INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

POLYMYXIN B (PX) is an amphipathic decapeptide that contains five positive charges, has a fatty acid tail, and has a molecular mass of ~1,400. It is used in topical ointments and sometimes as a bactericidal agent in irrigating solutions during long-term catheterization of the urinary bladder and is most active against Gram-negative bacteria. One of the reasons for the limited use of systemic PX is its nephrotoxicity. The mechanism of the bactericidal action of PX on Gram-negative bacteria is to increase the permeability of the bacteria's cytoplasmic membrane to ions and other cellular constituents. Of interest is that the bactericidal potency of PX to Gram-negative bacteria requires the fatty acid tail and is sensitive to the concentration of divalent cations in the bathing medium (23). It has been proposed that the effect of divalent cations is due to a competitive interaction between PX and the divalent cation for membrane binding such that the divalent cations form an electrostatic interaction with the negative charges on the phospholipid head groups. These interactions between divalent cations and phospholipid head groups alter the ability of PX to form an electrostatic bond with the negatively charged phospholipid head groups.

A recent study on rabbit urinary bladder (2) demonstrated that addition of PX to the lumen increased the apical membrane conductance to small monovalent cations and anions. The increase in the membrane conductance by PX required that the membrane potential be cell interior negative, with the magnitude of the conductance change being an exponential function of the applied voltage. Such a sensitivity to membrane potential might explain why a molecule like PX is more toxic to Gram-negative bacteria than to mammalian cells. Exposure of the urinary bladder to low concentrations of PX for short times could be reversed by either clamping the epithelium such that the apical membrane potential was cell interior positive or by washing the PX out of the luminal bath. However, when the bladder was exposed, at cell interior negative potentials, to high concentrations of PX or for long periods of time (tens of minutes) PX caused an irreversible increase in the transepithelial conductance of the urinary bladder epithelium.

This study investigated whether divalent cations alter the ability of PX to increase the apical membrane conductance of the rabbit urinary bladder to PX. In brief, we show that divalent cations modify the PX-induced conductance (G^PX_t) not only by competitive interaction with a membrane binding site (a negatively charged phospholipid as suggested for bacteria; Ref. 23), but also by acting as an open-channel blocker (both Ca²⁺ and Mg²⁺) as well as by changing the rate at which the G^PX_t falls out of the membrane (fall out, only Ca²⁺). It is shown that PX requires both a positive charge and a fatty acid tail to increase the membrane conductance. Last, it is demonstrated that lowering luminal pH from 7.8 to 6.5 decreases the ability of PX to induce a membrane conductance by two of the above mechanisms.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Urinary bladders were extracted from male New Zealand White rabbits. After removing the underlying smooth muscle layers (12), the epithelium was mounted between temperature-controlled modified Ussing chambers designed to reduce edge damage (13). The mucosal and serosal solutions were stirred by Teflon-coated magnetic spin bars driven by an external magnet coupled to a motor.

Solutions

View larger version (13K): [in this window] [in a new window]   Fig. 1.   A: structure of polymyxin B (PX). PX is a decapeptide with a fatty acid tail (R). The decapeptide has 6-diaminobutyric acid (DAB) molecules, and at pH 7.4, PX has 4.8 positive charges. F, phenylalanine; L, leucine; T, threonine. The fatty acid tail for polymyxin B1 is 6-methyloctanoyl and for polymyxin B2 is 6-methylheptanoyl. B: structure of PX nonapeptide (NP). NP has the same cyclic structure as PX but is missing the fatty acyl tail and one DAB molecule. This molecule has the same number of positive charges that PX has at a pH of 7.4.

Electrical Measurements

The current and voltage outputs of the clamp were connected via a variable gain amplifier (model LPF-202; Warner Instruments) to an analog-to-digital converter (model PP-50 LAB; Warner Instruments), which was interfaced to a computer. These parameters were digitized, stored on hard disk, and logged on a printer along with the time of data acquisition and the calculated values for R_t, G_t, and short-circuit current (I_sc). During open-circuit conditions I_sc = V_t/R_t. Data acquisition was at a maximum rate of five per second. In addition, V_t and I were continuously monitored on an oscilloscope and paper strip-chart recorder.

