Recombinant yeast VDAC2: a comparison of electrophysiological features with the native form

Voltage‐dependent anion channel isoform 2 of the yeast Saccharomyces cerevisiae (yVDAC2) was believed for many years to be devoid of channel activity. Recently, we isolated yVDAC2 and showed that it exhibits channel‐forming activity in the planar lipid bilayer system when in its so‐called native form. Here, we describe an alternative strategy for yVDAC2 isolation, through heterologous expression in bacteria and refolding in vitro. Recombinant yVDAC2, like its native form, is able to form voltage‐dependent channels. However, some differences between native and recombinant yVDAC2 emerged in terms of voltage dependence and ion selectivity, suggesting that, in this specific case, the recombinant protein might be depleted of post‐translational modification(s) that occur in eukaryotic cells.

Voltage-dependent anion channel isoform 2 of the yeast Saccharomyces cerevisiae (yVDAC2) was believed for many years to be devoid of channel activity. Recently, we isolated yVDAC2 and showed that it exhibits channel-forming activity in the planar lipid bilayer system when in its so-called native form. Here, we describe an alternative strategy for yVDAC2 isolation, through heterologous expression in bacteria and refolding in vitro. Recombinant yVDAC2, like its native form, is able to form voltage-dependent channels. However, some differences between native and recombinant yVDAC2 emerged in terms of voltage dependence and ion selectivity, suggesting that, in this specific case, the recombinant protein might be depleted of post-translational modification(s) that occur in eukaryotic cells.
The voltage-dependent anion channel (VDAC) proteins, also known as mitochondrial porins, are the most abundant proteins in the mitochondrial outer membrane (MOM) of all eukaryotes. Characterized by a molecular mass of 30-32 kDa, VDAC proteins allow for the continuous exchange of metabolites (ATP/ ADP, NAD + /NADH, Krebs cycle intermediates) and ions (Na + , Cl À , Mg 2+ , Ca 2+ ) between mitochondria and cytosol. Therefore, they play a crucial role in mitochondrial bioenergetics [1][2][3]. In mammals, where three different but conserved isoforms are expressed (VDAC1, VDAC2 and VDAC3) [4], evolution has conferred to each isoform peculiar additional functions beyond the metabolic role. For example, VDAC1 and VDAC2 participate differently in the regulation of cell death [5][6][7][8], while recent literature suggests the involvement of VDAC3 in reactive oxygen species (ROS) homeostasis [9,10]. The three-dimensional structure of mammalian VDAC1 has been shown to involve a transmembrane b-barrel structure, formed by 19 amphipathic, mostly anti-parallel b-strands, with the exception of the N-terminal domain, which is structured as an a-helix and partially exposed to the cytosol [11][12][13]. A similar result was obtained for zebrafish VDAC2 [14] and for other species by homology modeling studies [15], suggesting a conserved structure of VDAC across evolution.
Differently from mammals, the budding yeast Saccharomyces cerevisiae has only two distinct VDAC genes, POR1 and POR2, encoding two different VDAC isoforms. However, only yVDAC1 is widely considered essential for mitochondrial metabolism. Indeed, the inactivation of the POR1 gene, as in the Dpor1 mutant strain, resulted in a strong inhibition of cell growth in non-fermentable conditions, especially at the restrictive temperature of 37°C [16]. On the contrary, the inactivation of the POR2 gene resulted in an undetectable phenotype, since Dpor2 cells were able to grow in a comparable way to the wild-type [16]. Overall, the channel-forming ability of yeast voltagedependent anion channel isoform 2 (yVDAC2) has been questioned for many years. Contrasting results were indeed obtained upon POR2 overexpression; namely, a complete recovery of Dpor1 yeast growth on glycerol was detected [16,17], suggesting a channel activity for yVDAC2, at least under particular conditions. Only recently, an electrophysiological analysis has clearly shown that yVDAC2 is able to form channels in artificial membranes, with features resembling many other members of the VDAC family [18].
