids (p ≤ 0.0001) (Figure 4d). The time course of UCP1 expression is not modified by the addition of amino acids; at 0.003% galactose, UCP1 is detected 4 h after induction (Figure S2b). Next, we investigated the effect of galactose concentration and amino acids supplementation on cell growth using a 48‐well plate. Figure 4e shows that cell growth of UCP1‐expressing yeasts is inversely correlated with galactose concentration. At 1% galactose, yeast growth is strongly delayed and decreased, while at 0.003%, cell growth is restored and becomes identical to cells not expressing any heterologous protein (empty vector 0%). Significant growth inhibition is seen at a high galactose concentration (1%) on yeast cells harboring the empty vector. Amino acids supplementation increases the exponential phase length, and consequently the final cell density.

FIGURE 4.

Amino acid supplementation at induction stimulates yeast growth and UCP1 production. (a) Yeast cells are induced at 0D_600nm = 4 with a range of galactose concentrations from 0.003 to 1% in (a) fresh S‐lactate media (R‐culture), (b) the same medium, and (c) the same medium with amino acid supplementation (0.1%) at induction. (b) Optimization of amino acid supplementation on UCP1 levels using protocol (Figure 4a) (c). Amino acids (0.2%) are added at inoculation (OD_600nm = 0.03) (a) or at induction (OD₆₀₀ = 1) at 0.1% (b) or 0.5% (c). (c) Application of amino acid supplementation to P‐culture. UCP1 is induced using 0.003, 0.01, and 1% galactose ± 0.1% amino acids at induction (OD_600nm = 1). (d) Characterization of the P‐optimized protocol. UCP1 is induced with 0.003% galactose without (blue) or with (green) 0.1% amino acids. Left panel shows UCP1/VDAC intensities ratios on TCA extracts (see Figure S2). Right panel shows the final OD_600nm of cell cultures. Experiments were conducted with n = 5; bars indicate means ± SEM. Statistical analyses with multiple unpaired t‐tests were done on PRISM (see Tables S1 and S2). (e) Effects of amino acids supplementation and galactose on yeast cell growth. Representative growth curves were monitored with a CLARIOstar Plus Microplate Reader. Cells from preculture are diluted to OD_600nm = 0.1 in S‐lactate media supplemented with 0.1% glucose and 1% (red), 0.003% galactose (brown and dark blue). Control cells harboring an empty pYeDP60 vector are shown in gray (0% galactose) and pink (1% galactose). The optimized P‐culture (0.003% galactose, 0.1% amino acids) is in dark blue. Data are representative of two to three independent experiments.

In the R‐optimized protocol with 0.003% galactose, the relative amount of UCP1 is 28% of VDAC expression level (UCP1/VDAC = 0.277 ± 0.017; Figure 5a), and the solubilization efficiency is below 10% with mild detergents and reaches 62% with DPC (Figure 5b). In contrast, at the same galactose concentration, the P‐optimized protocol yields a lower UCP1 expression level (UCP1/VDAC = 0.179 ± 0.020, p ≤ 0.05) but higher biomass (Figures 4d and 5a), and enhanced solubilization ranging from 48% to 79% using DDM and Cymal‐7, without further improvement by DPC (Figure 5b). As expected, at 1% galactose, UCP1 expression increases significantly (UCP1/VDAC = 1.182 ± 0.246, p ≤ 0.05) (Figure 5a); however, solubilization remains below 6% with mild detergents and reaches only 27% with DPC (Figure 5b).

FIGURE 5.

Quantification of UCP1 levels and solubilization efficiencies in large‐scale cultures. Culture conditions are: P‐optimized culture (P‐opti in purple) − induction with 0.003% galactose at OD_600nm = 1 with 0.1% amino acids for 16 h; Control P‐culture (P‐control in blue) − induction with 1% galactose at OD_600nm = 1 for 16 h; R‐AA culture (R‐AA in green) − induction with 0.003% galactose at OD_600nm = 4 and with 0.1% amino acids for 4 h in the same medium. (a) Expression levels of UCP1 in crude mitochondria: the relative amount UCP1/VDAC (intensity band ratio) is indicated (see Figure S3). (b) Solubilization efficiencies: detergents were used at 1% with protein concentration of 10 mg/mL (P‐opti and R‐AA) or 1 mg/mL (P‐control) (see Figure S4). Experiments were conducted with n = 3 and bars indicate means ± SEM. Statistical analyses with unpaired t‐tests were done on PRISM (Tables S3 and S4).

2.4. Large‐scale purification of functional UCP1

Starting from 30 mg of yeast mitochondria solubilized with 1% DDM, 200 μg of UCP1 is purified as described in section 4. SDS‐PAGE analysis shows a rather pure UCP1 sample (Figure 6a) after concentration at 1 mg/mL. Sample oligomeric state and mass of protein are assessed by SEC‐MALLS (Figure 6b). In the first peak (blue curve) of the elution profile, corresponding to the UCP1‐DDM‐Cardiolipin (CL) complex, the experimental weight‐averaged molar mass of UCP1 (43 kDa) is higher than the expected value for the monomeric form (32 kDa). This discrepancy might be due to a monomer–dimer equilibrium of UCP1 that could not be resolved with the size exclusion column used. The second peak (the red curve) corresponds to the DDM‐CL‐free micelles. The experimental weight‐average molar mass of the DDM‐CL‐free micelles is around 93 kDa, a value superior to the expected weight‐average molar mass of the DDM‐free micelles, which is around 65 kDa (Barret et al. 2013). In this case, the mass difference is explained by the insertion of the CL in the DDM micelles.