Current-Voltage Relationship

Standard Protocol

Statistics

	RESULTS

Top Abstract Introduction Methods Results Discussion References

In this section, it is first demonstrated that the magnitude of the increase in the transepithelial conductance by PX is dependent upon the luminal Ca²⁺ and Mg²⁺ concentrations. Then it is shown that the modulation of the PX-induced conductance by divalent cations occurs by at least two different mechanisms. Next, it is demonstrated that both the positive charge and the fatty acid tail of PX are required to increase the transepithelial conductance. Last, the effects of protons on the PX-induced conductance are studied.

Effect of Divalent Cations on PX-induced Conductance

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Effect of luminal Ca2+ on the ability of PX to increase transepithelial conductance (Gt). In this experiment PX was added to the luminal solution (with a Ca2+ concentration of 0 mM, 2 mM, or 10 mM) when transepithelial potential (Vt) was clamped at 70 mV. After a 5-min incubation period, Vt was clamped to 0 mV, and the change in PX-induced conductance (GPXt) was calculated. Note that when the bath contained 2 and 10 mM Ca2+, it took 5 and 20 times the concentration of PX to achieve the same GPXt as in the nominally Ca2+ free bath. These results suggest that the interaction between PX and Ca2+ is competitive.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   A: effect of luminal Ca2+ on GPXt. Experimental protocol was to add 400 U/ml of PX to a Ca2+/Mg2+-free luminal solution when Vt was clamped at 70 mV. After a 5-min incubation, Vt was clamped to 0 mV, and the GPXt was calculated. Next, Vt was clamped to 70 mV, and the luminal calcium concentration was increased. After a 5-min incubation, Vt was again clamped to 0 mV, and the GPXt was calculated. The smooth curve through the points was fit by hand. B: effect of increasing luminal Mg2+ on GPXt when the tissue was clamped from 70 mV to 0 mV. Experimental protocol was similar to that described for Ca2+. The smooth curve through the points was fit by hand.

Three potential mechanisms by which divalent cations might inhibit the ability of PX to increase the membrane conductance are: 1) block of the conductance once it has been induced (conductive block), 2) increase the rate of loss of the conductance from the membrane (increase the rate of fall out), and 3) competitive interaction with the protein for a membrane binding site.

Conductive block. To investigate conductive block, the standard protocol was used. Once G_t had increased by 100-200 µS/cm², the PX was washed from the luminal solution. In the absence of bath Ca²⁺, the loss of the PX-induced conductance is slow (see below). Next, the luminal concentration of Ca²⁺ or Mg²⁺ was increased in a step-wise manner, and the resulting steady-state conductance was measured. Following the step-wise additions and subsequent washout of the divalent cation, 67 ± 10% of the conductance decrease was recovered after Ca²⁺ addition, and 73 ± 11% was recovered after Mg²⁺ addition. Thus the Ca²⁺- or Mg²⁺-dependent decrease in conductance was reversible. Figure 4, A and B, shows the dose-response curves for Ca²⁺ and Mg²⁺ respectively. The data were fit by the sum of the Hill equation plus a divalent cation-insensitive PX-induced conductance (G_iX). The best fit values are shown in Table 1. Ca²⁺ blocked all of the PX-induced conductance, whereas Mg²⁺ only partially blocked the PX-induced conductance. Addition of 10 mM Ca²⁺ blocked the remainder of the Mg²⁺-insensitive component of the G^PX_t (data not shown). The above data suggest that there could be at least two places for conductive block and that Ca²⁺ can act at both, whereas Mg²⁺ only acts at one.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   A: effect of luminal Ca2+ concentration on GPXt. Experimental protocol was to add PX to the Ca2+/Mg2+-free mucosal bathing solution of a tissue clamped at 70 mV, incubate for 5 min, and voltage clamp the tissue to 0 mV. Conductance was allowed to increase, and the PX was then washed out of the lumen. Increments of Ca2+ were added to the luminal solution in a stepwise manner. The resulting decrease in GPXt by Ca2+ was then fit by the Hill equation. The best fit Ki was 0.56 ± 0.18 mM, and N was 1.02 ± 0.1 (n = 4). Ca2+ inhibited all of the GPXt. B: effect of luminal Mg2+ concentration on the GPXt. Experimental protocol was similar to that for Ca2+, except luminal Mg2+ was increased. The resulting decrease in GPXt by Mg2+ was fit by the Hill equation. The best fit Ki was 0.37 ± 0.16 mM (n = 5), and the Hill coefficient was 0.68 ± 0.02. Magnesium only blocked 44 ± 6% of the GPXt.