In a previous work, we purified yVDAC2 directly from mitochondria extracted from Dpor1 cells overexpressing POR2, under native conditions [18]. In parallel, the same protein was prepared by heterologous expression in a bacterial system, purified and refolded as previously performed for many other VDAC isoforms and mutants [19][20][21][22]. In this work, we analyzed the electrophysiological features of the recombinant yVDAC2. Our results indicate that the protein is able to form channels in artificial membranes, with similar features to native one but, at the same time, with differences in terms of voltage sensitivity and ion selectivity.

Cloning of yVDAC2
The sequence encoding yVDAC2 was obtained by reverse transcription from the total RNA previously extracted from S. cerevisiae BY4742 (EUROSCARF, Frankfurt, Germany). Reverse transcription was performed by using the AffinityScript Multi-Temp RT and RT-PCR system (Agilent, Santa Clara, CA, USA) according to the manufacturer's protocol. The corresponding cDNA was amplified by PCR using the following primers: forward 5 0 -TTTTGCTAGCATGGCACTACGATTTTTCAACGAT-3 0 and reverse 5 0 -TTTTCTCGAGGGGCGAGAACGATA-GAGACCA-3 0 . The yVDAC2 cDNA was cloned in the bacterial expression vector pET-21b (Novagen, Madison, WI, USA) by NheI/XhoI digestion, in frame with the Histag at the C-terminal part. The construct was verified by sequencing.
Expression, purification and refolding of yVDAC2 protein Escherichia coli BL21 (DE3) cells were transformed with pET-21b plasmid containing the yVDAC2 sequence. Transformed cells were grown to an optical density (k = 600 nm) of 0.6, and the protein expression was induced by addition of 0.4 mM isopropyl-b-D-thiogalactopyranoside (IPTG) (Sigma-Aldrich, Saint Louis, MO, USA), for 3 h at 30°C. Cells were harvested by centrifugation and lysed in buffer B (8 M urea, 1 mM NaH 2 PO 4 , 0.01 mM Tris/HCl, pH 8.0) overnight at 4°C. The total protein lysate was clarified by centrifugation and loaded onto Ni-NTA agarose (Thermo Fisher Scientific, Waltham, MA, USA) packed column, previously equilibrated with buffer B. The column was then washed twice with five volumes of buffer C (8 M urea, phosphate buffer, pH 6.3), and proteins were purified by elution with five volumes of buffer E (8 M urea, phosphate buffer, pH 3.5). The eluted protein was refolded by incubation at 4°C in a 10-fold volume of refolding buffer [25 mM Tris, 100 mM NaCl, 1 mM EDTA, 1% (v/v) lauryldimethylamine oxide (LDAO, Sigma-Aldrich), pH 7.0]. Then, the protein solution was dialyzed against 100 volumes of a dialysis buffer (25 mM Tris, 1 mM EDTA, 0.1% LDAO, pH 7.0) with Thermo Scientific Slide-A-Lyzer Dialysis Cassettes (Thermo Fisher Scientific) (3.5 K MWCO). The solution containing the refolded proteins was clarified by centrifugation (60 000 g, 1 h) and loaded into a size exclusion chromatography column (Superdex 200 Increase 30/100, GE Healthcare, Chicago, IL, USA) in SEC buffer (150 mM NaCl, 20 mM Tris/ HCl, pH 8.0, 1 mM DTT 0.1% LDAO) to separate aggregates from monomers. Elution fractions, corresponding exclusively to the monomer peak, were collected.

