INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

Mechanosensation is an essential and diverse type of sensory transduction that is widely spread in living cells belonging to organisms of various phylogenetic origin. Mechanosensitive (MS) ion channels have been thought to be the primary molecular biosensors that may function as mechanoelectrical switches at the basis of mechanosensation in such diverse physiological processes as touch, hearing, proprioception, or embryogenesis, as well as turgor control in plant cells and osmoregulation in bacteria (Sachs, 1992; Martinac, 1993; Sackin, 1995; García-Anoverños and Corey, 1997). The ubiquity of MS channels further supports the notion of an important physiological role for this type of channels in these cellular processes (Sachs, 1988, 1992; Morris, 1990; Martinac et al., 1992; Martinac, 1993; Sackin, 1995; Hamill and McBride, 1996).

The MS channels have been extensively studied in both Gram-negative and Gram-positive bacteria (Martinac et al., 1992). Three types of MS channels have been documented in Escherichia coli: 1) Mechanosensitive channel of Large conductance (MscL), 2) Mechanosensitive channel of Small conductance (MscS), and 3) Mechanosensitive channel of Mini conductance (MscM) (Martinac et al., 1987, 1992; Sukharev et al., 1993, 1994a, 1997; Berrier et al., 1996). In particular, since it is the only MS ion channel with the known primary amino acid sequence and the corresponding gene, mscL (Sukharev et al., 1994a,b; Hamill and McBride, 1994) encoding the channel protein whose mechanosensitivity has been unambiguously documented (García-Anoverños and Corey, 1997), the MscL has been well characterized at the molecular level.

To understand the MscL mechanosensitivity, structure and function relationship of the wild-type and various recombinant MscL mutants have been studied by the patch clamp technique (Hamill et al., 1981) using both in situ (Blount et al., 1996a,b) and in vitro preparations (Häse et al., 1995, 1997). Deletion or substitution of the first eight amino acids in the N-terminus (Häse et al., 1997) or deletion of 12 initial N-terminal residues (Blount et al., 1996a) resulted in channels exhibiting altered gating and pressure sensitivity. Site-directed N-terminal mutations in the G14 residue also resulted in channels with altered gating and pressure sensitivity. Moreover, a deletion of the G14 residue caused a complete loss of mechanosensitivity of MscL, since the activity of these mutant channels was independent of the applied pressure (Liu, Gu, Deitmer and Martinac, in preparation). These results indicated the importance of the N-terminus, and of the G14 residue in particular, for gating and pressure sensitivity of MscL. A deletion of 27 amino acids for the C-terminus did not affect the function of the MscL channels, whereas a deletion of 33 C-terminal amino acids that included a charged cluster of five amino acids (RKKEE) abolished the channel activity (Blount et al., 1996b; Häse et al., 1997). This result indicated a particular importance of this group of charged amino acids for channel activity. Also, site-directed mutagenesis of other charged and polar residues present in the MscL amino acid sequence, such as the lysine K31 of the first transmembrane helix M1 or the glutamine Q56 of the periplasmic S2-S3 loop, revealed the overall importance of charged residues for the mechanosensitivity of the MscL (Blount et al., 1996b,c).

How the MscL is operated by the mechanical force transmitted exclusively via membrane lipid bilayer remains poorly understood. In the present study, we propose the electromechanical coupling (EMC) model of MscL gating. By emphasizing the significance of the N- and C-termini for channel gating as well as the importance of the attractive and repulsive coulomb forces between the various MscL structural domains, the model suffices to explain most of the pressure-sensitive behavior of the wild-type and several mutants of the MscL that have been investigated to date.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Model prerequisites

The EMC model is based on several assumptions, proposed MscL tertiary structure, and presently available experimental evidence that all can be summarized as follows:

