Lens Gap Junctional Coupling Is Modulated by Connexin Identity and the Locus of Gene Expression

Abstract

purpose. To investigate the effects of reducing connexin (Cx) diversity in the lens when the amount of connexin protein is nearly constant.

methods. Lenses in which the Cx50 coding region was replaced by that of Cx46 (knockin [KI]), were compared with wild type (WT) and Cx50-knockout (KO) lenses. Gap junctional conductance (G_j), and membrane conductance were evaluated by using frequency domain impedance of intact lenses.

results. KO of Cx50 produced small depolarized lenses with central opacities. KI of Cx46 did not restore growth, but rescued resting voltage and eliminated opacities. In WT lenses, the average G_j was approximately 1 S/cm² of cell-to-cell contact in the outer shell of differentiating fibers (DFs), whereas it was approximately half that value in the core of mature fibers (MFs). KO of Cx50 reduced G_j in DF to 44% of normal, whereas KI of Cx46 restored G_j to approximately 60% of normal. In addition, KI of Cx46 markedly increased G_j in MFs. In WT lenses, all gap junction channels in DFs close when pH is reduced, whereas those in MFs are insensitive to pH. KO of Cx50 made both DF and MF channels pH insensitive, whereas KI of Cx46 restored pH sensitivity of all DF channels without altering MF pH insensitivity

conclusions. Lens size and fiber cell coupling conductance depended on which connexin was expressed on the Cx50 gene locus, whereas homeostasis of central fibers and normal gap junction gating were maintained when either connexin was expressed. The authors conclude that the roles of lens gap junction channels depend not only on the primary sequence of the expressed connexin, but also on the gene locus that expresses the connexin.

The lens has a single layer of epithelial cells covering its anterior surface. At the equatorial region, these epithelial cells begin to elongate and differentiate into the fiber cells that make up the bulk of the lens. The fiber cells can be divided into two zones: differentiating fibers (DFs), which make up the outer ∼20% of the radius of the lens, and mature fibers (MFs), which make up the inner ∼80%, and where there are no organelles and low metabolic activity (Fig. 1) . The transition from DF to MF occurs in a few cell layers, ¹ where the organelles disappear and many membrane proteins, including connexins (Cx), undergo cleavage. ^{2 3 4} These changes are related to the lens’ main physiological role, which is to help focus light on the retina. The organelles are eliminated because they scatter light and would render the lens translucent rather than transparent. Given the lack of organelles, protein turnover in MF is not possible, but the controlled proteolysis at the DF-to-MF transition appears to stabilize membrane proteins. For example, lens gap junctions remain functional all the way to the lens center, at least in small lenses where such studies could be performed. Because the central fibers are the first to be formed, their connexin proteins are as old as the animal. The transition from DF to MF is therefore a carefully orchestrated event, aimed at maintaining the homeostasis of the central fibers and lens transparency. 

Rae et al. ⁵ were the first to suggest that gap junction coupling might change at the DF-to-MF transition. They found that 2-4-dinitrophenol induces closure of fiber cell gap junction channels only in the outer shell of fibers; however, they could not rule out a diffusion limitation of dinitrophenol rather than a change in gap junction sensitivity to the drug. Mathias et al. ⁶ induced the closure of fiber cell gap junction channels by bubbling the bath with CO₂, which caused a decrease in intracellular pH. They also found that uncoupling was limited to an outer shell of fiber cells. Changes in intracellular pH, however, could be detected using pH-sensitive microelectrodes, and they were able to demonstrate that significant pH changes were occurring in MF where no gap junction uncoupling was occurring, suggesting a change in gap junction gating properties at the DF-to-MF transition. Although the functional significance of the dinitrophenol treatment or pH changes is not clear, the ability of DF junctions to gate in response to these factors suggests that they may also respond to more physiologically relevant signals. 

Mathias et al. ⁶ and Baldo and Mathias ⁷ reported a change in the distribution and average value of coupling conductance at the DF-to-MF transition. Coupling conductance in DFs varies from a minimum at either pole, to a maximum at the equator, with the average being approximately 1 S/cm² of cell-to-cell contact. In the MFs, the conductance appears to be much more evenly distributed, with an average value of 0.4 to 0.5 S/cm². The decrease in conductance at the DF-to-MF transition was thought to be related to protein cleavage; however, the redistribution of conductance at the DF-to-MF transition cannot be so simply explained. More recent studies have shown there is a dramatic reorganization of gap junction plaques at this transition. ^{2 3} In addition, heterologous expression of C-terminal cleaved forms of Cx46 (Martinez-Wittinghan FJ, et al. IOVS 2001;42:ARVO Abstract 4696) and Cx50 ³ resulted in functional gap junctional channels. Thus, the change in conductance at the DF-to-MF transition may involve additional processing of connexins beyond the simple removal of their C termini. 

