Unfolding Individual Domains of BmrA, a Bacterial ABC Transporter Involved in Multidrug Resistance

The folding and stability of proteins are often studied via unfolding (and refolding) a protein with urea. Yet, in the case of membrane integral protein domains, which are shielded by a membrane or a membrane mimetic, urea generally does not induce unfolding. However, the unfolding of α-helical membrane proteins may be induced by the addition of sodium dodecyl sulfate (SDS). When protein unfolding is followed via monitoring changes in Trp fluorescence characteristics, the contributions of individual Trp residues often cannot be disentangled, and, consequently, the folding and stability of the individual domains of a multi-domain membrane protein cannot be studied. In this study, the unfolding of the homodimeric bacterial ATP-binding cassette (ABC) transporter Bacillus multidrug resistance ATP (BmrA), which comprises a transmembrane domain and a cytosolic nucleotide-binding domain, was investigated. To study the stability of individual BmrA domains in the context of the full-length protein, the individual domains were silenced by mutating the existent Trps. The SDS-induced unfolding of the corresponding constructs was compared to the (un)folding characteristics of the wild-type (wt) protein and isolated domains. The full-length variants BmrAW413Y and BmrAW104YW164A were able to mirror the changes observed with the isolated domains; thus, these variants allowed for the study of the unfolding and thermodynamic stability of mutated domains in the context of full-length BmrA.


Introduction
Approximately 20-30% of all genes in any genome code for membrane-integral proteins [1], and the proper folding and assembly of membrane proteins is imperative for their function [2][3][4][5]. Missense mutations in human membrane proteins often result in malfunction, protein destabilization and/or misfolding, or improper intracellular protein trafficking [6][7][8][9]. Yet, mostly due to experimental difficulties, the (impaired) folding and stability of transmembrane (TM) proteins is far less studied and understood when compared to the knowledge of these processes in soluble proteins [10,11].
The thermodynamic stability and the folding kinetics of soluble proteins are often studied via the unfolding (and refolding) of proteins by chaotropic agents, such as guanidinium hydrochloride or urea [12]. Yet, with some exceptions [13], only the stability of water soluble proteins or domains can be examined using urea [4,12,14], while the hydrophobic domains of α-helical membrane proteins are typically protected by the lipid bilayer or membrane-mimicking detergents and are, therefore, inaccessible for urea [12,[15][16][17]. However, α-helical membrane proteins can be unfolded in vitro by the anionic detergent sodium dodecyl sulfate (SDS) [17][18][19]. Upon the addition of SDS to a membrane protein dissolved in a mild, non-ionic detergent, mixed micelles form. At increasing SDS concentrations, the 3 of 13 TMD interactions. Therefore, we first needed to establish a system to selectively monitor the stability of one specific domain. We investigated the stability of full-length BmrA and isolated domains by monitoring changes in the intrinsic Trp fluorescence emission signal upon protein unfolding induced by urea or SDS. Upon the replacement of the BmrA Trp residues, we finally generated BmrA variants with wild-type activity that allowed us to study the stability of the TM or soluble BmrA domain, respectively, in the context of the full-length protein under defined conditions.

Results and Discussion
2.1. Urea-Induced Trp Fluorescence Changes Originate Exclusively from the NBD An established method for studying the thermodynamic stability and/or the pathway of protein (un)folding is to monitor changes in the intrinsic fluorescence emission of a protein. Naturally occurring Trp residues can be used as sensors for a protein's stability since changes of the polarity in a Trp environment alter its fluorescence emission characteristics [58]. Yet, when multiple Trp residues exist in multi-domain proteins, the impact of each individual Trp residue on the observed Trp fluorescence changes remains unclear. In a BmrA wild-type (wt) monomer, three Trp residues naturally occur ( Figure 1A, Trps shown in red), two of which are located in the TMD. More precisely, W104 is localized in the TMD, whereas W164 is part of a short, unstructured loop that connects TM3 with TM4 at the extracellular side (extracellular loop 2) close to the head groups of the lipid bilayer. W413 is located at the C-terminus of the cytosolic NBD. As BmrA is a multi-domain protein, we aimed at generating a BmrA variant that allows for the study of the unfolding of a selected domain within the full-length protein using Trp fluorescence. Subsequently, this involved deleting the Trp contributions that were outside of the domain of interest. mediated by ATP binding and/or hydrolysis in the NBDs to the TMDs [51]. We aim (on long term) to analyze the functional interaction of individual domains and the contributions of specific NBD-TMD interactions. Therefore, we first needed to establish a system to selectively monitor the stability of one specific domain. We investigated the stability of full-length BmrA and isolated domains by monitoring changes in the intrinsic Trp fluorescence emission signal upon protein unfolding induced by urea or SDS. Upon the replacement of the BmrA Trp residues, we finally generated BmrA variants with wild-type activity that allowed us to study the stability of the TM or soluble BmrA domain, respectively, in the context of the full-length protein under defined conditions.