Electrophysiological analysis of recombinant yVDAC2
The refolded yVDAC2 was reconstituted into a planar lipid bilayer (PLB) system as previously described in [20][21][22]. Bilayers of approximately 80-100 pF capacity made of asolectin from a soybean phospholipid mixture (Sigma-Aldrich) at a concentration of 20 mgÁmL À1 in n-decane were used. Channel insertion was obtained by addition of 0.5-5 lL of protein solution to the cis side of the cuvette containing approximately 3 mL of aqueous solution (1 M KCl, 10 mM Hepes, pH 7.0). Data were acquired using a Bilayer Clamp amplifier (Warner Instruments, Hamden, CT, USA) at 100 ls per point, filtered at 300 Hz and analyzed offline using the PCLAMP program set (version 10; Molecular Devices, San Jose, CA, USA). Channel conductance (G) was calculated from current (I) measurements in the presence of an applied constant voltage (V m ) of +10 mV, as the I/V m ratio. Distribution of conductance was obtained from six independent reconstitution experiments, each showing several channel insertions.

Voltage dependence analysis of yVDAC proteins
Analysis of voltage dependence of recombinant yVDAC2 was performed in 1 M KCl, 10 mM Hepes, pH 7.0, by application of two distinct voltage ramps including positive and negative potentials (amplitude AE90 mV, time 100 s). At least three independent multi-channel experiments were performed. Values of G and G max , indicating respectively the conductance value at a given V m and the maximal conductance, were calculated for the recombinant yVDAC2. Control experiments were performed using the native yVDAC1 and yVDAC2 produced as previously described [18]. The relative conductance was calculated as the G/G max ratio and plotted as a function of the voltage using PRISM 7 software (GraphPad Software Inc., La Jolla, CA, USA). Three independent experiments were performed for each protein and data are shown as the mean AE SEM.

Ion selectivity measurement of yVDAC2
Selectivity measurements were performed for the recombinant yVDAC2 in a 0.1 M/1 M cis/trans gradient of KCl in a voltage range of AE90 mV, by increasing V m with discrete steps of AE5 mV for 45 s. Values of conductance were plotted as a function of V m , and linear regression was applied with PRISM 7 software. The permeability ratio of cation K + (P C ) over anion Cl À (P A ) was calculated from the reversal potential (V r ) using the Goldman-Hodgkin-Katz equation, according to [23]. Six independent selectivity measurements were performed with both a single-and a multi-channel approach.

Reconstituted recombinant yVDAC2 forms channels of conductance similar to the native protein
To analyze the electrophysiological properties of recombinant yVDAC2, the corresponding encoding sequence was introduced into E. coli BL21 (DE3) cells and the protein expression was achieved by IPTG induction, as verified by SDS/PAGE (Fig. 1A). The recombinant protein was extracted directly from inclusion bodies and purified by nickel affinity chromatography, exploiting the 69His tag (Fig. 1B). Since the heterologous expression of membrane proteins in bacterial systems leads to loss of proper conformation and activity, the purified yVDAC2 was refolded and dialyzed. In order to avoid the presence of aggregates in the final protein solution, size exclusion chromatography was performed and the elution fractions, corresponding to the monomers (Fig. 1C), were collected and used in our study. Finally, the electrophysiological features of recombinant yVDAC2 were analyzed at the PLB.
The representative current trace displayed in Fig. 2A clearly shows that recombinant yVDAC2 was able to form channels in artificial membrane, since insertion of three different channels of similar size in a short time was obtained after the addition of the protein solution to the cis side of the cuvette. The histogram in Fig. 2B shows the distribution of conductance values calculated for n = 36 channels. The analysis, performed in 1 M KCl, 10 mM Hepes, pH 7, in the presence of an applied constant voltage of +10 mV, allowed for calculation of the average conductance, estimated as 3.76 AE 0.94 nS (SD, n = 36). This value perfectly overlaps our previous measurements showing an average conductance for the native yVDAC2 of about 3.6 nS in the same experimental conditions [18]. Overall, our results indicate that, in artificial membrane, recombinant yVDAC2 forms channels of conductance similar to the native protein.