Generation of the lens fiber cell connexin knockout (KO) mice ^{8 9} and studies of fiber cell coupling conductance in these lenses ^{10 11} have provided new information that has led to some new hypotheses. These studies suggested that the reduction in coupling conductance at the DF-to-MF transition might reflect the loss of functional Cx50 channels, whereas Cx46 channels remain functional in the MF. However, this loss of functionality of Cx50 cannot be explained as a result of cleavage alone, since cleaved Cx50 is functional in vitro, ³ and truncated Cx50 is found in the MFs of wild-type (WT) lenses. ¹² Moreover, the Cx46 channels in the DFs of Cx50 KO lenses were not pH-sensitive, suggesting that Cx50 channels were necessary for gating of the DF coupling conductance, possibly through cooperative interactions between neighboring channels, and loss of functional Cx50 channels, and/or cleavage of the Cx50 channels ^{2 3} at the DF-to-MF transition may cause the loss of pH sensitivity. Truncation of Cx46 is not thought to be the cause of this loss, because in vitro truncated Cx46 is capable of pH gating ¹³ (Martinez-Wittinghan FJ, et al. IOVS 2001;42:ARVO Abstract 4696). 

These published studies of KO lenses reveal surprising complexity in the regulation of gap junction conductance and gating in the lens. The main limitations of these KO studies are: Both the diversity and quantity of connexin subunits were simultaneously changed; and total loss of an important channel protein generally causes loss of homeostasis, which leads to indirect effects that have to be separated from the direct effects of knocking out the protein. The studies described in this article used knockin (KI) lenses in which the coding region of Cx50 is replaced with that of Cx46. ¹⁴ KI lenses provide information about gap junction coupling of fiber cells when diversity is reduced but the total number of connexin subunits remain unchanged. 

Materials and Methods

Unless otherwise noted, chemicals and supplies were acquired from Sigma-Aldrich (St. Louis, MO). 

Solutions

Tyrode solution contained (in mM) 137.7 NaCl, 5.4 KCl, 0.3 NaOH, 1 MgCl₂, 10 glucose, and 5 HEPES (pH 7.4) with NaOH. 

KI Mice

The methods for the creation of these animals are described in White. ¹⁴ However, a brief description of the genetically engineered mice is provided: WT mice have two alleles that code for Cx46 and two alleles that code for Cx50. In the KI construct, the coding region of Cx50 was replaced with the coding region of Cx46, by using homologous recombination, with little disruption of the Cx50 locus. Thus, a heterozygous KI mouse KI(50/46) had three genetic copies of Cx46 and one of Cx50. A homozygous KI mouse had four genetic copies of Cx46; however, two had the expression pattern of Cx50. 

Lens Extraction

All the procedures involving the use of animals were in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Division of Laboratory Animal Resources (DLAR) of SUNY at Stony Brook. The mice were killed by peritoneal injection of pentobarbital (100 mg/kg body weight). The lenses were dissected from the eye and placed in a silicone-lined (Sylgard; Dow Corning, Midland, MI) chamber in normal tyrode solution at 36°C to 37°C. ^{7 10}  

Impedance Studies

The impedance studies were performed with two microelectrodes shielded with grounded silver paint to within 300 μm of the tip and coated with silicone (Sylgard; Dow Corning) to avoid having the paint in contact with the bath. One intracellular microelectrode was used to inject a stochastic current, composed of a wide band of sinusoidal frequencies, into a central fiber cell. A second intracellular microelectrode recorded the induced voltage at a distance r (cm) from the lens’ center. The current and voltage signals were sent to a fast Fourier analyzer for analysis and to compute the impedance in real time. A measurement of the high-frequency series resistance (R_S) in either DFs or MFs was used to estimate gap junction coupling. To test the sensitivity of coupling to pH, R_S was determined while superfusing the lens with tyrode solution that had been saturated with 100% CO₂. 