Urea-Induced Trp Fluorescence Changes Originate Exclusively from the NBD
An established method for studying the thermodynamic stability and/or the pathway of protein (un)folding is to monitor changes in the intrinsic fluorescence emission of a protein. Naturally occurring Trp residues can be used as sensors for a protein's stability since changes of the polarity in a Trp environment alter its fluorescence emission characteristics [58]. Yet, when multiple Trp residues exist in multi-domain proteins, the impact of each individual Trp residue on the observed Trp fluorescence changes remains unclear. In a BmrA wild-type (wt) monomer, three Trp residues naturally occur ( Figure 1A, Trps shown in red), two of which are located in the TMD. More precisely, W104 is localized in the TMD, whereas W164 is part of a short, unstructured loop that connects TM3 with TM4 at the extracellular side (extracellular loop 2) close to the head groups of the lipid bilayer. W413 is located at the C-terminus of the cytosolic NBD. As BmrA is a multi-domain protein, we aimed at generating a BmrA variant that allows for the study of the unfolding of a selected domain within the full-length protein using Trp fluorescence. Subsequently, this involved deleting the Trp contributions that were outside of the domain of interest.
First, we investigated the unfolding of the purified full-length protein and the isolated domains using urea ( Figure 1B). Since the surfaces of soluble proteins are exposed to water to a greater extent than TMDs, we expected larger differences upon the addition of increasing urea concentrations to the full-length protein as well as in case of the isolated NBD compared to the isolated TMD.  First, we investigated the unfolding of the purified full-length protein and the isolated domains using urea ( Figure 1B). Since the surfaces of soluble proteins are exposed to water to a greater extent than TMDs, we expected larger differences upon the addition of increasing urea concentrations to the full-length protein as well as in case of the isolated NBD compared to the isolated TMD.
The normalized emission spectra of the full-length BmrA wt protein shows a fluorescence emission maximum at 322 nm, indicating that the Trp residues are mainly located in a rather hydrophobic environment [59]. When exposed to 6.5 M urea, the Trp fluorescence emission maximum shifted (6 nm) to longer wavelengths (Figure 2A, black), indicating that the environments of some Trps became more hydrophilic. When varying the urea concentration between 0 and 6.5 M, the fluorescence intensity initially remained at a constant level until, starting at concentrations >1.5 M urea, a final decrease of around 42% at 6.5 M urea was observed ( Figure 2B, black). creased with increasing urea concentrations, indicating that the changes in the Trp fluo-rescence emission observed with full-length BmrA mainly evidence changes in the NBD structure. Since the TMD is surrounded by a n-dodecyl-β-D-maltoside (DDM) detergent micelle, the Trps of the TMD are likely protected against urea-induced unfolding [4]. Accordingly, with the isolated TMD, no decrease in the maximum fluorescence intensity was observed when this domain was exposed to increasing urea concentrations (Figure 2A,B, red), and changes in the Trp fluorescence characteristics observed upon urea denaturation of the isolated NBD mixed with the isolated TMD were essentially identical to the fluorescence emission changes observed with the NBD (Figure 2, green). Taken together, it was ascertained that the Trp fluorescence emission changes detectable upon the ureabased denaturation of the full-length BmrA wt protein originate from the unfolding of the soluble NBD, with little or no contribution from the TMD.