1.	The MscL protein alone is necessary and sufficient for the activity of the large conductance MS ion channel of E. coli (Sukharev et al., 1994a);
2.	The membrane tension  that gates the MscL can be calculated for an ideal biological membrane described by lipid bilayer alone, since the MscL remains fully functional upon reconstitution into liposomes (Häse et al., 1995);
3.	The proposed membrane spanning model of the MscL monomer consists of five domains that include the S1 amphipathic N-terminal domain, two membrane-spanning -helical domains M1 and M2, another amphipathic S2-S3 domain, and a hydrophilic C-terminal domain (Fig. 1) (Blount et al., 1996a; Sukharev et al., 1996, 1997); consequently the MscL belongs to a new family of structurally related ion channels having two membrane spanning -helices (North, 1996);
4.	The functional MscL channel is a homohexamer (Blount et al., 1996a; Sukharev et al., 1996, 1997) with a pore size of ~40 Å (Cruickshank et al., 1997);
5.	All 12 -helices of the MscL homohexamer line the pore of the channel (Cruickshank et al., 1997);
6.	Deletions and amino acid substitutions in the N-terminal domain strongly affect the channel pressure sensitivity and gating properties (Blount et al., 1996a; Häse et al., 1997);
7.	Deletions in the C-terminal domain region affect gating properties of the channel to a lesser extent, except when a deletion included a charged group of five amino acids (RKKEE) that resulted in complete abolishment of channel activity (Blount et al., 1996a; Häse et al., 1997);
8.	Site-directed mutations that affect the overall net electric charge of any of the channel domains (Blount et al., 1996b), may cause changes in the pressure sensitivity as well as channel gating kinetics of the MscL;
9.	Structural flexibility of the N-terminal domain may be provided by the glycine G14 located at the interface between the N-terminus and M1 -helix, since glycine residues may exhibit many different conformations in various unfolded protein structures (Branden and Tooze, 1991), such as the putative link between the N-terminus and the M1 helix (Fig. 1);
10.	Attractive and repulsive electrostatic coulomb forces exist between various domains of the channel.

View larger version (31K): [in this window] [in a new window]   FIGURE 1   Structure of the MscL monomer. (A) A working topological model of the MscL monomer. The N-terminus (S1 domain) is proposed to form a cytoplasmic amphipathic -helix; M1 and M2 domains represent the hydrophobic transmembrane -helices. A glycine (Gly14, black circle), which is located near the N-terminus and M1 domain interface, is assumed to serve as a hinge in the proposed EMC gating model. (B) Amino acids of the N-terminus and M1 and M2 domains are arranged in an -helical wheel of ~3.9 turns for the N-terminus and 7.4 turns for each of the M1 and M2 helices. Left sides in the N-terminus and M2 helix, and the right side in the M1 helix (dashed line) are more hydrophilic than the opposite side. Charged amino acids are marked by superscripts. The aspartate D67 (underlined) in M2 is the charged amino acid of the upper neighboring turn of the -helix.

In the proposed EMC gating model the strategically positioned equivalent net charges within each single domain of MscL are considered to be the source of coulomb forces responsible for conformational changes underlying closed-open transitions of the channel when the membrane is stretched. From the helical wheel representation of M1 and M2 -helices (Fig. 1 B), one side of the -helix (right side of the M1 domain and left of the M2 domain relative to the dashed line in Fig. 1 B) is more charged and therefore possibly overall more hydrophilic than the other side of the helix. The hydrophilic side may be envisioned as facing the aqueous phase inside the pore of the functional MscL channel that was proposed to be a homohexamer (Sukharev et al., 1996, 1997). Similarly, the N-terminus has amphipathic properties by being hydrophilic on one side and less hydrophilic on the other, as shown in its helical wheel representation (Fig. 1 B). According to the EMC model proposed in this study, the hydrophilic side of N-terminus faces the aqueous extramembranous environment and the less hydrophilic side is oriented toward the pore (Fig. 2 B). Although the water in the large MscL pore may be expected to be equivalent to bulk water, this orientation of the N-terminus is reasonable to assume, since in this orientation arginine R8 provides the N-terminus with a net positive charge. Consequently, the less hydrophilic side can be kept in a position parallel to the membrane bilayer by attractive electrostatic forces inside the channel pore, thus keeping the channel closed.