Model for Prediction of Gap Junction Coupling

Mathias et al. ⁶ described the mathematical method to calculate gap junction coupling from impedance measurements. Briefly, when a high-frequency current is injected into a lens cell, membranes are effectively shorted out by their capacitance. In this situation, the induced intracellular and extracellular voltages within the lens are essentially identical, and the properties of the lens become similar to those of a conductive sphere, which is grounded at its surface. The effective resistivity of this “sphere” is the parallel combination (R_P) of the resistivities for the extracellular (R_e) and intracellular (R_i) pathways: R_P = R_iR_e/(R_i + R_e) Ω/cm. 

The value of R_e depends on the small volume fraction and tortuosity of the extracellular clefts, and therefore it is on the order of 50 kΩ/cm. ¹⁵ The value of R_i depends on gap junction coupling and it is normally on the order of 2 to 4 kΩ/cm. Thus, in normal situations, R_P ≈ R_i, but when acidification causes uncoupling of lens gap junctions, R_i can become quite large, producing a situation where R_P ≈ R_e. For a symmetric, conductive sphere with radius a (in centimeters) and resistivity R_P, the induced voltage ψ (volts) is related to the injected current I (in Amperes) by:  

\[\frac{{\psi}}{\mathrm{I}}\ {=}\ \frac{1}{4{\pi}}\ {{\int}_{r}^{a}}\ \frac{\mathrm{R}_{\mathrm{P}}(r)}{r}dr\]

In the lens, gap junction coupling conductance changes at the DF-to-MF transition, causing an abrupt change in R_P, which we assign values of R_DF and R_MF. Incorporating this change and performing the integral in equation 1 yields:  

\[\mathrm{R}_{\mathrm{S}}\ {=}\ {\psi}/\mathrm{I}\]

 

\[\mathrm{R}_{\mathrm{S}}\ {=}\ \frac{\mathrm{R}_{\mathrm{DF}}}{4{\pi}}\left(\frac{1}{r}\ {-}\ \frac{1}{a}\right)\ b{\leq}r{\leq}a\]

 

\[\mathrm{R}_{\mathrm{S}}\ {=}\ \frac{\mathrm{R}_{\mathrm{DF}}}{4{\pi}}\left(\frac{1}{b}\ {-}\ \frac{1}{a}\right)\ {+}\ \frac{\mathrm{R}_{\mathrm{MF}}}{4{\pi}}\left(\frac{1}{r}\ {-}\ \frac{1}{b}\right)\ 0{\leq}r{\leq}b\]

 

By mapping the induced voltage at various depths into the lens and comparing the results with equation 2 , we deduced the values of R_DF and R_MF. (R_S is the measured series resistance, and b is the radial location of the DF-to-MF transition.) 

Equations 1 and 2 2 describe a sphere with resistivity that does not vary in the angular dimension. If the resistivity varies with angular location, as it does in the DFs of the lens, then equations 1 and 2 2 describe the angular average of the induced voltage. Our measurements are generally made at 45° from the equator, where the induced voltage is equal to its angular average; hence, the resistivity reported herein reflects the angular average. 

Equivalent Circuit

Mathias et al. ¹⁶ showed that the lens behaves as a nonuniform spherical syncytium. The impedance data can then be fitted to obtain the surface cell conductance (G_S), surface cell capacitance (C_S), inner fiber cell membrane conductance (G_m), and effective extracellular resistivity (R_e), using the following equations:  

\[Y\mathrm{e}\ {=}\ \frac{{\gamma}}{\mathrm{R}_{\mathrm{i}}\ {+}\ \mathrm{R}_{\mathrm{e}}}\ (\mathrm{coth}({\gamma}\mathrm{a})\ {-}\ \frac{1}{{\gamma}a}).\]

R_i is the effective radial intracellular resistivity in the outer shell.  

\[{\gamma}\ {=}\ \sqrt{(\mathrm{R}_{\mathrm{i}}\ {+}\ \mathrm{R}_{\mathrm{e}})\ \frac{\mathrm{Sm}}{\mathrm{Vt}}\ (\mathrm{G}_{\mathrm{m}}\ {+}\ j{\omega}\mathrm{C}_{\mathrm{m}})}.\]

C_m is the fiber cell membrane capacitance (1 μF/cm²), Sm/Vt (6000 cm⁻¹) surface-to–volume ratio, and ω is the frequency in radians/second (j is the imaginary operator of this complex equation, \(j\ {=}\ \sqrt{{-}1}\) :  

\[Y\mathrm{s}\ {=}\ \mathrm{G}_{\mathrm{s}}\ {+}\ j{\omega}\mathrm{C}_{\mathrm{s}},\]

 

\[Z\ {=}\ \frac{1}{4{\pi}a^{2}(Y\mathrm{s}\ {+}\ Y\mathrm{e})}\ {+}\ \mathrm{R}_{\mathrm{S}},\]

where Z is the impedance (complex) and R_S is determined by equation 2 . 