SDS-Induced Trp Fluorescence Changes Originate Mainly, but Not Exclusively, from the NBD
While urea appeared to selectively unfold the NBD, SDS might unfold the soluble and TM domains (as outlined in the Introduction). Therefore, we investigated the SDSinduced unfolding of the full-length BmrA wt and the isolated domains. Protein unfolding In the absence of urea, the isolated NBD had a fluorescence emission spectrum that was essentially identical to the full-length BmrA protein, and upon urea denaturation, comparable changes in the fluorescence emission characteristics were observed: the fluorescence intensity decreased (~67%) and the absorption maximum shifted to longer wavelengths by about 15 nm (Figure 2A,B, blue). In addition, in case of the isolated NBD, the fluorescence emission intensity remained constant up to 1.5 M urea and thereafter decreased with increasing urea concentrations, indicating that the changes in the Trp fluorescence emission observed with full-length BmrA mainly evidence changes in the NBD structure. Since the TMD is surrounded by a n-dodecyl-β-D-maltoside (DDM) detergent micelle, the Trps of the TMD are likely protected against urea-induced unfolding [4]. Accordingly, with the isolated TMD, no decrease in the maximum fluorescence intensity was observed when this domain was exposed to increasing urea concentrations (Figure 2A,B, red), and changes in the Trp fluorescence characteristics observed upon urea denaturation of the isolated NBD mixed with the isolated TMD were essentially identical to the fluorescence emission changes observed with the NBD (Figure 2, green). Taken together, it was ascertained that the Trp fluorescence emission changes detectable upon the urea-based denaturation of the full-length BmrA wt protein originate from the unfolding of the soluble NBD, with little or no contribution from the TMD.

SDS-Induced Trp Fluorescence Changes Originate Mainly, but Not Exclusively, from the NBD
While urea appeared to selectively unfold the NBD, SDS might unfold the soluble and TM domains (as outlined in the Introduction). Therefore, we investigated the SDS-induced unfolding of the full-length BmrA wt and the isolated domains. Protein unfolding was initiated by the addition of increasing amounts of SDS to a solution of BmrA dissolved in DDM. The anionic detergent SDS is able to form mixed micelles with the mild detergent DDM, eventually resulting in the unfolding of the soluble and TM domains. Noteworthily, the α-helical structure of hydrophobic TM helices, designed by nature to reside within the hydrophobic membrane's core region, typically remains preserved in membrane-mimicking micellar environments, which prevents the study of unfolding by following changes in the secondary structure via techniques such as circular dichroism (CD) spectroscopy.
First, we analyzed the influence of increasing SDS concentrations on the Trp fluorescence emission characteristics using the full-length BmrA wt protein. With an increasing χ SDS , the monitored fluorescence emission intensities initially remained constant until χ SDS = 0.08; subsequently, the Trp fluorescence intensity at 322 nm was reduced by~45% at the highest SDS mole fraction (χ SDS = 0.95), with most of the change occurring between 0.08 and 0.2 χ SDS ( Figure 3A,B, black). When further increasing χ SDS > 0.2, the Trp fluorescence intensities gradually decreased. Thus, SDS clearly alters the structure of BmrA. = 0.08; subsequently, the Trp fluorescence intensity at 322 nm was reduced by ~45% at the highest SDS mole fraction (χSDS = 0.95), with most of the change occurring between 0.08 and 0.2 χSDS ( Figure 3A,B, black). When further increasing χSDS > 0.2, the Trp fluorescence intensities gradually decreased. Thus, SDS clearly alters the structure of BmrA.