Recombinant yVDAC2 is less sensitive to the applied voltage than the native protein It is known that the conductance of VDAC proteins changes depending on the applied voltage. This feature, well conserved through evolution from yeast to human, is known as voltage dependence. In general, yeast or mammalian VDAC1 persists at low potentials in a stable highly conductive state, called the 'open' state. However, as the applied voltage rises (starting from AE20-30 mV), VDAC1 switches to several lowconducting states, called 'closed' states [24][25][26][27]. We demonstrated previously that the native yVDAC2 shows a less pronounced voltage dependence in comparison to yVDAC1, because yVDAC2-formed channel closure began from AE40-50 mV of applied voltage [18].
In order to analyze the effect of voltage on recombinant yVDAC2 conductance, PLB experiments were performed by the application of voltage ramps from 0 to AE90 mV. The curves, displayed in Fig. 3A, indicate that recombinant yVDAC2 is completely insensible to low positive and negative voltages, since no closure was detected either at potentials that normally close yeast or mammalian VDAC1 or at voltages able to close native yVDAC2. Only the application of high voltages, starting from AE70-80 mV, induced a channel closure, which occurred symmetrically at both positive and negative potentials, suggesting that recombinant yVDAC2 is significantly less voltage sensitive than the native protein. Furthermore, current across recombinant yVDAC2 was monitored upon application of the constant voltage +90 mV. As shown in the representative trace in Fig. 3B, a complete closure of the channels occurred only after a prolonged exposure to high potential, highlighting the recombinant yVDAC2 limited sensitivity to voltage.
The electrophysiological behavior of the native protein was then analyzed in the presence of similar high voltages, by applying a voltage ramp from 0 to +90 mV. According to our previous data [18], native yVDAC2 began to close around +40-50 mV, as shown by the trace in Fig. 4A. However, the application of higher potentials (i.e. +80 mV) promoted a further channel closure, similarly to that previously shown by the recombinant  protein (see Fig. 3A). To deeply analyze this aspect, the values of conductance (G) obtained at each applied voltage for both native and recombinant proteins, as well as for the native yVDAC1, were normalized relative to the maximal conductance (G max ). Then, as shown in Fig. 4B, relative conductance values (G/G max ) were plotted as a function of the applied voltage (amplitude of AE90 mV). As expected, yVDAC1 conductance (pale green circles) significantly decreased starting from AE20-30 mV, reaching a stable level of closure characterized by a decrease at about 50% of the maximal conductance. Native yVDAC2 (orange circles) began to close around AE40-50 mV and its conductance at these voltages showed a decrease of about 20%, in a similar manner to that shown previously [18]. It is clear that this channel is not able to reach a channel closure similar to that observed for yVDAC1. However, the application of high voltages, up to AE80-90 mV, promoted a further 10-15% decrease of its relative conductance.
The results shown in Fig. 4B also confirmed the overall attenuation of voltage dependence in the case of recombinant yVDAC2 (red circles) observed in Fig. 3. Furthermore, the application of voltages up to AE60-70 mV did not promote any significant decrease in the channel conductance. Conversely, the application of potentials AE80-90 mV caused channel closure, as G/G max values almost overlap those calculated for both native yVDAC1 and yVDAC2 at similar voltages. Although the recombinant yVDAC2 still shows voltage dependence, our results suggest a decrease of voltage sensitivity even with respect to the native form.

Recombinant yVDAC2 displays exclusively cation-selective states
It is known that VDAC displays an anionic selectivity in the high-conducting 'open' state and a less anionic or more cationic selectivity in the low-conducting 'closed' state [28,29]. In the previous work, the presence of 1 and 0.1 M KCl in the cis and trans side of the cuvette, respectively, allowed for detection of the presence of different selectivity states, three in total for native yVDAC2. Two of them appeared to be highconducting states but with an opposite selectivity, while the third was a very low-conducting state characterized by an extremely pronounced cation selectivity [18]. In order to analyze the ion selectivity of recombinant yVDAC2, current values, measured in a 10-fold gradient of KCl and obtained upon application of discrete voltages in the range of AE90 mV, were plotted as a function of the applied voltage. Linear regressions of the obtained data (Fig. 5) indicated the presence of only two distinct states of recombinant yVDAC2, i.e. a main high-conducting state (called state 1 in Fig. 5) and a low-conducting state (state 2 in Fig. 5) and that both states displayed a cation selectivity.