FIGURE 6.

Purification, biophysical characterization, and functional analysis of UCP1 from P‐optimized protocol. (a) Two‐step purification of UCP1 in DDM on Ni‐NTA affinity chromatography. The 16% SDS‐PAGE gel shows, respectively, crude mitochondria (CM), DDM‐solubilized fraction (S), first Ni‐NTA elution (N1 elution), cleaved UCP1 by TEV protease overnight (OVN TEV), flow‐through of the second Ni‐NTA (N2 FT), concentrated UCP1, and agarose beads of the second Ni‐NTA (N2 bound). (b) SEC‐MALLS analysis of UCP1 in DDM‐Cardiolipin micelles (blue) and protein‐free micelles (red). Samples are eluted using a Superdex 200 Increase (10/300) column. Molar masses are calculated and depicted for UCP1‐DDM‐CL complex (■), UCP1 protein (+), and its bound detergent‐CL complex (x). (c) Representative UCP1 protonophoric activity. Kinetics curves of the safranine polarization were monitored at 520 nm. The arrows indicate the addition of nigericin and CCCP. Ligands were added before nigericin: 30 μM lauric acid (black), or 30 μM lauric acid with 1 mM GDP (pink). The green curve corresponds to basal activity. See Table S5 for statistical analysis of UCP1 activity and regulation. (d) UCP1 protonophoric activity in liposomes in response to 30 μM lauric acid (LA) or 1 mM GDP or both (blue). Empty liposomes serve as a negative control (purple). Slopes were normalized to liposome volume (estimated from CCCP values). Nigericin is added at the start of each measurement. Sample sizes: for UCP1 liposomes‐nigericin (n = 6), GDP (n = 5), LA (n = 8), LA and GDP (n = 8); for empty liposomes‐nigericin (n = 8), GDP (n = 2), LA (n = 6), LA and GDP (n = 9). Boxplots represent the first quartile, median, and third quartile. The whisker length is 1.5 times the interquartile range. Statistical analyses with unpaired t‐tests were done on PRISM (Table S5).

As for UCP1 activity assessment, proton transport was measured upon its reconstitution in liposomes (Figure 6c,d) as previously described (Zoonens et al. 2013). UCP1 is well activated by 30 μM lauric acid (LA) and inhibited by 1 mM GDP until the basal activity level. The basal activity of proton transport is observed when no fatty acids are added, and it is completely inhibited by the addition of GDP. Control empty liposomes show that at the used concentration of LA to activate UCP1, they have no destabilizing effects on lipids (Figure 6d).

2.5. Generalizing the yeast GAL10 optimized protocol for recombinant membrane protein purification

To extend the membrane protein coverage of our approach, we titrated the galactose promoter on yeast cells harboring the pYeDP60‐GFP‐ATP7B encoding a copper‐transporting ATPase 2 (Bitter et al. 2022) fused to a GFP‐tag (generously given by Thibaud Dieudonné and Tomàs Heger). As initial screening, cells were grown on the 48‐well plate with an adapted autoinduction protocol (see section 4). A range of decreasing concentrations of galactose was added at the beginning of the culture in the presence of 0.1% glucose. Cell growth monitoring shows that 1% galactose impairs yeast cell growth (Figure 7a) as observed with UCP1. In the case of ATP7B, it is necessary to decrease galactose concentration up to 0.0015% to suppress the toxicity associated with the production of the GFP‐ATP7B protein. GFP fluorescence monitoring shows a saturation curve of fluorescence intensity as a function of galactose concentrations in S‐lactate in the presence or absence of amino acids. Since maximal fluorescence values are normalized by OD_600nm, the curve in Figure 7b reflects expression levels of ATP7B per cell. Amino acids supplementation slightly increases protein expression at low concentrations of galactose. At 0.003% galactose, the same expression levels are observed with or without amino acids supplementation. Above 0.003%, GFP fluorescence reaches its maximal level, and the effect of amino acids addition is weaker. Of note, cells with no addition of galactose exhibit a GFP fluorescence twice above the autofluorescence of yeast cells harboring the control empty vector.

FIGURE 7.

Application of P‐optimized protocol to GFP‐ATP7B transporter. (a) Growth curves of cells expressing GFP‐ATP7B in S‐lactate media monitored using a CLARIOstar Plus Microplate Reader (as described in Figure 4e). GFP‐ATP7B was induced by different galactose concentrations: 1% (black and red), 0.01% (gray), 0.0015% (green), and 0% (purple). Amino acids are supplemented in all cases except for the one represented by the red curve. The experiment was repeated twice with similar results. (b) Green fluorescence intensities per cell of GFP‐ATP7B as a function of galactose concentration. Maximum fluorescence is recorded after 35 h of culture. The effect of amino acids supplementation is shown (red: without, blue: with). The x‐axis is represented by a log scale where 0.0001% galactose corresponds to 0%. The dotted line represents control cells harboring the empty pYeDP60 vector, with amino acid addition. The experiment was repeated twice with similar results. (c) Comparison of ATP7B solubilization efficiencies from light (top) and heavy membranes (bottom) using Cymal‐7 supplemented with CHS. GFP‐ATP7B is produced following P‐culture conditions with either 1% galactose or a range of galactose concentrations (0.0015–0%) with amino acids supplementation. Membranes (Mb) correspond to the starting material while S corresponds to the solubilized fraction. GFP fluorescence intensities on 10% SDS‐PAGE gels are assessed by a typhoon laser scanner (GE Healthcare). (d) Quantification of GFP‐ATP7B solubilization efficiency from light (light color) or heavy membranes (dark color). GFP‐ATP7B was induced according to the P‐optimized conditions (except for galactose concentration with 0.0015% instead of 0.003%, green) or the P‐control conditions (1% galactose, blue). Solubilization is performed with 1% Cymal‐7‐CHS or DPC‐CHS for 1 mg/mL of proteins. Experiments were conducted with n = 4, and bars indicate means ± SEM (see Figure S5 and Table S6).