To examine the contribution of each domain to the overall fluorescence signal changes, we next investigated the impact of an increasing χSDS on the isolated NBD and the isolated TMD. The normalized emission spectra obtained using full-length BmrA wt (black), the isolated TMD (red), and the isolated NBD (blue) have a similar shape ( Figures  2A and 3A), albeit the positions of the fluorescence emission maxima differ to a certain extent (BmrA wt: 322 nm, TMD: 328 nm, NBD: 324 nm). Upon the unfolding of the isolated NBD by SDS, the fluorescence emission intensity overall decreased by about two thirds compared to the intensity measured for the NBD in the absence of SDS ( Figure 3B, blue). Between 0.08-0.2 χSDS, the most significant changes were observed, and with a further increasing χSDS, the fluorescence emission intensity decreased gradually, yet only to a minor extent. Noteworthily, analyses of the water-soluble isolated NBD would not necessarily have required detergent. Yet, to allow for the formation of mixed micelles and thus a direct comparison of the experiments, DDM was present in the buffer. When the isolated TMD was unfolded via the addition of increasing amounts of SDS, the fluorescence emission intensity first slightly increased by ~5% until a maximum was reached at χSDS = 0.16. Similar observations were made with the α-helical TM protein GlpG, where an early "unfolding" transition state with increased fluorescence emission To examine the contribution of each domain to the overall fluorescence signal changes, we next investigated the impact of an increasing χ SDS on the isolated NBD and the isolated TMD. The normalized emission spectra obtained using full-length BmrA wt (black), the isolated TMD (red), and the isolated NBD (blue) have a similar shape (Figures 2A and 3A), albeit the positions of the fluorescence emission maxima differ to a certain extent (BmrA wt: 322 nm, TMD: 328 nm, NBD: 324 nm). Upon the unfolding of the isolated NBD by SDS, the fluorescence emission intensity overall decreased by about two thirds compared to the intensity measured for the NBD in the absence of SDS ( Figure 3B, blue). Between 0.08-0.2 χ SDS , the most significant changes were observed, and with a further increasing χ SDS , the fluorescence emission intensity decreased gradually, yet only to a minor extent. Noteworthily, analyses of the water-soluble isolated NBD would not necessarily have required detergent. Yet, to allow for the formation of mixed micelles and thus a direct comparison of the experiments, DDM was present in the buffer.
When the isolated TMD was unfolded via the addition of increasing amounts of SDS, the fluorescence emission intensity first slightly increased by~5% until a maximum was reached at χ SDS = 0.16. Similar observations were made with the α-helical TM protein GlpG, where an early "unfolding" transition state with increased fluorescence emission was observed at a low χ SDS , whereas the fluorescence emission decreased when χ SDS was further increased [60]. Similarly, the fluorescence emission of the TMD steadily decreased when χ SDS was further increased, and the overall reduction in the fluorescence emission intensity between χ SDS = 0 and χ SDS = 0.95 was approx. 25% ( Figure 3B, red). Thus, SDS appears to have a stronger effect on the structure of the soluble NBD than on the TMD, at least when Trp fluorescence changes were monitored. Finally, we mixed the isolated NBD with the isolated TMD and unfolded the mixture via increasing the SDS mole fraction. The observed changes in the fluorescence emission intensities ( Figure 3A,B, green) were similar to the changes observed with the full-length BmrA wt protein. These observations suggest that changes in the Trp fluorescence emission with the full-length BmrA wt protein upon SDS-induced unfolding are the sum of the isolated NBD and TMD, albeit the signal is dominated by the single Trp located in the NBD.