View larger version (43K): [in this window] [in a new window]   FIGURE 2   Model comparison. (A) A model in which the simultaneous "en bloc" displacement of all six monomers upon membrane stretch is responsible for the opening of the channel pore (Sukharev et al., 1997). (B) The ECM model proposes that a swing-like movement of the N-terminus is the major component responsible for the opening of the channel. (C) Net equivalent charges and their positions used for calculations in the ECM model: 2r = 40 Å, a = 11.25 Å, b = 2 Å, h = 40 Å, hM1 = 4.5 Å, hM2 = 18.75 Å (see Table 1). Only three MscL subunits are shown for clarity. The electrostatic forces acting on the N-terminus are illustrated for one subunit of the MscL. To illustrate the electrostatic forces clearly, N- or C-termini are represented by dashed lines in the closed (C) and open (D) channel configuration.

Another assumption required by the EMC model is the rotational flexibility of the N-terminus. According to the membrane spanning model of the MscL (Fig. 1) the N-terminus is linked to the M1 domain through the glycine residue G14. Generally, glycine residues provide proteins with flexibility, since they may exhibit many different conformations in various unfolded protein structures (Branden and Tooze, 1991). Such may be the putative link between the N-terminus and the M1 helix. Also, although possibly a gross oversimplification, the C-terminus may or may not move from the closed position when the channel opens (Fig. 2 B).

Calculation of equivalent net charges of single domains of the MscL monomer

The MscL monomer consists of 136 amino acid residues deduced from its gene (Sukharev et al., 1994a). Hydropathy analysis revealed a highly hydrophobic protein with an amphipathic N-terminus (residues 1-15), followed by a highly hydrophobic segment (19-38), an amphipathic segment (50-69), a second highly hydrophobic segment (70-96), and a hydrophilic C-terminus (97-136) containing a cluster of charged residues RKKEE (104-108) (Sukharev et al., 1997). Secondary structure analysis (Arkin et al., 1997) together with the PhoA-fusion method analysis (Blount et al., 1996a) led to a working membrane spanning model of the MscL monomer comprising five structural domains denoted as S1, M1, S2-S3, M2, and C domain (Fig. 1 A) (Sukharev et al., 1996, 1997; Blount et al., 1996a). For the purposes of the EMC model we calculated the overall electric charge of each of the five domains by representing each of the S1, M1, M2, and S3 domain by a helical wheel (Fig. 1 B), and assuming no particular secondary structure for S2 and C domain. Also, we estimated the relative positions of these charges within each domain of the membrane-spanning MscL model (Fig. 2, C and D) by taking into account that an -helix has an interturn distance of 5.4 Å corresponding to an advancement of ~1.5 Å per amino acid residue along the helix (Sybesma, 1977; Geoffrey et al., 1988). The calculations of these model parameters are summarized in Table 1.

Using the helical wheel presentations in Fig. 1 B we estimated the net charges within the MscL monomer to be +1e and 1e for N-terminus and M1 domain, respectively. The M1 domain, S2-S3 loop, and M2 domain most likely follow the general helix-loop-helix model (Sukharev et al., 1997). Consequently, the aspartate D67 in the S3 domain may electrically neutralize the histidine H74 in the M2 domain, as they are in a close apposition to each other in the neighboring helices (Fig. 1 B). In that case the net charge of the M2 domain is 1e, and the net charge of the S2-S3 loop is zero. Otherwise, the net charge of M2 will be zero, and 1e will be the net charge of the loop domain. Its equivalent position would then be determined by the aspartate D67, as the other oppositely charged amino acids are close to each other according to their -helical structure. Therefore, we calculated the electrostatic forces by considering both possibilities. In addition, it is not known how deep the S2-S3 loop may protrude into the channel pore inside the membrane bilayer. Thus, when the loop was considered for calculation, two possibilities were taken into account for the position of the loop S2 and the helix S3 relative to the membrane surface (see Table 3). Also, to simplify the calculation, the aspartate D67 was assumed to be equally distant from M1 and M2 domains, so that the horizontal component of the electrostatic force it exerts on the N- or C-terminus could be ignored. The C-terminus is highly hydrophilic and most likely located intracellularly (Fig. 1). Deletion of amino acid residues of 110-136 had no effect on channel function and gating (Blount et al., 1996b; Häse et al., 1997). However, the charged residues R104, K105, K106, E107, and E108 were found to be very important for channel function, because a deletion of residues from 104 on in the 104 C-terminal deletion mutant abolished the channel activity (Blount et al., 1996b; Häse et al., 1997). Therefore, we only considered the residues 96-110 to calculate the net equivalent electric charge of +3e for the C-terminus (Table 1).