Rate of reversal. The experimental protocol for investigating the ability of divalent cations to reverse the G^PX_t was as follows. After G_t had increased by 100-200 µS/cm² using the standard protocol, PX was washed out of the luminal compartment, and the time-dependent change in conductance was monitored. A kinetic model to describe the time course of the change in G^PX_t is shown below.

View larger version (308K): [in this window] [in a new window]

where G^PX_p is the value for the bath and membrane-associated (bound) PX (which can either be washed out of the bath or become active) at the beginning of the mucosal solution wash, G^PX-A_t is the active G^PX_t present at the beginning of the mucosal wash, and G^PX-S_t is the stable G^PX_t. This latter conductance is not removed from the membrane after washing PX out of the mucosal bath. It can, however, be removed by clamping back to 70 mV (data not shown) or by adding a divalent cation to the mucosal bath. k_{pool Gtactive Gt} (k_pa) is the rate constant for the formation of an active conductance from the bath and membrane-associated PX, k_{active Gtout of membrane} (k_ao) is the rate constant for the loss of G^PX-A_t from the membrane, k_{active Gtstable Gt} (k_as) is the rate constant at which G^PX-A_t is transformed to G^PX-S_t (the stable form of the G^PX_t), and k_{pool Gtout of bath} (k_po) is the rate constant for washout of PX from the mucosal bath. The equation that describes the time-dependent change in G^PX_t during the wash of PX from the mucosa, G^PX_t(t), is

(1)

1

1

1

1

1

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Time course of the change in the Gt after washing PX out of the luminal solution. PX was added to the mucosal bath while the tissue was clamped at 70 mV and incubated for 5 min. Tissue was then clamped to 0 mV, and the Gt was allowed to increase by 50-100 µS. PX was then washed out of the bath (while clamped at 0 mV), and the change in Gt was measured over time. Smooth curve through the data points is the best fit of Eq. 1 to the data. For this example the rate constant for the loss of GPX-At from the membrane (kao) = 0.04 s1, and the rate constant at which GPX-At is transformed to GPX-St (the stable form of the GPXt) (kas) = 0.01 s1.

DEPENDENCE OF REVERSAL ON CALCIUM. The effect of luminal Ca²⁺ on the reversal of the G^PX_t was studied using the protocol for reversal described above. The best fit values for k_ao and k_as as a function of luminal Ca²⁺ are shown in Fig. 6A. Of interest is that increasing luminal Ca²⁺ concentration caused a decrease in k_ao. The values for k_as at 1, 2, and 10 mM Ca²⁺ concentrations were not statistically different from zero; however, the k_as value for 0 Ca²⁺ is significantly different from 0. This suggests that luminal Ca²⁺ inhibits the formation of the stable form of the conductance.

View larger version (14K): [in this window] [in a new window]   Fig. 6.   A: rate constants, kao () and kas (), as a function of the luminal Ca2+ concentration were obtained by fitting Eq. 1 to the data. An increase in the Ca2+ concentration caused a decrease in the kao, indicating that the presence of Ca2+ slowed the process of GPX-At leaving the membrane. The increase in kao from 2 to 10 mM Ca2+ was not significant. The kas values were not significantly different from 0 except at 0 mM Ca2+ concentration, suggesting that luminal Ca2+ inhibits the formation of the stable conductance. B: rate constants, kao () and kas (), as a function of the luminal Mg2+ concentration. These values were obtained in the same manner as for Ca2+. Luminal Mg2+ did not decrease kao or kas.