Monitoring the Unfolding of the NBD in the Context of the Full-Length Protein
Thus far, the unfolding of the full-length wt protein was compared to the isolated TMD and NBD. To selectively monitor the unfolding of one selected domain within the context of the full-length protein, we attempted to generate BmrA variants that contain Trps exclusively in the NBD or TMD, respectively. Therefore, we replaced each of the three Trp residues individually by Ala, Phe, and Tyr and determined the Hoechst transport activity of the altered proteins using inverted vesicles containing the expressed proteins. While the replacement of amino acids by Ala is common, as Ala's small side-chain typically does not disturb a protein's structure, the mutations should not affect the protein activity. Yet, in some cases, Trp's indole ring established hydrophobic, Van der Waals, and/or polar interactions with the surrounding residues; thus, we also replaced the Trp residues by Phe and Tyr. When the NBD-localized Trp 413 was replaced, solely the BmrA W413Y variant showed similar transport activity to that of the wt protein ( Figure 4); thus, this protein was subsequently used to selectively monitor structural changes in the TMD. For the construction of a BmrA variant that contains solely a Trp (W413) within the NBD, the Trp residues 104 and 164 were individually replaced with Ala, Phe, or Tyr. The three variants W104Y, W104F, and W164A, as well as the variant BmrA W104YW164A , showed wt-like activity ( Figure 4). Thus, the two variants BmrA W413Y and BmrA W104YW164 enabled us to selectively monitor the influence of increasing amounts of SDS on the structure and stability of the TMD or NBD, respectively, within the context of an active full-length BmrA protein.   Compared to wt BmrA, the maxima of the fluorescence emission spectrum of both variants ( Figure 5A, ocher/ turquoise) were slightly shifted to a shorter wavelength, with a maximum fluorescence intensity for BmrA W413Y at 320 nm and BmrA W104YW164A at 318 nm. Additionally, the fluorescence emission peak of BmrA W413Y appeared to be slightly broadened. Noteworthily, the fluorescence emission maximum of the isolated TMD was at 328 nm ( Figures 3A and 5A). The altered spectral shape of the mutated full-length proteins compared to the isolated domains was caused by the increased number of Tyrs in the sequence. In case of the NBD-domain, the spectrum is the sum of one Trp and seven Tyrs. In the case of the mutant BmrA W104YW164A , it is the sum of one Trp and fifteen Tyrs due to the additional eight Tyrs from the TMD. The increased number of Tyrs led to an overall blue-shift of the spectrum. However, since the fluorescence of Tyr is not altered upon changes in the polarity of the environment, their spectral contribution is constant, and thus does not have an effect on the shape of the denaturation curves. In addition, the contribution is relatively small since the denaturation curves are constructed based on the intensity above 320 nm, where the Tyr fluorescence is relatively low. Upon the addition of increasing SDS concentrations, the fluorescence emission maximum of BmrAW104YW164A changed nearly identically to that of the wt protein ( Figure 5B, ocher). For this variant, where exclusively structural changes occurring in the NBD were monitored, the fluorescence intensities remained constant until χSDS = 0.08. Thereafter, the intensity decreased by about 23% at χSDS = 0.2. Further increasing χSDS led to a small but steady decrease in the fluorescence emission intensity until at χSDS = 0.95 a reduction of ~45% compared to the intensity observed in the absence of SDS was achieved. In contrast, the fluorescence emission changes observed upon the unfolding of BmrAW413Y with increasing SDS concentrations ( Figure 5B, turquoise) differed from the full-length BmrA wt, as at this point the Trp fluorescence intensities initially increased until at χSDS = 0.16 a Upon the addition of increasing SDS concentrations, the fluorescence emission maximum of BmrA W104YW164A changed nearly identically to that of the wt protein ( Figure 5B, ocher). For this variant, where exclusively structural changes occurring in the NBD were monitored, the fluorescence intensities remained constant until χ SDS = 0.08. Thereafter, the intensity decreased by about 23% at χ SDS = 0.2. Further increasing χ SDS led to a small but steady decrease in the fluorescence emission intensity until at χ SDS = 0.95 a reduction of 45% compared to the intensity observed in the absence of SDS was achieved. In contrast, the fluorescence emission changes observed upon the unfolding of BmrA W413Y with increasing SDS concentrations ( Figure 5B, turquoise) differed from the full-length BmrA wt, as at this point the Trp fluorescence intensities initially increased until at χ SDS = 0.16 a maximum was reached, similar to the results obtained with the isolated TMD ( Figure 3B, red). Thereafter, the fluorescence intensities linearly decreased with increasing SDS concentrations. The final fluorescence emission intensity determined at χ SDS = 0.95 was slightly decreased (approx. 15%) compared to the native state (without SDS). The differences in the alteration of the maximum fluorescence intensities observed with BmrA W104YW164A and BmrA W413Y suggest that the Trp environment of the soluble and the membrane integral domains of the full-length BmrA protein are differently affected by SDS. Furthermore, the high degree of similarity of the BmrA wt and BmrA W104YW164A denaturation curves indicate that the altered Trp fluorescence characteristics observed with increasing χ SDS in the case of the wt are dominated by changes in the NBD-located W413 fluorescence, which is in agreement with the results obtained when the isolated domains were analyzed (Figure 3). Thus, the altered fluorescence characteristics mainly reflect changes in the environment of W413, and any changes in the structure and stability of the TMD, e.g., those caused by mutations or substrate binding, can only be observed in the BmrA W413Y background.

