C. Aflalo, Department of Life Sciences, Ben Gurion University, P.O. Box 653, Beer Sheva 84105, Israel. Fax: + 972 7 6472890, Tel.: + 972 7 6472118, E-mail: email@example.com
The association of rat brain hexokinase with heterologous recombinant yeast mitochondria harboring human porin (Yh) is comparable to that with rat liver mitochondria in terms of cation requirements, cooperativity in binding, and the effect of amphipathic compounds. Mg2+, which is required for hexokinase binding to all mitochondria, can be replaced by other cations. The efficiency of hexokinases, however, depends on the valence of hydrophilic cations, or the partition of hydrophobic cations in the membrane, implying that these act by reducing a prohibitive negative surface charge density on the outer membrane rather than fulfilling a specific structural requirement. Macromolecular crowding (using dextran) has dual effects. Dextran added in excess increases hexokinase binding to yeast mitochondria, according to the porin molecule they harbor. This effect, significant with wild-type yeast mitochondria, is only marginal with Yh as well as rat mitochondria. On the other hand, an increase in the number of hexokinase binding sites on mitochondria is also observed. This increase, moderate in wild-type organelles, is more pronounced with Yh. Finally, dextran, which has no effect on the modulation of hexokinase binding by cations, abolishes the inhibitory effect of amphipathic compounds. Thus, while hexokinase binding to mitochondria is predetermined by the porin molecule, the organization of the latter in the membrane plays a critical role as well, indicative that porin must associate with other mitochondrial components to form competent binding sites on the outer membrane.
yeast mitochondria harboring recombinant tagged human VDAC1
yeast mitochondria harboring recombinant native human VDAC1.
Mammalian hexokinase in brain (type I)  and muscle (type II)  reversibly associates with mitochondria; it is released in the presence of excess glucose 6-phosphate (Glc6P) . The association represents a classic example of a regulated dynamic rearrangement of catalytic components in cells according to metabolic need [1–5]; it allows the preferential access of hexokinase to mitochondrially generated ATP, as well as a steady ADP supply to respiring mitochondria. The outer mitochondrial membrane protein porin (or VDAC, voltage-dependent anion channel) has been identified as the receptor for hexokinase [6–8].
The association, assumed to be through molecular interaction between brain hexokinase and porin specifically , may involve other mitochondrial membrane components as well [3,5,6,10]. A short hydrophobic N-terminal peptide of hexokinase, which was identified as a major molecular determinant of the association, has been proposed to be inserted in the outer membrane as an early step .
The molecular requirements of the porin molecule are more difficult to delineate. The heterologous expression of foreign VDAC genes in yeast mitochondria lacking porin , originally designed for functional studies of VDAC as a channel, enabled a better assessment of the involvement of porin in brain hexokinase binding to isolated recombinant yeast mitochondria. The initial results indicated that the recombinant human VDAC1 performed better than the human VDAC2 isoform in this heterologous reconstituted binding system . The extension of these studies indicated a general propensity for brain hexokinase to associate with yeast mitochondria . Indeed, hexokinase binding to mitochondria is correlated with the similarity of the recombinant porin gene to that of the mammalian VDAC1 isoform. Accordingly, a preliminary classification of various porin molecules to support this heterologous association has been proposed . Nevertheless, a significant and saturable (specific) binding can be achieved  in the presence of dextran when used as a macromolecular crowding agent  in the incubation mixture. Macromolecules have also been proposed to induce structural changes in mammalian mitochondria that were shown to be favorable for hexokinase binding . Moreover, the general properties of the association inferred from the homologous system (rat brain hexokinase and either liver or brain mitochondria in the absence of dextran ) seem to characterize the purely heterologous systems (i.e. yeast mitochondria with wild-type yeast porin) as well when dextran is added . These include an apparent positive cooperativity in hexokinase binding, the absolute requirement for Mg2+ ions  and the N-terminal hydrophobic peptide of brain hexokinase . Finally, significant binding of rat hexokinase to yeast mitochondria harboring mammalian VDAC1 could be achieved in the absence of dextran .