Based on this analysis, galactose concentrations below 0.003% together with supplementation with amino acids were the culture conditions selected to avoid any negative side effects on cell growth and to test the solubilization efficiency of the recombinant transporter. Light and heavy membranes were prepared as described in section 4. Decreasing galactose concentration to 0.0015% does not affect GFP‐ATP7B expression levels compared to 1% galactose, but it increases solubilization efficiencies of the recombinant protein from light membranes (from 7.8% to 33.5%) using Cymal‐7‐CHS detergent (Figure 7c,d). Induction with less than 0.0015% galactose is not beneficial as total protein levels start to decrease (Figure 7c). Moreover, while heavy membranes are completely insoluble when inducing with 1% galactose, 25 to 30% of solubilization is achieved using 0.0015% galactose (Figure 7d). DPC‐CHS solubilization was effective no matter the galactose concentration (1% and 0.0015%), using both heavy and light membranes, which, as in the case of UCP1, suggests a misfolding of the recombinant protein when using high concentrations of galactose. Similarly to the production of UCP1, the P‐protocol optimized by the addition of amino acids restores cell growth resulting in a final OD_600nm of 9 (see Table S7).

2.6. PDB analysis of culture conditions used with S. cerevisiae as expression host

To assess the novelty of our optimized protocol, we analyze the PDB as previously described (Dilworth et al. 2018) focusing on expression conditions with S. cerevisiae expression system. An up‐to‐date analysis of the PDB (see Data S2) shows 76 membrane protein structure entries obtained with the yeast S. cerevisiae as the expression host. The inducible GAL promoter is highly used (73.7%), with three different variants: GAL1 (63.2%), GAL10/CYC1 (10.5%, this study), and GAL10 (4%) (Table 1). The strength of the latter promoter is weaker than that of GAL1 (Da Silva and Bailey ¹⁹⁹¹). The galactose concentrations added for induction of the expression of the recombinant protein range from 0.4% to 2% galactose, with 2% being the most common. These concentrations are 100–1000 times higher than the optimized amounts used in this study. Induction periods range from 20 to 72 h, and for 12 occurrences, yeast cells were resuspended in a fresh medium at the time of induction (Figure 8a).

TABLE 1.

Distribution of promoters by family and occurrences.

Promoter family	Promoter name	Occurrences	Frequency (%)
Galactose	GAL1	48	63.2
	GAL10/CYC1	8	10.5
	GAL10	3	4.0
Constitutive	PMA1	12	15.8
	PDR5	1	1.3
	GAP	1	1.3
Other promotors	Native promotor	1	1.3
	ADH2	2	2.6
Total		76	100

FIGURE 8.

Comprehensive description of culture conditions and S. cerevisiae strains from the PDB‐MP database. (a) Schematic representation of induction period as a function of galactose concentration and temperature of culture. The symbol size is proportional to the number of occurrences. Fresh medium used before induction is indicated by a circle, while unchanged medium is indicated by a square. This set of complete data consists of 40 entries from Data S2. (b) Induction temperature usage by strain. The size of the symbol is relative to the number of occurrences. The complete data set (strain, induction temperature) consists of 44 entries. Strains with the Gal4 enhancer are indicated in bold.