Electrostatic force analysis

Table 1 lists the parameters used in the model calculations. See also Fig. 3, A and B.

View larger version (17K): [in this window] [in a new window]   FIGURE 3   Electrostatic force analysis. (A) Top view of the coulomb forces and their corresponding horizontal components acting upon a single N-terminus of the MscL homohexamer in the plane of the membrane. (B) Cross-sectional view of the coulomb forces and their corresponding vertical components acting upon a single N-terminus.

In the closed state, the main electrostatic forces acting on one N-terminus come from the other five N-termini, six C-termini, six M1 domains, and six M2 or S2-S3 (loop) domains in a closed channel. As shown in Results, in the case that the net charge of 1e for M2 is considered, the effect of S2-S3 can be neglected, while in the case that the contribution of S2-S3 is taken into account, the net charge of M2 can be neglected.

The sum of horizontal coulomb forces F_H is composed of each of the forces originating in other N-termini (F_HN), C-termini (F_HC), M1 (F_HM1), and M2 (F_HM2): In the case that the net charge of M2 is zero, and S2-S3 net charge of 1e is considered, the horizontal component of S2-S3 to the F_H can be ignored, so that it follows: The sum of vertical components of coulomb forces F_V derives from the contribution of M1 (F_VM1), and M2 (F_VM2), or S2-S3 (F_loop): or All the electrostatic components are calculated in the following section.

Fig. 3 A shows the plane view of electrostatic interactions within the MscL homohexemeric pore. F_HN, the sum of the coulomb forces originating in five other N-termini and acting on each N-terminus, is given as: where where l_Nⁱ denotes the distance between N₁ and N₂ and _Nⁱ denotes the angle between the coulomb force and its horizontal component f_HNⁱ. All five horizontal components are described as follows: The total contribution of six C-termini to the horizontally oriented forces acting on each N-terminus is: where Each component can be written as Because N- and C-termini are both net positively charged, the total horizontal repulsive force between them drives the N-terminal domains out of the channel pore.

Fig. 3 B shows the electrostatic field analysis along the vertical axis of M1 and M2 domains. Each N-terminus (positively charged) is electrostatically attracted by the resultant force F_M1M2 originating in six net negatively charged M1 and six M2 transmembrane domains. F_M1M2 can be decomposed into horizontal (along the membrane surface) and vertical (membrane orientation) components: F_HM, the total contribution of horizontal components originating from M1 and M2 transmembrane domains, and F_VM, the total contribution of vertical components originating from M1 and M2 domains, are given by where f_HM1ⁱ and f_HM2ⁱ are the individual horizontal components of electrostatic forces between N-termini and M1 and M2 domains. and and where f_VM1ⁱ and f_VM2ⁱ are the individual vertical components of electrostatic forces between N-termini and M1 and M2 domains. and The attractive coulomb force between the loop S2-S3, F_loop, and the N-terminus is determined as follows: The maximal horizontal component of the attractive coulomb forces between the S2-S3 region and one N-terminus would be equal to 0.004 e²/4₀ if S2-S3 domains are positioned deep into the channel pore closest to the N-termini. In that extreme case the strength of the horizontal components of the coulomb forces (0.004 e²/4₀) is an order of magnitude smaller than the strength of the vertical components (0.165 e²/4₀  0.168 e²/4₀) (Table 2). Therefore, the contribution of the horizontal electrostatic components was ignored for further model considerations.