DEPENDENCE OF REVERSAL ON MAGNESIUM. The effect of Mg²⁺ on reversing the G^PX_t was studied in a manner identical to that for Ca²⁺. The time course of the conductance change from the initiation of mucosal wash was curve fit by Eq. 1. Figure 6B shows the effect of increasing luminal Mg²⁺ on k_ao and k_as. Of interest is that Mg²⁺ did not alter the rate at which G^PX-A_t entered the stable state G^PX-S_t, nor did it decrease the rate at which the G^PX_t left the membrane.

Binding site. To investigate a possible interaction of these divalent cations at a membrane binding site, the measured G^PX_t in the presence of divalent cations must be corrected for the rate of fall out as well as conductive block. Since the rate constant for loss of G^PX-A_t (k_ao) is slow compared with the time period over which G^PX_t is measured, fall out will not significantly alter the magnitude of G^PX_t. The G^PX_t can be reduced to two components, the conductance formation and the conductive block. Using the relationship between divalent cation concentration and conductive block (Fig. 4), the G^PX_t (Fig. 3) was corrected for the conductance that was being blocked by the divalent cation, and this corrected conductance is G^PX_t(X).

A kinetic model was developed to describe the binding of PX to the membrane binding site in the presence of divalent cations to determine the dissociation constant for Ca²⁺ and Mg²⁺ to the membrane binding site.

View larger version (7K): [in this window] [in a new window]

where S is an unoccupied membrane binding site, B is a binding site that has been bound by a divalent cation and is unavailable to bind a PX molecule, and O is a binding site that has been bound by PX and has formed a conductance. K_{Ca,Mgbinding site} (K_i) is the dissociation constant for either Ca²⁺ or Mg²⁺ for the binding site, and K_{PXbinding site} (K_PX) is the dissociation constant for PX for the binding site. The equation that describes the divalent cation effects on the G^PX_t is as follows

(2)

G^PX_t(max X) is the maximum rate of increase of the G^PX_t in Ca²⁺- and Mg²⁺-free solution, [X] and [PX] are the concentrations of the divalent cation and PX, respectively, and K_PX and K_i are dissociation constants as previously described in the kinetic scheme. The normalized data sets, corrected for conductive block as described above, were fit by Eq. 2. The means of the fit values for the data set are shown in Table 2. Figure 7, A and B, shows the relationship of the G^PX_t, which has been corrected for block as a function of the divalent concentration.

View this table: [in this window] [in a new window]   Table 2.   Effects of divalent cations on GPXt(X)

View larger version (14K): [in this window] [in a new window]   Fig. 7.   A: effect of luminal Ca2+ on the GPXt, which has been corrected for conductive block. The data from Fig. 3 were corrected for block. These data were fit by Eq. 2. Smooth curve was generated from the mean of the best fit values from these data sets. See Table 2 for best fit values. B: effect of luminal Mg2+ on the GPXt, which has been corrected for block. Data and curve fitting were treated identically to that for Ca2+. See Table 2 for best fit values.

Given the ability of Ca²⁺ and Mg²⁺ to alter G^PX_t, we determined the selective permeability (using I-V relationships, see METHODS) to Cl, K⁺, and Na⁺ of the G^PX_t in Ca²⁺- and Mg²⁺-containing Ringer solution, in Ca²⁺-free, Mg²⁺-containing Ringer solution, and in Ca²⁺- and Mg²⁺-free Ringer solution. The concentration of PX was adjusted among these conditions such that the G^PX_t was similar. Table 3 shows that the selective permeability of the G^PX_t to Cl, K⁺, and Na⁺ was not altered by the presence or absence of either Ca²⁺ or Mg²⁺. This suggests that the induced conductance is of a uniform type, since blocking some of the conductance does not affect the ion flow through the remainder of the induced conductance.

Effects of Structural Components of PX

The addition of up to 185 µM of nonapeptide (5 times the amount of PX on a mole:mole basis) to the luminal Ca²⁺/Mg²⁺-free KCl Ringer solution did not alter G_t when V_t was clamped from 70 mV to 0 mV (data not shown). Preincubation of the tissue with nonapeptide did not change the ability of PX to increase G_t, as shown in Table 4. This not only suggests that PX is inactive in the absence of the fatty acid tail but also that nonapeptide is not competing for a binding site with PX.