The dependence of the binding on Mg2+ ions indicates direct electrostatic interactions between hexokinase and the surface of the outer membrane (phospholipids or other components) [3,17]. Mg2+ has been proposed to form a bridge between repulsive negative charges on both the enzyme and the outer mitochondrial membrane . The effective concentration range for Mg2+ in supporting binding in vitro being comparable with its intracellular concentration, it has been proposed to be the physiological effector [1,17]. However, in addition to electrostatic screening, Mg2+ may also be needed as a structurally specific bridge between hexokinase and its mitochondrial receptor. While monovalent cations act as releasing agents for bound hexokinase in the presence of Mg2+, a low pH seems to stabilize the interaction . Moreover, monovalent cations , polyamines  and Ca2+ by themselves may replace Mg2+ to promote binding in the homologous system. It thus seems that electrostatic interactions affect the association in variable ways, and different modes of binding have been proposed [1,3].
Finally, as the association occurs at a lipid–water interface, one may expect that amphipathic compounds would interfere with binding, as was demonstrated with the organization of cytosolic proteins upon the cytoskeleton , also inferable from basic hydrophobic interactions . Such components have recently been reported to influence the degree of hexokinase binding in vivo. The present study complements earlier reports characterizing the association in a yeast-based heterologous system as a potential reconstituted nonanimal model for intracellular organization of metabolic entities. Such a system is readily amenable to genetic engineering in terms of VDAC [9,12,24,25], as well as a heterologous localized enzymic probe for ATP, to reveal and study the heterogeneous character of the system [26–29], as well as the functional consequences of its occurrence [4,5,8,10,15,16,20]. We address here the mode of binding of hexokinase to recombinant yeast mitochondria from a physico-chemical point of view, in comparison with that occurring in the homologous system. Our purpose is to identify possible artifacts in binding introduced by the yeast mitochondrial environment, in preparation for a detailed catalytic study of the mitochondria–hexokinase coupled system. We specifically investigated the effects of macromolecular crowding agents and amphipathic molecules, and the alteration of the surface charge density on the outer mitochondrial membrane.
Leuconostoc Glc6P dehydrogenase as well as ATP, ADP, NAD+, dextran (average Mr = 40 000), hydrophobic cations and amphipathic drugs (see Fig. 1) were from commercial sources. The latter were used as concentrated solutions in ethanol.
Adult white rats (var. Sprague–Dawley) were the source of brains and livers for hexokinase (type I, rHK) and mitochondria (R+) preparations, respectively. Saccharomyces cerevisiae strains M3  and its por1– mutant M22–2  were used for wild type (Y+) and porinless (Y–) yeast mitochondria preparations, respectively. Yeast mitochondria harboring native human porin HVDAC1 (Yh) , or its engineered version with a short hemagglutinin tag at the C-terminus HVDAC1-HA (Yh′) , were isolated from the M22–2 strain expressing the engineered human porin gene, under control of the yeast porin promoter. Rat liver  and yeast  mitochondria were isolated according to published procedures; they were resuspended at 20 mg·mL−1 in isosmotic medium (0.3 m or 0.6 m sorbitol, respectively) containing 1 mg·mL−1 BSA and stored at −70 °C.
We used the bindable form of rHK isolated from brain particulate fraction and purified by affinity chromatography according to the procedure published by Wilson  with slight modification .
Binding of rat brain hexokinase to mitochondria
Yeast or rat liver mitochondria (2 mg·mL−1) were incubated on ice for 1 h with relatively limiting rHK (200–300 mU·mL−1) in 0.1 mL of isosmotic medium supplemented with Mg/Hepes 5 mm, pH 7.8 and 1 mg·mL−1 BSA. Dextran (25% w/v) was routinely added to yeast mitochondria incubations to enhance rHK binding . After separation of soluble and mitochondria fractions by centrifugation and solubilization of the mitochondrial pellet (2% Triton X-100) as described , aliquots were analyzed for free and bound hexokinase, respectively. Hexokinase binding is calculated as the percent activity in the solubilized bound fraction (routinely 70–85% maximal binding of limiting hexokinase), relative to the sum of the activities found in the bound and free fractions (typically 95–105% of the initially added activity). Activities assessed this way do not necessarily reflect the respective activities of free or bound hexokinase during the incubation with mitochondria (conducted in the absence of substrates).
Assay of hexokinase activity
Hexokinase activity was measured spectrophotometrically at room temperature by coupling NADH formation by Glc6P dehydrogenase to Glc6P production by hexokinase. NADH formation was measured at 340 nm, as described .