3. DISCUSSION

Promoter engineering is a common approach to produce metabolites and proteins in microbial systems. However, the design of expression systems is often motivated by maximizing the amount of recombinant protein at the expense of the quality. In E. coli, the T7 RNA polymerase‐based expression system is one of the most powerful systems to overproduce recombinant proteins. In the T7 BL21(DE3) host, the addition of the IPTG inducer most of the time kills the cells, and recombinant membrane proteins have difficulties folding and inserting into the E. coli membranes. Attenuated T7 hosts restore cell growth, delay the time course of accumulation of the recombinant membrane protein, and favor its folding and insertion (Angius et al. 2018; Schlegel et al. ²⁰¹²). Yeast expression systems are widely used for heterologous membrane protein production, including from human origin (Carlesso et al. 2022; Cimadamore‐Werthein et al. ²⁰²⁴; Fiorini and Mus‐Veteau ²⁰¹⁶; Kintzer and Stroud ²⁰¹⁶; Lenoir et al. ²⁰⁰²; Moncoq et al. ²⁰⁰⁸; Pyrihová et al. ²⁰²⁴; Tavoulari et al. ²⁰²⁴). In the yeast S. cerevisiae system, increasing the strength of GAL promoters is achieved by controlling the sequence, number, and substitution of upstream activating sequences of the GAL promoter (Deng et al. 2021). In our study, the GAL promoter‐based expression system is enhanced by three means: (i) the pYeDP60 plasmid contains a promoter‐enhancing sequence of the cytochrome C oxidase gene that is present immediately upstream of the GAL10 promoter, (ii) the modified W303.1b/Gal4 host strain contains a supplementary copy of the GAL4 transactivator protein (a, leu2, his3, trp1::TRP1‐GAL10‐GAL4, ura3, ade2‐1, canr, cir+), (iii) at the time of induction, the low amount of glucose (0.1%) initially present in the S‐lactate and which suppresses the GAL promoter activity, has been consumed. In the presence of glucose, GAL4 is indeed inactive. When glucose is catabolized, the expression system is set up at its maximal capacity with 2% galactose for induction, a culture temperature of 30°C, and 16–24 h of induction. This combination of induction and culture conditions is the most widely used in the structural biology of membrane protein (Figure 8a), and contributed to 3% of unique membrane protein structures in the PDB (Carlesso et al. 2022; Dilworth et al. ²⁰¹⁸). Among them, the yeast expression system has succeeded in the biochemical and structural analyses of the human mitochondrial carriers. Efficient heterologous expression in yeast has been observed for the human citrate carrier (Cimadamore‐Werthein et al. ²⁰²⁴), the dicarboxylate carrier and oxoglutarate carrier (Cimadamore‐Werthein et al. ²⁰²⁴; Pyrihová et al. ²⁰²⁴), aspartate/glutamate carrier (Cimadamore‐Werthein et al. ²⁰²⁴; Tavoulari et al. ²⁰²⁴), and the human uncoupling protein UCP1 (Jones et al. 2023), all induced with 0.4% galactose. In contrast, it has also been reported that UCP1 and UCP3 are prone to aggregation and uncontrolled uncoupling of respiration when expressed at high levels in the yeast S. cerevisiae and even at close to physiological concentrations in the muscle L6 cell line (Echtay et al. 2000; Harper et al. ²⁰⁰²; Stuart et al. ²⁰⁰¹). Comparing galactose concentrations between studies is difficult because yeast media and vector/host strain combinations are different (Peng et al. 2015). For instance, we use a variant of the W303.1b yeast strain harboring a supplementary copy of the Gal4 trans‐activator, which strongly enhances the activity of the promoter. We grow yeast in minimal lactate medium with low (0.1%) glucose concentration and not in a rich medium. Varying inducer concentration and time of induction is a common practice during the optimization of membrane proteins production in microbial systems. Nevertheless, here we show that downregulation of the promoter by very low concentration of inducer decreases the time course of expression of the recombinant membrane protein, its final accumulation levels, and ultimately, its solubilization efficiency. We used the natural ability of the GAL10 promoter to be titrated with very low (down to 0.009%) galactose concentrations (Ah Kang et al. 2005; Lee et al. ²⁰¹⁵). In the W303.1b/Gal4 strain and in lactate minimal medium, decreasing galactose to 0.025% does not affect UCP1 or GFP‐ATP7B protein levels. This is consistent with the results of Hovland et al. and Ah Kang et al., which have used galactose concentration down to 0.05% for the recombinant production of human soluble proteins (Ah Kang et al. 2005; Hovland et al. ¹⁹⁸⁹). This first level of optimization not only decreases the cost of large‐scale fermentation of yeast but also avoids the side effects of high galactose concentrations on the replicative lifespan of yeast cells and oxidative stress (Liu et al. 2015). We indeed observed a negative effect of 1% galactose on the growth of control yeast cells (Figure 4e). A second level of optimization is to systematically compare the amount of inducer with the solubilization efficiency of the recombinant membrane protein. For UCP1 and GFP‐ATP7B, the optimal galactose concentrations are 0.003% and 0.0015%, respectively, which are 1000 times less than the concentration we and other groups have used for functional studies (Cimadamore‐Werthein et al. ²⁰²⁴; Fiorini and Mus‐Veteau ²⁰¹⁶; Gagelin et al. ²⁰²³; Kintzer and Stroud ²⁰¹⁶; Lenoir et al. ²⁰⁰²; Moncoq et al. ²⁰⁰⁸; Piel et al. ²⁰²¹; Pyrihová et al. ²⁰²⁴; Tavoulari et al. ²⁰²⁴), and the most frequent galactose concentration observed in the PDB‐Membrane proteins (this study). This suggests that yeast processes correctly a limited amount of recombinant membrane proteins and that the remaining excess of the overproduced protein is misfolded or aggregated. The differential solubilization efficiency of Cymal‐7 and DPC detergents supports this analysis. DPC does not differentiate between folded and misfolded alpha‐helical membrane proteins (Chipot et al. 2018). In the case of GFP‐ATP7B, it readily solubilizes the recombinant protein no matter the concentration of galactose, while Cymal‐7 solubilizes the membrane protein only at low galactose concentrations (Figure 7d). A third level of optimization relates to nutrient conditions. At 0.003% galactose concentration, the time course of UCP1 production is delayed to a late stationary phase where nutrients in S‐lactate minimal medium become limiting. Here we show that supplementation with amino acids at the induction time resumes cell growth and the synthesis of the recombinant protein. It is as efficient as changing the culture medium, a usual practice, 12 out of 59 entries for the GAL promoter system found in the yeast PDB‐MP (Figure 8a). A fourth parameter, not investigated in this study, is the culture temperature. Decreasing the culture temperature at induction time in microbial systems is a frequent practice (Dilworth et al. 2018). In the yeast PDB‐MP, we found that 15 out of 59 entries had an induction temperature below 30°C (Figure 8a). Interestingly, the expression protocols for AQP proteins, Ca²⁺‐ATPase, and lipid flippase complexes Drs2p/Cdc50p and ATP8B1/Cdc50p, which were produced at the lowest temperatures in S. cerevisiae PDB‐MP entries (15 and 18°C), employed strains with the Gal4 enhancer (Figure 8b). The lower temperature may compensate for the overproduction driven by the Gal4 enhancer, promoting the accumulation of correctly folded membrane proteins (Azouaoui et al. 2014; Bjørkskov et al. ²⁰¹⁷; Lenoir et al. ²⁰⁰²). However, in contrast to the fine‐tuning of the galactose promoter, lowering culture temperature has the drawback of decreasing cell machinery and biomass.