View this table: [in this window] [in a new window]   TABLE 2   Calculated horizontal and vertical electrostatic components (in e2/40)

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Coulomb forces governing the MscL mode of operation

The major events that may be occurring when the channel undergoes the closed-to-open transition can be briefly summarized as follows. In the model the channel pore of 40 Å is assumed to be formed by 12 -helices of the channel homohexamer (Cruickshank et al., 1997) and has approximately the same diameter in both closed and open channel configuration. Glycine residue G14 is supposed to function as a hinge around which the N-terminus can rotate out of the pore within certain range of space angles. Although the C-terminus could be moving too, the N-terminus is more likely to undergo a swing-like movement than the C-terminus, due to the rotational flexibility of the glycine G14. Therefore, for the sake of simplicity we calculated the resulting coulomb forces as acting exclusively on the N-terminus. All parameter calculations in this study were based on this simplifying assumption. Acting as a gating arm, the N-terminus is supposed to be positioned parallel to the membrane when the channel is closed (Fig. 4). In this conformation, both N- and C-termini form six gating pairs to keep the channel closed. When pressure is applied and the membrane is stretched, the six N-termini of the homohexamer are forced out of the channel pore by swinging around their glycines G14. This simultaneous outward movement of six N-termini leads to the "unplugging" of the channel pore providing for ions to flow down their electrochemical gradients.

View larger version (22K): [in this window] [in a new window]   FIGURE 4   Electromechanical coupling mechanism. (A) No external pressure is applied. The resultant electrostatic force FR is oriented in such a way that it encloses a very small angle  ~ 3.8° with the gate plane formed by N-termini. The major effect of this force is to keep the gate closed; only two monomers of the MscL hexamer are shown for clarity. (B) External pressure is applied. The induced membrane tension  pulls the monomers apart. The asymmetry of the charge distribution at the bottom and the top end of the channel hexamer causes them to tilt slightly at an angle . As a consequence, the N-terminus will follow the M1 tilt at the same angle with a result that the angle between the N-terminus and the resultant coulomb force FR becomes   . (C) The increase in membrane tension pulls the single monomers apart at the angle  = . The orientation of the N-terminus and the resulting electrostatic force FR are in the same direction. FR pushes the N-terminus outward to rotate around G14 (shaded circle) out of the channel pore. The channel undergoes a transition from the closed to the open state. In this situation Po = 0.5. (D) Larger membrane tension will cause tilting of the monomers at angles  > . FR is more likely to induce the closed-open channel transition, such that open probability increases (Po > 0.5).

In the closed state, the electrostatic forces acting on one of the six N-termini come from the other five N-termini, six C-termini, and six of each M1 and M2 domains (see Methods). Since the N- and C-termini are both positively charged and are assumed not to be compressed by the repulsive electrostatic forces they exert on each other in the membrane plane, the resulting effect of the total repulsive force acting between them is to push the N-termini out of the membrane plane and consequently unplug the channel when the membrane is stretched (see next section). The horizontal repulsive force resulting from the other five N-termini is designated F_HN. The total repulsive force acting on one N-terminus that results from six C-termini is designated F_HC. Each N-terminus with one equivalent positive charge is electrostatically attracted by six of each negatively charged M1 and M2 helices. The corresponding resulting attractive force is designated F_M1M2. By constructing a parallelogram of coulomb forces, F_M1M2 can be presented as a sum of a horizontal component, F_HM, along the membrane surface and a vertical component, F_VM, orthogonal to the membrane plane: The horizontal resultant coulomb force acting on one N-terminus is therefore Another component, designated F_loop, that contributes to the electrostatic forces in a direction orthogonal to the membrane plane, originates from the net charge of six S2-S3 loops linking M1 and M2 domains. Thus, the sum of the orthogonal forces acting on one N-terminus is The values of all electrostatic components (F_HN, F_HC, F_HM, F_VM, and F_loop), and resulting horizontal (F_H) and vertical (F_V) forces (Table 2) were calculated using the parameters obtained from the working model of the MscL (Fig. 2, C and D). Details of the analysis are described in Methods.