Cloning
A pET303-CT/His-BmrA wt [61] plasmid encoding a BmrA protein (UniProtKB accession no. O06967) with a C-terminal His 6 -tag was used for expression of the wt protein and as a template for the generation of plasmids used for expression of modified genes. The Trp residues of the full-length BmrA wt protein were replaced by Ala, Phe and Tyr via sitedirected mutagenesis (primers are listed in Table 1). The pET303-CT/His-TMD (residues M1-G331) and pET303-CT/His-NBD (residues K332-E591) plasmids were generated by introducing restriction sites (XbaI or XhoI) followed by restriction digestion and re-ligation. The resulting plasmids were utilized for expression of the isolated TMD and NBD, each containing a C-terminal His 6 -tag.

Protein Expression and Purification
The different proteins were expressed in Escherichia coli (E. coli) C41(DE3) cells upon induction of protein expression via 0.7 mM isopropyl-β-D-thiogalacto-pyranoside (IPTG). Pelleted cells were resuspended in buffer A (50 mM phosphate buffer, 150 mM NaCl, 10% glycerol (v/v), and pH = 8.0) and were lysed by three successive passages through a microfluidizer (LM20, Microfluidics, Westwood, USA; 18,000 psi). The lysed cells were centrifuged (12,075× g, 10 min, 4 • C) and the supernatant was again centrifuged (165,000× g, 1 h, 4 • C) to maintain the membrane fraction (except for the NBD-expressing cells). For solubilization of membrane proteins, the pellet containing the membrane fraction was incubated for 1 h at room temperature (RT) in solubilization buffer (buffer A with 1% (w/v) DDM). The solubilized proteins were further incubated with equilibrated Protino ® Ni-NTA agarose (Macherey-Nagel GmbH & Co., KG, Düren, Germany) for 1 h at RT. The Ni-NTA agarose resin with bound protein was washed with 25 mL of washing buffer 1 (buffer A with 0.1% DDM (w/v), and 10 mM imidazole), 50 mL of washing buffer 2 (buffer A with 0.1% DDM (w/v), and 35 mM imidazole), and 35 mL of washing buffer 3 (buffer A with 0.1% DDM (w/v), and 45 mM imidazole). The proteins were finally eluted with 5 mL elution buffer (buffer A with 0.1% DDM (w/v), and 400 mM imidazole). A PD-10 desalting column (Macherey-Nagel GmbH & Co., KG, Düren, Germany) was used for desalting and to exchange the buffer for the assay buffer (buffer A containing 5 mM DDM).
For purification of the soluble NBD, the supernatant obtained after cell lysis and the initial centrifugation was directly incubated with the equilibrated Ni-NTA resin. After 1 h incubation, the matrix was washed with 50 mL of buffer A followed by 50 mL of buffer A with 10 mM imidazole, and 25 mL of buffer A with 40 mM imidazole. The protein was eluted with 5 mL buffer A containing 400 mM of imidazole, and the imidazole was removed by exchanging the buffer for the assay buffer using a PD-10 desalting column (Macherey-Nagel GmbH & Co., KG, Düren, Germany).
Protein concentrations were determined photometrically by measuring the absorbance at 280 nm using the following extinction coefficients calculated using ExPASy [62]:

Preparation of Inverted E. coli Membrane Vesicles
For preparation of inverted E. coli membrane vesicles [62], C41(DE3) cells overexpressing the BmrA variants were used. A cell pellet of a 2 L expression culture was resuspended in buffer B (50 mM Tris-HCl, 5 mM MgCl 2 , 1 mM DTT, 1 mM PMSF (phenylmethylsulfonyl fluoride), and pH = 8.0) and lysed using a microfluidizer (3 × 18,000 psi). After lysis, EDTA (ethylenediaminetetraacetic acid, pH = 8.0, and 10 mM final concentration) was added to the cells, and cell debris and unbroken cells were removed by centrifugation (10,000× g, 30 min, and 4 • C). Cell membranes containing the overexpressed proteins were isolated via centrifuging the supernatant (140,000× g, 1 h, 4 • C). Then, the supernatant was discarded, the pellet was resuspended in 20 mL buffer C (50 mM Tris-HCl, 1.5 mM EDTA, 1 mM DTT, 1 mM PMSF, and pH = 8.0), and, subsequently, centrifuged again (140,000× g, 1 h, 4 • C). The pellet containing the membrane vesicles was resuspended in 4 mL buffer D (20 mM Tris-HCl, 300 mM sucrose, 1 mM EDTA, and pH = 8.0). Small aliquots were shock-frozen in liquid nitrogen and stored at −80 • C until use. The protein concentration was determined with the BCA protein assay kit (Thermo Fisher Scientific Inc., Waltham, MA, USA) following the manufacturer's instructions.