To conclude, the work presented here shows, in two independent examples, that titration of the gal promoter with very low concentrations of galactose is a useful parameter to test for improving the solubilization of some recombinant membrane proteins. Additionally, it reduces the toxicity associated with the expression of the target gene, and it is cost‐effective for industrial large‐scale cultures.

4. MATERIALS AND METHODS

4.1. Reagents

All reagents were purchased from Sigma‐Aldrich unless specified in the text. All detergents were purchased from Anatrace except for DDM (Glycon).

4.2. Yeast strain, expression vector

The GGATCCATGCATCACCATCACCATCATCACCATGAAAACTTGTACTTTCAATCAGGGTCGAC sequence was cloned between BamH1 and Sal1 of the pYeD60 vector. It is coding for the MHHHHHHHHENLYFQSGST His8‐TEV N‐Terminal Tag peptide. The Rattus norvegicus UCP1 sequence (NCBI accession number NP‐036814.1) was cloned between Sal1 and Kpn1 without codon optimization in frame with the N‐Ter peptide. GFP‐ATP7B is cloned in the pYeDP60 yeast expression vector as described in Urban et al. (1994), a gift of Denis Pompon (TBI, Toulouse, France). Yeast S. cerevisiae strain W303.1b/Gal4 (a, leu2, his3, trp1::TRP1‐GAL10‐GAL4, ura3, ade2‐1, canr, cir+) (Figure S1) is transformed following the lithium acetate/single‐stranded carrier DNA polyethylene glycol method (Daniel Gietz and Woods ²⁰⁰²; Piel et al. ²⁰²¹).

4.3. UCP1 production in yeast

UCP1 expression is performed according to two different protocols (Figure 1). Respiration assay culture (R‐culture) was previously validated by spheroplast respiration for the functional expression of UCP1 in yeast mitochondria (Gagelin et al. 2023; Piel et al. ²⁰²¹). Overnight yeast preculture is done at 30°C under 200 rpm agitation in S‐lactate medium (2% lactate, 0.67% yeast nitrogen base without amino acids, 0.1% casamino acids (Fisher Scientific, Hampton, NH), 9 mM (NH₄)₂SO₄, 7.3 mM KH₂PO₄, 98 μM tryptophan, pH 4.5) supplemented with glucose (0.1%). The starter culture is diluted at OD_600nm = 0.03 in S‐lactate medium with 0.1% glucose, and yeast cells are grown at 30°C under 200 rpm agitation. After 19 h of incubation, the yeast culture is at OD_600nm = 4. The medium is exchanged into a new S‐lactate medium supplemented with different galactose concentrations for a 4 h induction at 30°C under 200 rpm agitation. On the other hand, the culture for protein purification (P‐culture) was designed for large‐scale cultures. To shorten the time of growth in S‐lactate media, preculture is done overnight at 30°C under 200 rpm agitation in S‐glucose medium (2% glucose, 0.67% yeast nitrogen base without amino acids, 0.1% casamino acids (Fisher Scientific, Hampton, NH), 9 mM (NH₄)₂SO₄, 7.3 mM KH₂PO₄, 98 μM tryptophan, pH 5.5). The starter culture is diluted at 0D_600nm = 0.3 in S‐lactate medium supplemented with 0.1% glucose. After 6 h of incubation at 30°C, UCP1 expression is induced at OD_600nm = 1, with the addition of galactose to the culture. UCP1 is produced for 16 h at 30°C under 200 rpm agitation.

4.4. Preparation of yeast spheroplasts and isolation of mitochondria

Yeast cells are harvested by centrifugation (5000g for 10 min), washed in water, and resuspended (10 mL/g of wet yeast pellet) in SED buffer (1M sorbitol, 25 mM EDTA, and 50 mM 1.4‐dithiothreitol). After a 10‐min incubation at 30°C, cells are washed with zymolyase buffer (sorbitol 1.2M, KH₂PO₄/K₂HPO₄ 20 mM, pH 7.4), and resuspended (10 mL/g of yeast cells) in zymolyase buffer supplemented with 0.5 mg/mL Zymolyase‐20T (Amsbio). Digestion of yeast cell walls is achieved during an incubation at 30°C for 45 min. The resulting spheroplasts are washed three times (1500g centrifugation, 5 min) with respiration buffer (1M sorbitol, 0.5 mM EDTA, 2 mM MgSO₄, 1.7 mM NaCl, 10 mM KH₂PO₄/K₂HPO₄, 0.1% Bovine Serum Albumin fatty acids free, pH 6.8).