Electrophysiological Data Analysis

Data analysis was performed on computer (Sigma Plot; Sigma Stat Software, ver. 4.0, MathCAD 11; SPSS, Chicago, IL; and Excel; Microsoft, Redmond, WA). 

Results

This article reports studies of lenses from WT mice and from KI mice, where Cx46 replaced Cx50. Heterozygous KI, KI(50/46) lenses, contained three Cx46 alleles and one Cx50 allele; homozygous KI, KI(46/46) lenses, contained four Cx46 alleles. 

Physical Properties

We have shown that knockout of Cx50 results in smaller lenses. ⁸ However, the KO lenses also had mild cataracts and depolarized resting voltages and were generally in very poor condition. ^{8 11} The small size could therefore have been a secondary effect of a general loss of homeostasis. However, White ¹⁴ showed that the KI(46/46) lenses lacked the cataract in the Cx50 KO lenses, yet the growth defect was still present. We verified that these previously reported effects on growth and transparency were present in the lenses used in the present study. In the next sections, we report new data on gap junction coupling and other physiological properties of these lenses. 

When KI and WT lenses from littermates of the same age (approximately 3 weeks) were compared, the average radius of homozygous KI lenses (0.091 ± 0.004 cm) was significantly smaller than WT (0.102 ± 0.005 cm; P < 0.001). The ratio of radii KI:WT was 0.89, resulting in a calculated 30% reduction in lens volume, a value in good agreement with the 34% reduction in lens mass previously reported. ¹⁴ Baldo et al. ¹¹ also observed smaller radii for Cx50 KO lenses, with the ratio of radii KO:WT being 0.79. Moreover, consistent with the data in Martinez-Wittinghan et al., ¹⁷ the size of heterozygous KI lenses (a = 0.102 ± 0.004 cm; n = 13) was not significantly different (P = 0.803) from that of WT lenses. Thus, in the absence of Cx50, lenses were consistently undersized, although the KI lenses were more transparent and slightly larger than the KO lenses. 

To verify that as Cx46 replaced Cx50, the only physiological changes would be in gap junction coupling conductance, we first examined other lens biophysical properties. Figure 2 compares the radial distribution of intracellular voltage in the various types of lenses. The data are a scatter of recordings of steady state voltage, measured with respect to the bathing solution, at several radial locations from several lenses of each type. The lens radius is a (in centimeters) and r/a is the normalized distance from the lens center. The smooth curves were generated based on a model of circulating current flow causing the radial gradients in voltage. ¹⁵ The voltage appears to be flatter in the KI lenses, as one would expect, since the values of R_MF are significantly smaller in the KI than WT lenses (see Fig. 4 ). Thus, for the same ionic current densities, one expects there would be smaller voltage gradients in the KI than WT lenses. Indeed, the three theory curves were generated using the same parameters except for the average voltage at r = a (−70 mV for WT, −69 mV for KI[50/46], and −60 mV for KI[50/46]) and the values of R_MF, which are given in the legend to Figure 4 . The fits suggest that the circulating ionic currents are quite similar in the three types of lenses. Moreover, because resting voltage depends on the transmembrane Na⁺ and K⁺ gradients, which are established by the Na/K ATPase, as well as the selectivity of the membranes for Na⁺ and K⁺, these transport properties seem to be similar in the three types of lenses. Last, the resting voltage in the KI(50/46) lenses is somewhat more negative than in KI(46/46), even though the KI(50/46) lenses have the cataract, so the cataract does not appear to be due to compromise of any of the ion transport systems mentioned earlier. 