Proton Effects on G^PX_t

Figure 8A is a titration curve for PX and demonstrates that PX has a pK_a of 8.8, i.e., at pH 8.8, PX will have on average 2.5 charges. Changing the pH on either side of this pK value would change the number of charges on the PX molecule, providing an opportunity to determine the effects (if any) of charge on the ability of PX to induce a conductance. The protocol for this experiment was as follows. The pH of the luminal solution was adjusted to pH 9.8 in a Ca²⁺/Mg²⁺-free KCl Ringer solution, and V_t was clamped at 70 mV. After a 5-min incubation with PX, V_t was clamped to 0 mV and the G^PX_t was measured. G_t was clamped back to 70 mV, and the conductance was allowed to recover to the baseline value. The pH of the solution was decreased by the addition of HCl (the pH was monitored by a pH electrode in the luminal compartment). After 5 min at the lowered pH, V_t was clamped to 0 mV and the G^PX_t was measured. As shown in Fig. 8B, as the luminal pH was decreased from 9.8, the G^PX_t increased up to a pH of 7.8. However, as the pH decreased below 7.8, the G^PX_t also decreased.

View larger version (12K): [in this window] [in a new window]   Fig. 8.   A: titration curve for 250 µM PX. Peak in the curve yields a pKa for PX of 8.8, which is a Ka of 1.6 nM. B: effect of charge of the PX molecule on GPXt. GPXt was normalized to the conductance change at pH 7.7. A minimum of three charges is required to produce a measurable GPXt. GPXt reached a maximum at a pH of 7.8. A further decrease in luminal pH resulted in a decrease in GPXt. Solid circles, GPXt values that have not been corrected for conductive block. Solid line, best fit of Eq. 3 to the uncorrected data. In this example, the best fit values for the uncorrected data are GPXt(max H+) (the maximum GPXt at a proton concentration of 0) = 6.27 and Ksh (the dissociation constant for a proton to the binding site) = 23 nM. Open circles, GPXt values that have been corrected for conductive block. Dashed curve, best fit of Eq. 3 to the corrected data. In this example, the best fit values are GPXt(max H+) = 5.5 and Ksh = 50 nM.

To determine whether the increase in G^PX_t when changing pH from 9.8 to 7.8 was due to the number of charges on the PX molecule and not to a pH-dependent alteration of the luminal membrane, the same experiment was performed using protamine sulfate (PS). Protamine is a cationic protein of 4,000 Da, is composed of 67% arginine, and has been shown to induce a conductance in apical membrane of the urinary bladder (20, 21). Guanidinium is the residue that gives arginine a net positive charge and has a pK of 12.5. Thus the net charge on protamine will be constant over the pH range studied (i.e., pH 9.8 to 7.8). There was no effect of pH from 9.8 to 7.8 on the ability of PS to induce a membrane conductance (data not shown). Thus pH over this range does not seem to alter the properties of the luminal membrane.

The possible effects of protons on the PX-induced conductance were studied in a manner similar to that for divalent cations.

Conductive block. To determine how much of the induced conductance was being blocked by protons, we added PX to the luminal compartment. The pH of the Ca²⁺/Mg²⁺-free, carbonate-buffered KCl Ringer solution was adjusted to 8.2 while V_t was clamped to 70 mV. After a 5-min incubation, V_t was clamped to 0 mV, and the conductance was allowed to increase to a maximum of 200 to 300 µS/cm². The mucosa was then washed with a PX-free solution of pH 8.2. After the transepithelial conductance reached a steady state, the mucosal bathing solution pH was decreased (in a stepwise manner) using predetermined aliquots of 1.8 mM H₂SO₄. After the luminal solution had been titrated to a pH value of 6.0, the mucosal pH was returned to 8.2 to determine whether the conductance loss was reversible. In these experiments, 91.5 ± 21% (n = 6) of the conductance was recovered. The resultant conductance change was determined for each pH value, and the normalized data were fit by the sum of the Michaelis-Menten equation plus a proton-insensitive PX-induced conductance (G_iH+). Figure 9 shows the concentration-conductance relationship for protons. The best fit values are as follows (n = 6): G^PX_t(max bH⁺) = 0.57 ± 0.13, K_bH+ = 55.7 ± 10.8 nM (equivalent to a pH of 7.25), and G_iH+ = 0.49 ± 0.13. Thus about half of the induced conductance is not blocked by protons. Since the dissociation constant for proton conductive block is a pH of 7.25, the increase in the rate of PX-induced conductance as pH is decreased from 9.8 to 7.8 cannot be accounted for by release from conductive block. These data suggest that most of the change over the pH range of 7.8 to 9.8 must be due to the charge on the PX molecule.