The final washed pellet is resuspended in respiration buffer, and cell lysis is performed using the Yeda Press (Nitrogen cavitation). The cell suspension is collected into a pre‐cooled cavitation chamber. Under stirring at 4°C, it is subjected to 500 psi for 10 min. At the end of this step, the pressure in the chamber falls to approximately 350 psi for another 10 min. The cell suspension is then released through outflow tubing attached to the bottom valve, and the depressurization of the cell suspension from 350 psi to normal atmospheric pressure led to a highly efficient cell breakage (Wang et al. 2014). The integrity of mitochondria issued from this method is preserved, as studied by respiration assays (data not shown). To get rid of unbroken and partially disrupted cells, cell homogenate is centrifuged at 1000g at 4°C for 10 min. The resulting supernatant is centrifuged at 12,600g for 30 min to sediment mitochondria. The obtained crude mitochondrial fraction is resuspended in a freezing buffer (10% Trehalose, 100 mM Tris–HCl pH 7.5) and stored at −70°C after flash‐freezing with liquid nitrogen. Proteins in the mitochondrial fraction are quantified by the BCA method with BSA as the standard.

4.5. Extraction of total yeast proteins

Volumes of cultures (mL) corresponding to a constant OD_600nm of 5 (V = 5/OD_600nm) are centrifuged at 5000g for 10 min. Harvested cells were resuspended with 400 μL of cold 5% TCA. Upon the addition of glass beads, yeast cells are broken using a vortex for 5 min at maximum speed. The supernatant is collected and beads are washed two times with 500 μL cold 2% TCA. All supernatants are pooled and chilled on ice for 15 min. Proteins are pelleted by centrifugation at 16,000g for 15 min at 4°C. Pellets are resuspended with 100 μL of 100 mM Tris–HCl pH 8 and 8M urea.

4.6. Immunodetection of UCP1

The same amount of total protein extracts (1 μg) or mitochondrial proteins (10 μg) among all expression protocols and all biological replicates is loaded on a 12% SDS‐PAGE gel and transferred onto a nitrocellulose membrane (GE Healthcare). UCP1 is revealed using a mouse anti‐penta‐Histidine HRP‐conjugate 0.2 μL/mL (Bio‐Rad, catalogue number: MCA5995P, clone: ABD 2.2.20). VDAC1 Porin is used as a mitochondrial loading control and detected by a mouse anti‐VDAC1 0.1 μL/mL (Invitrogen, catalogue number: 459500, clone: 16G9E6BC4) and an HRP‐coupled goat anti‐mouse antibody 0.1 μL/mL (Promega, catalogue number: W4021, polyclonal). Peroxidase activity is detected using the Pierce ECL Western Blot Kit (Thermo Fisher Scientific), and the intensities of the bands are calculated using Image Lab software (6.1; Bio‐Rad, Hercules, CA). The relative amount of UCP1 protein is quantified using the UCP1/VDAC band intensity ratio. Uncropped scans of all blots are provided as a Source Data file.

4.7. Solubilization assays

Mitochondria are thawed and buffer is exchanged for solubilization buffer (150 mM NaCl, 50 mM HEPES, 20 mM imidazole) to a final concentration of 1 mg/mL for a detergent to protein ratio = 10 or 10 mg/mL for the detergent to protein ratio = 1, as the final concentration of detergent is 10 mg/mL (1%). Crude mitochondria are solubilized at 4°C for 1 h. Ultracentrifugation at 100,000g for 1 h separates the insoluble proteins (pellet) from the soluble ones (supernatant). Pellets are resuspended in the same initial volume before centrifugation in 20% SDS and 8M urea. The same volume of Total, Supernatant, and Pellet fractions is loaded on SDS‐PAGE.

4.8. Purification of UCP1

The Ni‐NTA resin (Thermo Fisher Scientific) is pre‐washed with buffer A (300 mM NaCl, 50 mM HEPES, 20 mM imidazole, 0.03% DDM, 0.1 mg/mL Cardiolipin (Avanti Polar Lipids)). The solubilized mitochondrial fraction is supplemented with 150 mM NaCl before batch binding to 1 mL of Ni‐NTA resin for 1 h at 4°C. The resin is washed under gravity flow with 20 mL of buffer A, followed by 20 mL of buffer B (150 mM NaCl, 50 mM HEPES, 20 mM imidazole, 0.03% DDM, 0.03 mg/mL Cardiolipin). Proteins are eluted from the Ni‐NTA resins with elution buffer (150 mM NaCl, 50 mM HEPES, 300 mM imidazole, 0.03% DDM, 0.03 mg/mL Cardiolipin), supplemented with 50 units of histidine‐tagged TEV protease. Elution fractions are pooled and dialyzed (membrane cut‐off = 10 kDa) in dialysis buffer: (150 mM NaCl, 50 mM HEPES, 0.03% DDM, 0.5 mM TCEP, and 0.1 mM EDTA) at 4°C. The resulting TEV‐cleaved UCP1 protein is loaded onto a second 500 μL Ni‐NTA Resin equilibrated with dialysis buffer. After incubation at 4°C for 1 h, the flow‐through containing the TEV‐cleaved UCP1 is collected and concentrated. The final UCP1 concentration was determined by spectrophotometry according to the Beer–Lambert law.

4.9. Reconstitution in liposomes and proton flux assay

For liposomes preparation, 5 mg of phosphatidylethanolamine (PE) (Egg PE sigma P7943, DPPE Avanti 850705), 5 mg of Cardiolipin, and 40 mg of phosphatidylcholine (PC) (Avanti 840051) are solubilized in 5 mL of internal buffer (70 mM K₂SO₄, 5 mM KH₂PO₄ pH 6.8). The suspension is homogenized by vortexing and subjected to ultrasounds on ice for 2 × 5 min at 70 Watts using Bioblock Scientific VibraCell 72446.