We also used the lens impedance data to evaluate the values of surface cell membrane conductance (G_S), fiber cell membrane conductance (G_m), and the effective extracellular resistivity (R_e) in WT and KI lenses (Fig. 3) . There were no significant differences in these parameters between the three types of lenses. These results and the data on resting voltage are consistent with the conclusion that gap junctional conductance was the only electrical parameter that changed in these lenses as a result of knocking in Cx46. 

Gap Junction Coupling

Figure 4 illustrates the series resistance (R_S Ω) due to the resistance of all gap junctions between the point of recording (at a radial distance r centimeters from the lens center) and the surface of the lens (at a centimeters from the center; see equations 1 and 2 2 and related discussion). When the R_S data were plotted as a function of normalized radial location (r/a), it was evident that as Cx46 replaced Cx50, the effective intracellular resistivity increased in DFs but decreased in MFs (see the slope of best fit R_S in Fig. 4D ). Analysis of the data provided regional resistivity values in DF and MF for the different mice (see Figs. 4A 4B 4C ). The coupling conductance per unit area of cell-to-cell contact was calculated for DFs (G_DF) and MFs (G_MF), from conductance = 1/(resistivity × cell width). A decrease in G_DF was found as Cx50 was replaced by Cx46: WT G_DF = 0.81 S/cm²; KI(50/46) G_DF = 0.66 S/cm²; KI(46/46) G_DF = 0.47 S/cm². We also measured G_DF as a function of position around the lens and found the same pattern in WT and KI lenses: G_DF is maximum at the equator and minimum at either pole. ⁷ In MF, an increase in G_MF was observed as Cx50 was replaced by Cx46: WT G_MF = 0.43 S/cm²; KI(50/46) G_MF = 0.88 S/cm²; KI(46/46) G_MF = 1.05 S/cm² (see Fig. 5 ). Table 1 shows the normalized coupling conductance and compares it with previously reported data. 

The increase in G_MF when Cx46 replaced Cx50 suggested that the KO Cx46 was functional in the MF whereas Cx50, which was replaced, was not. This was consistent with our previous hypothesis that G_MF depends only on Cx46 channels. ^{10 11} To illustrate the overall basis for this hypothesis, the bottom panel of Figure 5 shows the normalized value of G_MF (normalized to WT G_MF = 0.4–0.5 S/cm², depending on the litter) as a function of the number of alleles (n) expressing Cx46. One sees a continuous progression in the coupling conductance, starting at zero in the Cx46^−/− lenses ¹¹ and reaching a maximum in the KI(46/46) lenses. Moreover, as seen in Table 1 , G_MF in Cx50^+/− lenses was the same as in WT, ¹⁰ also suggesting a lack of functional Cx50 channels in MFs. The only inconsistency was the reduction in G_MF in the Cx50^−/− lenses. However, as previously stated, ¹¹ this could be an indirect effect due to the poor health of these lenses. Thus, the pattern of MF conductance change in the previous and present studies has led to the hypothesis that Cx46 alone is responsible for the coupling of MF. 

The overall pattern of conductance changes in the KI lenses, however, was not so simply explained. In WT lenses, the average conductance drops in the transition from DF to MF (see Table 1 ). Such a decrease is consistent with loss of functional Cx50 channels, but in both the heterozygous and homozygous KI lenses, the measured conductance increased at the DF-to-MF transition to levels that were significantly greater than the DF conductance. Because the MFs today were DFs in the mouse last week, and because the DF-to-MF transition involves protein cleavage, not synthesis, one expected either the same average conductance in the DFs and MFs, or a lower conductance in the MFs than DFs, if some connexins were degraded in the transition. 

Gap Junction Gating

The pH gating of fiber cell gap junctions was assessed by acidification with CO₂ and the results from KI compared with WT lenses. Our impedance method essentially measures the series resistance of all gap junctions between the point of recording (r cm from the center) and the lens surface, which is at a radius of a cm from the center (Fig. 6A) . In WT lenses, when the point of recording was in DFs (b ≤ r ≤ a), all the channels were pH sensitive, and the relative increase in series resistance (R_S Ω) was large as the channels closed. Conversely, when the point of recording was in MF (r ≤ b), the majority of the resistance was due to pH-insensitive channels of the MF; hence, the relative increase in R_S was much less. This is illustrated in Figure 6A . In this particular study, we were primarily interested in whether KI of Cx46 for Cx50 altered the pH sensitivity of either DF or MF channels, and so this simple two-point assessment was used. In previous studies, we have carefully mapped the loss of fiber gap junction pH sensitivity to the DF-to-MF transition. ^{6 7}  