The protein concentration of the mitochondrial preparations was determined by the biuret method using BSA as standard . Protein in the hexokinase preparations was determined according to Bradford  using ovalbumin as standard.
RESULTS and DISCUSSION
The presence of a high concentration of Mg2+ was demonstrated to be crucial for the association of rHK to rat liver (R+) [3,17], and wild-type yeast mitochondria . The results presented in Fig. 2 show that Mg2+ is needed for the binding in all the mitochondrial preparations tested, independently of the presence or the origin of porin. However, both the binding ability and the requirement for Mg2+ vary with the mitochondria used. With mitochondria substantially depleted of endogenous cations, the relative inability of EDTA to decrease binding further in the absence of added Mg2+ indicates that any residual bound cation is ineffective by itself in promoting rHK binding. The experiment illustrated in Fig. 2 was done in the presence of dextran which has been reported to be required as a macromolecular crowding agent for rHK binding to yeast (Y+) but not mammalian (R+) mitochondria .
The effects of dextran, Mg2+ and the product Glc6P on the binding and retention of rHK to mitochondria lacking porin and recombinant yeast mitochondria harboring engineered (Yh′) or native (Yh) human VDAC1 were assessed (Fig. 3). The results indicate that while the addition of dextran significantly enhances the binding of rHK to Yh′ and Y–, as observed with most yeast mitochondria , it has little effect when native recombinant human porin is used in (Fig. 3A). On the other hand, no organelle-related or porin-related differential effect has been observed in relation to the known effects of Mg2+ (required) and Glc6P (lack of effect) on rHK binding, in both the homologous and heterologous systems .
Prebound rHK is readily released from mitochondria lacking porin resuspended in the absence of dextran (Fig. 3B), as was also observed with wild-type yeast mitochondria . However, in the presence of engineered or native mammalian porin, dextran has much less or no influence on the retention of rHK, respectively. The presence of native mammalian porin confers on yeast mitochondria a second notable characteristic. Prebound rHK is substantially retained in the presence of excess EDTA in the resuspension medium, as was found for mammalian mitochondria, but not wild-type yeast (Y+) or mitochondria lacking porin . This indicates that Mg2+ involved in rHK binding is readily removed by chelation from wild-type and yeast mitochondria lacking porin, while it seems to be rather buried in the case of mitochondria harboring native mammalian porin (in R+ and Yh mitochondria).
The retention of rHK by Y– and Yh′ is independent of Glc6P. However, the rHK product causes some release of the enzyme prebound to Yh mitochondria. Moreover, the removal of Mg2+ from the resuspension medium strengthens this differential effect of Glc6P on rHK retention. Again, this behavior is characteristic of the homologous system .
In summary, both the binding and retention of rHK to yeast mitochondria occurs in accordance with the presence and identity of the porin molecule. Indeed the effects of dextran, EDTA, Glc6P or combinations of these on rHK binding indicate that the best substitute for rat mitochondria in the homologous system is Yh, followed by Yh′ and Y– yeast mitochondria, in decreasing order. The current results indicate that alteration of the C-terminal but not the N-terminal  of mammalian porin are detrimental to rHK binding.
The results also confirm that the general capacity of mitochondria to bind rHK is not determined by the origin of the organelle as previously suggested , but rather involves a specific contribution of the porin molecule itself to rHK recognition .
Cation requirement for hexokinase binding in the homologous system
Being of metabolic significance in mammalian brain, kidney and muscle, the association of hexokinase with mitochondria should respond to the energetic status of the cell in general, and of the mitochondrion in particular. The surface charge density on the external face of the mitochondrial outer membrane may reflect changes in the inner membrane potential (for example, through contact sites), and thus represent a meaningful signal directly accessible to cytosolic catalytic systems. Assuming a specific interaction between rHK and a mitochondrial receptor, one may challenge the notion that Mg2+ represents an exclusive cofactor for the formation of a specific salt-bridge .