View larger version (16K): [in this window] [in a new window]   Fig. 9.   Effect of protons on the GPXt. In this experiment PX was added to the mucosal chamber of a tissue clamped at 70 mV and bathed in a Ca2+/Mg2+-free solution. After a 5-min incubation, Vt was clamped to 0 mV, and Gt was allowed to increase 100-200 µS before the PX was removed by wash. When a steady-state Gt was reached, the pH of the solution was lowered in a stepwise fashion, and the Gt was monitored. Resultant decrease in GPXt was fit by the Michaelis-Menten equation plus a constant (GiH+; the GPXt that was not blocked by protons). Best fit values are as follows: KbH+ (the dissociation constant for proton conductive block) = 55.7 ± 10.8 nM (pH of 7.25) and GiH+ = 0.49 ± 0.13. Maximum fraction of GPXt that protons blocked [GPXt(max bH+)] was 0.57 ± 0.13 (n = 6). Thus 54% of the GPXt was blocked by protons.

Rate of reversal. The effect of protons on the rate of fall out was studied by adjusting the pH of a Ca²⁺/Mg²⁺-free, carbonate-buffered mucosal KCl solution to a predetermined pH value between 6.5 and 8.2. (Above 8.2 and below 6.5 PX did not induce enough of a conductance to study its fall out rate.) PX was added while V_t was clamped at 70 mV. After a 5-min incubation period, V_t was clamped to 0 mV. A conductance increase of ~200 µS/cm² was induced before the mucosa was washed with a PX-free solution of the same pH. The resultant time-dependent conductance change was fit to Eq. 1 to determine the k_ao and k_as values (see Table 5). Protons did not alter the k_ao (rate constant for the loss of the active conductance from the membrane i.e., fall out), nor did they alter k_as (rate constant at which the active conductance is transformed into a stable conductance in the membrane). These data suggest that the decrease in the PX-induced conductance at an acidic pH is not due to an increased rate of fall out of the conductance.

Binding site. The following kinetic model was developed to determine the proton dissociation constant for a membrane binding site

View larger version (11K): [in this window] [in a new window]

where S is an unoccupied binding site on the membrane that can either be bound by PX and form a conductance O or be bound by protons that will block the binding of PX and subsequent conductance formation. This bound site is designated as B. [PX⁰⁺] is the concentration of the uncharged form of PX, and [PX⁵⁺] is the concentration of the fully charged form of PX. The dashed arrows indicate protons binding to and unbinding from the PX molecule, indicating the presence of all charged forms of PX. The sum of the concentrations of the forms of PX, i.e., PX⁰⁺ + PX¹⁺ + ... + PX⁵⁺ equals the total concentration of PX in the bath. [H⁺] is the concentration of protons that can affect the ability of PX to induce a conductance in two manners. First, by binding to the binding site(s) and preventing formation of a conductance and, second, by increasing the charge on the PX molecule, thereby increasing its effectiveness. K_{H+PX molecule} (K_a) is the dissociation constant of the proton binding to the PX molecule, K_{H+binding site} (K_sh) is the dissociation constant for a proton to the binding site, and K_{PX5+binding site} (K_PX) is the dissociation constant for PX to the binding site. The equation that describes this proton-dependent G^PX_t or G^PX_t(H⁺) is