As shown in Figures 6B and 6C , acidification caused closure of DF channels in KI(46/46) lenses (4.6-fold maximum increase in R_S) similar to that seen in WT (4.7-fold maximum increase). As shown in Figure 6D and 6E , gap junction channels of the MF were pH-insensitive in both WT (1.6-fold maximum R_S increase) and KI(46/46) (1.9-fold maximum R_S increase). When the average increases were compared, pH sensitivity of KI(46/46) lenses was not significantly different from that of WT lenses P > 0.05. Moreover, the KI(50/46) lenses had virtually identical gating properties (data not shown). Thus, the pH gating of channels in either type of KI lens was the same as in WT, even though the connexin composition of the channels in KI and WT was different. 

Discussion

In different tissues, cellular growth, differentiation, and homeostasis are based not only on the existence of intercellular communication but also on the specific type of connexins that are coupling the cells. We used lens Cx46 for Cx50 KI mice to assess the effect of changes in connexin diversity on the gross physiological properties of the lenses, on gap junction coupling in the DF and MF zones of these lenses, and on gating of DF and MF gap junction channels. 

Physiological Properties

The assessment of size, resting potential, and clarity was used to evaluate the physiological impact of changing lens connexin gene expression. As first described in White, ¹⁴ KI(46/46) lenses are smaller than WT. This correlated with the small size observed in Cx50 KO lenses, in which Cx50 was also absent. ^{8 11} However, the Cx50 KO lenses were rather unhealthy and had significantly depolarized resting voltages; thus, the possibility existed that the small size was an indirect effect of poor health. The KI(46/46) lenses were much healthier with near normal values of resting voltage, yet the small size persisted. These data are consistent with the hypothesis that channels formed from Cx50 mediate some biochemical signal that is necessary for proper development, and that signal does not pass through channels formed from Cx46; however, the signal mediated by Cx50 is not known. 

Coupling

In the MF zone, G_MF is normally 40% to 50% of the average G_DF. Previous studies of Cx46 or Cx50 KO lenses suggested Cx50 may not contribute to G_MF, presumably because of processing that occurs at the DF-to-MF transition. However, the precise nature of this processing is enigmatic, as the cleaved form of Cx50 is found in WT MFs, ¹² and cleaved Cx50 is functionally active when exogenously expressed. ³ This hypothesis implies the reduction in G_j at the DF-to-MF transition is due to the loss of functional Cx50 channels, and the G_MF is therefore due to the native Cx46 channels, which remain functional through the transition and apparently are stable for the life of the animal. The results on G_MF of KI lenses support this hypothesis, since G_MF increased as Cx46 replaced Cx50, suggesting that functional Cx46 channels were replacing nonfunctional Cx50 channels. The surprising result, however, was the amount of the increase in G_MF, which actually exceeded the average G_DF. 

In WT lenses the conductance contributed by Cx46 channels does not seem to change very much at the DF-to-MF transition. For example, in the Cx50 KO lenses, the G_DF was approximately 0.4 S/cm², the same as the level of 0.4 S/cm² for G_MF in WT lenses. Mefloquine, a quinine derivative, selectively blocks Cx50 channels and not Cx46 channels. ¹⁸ When lenses were treated with mefloquine, G_DF declined to approximately 0.5 S/cm², a level that is again similar to that of G_MF in these lenses (Mathias RT, et al. IOVS 2003;44:ARVO E-Abstract 4266). These results suggested that the coupling conductance due to native Cx46 channels is about the same in the DFs and MFs of WT lenses, and so processing at the DF-to-MF transition did not appear to alter Cx46 very much; it simply removed Cx50 channels from the functional pool. In this context, the results from the KI lenses suggest that when Cx46 was synthesized on the Cx50 gene locus, the resultant channels caused the large increase in conductance at the DF-to-MF transition. This implies that the different locus of gene expression in the knockin produces channels with altered processing at the DF-to-MF transition, and the result is a major increase in the number of open channels in the MF of KI lenses. Measurements of other lens biophysical properties confirmed that gap junctional conductance was the only electrical property that appeared to change in the fiber cells as a result of knocking in Cx46. 