The effects of various ionic species in the place of Mg2+ on rHK binding to rat liver (R+) mitochondria was tested under otherwise low ionic strength conditions. The results presented in Fig. 4 indicate that Ca2+ is as competent as Mg2+ in supporting the binding. Moreover, two trivalent cationic species (spermidine, an organic polyamine and TEC, a bulky chemically inert chelate of Co3+) also fulfil the requirement observed for Mg2+, but much more efficiently. Similar conclusions have been reached from independent studies with polyamines  and Ca2+ in the homologous system. This suggests that the cations operate in screening an excess negative charge at the surface of the mitochondrial outer membrane, rather than fulfilling a specific requirement dictated by macromolecular recognition. This relation between valence and concentration, essentially similar to that observed for ionic strength of dilute solutions, is compliant with the formalism of the GouY–Chapman diffuse double electrostatic layer around colloidal particles derived from the Debye–Hückel theory . Briefly, the formalism states that a particle suspended in an electrolyte and bearing a net negative charge at its surface is surrounded by a counterbalancing excess of positive charge in the neighboring solution. This phenomenon has been invoked to explain the ionic requirement of membrane-bound enzyme activities  and for the reconstitution of peripheral membrane proteins , in which the effect of the counterion valence was similar to that observed with cations in our case.
In order to evaluate the possible involvement of electrostatic repulsion of rHK by the mitochondrial surface charge, the effects of two hydrophobic monovalent cations, CT+ and TPA+ (see Fig. 1), were also tested. Both cations are expected to spontaneously partition into the lipid phase of the outer membrane, and thereby directly titrate fixed negative charges. However, the two cations interact with membranes in fundamentally different ways. The cationic detergent (CT+) is incorporated so that the aliphatic hydrocarbon chain is inserted into the lipid phase and its positive charge must remain exposed at the hydrophilic cytoplasmic face of the membrane . In contrast, the charge on the arsonium in TPA+, stabilized in the lipid phase, is free to migrate electrophoretically between the external and internal faces of the outer membrane under the influence of the neighboring inner membrane potential [38,39]. While the outer membrane itself is commonly considered as macroscopically devoid of significant electrical potential, the electrophoretic migration of TPA+ can be pertinent at the scale of a contact site.
The addition of CT+ up to 30 neq·mg−1 mitochondrial protein results in a steep, quasi-linear binding of all the limiting hexokinase in the mixture. The addition of slightly higher amounts of detergent results in a dramatic decrease (broken line in Fig. 4) that may represent crossing a critical point at which the outer membrane is destabilized and eventually disrupted by excessively dense positive charges. While TPA+ also promotes rHK binding, it is much less effective (by about two orders of magnitude) than CT+. We propose that the lipophilic cation diffuses further to internal locations and it is therefore less efficient than the impermeable detergent in reducing the excess negative charge at the mitochondrion–medium interface.
Cation requirement for hexokinase binding in the heterologous systems
In view of the apparent importance of the surface charge density of mitochondria for rHK binding, one may expect to find species-related differences between mitochondria from yeast and rat with respect to the cation requirements for binding. Tables 1 and 2 summarize the results obtained with aqueous and lipophilic cations, respectively, in experiments similar to these in Fig. 4, using different mitochondria. The approximate numbers given for the half binding concentration (C0.5), represent useful estimates for comparing the effectiveness of the cations in each system.
Table 1. Effect of hydrophilic cations on rHK binding to recombinant yeast mitochondria. rHK binding experiments (one or two for each condition) were carried out as described in Fig. 4 with the indicated mitochondria, in the presence of 25% dextran as indicated. C0.5 represents the cation concentration yielding the median value between minimal and maximal binding, derived from graphical analysis of curves similar to those presented in Fig. 4. Bm is the normalized maximal binding expressed as a percentage of the value observed under standard conditions (control, 5 mm MgCl2) with the same mitochondria and hexokinase preparations on the same day. Sper3+, spermidine; cont, control.
Bm (% cont)
Bm (% cont)
Bm (% cont)
Bm (% cont)
a Expressed as the percentage of the control in the presence of dextran.
R+ + dex
R+ − dex
Yh − dex
Yh′ − dex
Yh′ + dex
Y+ + dex
Y– + dex
Table 2. Effect of lipophilic cations on rHK binding to recombinant yeast mitochondria. The binding experiments were carried out as described in Table 1 in the presence or absence of 25% dextran as indicated. The concentration of BSA in the incubation mixtures was reduced to 0.1 mg·mL−1. Cont, control.