(3)

where G^PX_t(max H⁺) is the maximum G^PX_t at a proton concentration of 0, and [PX] is the concentration of PX (the sum of all the charged forms of PX, i.e., [PX⁰⁺], [PX¹⁺], ... [PX⁵⁺]), which was held at 400 U/ml in these experiments. K_a, K_sh, and K_PX are as explained above in Model 3. N was held at 5 during curve fitting; i.e., we assumed that PX had to be fully charged to induce a conductance. When fitting the data by Eq. 3, K_a was held at the value of the pK from the PX titration curve, pH of 8.8, or a proton concentration of 1.6 nM. K_PX was held at the value of 735 U/ml determined from the PX concentration-conductance relationship (2). Using the relationship between proton concentration and conductive block (Fig. 9), the G^PX_t(H⁺) (Fig. 8B, solid circles) was corrected for the conductance that was being blocked by protons. Thus only two parameters were varied G^PX_t(max H⁺) and K_sh. The best fit value for G^PX_t(max H⁺) was 8.6 ± 2.00 and for K_sh was 27.9 ± 7.49 nM or pH 7.6 (n = 6). This value represents the dissociation constant for protons to the binding site.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

This section will review divalent cation and proton modulation of the G^PX_t by three mechanisms, conductive block, fall out rate, and competitive interaction with a binding site. The data will be discussed in terms of the structure of PX and its effects on cell membranes. The evidence for a binding site and its possible identity will be addressed. Finally, a model for PX conductance formation and modulation will be presented.

Modulation of PX Effects

Divalent cations. The results from this study suggest that Mg²⁺ and/or Ca²⁺ modulated the PX-induced conductance increase by three mechanisms. First, Ca²⁺ and Mg²⁺ blocked the PX-induced conductance in a reversible manner (conductive block). Ca²⁺ completely blocked the induced conductance, whereas Mg²⁺ blocked only 44%. The Mg²⁺-insensitive conductance was blocked by the subsequent addition of Ca²⁺. Second, the addition of Ca²⁺ to the bathing solution slowed the loss of the PX-induced conductance from the membrane; however, Ca²⁺ was required for a complete removal of the conductance from the membrane. Mg²⁺ did not alter the fall out rate. Third, Ca²⁺ and Mg²⁺ compete with PX for a membrane binding site. The modulation by Ca²⁺ is in agreement with a previous study (22) of the cationic polypeptide PS. The voltage-dependent, PS-induced increase in the conductance of the rabbit bladder epithelium was also found to be modulated by Ca²⁺ and Mg²⁺.

Protons. This study also demonstrated that protons had a modulatory effect (both stimulatory and inhibitory) on the ability of PX to induce a membrane conductance. As the pH was changed from 9.8 to 7.8, the ability of PX to induce a conductance increased, whereas changing pH from 7.8 to 6.5 resulted in a decrease in the ability of PX to induce a conductance. This stimulatory effect of protons (at high pH) on the rate of formation of a PX-induced conductance was due to an increase in the number of charges on the PX molecule. The decrease in PX-induced conductance from pH 7.8 to 6.5 occurred by the same mechanisms as Ca²⁺ and Mg²⁺. Thus protons compete with PX for a membrane binding site and block (in a reversible manner) ~50% of the PX-induced conductance. As was found for Mg²⁺, protons did not alter the rate of loss of the PX-induced conductance from the membrane.

Identity of the Binding Site

A number of systems indicate that PX binding is cooperative (9, 17). This cooperative binding probably explains the sigmoidal nature of the conductance increase reported by Berg et al. (2) as an alternative to the number of PX molecules required to form a conductive unit.

Structure-Function Studies of PX and Related Peptides

Fatty acid tail. Addition of nonapeptide (at 5 times the molar concentration of PX) did not affect the tissue conductance. This indicates that without the fatty acid tail, this concentration of nonapeptide cannot form a conductance in the urinary bladder epithelium. In a study by Duwe et al. (7), a ⁵¹Cr release assay was used to determine the effectiveness of the nonapeptide and PX against a chronic myelogenous human leukemia cell line (K-562). It was found that at concentrations as low as 50 µg/ml of PX, there was a 21.6% ⁵¹Cr release, whereas at concentrations of nonapeptide as high as 3 mg/ml there was a minimal ⁵¹Cr release. In addition, the cells exposed to nonapeptide remained healthy over a period of 48 h as determined by morphological assessment and trypan blue exclusion. Using lipid bilayers, Schroder et al. (16) demonstrated that PX-induced single channel-like events, whereas nonapeptide at similar concentrations did not alter the bilayer membrane conductance. Taken together the above observations suggest that the fatty acid tail enhances the ability of PX to increase membrane conductance.