Taking into account the results from Table 2 we can consider three different possibilities:

First, the equivalent net charge of M2 is 1e and is positioned in the middle of M2, whereas the net charge of the S2-S3 loop is considered to be zero. In this case, the comparison of the relative strength of all electrostatic components indicates that of the three horizontal repulsive coulomb forces, the major repulsion comes from the electrostatic contribution of the N-termini (F_HN) and C-termini (F_HC). In comparison, both the horizontal (F_HM) and vertical (F_VM) attractive components coming from M1 and M2 are, in comparison, much weaker. The directions of F_H, F_V, and their resultant F_R are demonstrated in Fig. 4. The calculated angle  between F_H and F_R is very small: Second, the equivalent net charge of M2 is zero. The electrostatic attraction of M2 is therefore negligible. Instead, there is a net charge of 1e in the S2-S3 domains. We may assume a case in which the loop region enters halfway into the channel pore formed by transmembrane helices. In this situation the negative charge of the aspartate D67 will be located near the entrance of the channel pore and the resulting vertical electrostatic force will increase because of the shorter distance between the equivalent net charges of N-termini and the S2-S3 loop. As shown in the second raw b in Table 2 the angle  enclosed by F_H and F_R is in this case 4.1°, slightly larger than in the previous case (3.8°).

Third, the equivalent net charge of M2 is zero, the same as the former case, but the S2-S3 loops are assumed not to enter into the pore region. In this situation, in contrast to the previous case, the resulting vertical electrostatic force decreases due to the greater distance between the N-termini and the loops. The third raw c in Table 2 shows the calculation for this consideration. The angle  enclosed by F_H and F_R is 3.0°, slightly smaller than in the previous cases.

A comparison of the relative strengths of all electrostatic components points toward an important result. In all cases considered, the angle  between F_H and F_R varies within a very narrow range (~3.0°-4.1°). This result presents the basis for the EMC model of the MscL gating by mechanical force.

Tension-triggered electromechanical coupling mechanism

Fig. 2 B shows the diagram depicting the basic idea of the proposed EMC gating model. In the diagram the MscL channel pore with a diameter of 40 Å (Cruickshank et al., 1997) does not change significantly in size when the channel is closed or open. The swing-like movement of the N-terminus is assumed to dominate the gating process of the channel and should correspond to the open-close transition of the channel. In succession, the inclination of the single channel monomers caused by membrane tension (Fig. 4) acts as a "trigger" to initiate the N-terminal swing-like movement. The electrostatic coulomb forces existing between the single structural domains of the channel monomers carry sufficient energy to keep the channel closed, as well as to cause the swing-like movement of the N-termini. The net equivalent charges and their positions on the M1 and M2 domain, and N- and C-termini of the MscL homohexamer, are depicted in Fig. 2, C and D. Their corresponding values are given in Table 1.

Based on the distribution of coulomb forces in the closed state of the channel illustrated in Fig. 2, C and D, the MscL opening mechanism can be deduced as shown in Fig. 4. When no external pressure is applied to the patch pipette, the lipid membrane remains unstretched and forms an ideal planar lipid bilayer (Sokabe and Sachs, 1990; Sokabe et al., 1991). Twelve transmembrane helices (six M1 and six M2 -helical domains) of the six MscL monomers form the 40-Å channel pore (Cruickshank et al., 1997). The pore is shut by the combined gate of N- and C-termini, both of which are assumed to lie horizontally within the pore parallel to the patch membrane. The resultant electrostatic force F_R is oriented in such a way that it encloses a very small angle  with the gate plane formed by N- and C-termini. The resulting effect of this force is to keep the gate closed (Fig. 4 A).