Total coupling conductance depends on the number of channels (n) × open probability (P) × single channel conductance (γ). The open probability for gap junction channels is unusual in that time averages of single channel conductance give different values than ensemble averages based on morphometric estimates of the number of channels in a plaque. ^{15 19 20 21} The former suggest a probability near 1, whereas the latter suggest that only approximately 1 in 10 channels is open (i.e., an apparent probability near 0.1). Based on these data, gap junction plaques contain a large pool of latent channels, which are inhibited from opening by some factor, and a small population of active channels that seldom gate closed. A pool of latent Cx46 channels could be the source of the conductance increase observed at the DF-to-MF transition in the KI lenses. At this time, there is no simple explanation for these data, but they suggest we ought to know more about the molecular events that occur at the DF-to-MF transition. 

The studies presented here report a decrease in G_DF when Cx46 replaces Cx50, but the decrease is not as much as that in the Cx50 KO lenses. This decrease could reflect the lower value of γ for Cx46 (140 pS for Cx46 vs. 220 pS for Cx50 ^{22 23} ), assuming n and P are the same for either type connexin when expressed on the same Cx50 gene locus. Unfortunately, we do not know the value of either n or P in any of the conditions studied. Nevertheless, these results are not inconsistent with the hypothesis that in WT lenses, both connexins contribute similarly to G_DF. 

Gating

Gating of gap junction channels in the KI(46/46) or KI(46/50) lenses was indistinguishable from gating in WT lenses. This observation in isolation would not have been remarkable; however, in the context of results from previous studies, this result suggests our previous hypotheses on regulation of gap junction gating were too simple. 

Previous studies have shown that acidification causes G_DF to go to zero in WT lenses. ¹⁵ Studies of Cx50 KO lenses showed that removal of Cx50 reduced G_DF to 40% normal, and after acidification, G_DF remained at approximately 40% of normal, implying the remaining homotypic Cx46 DF channels were pH insensitive. ¹¹ When lenses were exposed to mefloquine, a selective blocker of Cx50 channels, G_DF dropped to approximately half normal and the remaining DF channels became pH insensitive (Mathias RT, et al. IOVS 2003;44:ARVO E-Abstract 4266). These results suggested two hypotheses on gating of DF channels: (1) in WT lenses, the Cx46 channels lack intrinsic pH sensitivity; (2) the pH-mediated total closure of all DF channels in WT lenses is due to cooperative interactions between Cx50 and Cx46 channels. The lack of intrinsic pH-sensitivity of Cx46 was surprising, because when it was exogenously expressed, it always formed pH-sensitive channels. ^{13 24 25} However, pH sensitivity can be altered by posttranslational processing, and so we assumed that the lens simply processed Cx46 differently from the exogenous expression systems. These previous results suggested a third interesting hypothesis linking the reduction in coupling and loss of gating at the DF-to-MF transition; (3) processing at the DF-to-MF transition renders Cx50 channels nonfunctional, leaving intrinsically pH-insensitive Cx46 channels to mediate coupling of MF. Although the data on gating of the KI lenses are not incompatible with these relatively simple hypotheses on WT lenses, they imply that there are additional complexities in the posttranslational processing of gap junction proteins in the lens. 

Acidification caused closure of DF channels in KI lenses similar to that in WT lenses, and in contrast to that in Cx50 KO or mefloquine-treated lenses. These new data suggest that it is the locus of gene expression that determines whether Cx46 channels are pH sensitive or insensitive, and it is the locus of gene expression, rather than connexin isoform, that regulates pH-gating of G_DF. In the lens, the stage of cell differentiation and protein expression depend on location. The two fiber cell connexins are expressed at different locations and stages of differentiation ^{9 12} ; hence, there are likely to be differences in posttranslational processing of Cx46 when it is expressed normally versus on the Cx50 gene locus. Posttranslational modifications can affect the pH sensitivity of Cx46, ²⁶ so this is one possible explanation for the locus dependent difference in gating properties of Cx46. More satisfying molecular explanations will require a great deal more work on processing of connexin proteins and on the relationship between the channels synthesized on the Cx50 and Cx46 gene loci in the lens. 