Purified UCP1, 40 μg, is incorporated into liposomes at 4°C for 1 h in the presence of 0.3% Triton X‐100. Detergent elimination is achieved by two incubations with 150 mg Bio‐Beads (Bio‐Rad) for 1 h 30 min, followed by a third incubation with 200 mg Bio‐Beads for 2 h.

UCP1 proton transport assay in liposomes is adapted from Mozo et al. (2006). Safranine probe aggregates as function of membrane potential (ΔΨ). Upon the addition of nigericin that exchanges H⁺ against K⁺ (H⁺ in, K⁺ out), a ΔpH is created. By letting protons out of liposome, UCP1 converts the ΔpH gradient into ΔΨ. The safranin absorbance at 520 nm is inversely proportional to ΔΨ; thus, it decreases due to UCP1 protonophoric activity. A control with empty liposomes ensures that the used concentration of fatty acids (30 μM of lauric acid) does not have any destabilizing effect on the lipidic membrane. When indicated, 1 mM of GDP is added to inhibit UCP1 activity. At the end of the experiment, a chemical uncoupler (CCCP) is added to let all protons go out to the external side of the liposome and to calculate ΔΨ_max, a value that is proportional to the internal volume of liposomes. Slopes of safranine curves are normalized by ΔΨ_max.

4.10. SEC‐MALLS analysis of the purified UCP1

Quality control of the purified UCP1 sample was checked by size exclusion chromatography (SEC) coupled to a multiple‐angle laser light scattering (MALLS). This experiment was carried out at room temperature on a Shimadzu HPLC coupled to three detectors: (1) a UV–VIS detector SPD‐20A (Shimadzu) for absorbance measurements at 280 nm, (2) an Optilab T‐rEX refractometer (Wyatt Technology) for the differential refractive index (dRI) measurements at 658 nm, and (3) a miniDawn TREOS detector (Wyatt Technology) for static light scattering intensity measurements at three different angles (44, 90, and 136°) at 658 nm. A 24‐ml Superdex 200 Increase column (10/300, Cytiva) was used to separate the protein‐detergent complex and protein‐free detergent micelles. The system was equilibrated overnight with a mobile phase containing 50 mM HEPES pH 7, 150 mM NaCl, 0.03% DDM, and 0.03 g/mL Cardiolipin at a flow rate of 0.3 mL/min. Once the UV, LS, and RI signals are stable, the buffer baseline is subtracted for each detector. Inter‐detector delay and peak‐broadening corrections are done using a sample of bovine serum albumin (i.e., 25 μL of BSA at 5 mg/mL dissolved in the mobile phase) as a monodisperse standard protein of known molar mass (66 kDa). Purified UCP1 (100 μL at 1 mg/mL) was then injected through the column. The absolute molar mass of UCP1‐DDM‐Cardiolipin complexes and protein‐free micelles was obtained without column calibration by using the module “Protein Conjugate” from the software Astra (Wyatt Technology). The molar mass of the complex is calculated from the following Rayleigh equation:

$$\mathrm{\Delta }R\left(\theta \right)=K·M_{w\text{complex}}·\left[\text{complex}\right]{\left(\frac{\partial n}{\partial c}\right)}_{\text{complex}}^2,$$

where K, M _w , ${\left(\frac{\partial n}{\partial c}\right)}_{\text{complex}}$ are, respectively, an optical constant, the weight‐average molecular mass of the complex, and the refractive index increment of the complex.

Knowing the respective extinction coefficients and refractive index increments of both the protein and the detergent micelles, we can obtain the molar mass fraction of UCP1 and protein‐bound detergent. The methodology of mass calculation is thoroughly described in Gimpl et al. (2016).

For UCP1, ε^0.1% at 280 equals 0.783 (mg/mL)⁻¹ cm⁻¹ and ∂n/∂c = 0.187 mL g⁻¹ calculated from sequence using SedFit software (sedfitsedphat.nibib.nih.gov), and for the DDM:CL micelles, ε^0.1% = 0 and ∂n/∂c = 0.146 mL g⁻¹ measured by injecting increasing concentrations of DDM:CL mixtures directly in the refractometer.

4.11. Yeast cell growth and fluorescence monitoring

Yeast cell growth expressing UCP1 or ATP7B was studied by Optical Density measurements at 600 nm using the CLARIOstar Plus Microplate Reader, which was also put into service for GFP of ATP7B fluorescence measurements, and the SPECTROstar Nano. Both machines are from BMG LABTECH (GmbH, Germany). After an overnight preculture of yeast cells in S‐lactate media supplemented with 0.01% glucose at 30°C at 200 rpm, the culture was diluted to OD_600nm = 0.1 in new S‐lactate media with 0.1% glucose and the appropriate galactose concentrations in a 48‐well plate. The plate is then incubated in the machine to assess OD_600nm and/or fluorescence (30 min/cycle) at 30°C with agitation before each reading. It is important to note that the auto‐induction and oxygenation conditions in this experiment differ from those applied to the protocols of Figure 1. S‐lactate media absorbance intensities were subtracted from raw data. Yeast cells with the empty pYeDP60 vector were taken as a negative control with no stress due to protein expression. For the fluorescence measurements, values were normalized by OD_600nm at each time point.