When the negative pressure is applied to the pipette, the area of the membrane patch becomes enlarged. The increase in membrane tension provides a stimulus to pull the channel monomers away from each other, such that they become inclined toward the membrane plane in the tension direction. The angle at which they become tilted relative to each other is  (Fig. 4). Consequently, the N-terminus will also become tilted at the same angle with a result that the angle between a single N-terminus and the resulting coulomb force F_R becomes    (Fig. 4 B). The hypothesis that membrane tension should cause the monomers to tilt in the particular direction is based on our calculations of electrostatic interactions between the channel domains. The calculations indicate that the repulsive coulomb forces at the N- and C-terminal end of the channel are larger than at the opposite end. When the membrane is stretched and the monomers are pulled apart, the asymmetry of the charge distribution at the two ends of the channel hexamer would be expected to cause a predominant spreading of the monomers at the bottom end such that they become tilted, as suggested in our model.

When the applied external pressure exceeds a certain level such that the pull on the single monomers caused by membrane tension tilts the M1 and M2 helices to a degree that  becomes equal to  (Fig. 4 C), the orientation of F_R to the N-terminus will start to change in a direction to pull the N-terminus downward out of the channel pore (Fig. 4 D). In this case, the channel undergoes a transition from the closed to the open state. In reality, because of the intrinsic thermal energy kT, the  and  defined in this model are the average values of corresponding angles over time. Therefore, even at lower applied pressures at which  is less than  (Fig. 4 B), the channel still has a finite probability to reach open state at certain time t, when its instantaneous value _t becomes greater than _t. This probability could be considered to correspond to the channel open probability (P_o) obtained at the particular negative pressure applied to the patch clamp pipette. Consequently, at pressures at which  is equal to , open probability should be P_o = 0.5.

In summary, the mechanosensation of MscL results from coupling of mechanical and electrostatic forces, such that relative movements of the channel domains affects electrostatic interactions between these domains. The applied pressure provides a mechanical external stimulus that causes an increased membrane tension. Due to lipid-protein interactions, membrane tension causes an inclination of the channel monomers toward each other. This slight tilting of the transmembrane domains may serve as a trigger to initiate the outward movement of the N-terminal domain caused by electrostatic repulsion between N- and C-termini. This results in the channel opening.

Activation pressure

To test the feasibility of the ECM model we calculated the pressure p needed to produce a tilt of the channel monomers at an angle  relative to each other (Fig. 4). For an ideal membrane patch, the relationship between external pressure p and the membrane tension  is

(1)

where R is the radius of curvature of a membrane patch under pressure p. The membrane tension  is determined by the elasticity constant K_L, and the fractional increase of membrane area A/A

(2)

The area of a relaxed lipid membrane patch having a diameter D equal to the pipette diameter is

(3)

When external negative pressure (suction) is applied to the patch pipette, the surface of the stretched membrane increases by a fraction corresponding to the area of a sphere with a radius of curvature R. The area of the spherical calotte is:

(4)

The fractional increase of membrane area (A/A) = (A'  A/A) due to the stretch of the membrane can be calculated and the relationship between applied pressure p and the radius of curvature can be obtained using Eqs. 1-4.

(5)

The increase of membrane patch area generates membrane tension , which causes the transmembrane domains of MscL to tilt at an inclination angle  along the tension direction, which results in an increase of the area of the channel pore s = s'  s, where s = r²   1200 Ų (Cruickshank et al., 1997). The fractional increase in pore area is assumed to be equal to the fractional increase of the membrane patch:

(6)

The limitation of this assumption is that the elastic properties of the channel protein and the surrounding membrane bilayer are assumed to be similar. Despite that limitation, we used this assumption to provide a bridge between structural changes within the channel molecule and macroscopic area changes in the membrane patch. The channel pore radius in the closed state is r = 20 Å. The area is

(7)

Taking into account that the tilt of the monomers relative to each other under membrane tension (Fig. 4) would cause a slightly conical shape of the channel pore, the average enlarged area of the pore, s', can be described as

(8)

where r is the increase of the pore radius, and is determined by

(9)

It follows:

(10)

As described above, when  = , activation pressure is assumed to cause the channel to be open 50% of the time (P_o = 0.5). Thus we can correlate the radius of curvature R and pressure p with differences in the patch diameter D (Table 3).