Bm (% cont)
Bm (% cont)
R+ − dex
Yh′ − dex
Yh′ + dex
Y+ + dex
Y– + dex
In general, no essential differences between yeast and mammalian mitochondria were observed in their requirement for aqueous cations (Table 1) to achieve rHK binding. This apparent similarity extends to mitochondria lacking porins, to which rHK binding has been proposed to be nonspecific . An interesting peculiarity is that spermidine seems slightly more effective than TEC with yeast, as compared to rat, mitochondria. Such an influence of the spatial distribution of the positive charge on the cation may reflect a different surface charge density and/or distribution on the outer membrane in each species, at least in the microenvironment of porin. The importance of such a consideration for general surface adsorption has been recently demonstrated in an artificial model system .
Most values for maximal binding (Bm) measured in the presence of different cations are in agreement with those obtained under standard conditions (5 mm MgCl2, defined as the control). Still, the results indicate a notable increase in relative maximal rHK binding to all yeast mitochondria in the presence of saturating TEC3+ (Table 1), CT+ or to a lesser extent TPA+ (Table 2). In these cases, most of the limiting rHK added to the mixture was bound, including a variable low amount (10–30% of total rHK) in the presence of saturating Mg2+, conditions under which it is not normally able to bind (Figs 1–3). The latter is presumably due to the loss of its short N-terminal peptide during preparation or storage . This observation may suggest that the mode of binding in these cases is a nonspecific, electrostatically driven adsorption of rHK to mitochondria bearing an excess of positive charge at their surface.
However, control experiments indicate that a normally nonbinding species of rHK (lacking the hydrophobic N-terminal peptide) binds only poorly (up to 20%) to R+ and Yh′ mitochondria in the presence of CT+ (not shown). This dose-dependent binding occurs at CT+ concentrations that promoted overbinding of native rHK (see above, and Fig. 4). Thus, a moderate electrostatic attraction by itself cannot compensate for the loss of hydrophobic interactions in our binding assay. It also appears that in the presence of CT+ but not Mg2+, the presence of bound intact rHK seems to facilitate the binding of nonbindable rHK. These phenomena were not observed with yeast hexokinase, a nonbindable species , which did not bind at all.
As dextran is needed to achieve maximal binding of rHK to yeast mitochondria (see Fig. 3A), its effect on the efficiency of cations to promote binding was also assessed. While dextran moderately increased the binding with rat and recombinant yeast mitochondria (Yh′), it had little or no influence on the requirement for Ca2+ (Table 1), CT+ or TPA+ (Table 2), as assessed by similar values of C0.5. This result indicates that while macromolecular crowding promotes the interaction between rHK and heterologous mitochondria (Y–, Y+ and Yh′), it has no net effect on the modulation of rHK binding by cations.
Binding of brain hexokinase to yeast mitochondria harboring human porin
In order to assess the contribution of porin in the interaction between mitochondria and rHK, the ability of yeast mitochondria harboring native human porin (Yh) to bind rHK was further investigated in the presence and the absence of dextran. The results of representative binding experiments are summarized in Fig. 5.