The Goldman-Hodgkin-Katz equation was then used to calculate the ratio between permeability of cations over anions (P C /P A ) for the two states of recombinant yVDAC2, using a reversal potential corresponding to the voltage at zero current [23]. The P C / P A value obtained for state 1 was 1.94 AE 0.19 and the value obtained for state 2 was 11.2 AE 2.8. For comparison, the corresponding P C /P A values obtained for native yVDAC2 were 2.17 AE 0.11 and 16.24 AE 6, respectively [18]. This clearly indicates that recombinant and native yVDAC2 are very similar in terms of cationic strength. In particular, the lowest conducting state determined for both variants of yVDAC2 is characterized by a pronounced selectivity for cations over anions, a feature found exclusively for yVDAC2. At the same time, a significant difference between native and recombinant protein emerged, since no anionic state was found for the recombinant one.

Discussion
So far, the second VDAC isoform of the yeast S. cerevisiae has been considered to be one of the most elusive members of the VDAC family. Indeed, since the POR2 gene discovery, the story of yVDAC2 has been controversial. Differently from POR1, the genetic inactivation of POR2 has no consequence for yeast growth on non-fermentable carbon sources [16]. Furthermore, no data about yVDAC2 reconstitution in liposomes and/or artificial membranes were available for a long time, raising the suspicion (now proved wrong [18]) that the yVDAC2 protein was completely devoid of channel function. On the contrary, the overexpression of the POR2 gene in Dpor1 cells promoted the restoration of the yeast wild-type growth phenotype. This effect can be achieved when the POR2 sequence is transcribed under the control of a strong promoter [16] or is activated by an external factor, such as the serendipitously found human SOD1, which stimulates POR2 overexpression [17]. These findings in turn suggest that the contribution of yVDAC2 to the permeability of the yeast MOM is a consequence of its expression level, providing a logical explanation for the recovery of Dpor1 cell growth [16] and of the mitochondrial functionality [17]. To definitively clarify the functional properties of yVDAC2, we have recently analyzed the electrophysiological features of this protein reconstituted into artificial membranes by using two parallel approaches. As shown in our previous work, we isolated yVDAC2 directly from Dpor1 mitochondria, after yeast cell transformation with an expression plasmid carrying the corresponding yVDAC2 cDNA sequence. This strategy allowed us to increase POR2 transcript concentration and consequently the corresponding protein amount, which simplified the protein isolation under the native condition [18]. Alternatively, yVDAC2 was produced as a recombinant protein in E. coli cells, a procedure commonly used for large-scale protein production, since it allows a high yield of protein to be obtained in a relatively short time with minimal cost and effort. The same approach was previously used by our group for the characterization of human VDAC3 [20] and its cysteine mutants [21], and for the production of human VDAC1 suitable for interaction studies with proteins involved in neurodegenerative diseases [22,30]. Furthermore, an identical approach was used by Hiller and coauthors to obtain the human VDAC1 preparation for NMR studies aimed at identifying its threedimensional structure [11,31] and by Abramson's group to purify the mouse VDAC1, later crystalized [13], confirming the strength of this method. With this perspective, the yVDAC2 cDNA was cloned into a specific expression vector and the protein was expressed in bacterial cells. Since membrane proteins localize into inclusion bodies, bacterial lysis occurred upon denaturing conditions, with the loss of the native conformation. Therefore, once isolated, in order to obtain a perfectly active protein, the recombinant protein must undergo a refolding procedure using a wellestablished protocol [11,[19][20][21][22].