The range of the calculated activation pressures is in good agreement with the negative pressure of ~65 mmHg required to activate the MscL 50% of the time in our experiments (Häse et al., 1995). However, we are uncertain about the radius of liposome patches formed inside our pipettes. The results in Table 3 indicate that the diameters of lipid membrane patches in pipettes formed in our experiments may range between 2 and 3 µm, which is in good agreement with the values reported in the literature (Sokabe et al., 1991). Also, the calculated values correspond well with our estimate of the size of the opening of the tip of pipettes of ~1 µm from the bubble number and pipette resistance (Cruickshank et al., 1997).

Energy calculation

The free energy G is a linear function of membrane tension  according to the model proposed by Howard et al. (1988),

(11)

where G⁰ is the difference in free energy between the closed and open conformations of the channel in the absence of membrane tension, and s is the difference in membrane area occupied by these two conformations at a given membrane tension. When the open probability P_o = 0.5, free energy change = 0. Using Eqs. 2, 6, and 15, it follows:

(12)

Thus, according to the EMC model the energy requirement to gate the MscL is minimal, since as little as 2 kT is sufficient to keep the channel open half of the time. This is consistent with a notion of very small molecular displacements of single channel domains underlying major conformational changes of the channel as a whole.

MscL mutants

We used the results obtained with several MscL mutants as a test for the consistency between the predictions of the ECM model and the experimentally observed results. An N-terminal deletion mutant NBE (1-8) and an N-terminal substitution mutant P6 (first eight N-terminal residues substituted by nine novel amino acids, resulting in a less charged N-terminus compared to that of the wild-type MscL) were examined for their pressure sensitivity and gating properties (Häse et al., 1997). Both mutants exhibited a marked alteration in channel activation by pressure. A site-directed mutation, Q56R, which made the S2-S3 loop more positively charged, resulted in a channel that became more sensitive to the applied pressure (Blount et al., 1996a). Although the experiments by Blount and coworkers (1996a,b) were performed in giant spheroplasts usually requiring higher pressures to activate the MscL, a comparison of experimentally obtained values for pressure sensitivity of the corresponding mutants with the theoretical predictions calculated according to the EMC model shows, for the most part, a good agreement between the two sets of data (Table 4). One exception is the site-directed mutant K31D in the M1 helix (Blount et al., 1996b). In its present form the model cannot account for the increase in the MscL pressure sensitivity observed in this mutant.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

How are MS ion channels gated by mechanical force? At present, two mechanisms of mechanosensitivity have been recognized (Hamill and McBride, 1997): the first, possibly more general mechanism, can be described according to the bilayer model (Martinac et al., 1990; Markin and Martinac, 1991; Opsahl and Webb, 1994), whereas the second, possibly more specialized mechanism, is best represented by the tethered model (Guharay and Sachs, 1984; Howard et al., 1988; Hudspeth and Gillespie, 1994; Hamill and McBride, 1996). The bilayer and the tethered model both were proposed at the time when no knowledge was available on structure and function relationships for any MS ion channel. Consequently, both models suffice only to account for the mechanosensitivity of MS channels at a phenomenological descriptive level, but cannot account for the underlying molecular mechanism(s).

In the case of MscL the gating mechanism complies with the bilayer model, since it was unambiguously demonstrated that MscL is gated by mechanical force that is exclusively transmitted via lipid bilayer alone (Sukharev et al., 1993<