The incubation of a limiting amount of rHK with increasing concentrations of Yh mitochondria results in binding displaying monophasic saturation behavior (Fig. 5A). As expected from the results in Fig. 3A, the presence of dextran does not affect the saturation value (binding percentage to mitochondria in excess). However, the apparent affinity assessed by this criterion is increased under macromolecular crowding conditions, as previously reported, when using recombinant yeast mitochondria harboring other porin molecules [13,14]. In order to determine the binding capacity of Yh mitochondria for rHK, the reciprocal experiment was done: a fixed limiting amount of mitochondria was incubated with increasing concentrations of rHK in the presence or absence of dextran. The results plotted as bound rHK versus free rHK (Fig. 5B) indicate that Yh mitochondria are saturated with rHK in both conditions. Hill analysis of the binding curve yielded a biphasic behavior similar to that reported and discussed for the purely homologous and heterologous systems , and indicates positive cooperativity in binding (Fig. 5B, inset). The addition of dextran results in a mild enhancement of the cooperativity, an increase in the apparent affinity to rHK and, most remarkably, a strong increase in the saturation value (Bmax). These effects of dextran are intermediate between those observed  with the homologous system (R+) and a purely heterologous system (Y+). The effect of dextran on the cooperativity and the apparent affinity with wild-type yeast mitochondria is attributable to a poor recognition a priori between rHK and the heterologous porin, in contrast with the optimal homologous system. In the case of binding to yeast mitochondria harboring native mammalian porin, the effect of dextran bears predominantly on Bmax, representing the number of competent rHK binding sites. Because the number of porin molecules is independent of the presence of dextran, macromolecular crowding must cause their reorganization in the membrane, resulting in the assembly of new competent binding sites. A similar proposal has been put forward by Brdiczka's group , stating that dextran promotes the formation of contact sites between the inner and outer mitochondrial membranes, which are enriched in porin molecules. Based on typical porin content (4–5 µg·mg−1 protein) of recombinant yeast mitochondria (H. Azoulay-Zohar and C. Aflalo, unpublished observations), a rHK : porin binding stoichiometry (in the presence of dextran) close to a consensus maximum of 4 [1,10] can be calculated. Such a value, much higher than that calculated for binding to rat liver (1.7) or wild-type yeast (1.4) mitochondria under similar conditions and porin content , suggests that heterologous porin has a higher tendency to localize in contact sites of the host mitochondria under macromolecular crowding conditions. On the other hand, purified porin reconstituted in phospholipid vesicles bind hexokinase less efficiently than in mitochondria . We propose that the interaction of porin with other mitochondrial proteins represents a necessary condition for efficient rHK binding. One can predict that impairing these interactions will result in less efficient binding.
Effect of amphipathic compounds
Clotrimazole and bifonazole were reported to induce detachment of bound HK from mammalian mitochondria in vivo. As amphipathic compounds, they should impair interactions between membrane proteins. Their effects were assessed in our homologous and heterologous systems in vitro.
The results presented in Fig. 6 indicate that rHK binding to yeast (Yh) as well as rat (R+) mitochondria is significantly perturbed in the presence of both these compounds. This suggests that the binding sites for rHK are included in a large set of amphipathic loci able to bind clotrimazole and bifonazole. The compounds at high concentrations may successfully compete with rHK binding sites, resulting in a decrease in the hexokinase binding. It should be noted that under the conditions we used (excess mitochondria), the effective concentration range of the drugs is underestimated. However, once the enzyme is bound to the membrane the association can not be reversed by either compound, indicating that the amphipathic drugs are no longer accessible to a properly assembled rHK binding site occupied by its ligand. This result, along with a relatively strong retention of bound rHK upon resuspension, especially in the presence of EDTA (see Fig. 3B), rules out the possibility of a simple, and reversible binding mechanism.
When a similar experiment was done in the presence of dextran (Fig. 7) a strong attenuation of the inhibitory effect of clotrimazole was observed. Again, this indicates a loss of accessibility of rHK binding sites to the drug in the presence of dextran. This result, along with the lack of effect on retention, suggests that dextran induces a major reorganization of the membrane components, thereby suppressing the inhibitory effect of the amphipathic drug.
We show that rat brain hexokinase binding to yeast recombinant mitochondria harboring mammalian VDAC1 does not differ from that in the homologous system. This supports a major contribution by the porin molecule for hexokinase recognition and only a minor contribution by other components of the heterologous organelle. In this quasi-homologous system, the extent of hexokinase binding is enhanced in the presence of dextran due to favorable membrane reorganization rather than a significant increase in the apparent affinity between the soluble enzyme and porin. Neutralization of negative charges at the yeast mitochondrial surface represents a condition necessary, but not sufficient, for rHK binding, as with wild-type rat mitochondria. Amphipathic molecules impair the process of binding in both the systems. However, the action of amphipathic compounds, Mg2+ chelation or mere dilution did not effect hexokinase release, as was observed in the purely heterologous systems. Thus the hexokinase/mammalian-porin/yeast mitochondria system represents a reliable model for further biochemical and molecular studies of not only hexokinase–mitochondria, but soluble enzyme–membrane interactions as well as their functional implications.
We wish to thank M. Forte for the gift of the yeast strains, as well as J. E. Wilson for advice and the gift of antibodies against rHK. This research was supported by grant no. 95-110 from the US–Israel Binational Science Foundation (BSF). C.A. gratefully acknowledges the partial support from the Black Center for Bioenergetics at BGU.