If PX and nonapeptide share the same binding site, then a reduction in the magnitude of the PX-induced conductance due to competition by nonapeptide might be predicted. Our results suggest that there was no effect on the PX-induced conductance when the epithelium was concomitantly treated with PX and up to five times its molar concentration of nonapeptide. A possible explanation is that the affinity of PX compared with the nonapeptide for a membrane binding site is potentiated by the presence of the PX hydrophobic tail. Buser et al. (6) have studied the effect of a fatty acid tail (myristol, 14 carbon tail) on the affinity of a cationic peptide (residues 2-16 of the Src protein) to charged lipid vesicles. These authors demonstrated that the myristylated cationic peptide binds four orders of magnitude more strongly to anionic phospholipid vesicles than the nonmyristylated cationic peptide. If the acyl tail of PX increases the affinity of PX compared with nonapeptide by a similar amount, then the concentration of nonapeptide would have to be many-fold higher (around 1,000 to 10,000) than that of PX before there would be a significant decrease in the ability of PX to induce a conductance. It is apparent that in both bacteria (24) and in urinary bladder epithelium that the fatty acid tail is needed to affect the cells in a lethal manner.

Positive charge. In addition to requiring the fatty acid tail domain, the necessity of a net positive charge for PX to induce a conductance was described. It is clear that as the pH of the solution was lowered from a value of 9.8, the ability of PX to induce a conductance increased (see Fig. 8B). This rate of conductance increase coincides with the increase in the number of charges on the PX molecule as determined from our titration study and the reported buffer capacity of PX (4). In a study of the susceptibility of Escherichia coli lipid liposomes containing glucose, a similar polymyxin response to pH was observed (8). As the pH was lowered from 10.5 to 7.8, the amount of polymyxin-induced glucose release increased. As the pH was lowered past 7.8, the amount of glucose released decreased. These data demonstrate the necessity of the charged portion of the molecule not only for conductance formation in the bladder but also for bactericidal action.

Model of PX Action on Bladder Epithelium

A tentative description of a model for the formation of a polymyxin-induced conductance is as follows. At a cell interior positive voltage, PX will associate with a membrane binding site but not induce a conductance. This association is composed of the insertion of the fatty acid tail of the polymyxin molecule into the cell membrane as well as the electrostatic interaction of the positive charges on the molecule with a negatively charged membrane binding site. Mg²⁺, Ca²⁺, and H⁺ can bind to the membrane binding sites and inhibit the formation of a conductance by PX. Once PX has associated with the membrane, it remains inactive until a cell interior negative voltage is applied. When this negative voltage is applied, the PX molecules move into the membrane forming, by an unknown mechanism, a conductive "pore." The lag phase in the conductance formation might be due to the time it takes the molecules to move into the membrane to the conductive configuration. The greater the PX concentration or the more negative the cell interior voltage, the greater will be the induced conductance and the shorter the lag phase of the induced conductance.

The reversibility of the polymyxin-induced conductance (in the absence of bath Ca²⁺) suggests that two types of conductances are formed, i.e., a stable and an unstable conductance. The unstable conductance (but not stable conductance) can be removed from the membrane by washing PX from the mucosal bath. This unstable conductance does not form in the presence of Ca²⁺. The stable conductance is defined as the PX-induced conductance that remains after PX is washed from the bath in the presence of a cell interior negative voltage. This stable conductance is blocked by divalent cations and protons. In addition, it can fall out of the membrane in the absence of Ca²⁺ when the voltage is at 70 mV and bath PX has been removed. This stable conductance can also be removed from the membrane (independent of voltage) by the addition of mucosal Ca²⁺ but not by Mg²⁺ or H⁺.

Studies are presently underway to determine how many different types of conductive units PX can form, whether these units are formed in a parallel or serial manner, and the influence of the type of lipid on the properties of the PX-induced conductance.