Hoechst 33342 Transport Assay
The activity of BmrA variants in inverted E. coli membrane vesicles was determined with a fluorescence-based transport assay with an ATP regeneration system. In this case, the hydrophobic dye 2 -[4-ethoxyphenyl]-5-[4-methyl-1-piperazinyl]-2,5 -bis-1H-benzimidazole (Hoechst 33342, Merck KGaA, Darmstadt, GER) was used as a BmrA substrate. For each sample, inverted membrane vesicles containing 50 µg of protein were used in a total volume of 200 µL. The samples were diluted in buffer E (50 mM Hepes-KOH, 2 mM MgCl 2 , 8.5 mM NaCl, 4 mM phosphoenolpyruvate, and 20 µg/µL pyruvate kinase (from rabbit muscle, Merck KGaA, Darmstadt, Germany)) and kept at 25 • C for 10 min. The fluorescence emission was measured for a total of 10 min at 457 nm using a FluoroMax-4 fluorometer (Horiba Instruments Inc., Edison, NJ, USA) upon excitation at 355 nm, with excitation and emission slit widths of 2 and 3 nm, respectively. First, the fluorescence was monitored for approx. 50 s and then the measurement was stopped. Subsequently, 2 µM Hoechst 33342 was added, the sample was mixed, and the fluorescence was measured again for approx. 50 s. Then, ATP was added (final concentration of 2 mM), and the sample was quickly mixed afterwards. The fluorescence was further monitored for overall~500 s. Data points were collected every 0.5 s and the initial slope of the measured fluorescence intensity after ATP addition was determined.

SDS-Induced Protein Unfolding
To unfold full-length BmrA variants as well as the isolated NBD and TMD, 2 µM of each purified protein in the assay buffer was exposed to increasing concentrations of SDS. The SDS mole fraction (χ SDS ) was used here to describe the detergent concentration due to the lack of information about the exact amount of SDS in the mixed micelles. The χ SDS was determined as in equation 1, with c SDS referring to the SDS concentration and c DDM to the concentration of DDM.
The samples were incubated for 1 h at 25 • C, and Trp fluorescence was measured from 290-450 nm upon excitation at 280 nm using a FluoroMax-4 fluorometer (Horiba Instruments Inc., Edison, NJ, USA) and a slit width of 3 nm.

Urea-Induced Protein Unfolding
The actual concentration of the urea stock solution was determined based on the refractive index of the solution after subtraction of the contribution of the buffer [63]. Purified proteins (2 µM final concentration) were exposed to increasing urea concentrations (ranging from 0 to 6.5 M urea in 0.5 M steps), and the samples were incubated at 25 • C for 1 h at RT. In all cases, the buffer solution contained 5 mM DDM to ensure comparable conditions. After incubation, Trp fluorescence spectra were recorded as described above.

Conclusions
Studying the unfolding of multi-domain TM proteins via following Trp fluorescence changes can be problematic if the protein contains multiple Trps that are located in different domains. One possible solution is to study the stability of the isolated domains individually. However, due to physiologically relevant interactions, the isolated domains do not necessarily behave as they would in the context of the full-length protein, and the stability of individual domains under conditions where the domains (putatively) differentially interact cannot be studied. To be able to study the stability of individual BmrA domains, we generated full-length BmrA variants via consecutively replacing all Trps within one domain by Phe, Tyr, or Ala and tested whether the variants structurally and functionally behaved like the wt protein. The stability of the full-length BmrA variants BmrA W413Y and BmrA W104YW164A , as determined via changes in the fluorescence intensity around 320 nm, closely mirror the behavior of the isolated NBD or TMD but have the advantage of being in the proximity of the full-length protein. Subsequently, these variants allow us to study the unfolding and the thermodynamic stability of the NBD or TMD as a part of the full-length protein to a greater extent upon mutations or the presence of substrates, nucleotides, or nucleotide analogs.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.