Recombinant yVDAC2 was then characterized by reconstitution in a PLB system and previous data were used as a reference. Our reconstitution experiments clearly indicate the ability of the recombinant protein to form channels in artificial membranes, with an average conductance value identical to that previously found for the native protein. This result confirmed also the appropriateness of our refolding method. The analyses of voltage dependence indicate that the recombinant yVDAC2 is less sensitive to voltage than native yVDAC1 and yVDAC2, although the voltage dependence of the latter is also weaker than that of the former. Accordingly, only application of a high potential is able to promote a strong attenuation (approximately halving) of relative conductance of native yVDAC2, allowing the protein to reach a similar closure level to that displayed by yVDAC1. Similarly, the recombinant yVDAC2 undergoes a complete and stable closure exclusively at high potentials, such as AE80-90 mV.
Another significant difference among native and recombinant yVDAC2 concerns ion selectivity. In our previous work, we found that native yVDAC2 displays the anion-selective state typical of most VDAC proteins, and two different cation-selective states [18]. Importantly, both cation-selective states are observed for recombinant yVDAC2 as indicated by values of P C /P A , but no anion-selective state was found for the recombinant protein. Although the presence of the highly conductive cationic state was already observed in VDAC1 extracted from Neurospora crassa or rat liver mitochondria [32], such a difference between native and recombinant VDAC1 from yeast or human has not been detected so far; rather many papers agree with the fact that both samples behave similarly upon PLB analysis in terms of conductance, voltage dependence and selectivity. Conversely, specific functional differences were noticed in the case of human VDAC3. The first electrophysiological characterization of VDAC3 was obtained by heterologous expression in bacteria, which produced proteins able to form pores in artificial membranes with a very low conductance (100 pS in 1 M KCl, channels about 35-40 times smaller than both VDAC1 and VDAC2) [20]. However, the work by Okazaki et al. showed that the recombinant VDAC3 could form channels with conductance similar to the other isoforms [33]. In this work, although the proteins were obtained with very similar protocols, the presence of reducing agents in preparation buffers and in PLB experiments led to the formation of channels of normal conductance. Moreover, channels of regular conductance were also obtained by using human VDAC3 directly extracted in the same conditions from yeast mitochondria [27]. At variance was that VDAC3 purified in the absence of reducing agents showed different electrophysiological features [21]. The reason for such differences was later attributed to the presence of specific post-translational modifications (PTMs), in particular to irreversible oxidation states of cysteine residues, as demonstrated by electrophysiological characterization in [21] and recently confirmed by high-resolution mass spectrometry [34,35].
We thus hypothesize that also the differences between native and recombinant yVDAC2 could be explained by naturally occurring PTMs. While the native protein is isolated directly from yeast mitochondria under native conditions, the recombinant protein is produced in a heterologous system lacking any eukaryotic PTMs. Beyond oxidation, VDAC proteins undergo phosphorylation of specific amino acid in both physiological and pathological conditions [36,37], which could be responsible for modulation of voltage sensitivity, as well as other electrophysiological parameters, as has been recently proposed [38,39]. Furthermore, it is known that modifications of amino acid residues change VDAC features. For instance, the treatment of VDAC1 with succinic anhydride abolishes the voltage dependence and changes its selectivity from an anionic to a cationic one [40].
In conclusion, the slight functional differences found in the specific case of S. cerevisiae VDAC2 between the native and recombinant proteins do not affect the biological significance of our findings, indicating that yVDAC2 is definitely another member of the VDAC family, being able to form voltage-dependent channel of about 3.6 nS conductance under our experimental conditions. The behavior of the recombinant protein supports our hypothesis of the presence of specific PTMs in the native one. As a further, intriguing consequence, our results indicate that such differences could be associated to the still unknown role of yVDAC2. Its encoding gene is under-expressed in normal conditions. It is not the first time, indeed, that an underexpressed isoform of VDAC was noticed, as was recently found in Drosophila melanogaster: in that case, a regulatory function has been associated to the alternative isoform [41].