4.12. ATP7B: Cell lysis and membrane preparation

For the copper transporter ATP7B experiments, cell culture was carried out according to the purification protocol. Harvested cells were directly subjected to lysis with Cell‐D disruptor, with no spheroplast preparation step beforehand. To ensure high efficiency of lysis, cells that were resuspended in 50 mM Tris–HCl pH 7, 20 mM EDTA, and 100 mM sucrose were broken during three repeated cycles with maximal pressure (40000 PSI). Unbroken cells and cell debris were sedimented and removed at 1500g for 10 min; then centrifugation at 20,000g for 30 min allowed obtaining heavy membranes in the pellet. The corresponding supernatant was finally subjected to a 120,000g ultracentrifugation for 1 h, and the light membrane fraction was therefore sedimented. The next steps were performed similarly to UCP1 protocols with the following exceptions: (1) MgCl₂ was added to the solubilization buffer, and both Cymal‐7 and DPC were mixed with CHS according to a 1:4 detergent to CHS ratio, (2) for the separation of soluble and insoluble parts after solubilization, the ultracentrifugation was done at 120,000g. Samples were run to SDS‐PAGE gels (10% acrylamide), and the fluorescence of GFP was monitored by typhoon biomolecular imager (GE Healthcare).

4.13. PDB analysis

Structural studies of membrane proteins need large quantities of homogeneous and correctly folded proteins. By focusing on PDB data, we can identify and analyze systems that optimize the production of the required quantity and quality of these challenging proteins. Therefore, the protein structures registered in “mpstruc membrane protein of known 3D structure,” Dr. Stephen White's database at UC Irvine (https://blanco.biomol.uci.edu/mpstruc/), were used as a basis to analyze expression and solubilization conditions, with data available up to the end of 2023. An entry corresponds to the structure of a membrane protein that has been published in a peer‐reviewed journal and is therefore identified by its PDB number. There is a distinction between the terms “unique protein” and “published report.” The date of the “unique protein” structure marks the anteriority; other similar protein structures (same organism of origin) filled and published later will be considered as “published reports.” As the same protein can be expressed, solubilized, purified, and characterized under different conditions, working on the set of unique proteins and published reports will enable us to work on a larger panel.

After having manually curated database entries (excluding structures without membrane protein domains), among the 3353 entries (for unique and published membrane proteins), 15.1% were obtained from native proteins, 38.3% were expressed in E. coli, 38.5% in eukaryotic systems, 5.4% in yeast, and 4.8% in other systems. Our study focused on the expression of membrane proteins in yeast and more specifically in S. cerevisiae. Of the 180 entries for yeast (5.4%), 76 correspond to S. cerevisiae (42.2%), 59 to K. Pastoris (32.8%), 41 to P. Pastoris (22.8%), and 4 to S. Pombe (2.2%). For this set of 76 membrane proteins expressed in S. cerevisiae, extracted variables from mpstruc database, Uniprot (https://www.uniprot.org/) and from the Protein Data Bank in Europe (https://www.ebi.ac.uk/pdbe/) were: PDB, Type, UniprotKB code, protein or complex name, and molecular weight. Additional variables on expression parameters (strain, plasmid, promotor, induction time, induction temperature, inductor concentration, medium) and solubilization parameters (detergent) were then consistently retrieved by reading the literature. All the variables that were analyzed are presented in Data S2.

4.14. Statistical analyses and replicates

Statistical analyses with multiple unpaired t‐tests were done on PRISM software: Tables S1 and S2 for Figure 4, Tables S3 and S4 for Figure 5, Table S5 for Figure 6d, Table S6 for Figure 7, and Table S7. The number of replicates is provided in the figure legend. Bars indicate means ± SEM in Figures 4d, 5, and 7d. For Figure 6d, boxes of the box plots indicate the first quartile, the median, and the third quartile. The whisker length is 1.5 times the interquartile range. Replicates for Figures 4d, 5a,b, and 7d are shown in Figures S2–S5, respectively. In the figures, the following correspondences may be found: ns = not significant; *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001; and ****, p ≤ 0.0001; p ≤ 0.05 was considered as statistically significant.

AUTHOR CONTRIBUTIONS

Rebecca Moussa: Conceptualization; writing – original draft; writing – review and editing; methodology; formal analysis; funding acquisition; investigation; validation; visualization. François Gellé: Writing – review and editing; funding acquisition; formal analysis; validation; investigation. Sandrine Masscheleyn: Validation; investigation. Alexandre Pozza: Writing – review and editing; formal analysis; investigation. Christel Le Bon: Writing – review and editing; investigation; visualization; resources. Karine Moncoq: Writing – review and editing; conceptualization; investigation; resources; visualization. Françoise Bonneté: Writing – review and editing; formal analysis; conceptualization; funding acquisition; investigation; visualization. Bruno Miroux: Writing – review and editing; supervision; conceptualization; funding acquisition; writing – original draft; project administration; validation; methodology; resources; visualization.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1. Supporting Information.

Data S2. Supporting Information.

Moussa R, Gellé F, Masscheleyn S, Pozza A, Le Bon C, Moncoq K, et al. Fine‐tuning the yeast GAL10 promoter and growth conditions for efficient recombinant membrane protein production and purification. Protein Science. 2025;34(5):e70125. 10.1002/pro.70125

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in Supporting Information of this article.