If hypothesis 3 on loss of gating at the DF-to-MF transition were strictly accurate, then in the KI(46/46) lenses, the channels coupling MF should have been pH sensitive. That is, because G_MF increases in KI relative to WT lenses, the pH sensitive Cx46 channels synthesized on the Cx50 gene locus appear to be functional in MF. Thus, in the KI lenses, rather than loss of functional pH-sensitive channels at the DF-to-MF transition, the pH-sensitive channels have been rendered pH insensitive at the DF-to-MF transition. This could be a new regulatory mechanism that is turned on in the KI lenses or it could be a sort of insurance policy that is always active and insures that MF junctions are not pH sensitive. Again, resolution of these questions will require extensive further work using ultrastructure and protein chemistry to understand what happens at the DF-to-MF transition. 

Summary

This study has correlated size and clarity with data on resting voltage, membrane conductance, gap junction coupling conductance, and gating of gap junction channels in the DFs and MFs of Cx46 KI lenses. We have unexpectedly found that the locus of gene expression is as important a determinant of connexin function as the protein sequence of the isoform that is being expressed. For example, normal growth and development depended on the presence of Cx50, and therefore protein sequence is critical for this function. However, either Cx50 or Cx46, when expressed on the Cx50 gene locus, produced clear lenses with normal resting voltages; thus, locus of gene expression is more important than connexin sequence for maintaining the homeostasis of central fiber cells. Moreover, the magnitude of coupling conductance depended on whether Cx46 or Cx50 was expressed on the Cx50 gene locus, and so this particular property depends on protein sequence rather than gene locus. In contrast, the gating of gap junction channels expressed on the Cx50 gene locus was the same in KI and WT, and so regulation of gap junction gating depends only on the locus of gene expression. 

In the lens there are at least two major sites that result in protein modifications. First, immediately after synthesis, each connexin is subjected to posttranslational modifications during channel assembly. ²⁷ Later during differentiation, at the DF-to-MF transition, further modifications are made to the lens connexins. ^{2 28} Cx50 and Cx46 are synthesized at different locations in the lens, the former primarily in the epithelium, and the latter primarily in newly DF cells. These different sites of synthesis involve different kinases, proteases, regulatory proteins and enzyme activity, and therefore it is not surprising that locus of gene expression affects connexin function in the lens. At the DF-to-MF transition, all connexins are exposed to the same set of enzymes. Nevertheless, the effect of this transition on the coupling conductance due to Cx46 channels depended on the gene locus of expression of this connexin. This suggests that Cx46 channels formed when synthesis is on the Cx50 gene locus are physically different from Cx46 channels formed when synthesis is on the Cx46 gene locus. These results raise important questions about the molecular mechanisms and roles of posttranslational modifications of connexin proteins in the lens. Although this study does not provide the answers, a question can never be answered until one knows the question exists, and progress has been made in identifying these new questions. 

 Supported by National Eye Institute Grants EY13163, EY06391, and EY02430.

 Submitted for publication April 19, 2004; revised May 21, 2004; accepted June 1, 2004.

 Disclosure: F.J. Martinez-Wittinghan, None; C. Sellitto, None; T.W. White, None; R.T. Mathias, None; D. Paul, None; D.A. Goodenough, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

The functional zones of the normal lens. The lens can be divided into epithelial and fiber cells; the lens fiber cells can be further divided into DFs and MFs. The lens radius is a ∼0.10 cm in WT mice; the radial location of the DF-to-MF transition is at b ∼ 0.8a cm.

Figure 2.

 

Resting voltages in the KI and WT lenses. The data are the recorded intracellular voltages in several lenses of each type at various normalized radial distances (r/a) from the lens center. The smooth curves are based on a model presented in Mathias et al. ¹⁵ for radial gradients in resting voltage due to flow of intracellular current from central to surface cells. The different types of lenses all had relatively normal resting voltages, even though the KI(50/46) lenses had a central opacity and the KI(46/46) lenses were undersized.

Figure 3.

 

Nonjunctional biophysical properties. The best fit parameters determined by fitting Equations 2-6 3 4 5 6 to the data are given in A–D. A) Surface cell conductance (mS/cm²): WT = 2.39 ± 0.75, KI(50/46) = 2.61 ± 0.55, and KI(46/46) = 1.74 ± 0.40. B) Surface cell capacitance (μF/cm²): WT = 7.05 ± 5.53; KI(50/46) = 6.22 ± 5.22; and KI(46/46) = 6.23 ± 5.93. C) Inner fiber cell membrane conductance (μS/cm²): WT = 6.36 ± 5.02; KI(50/46) = 4.91 ± 3.05; KI(46/46) = 4.51 ± 3.64. D) Effective extracellular resistivity (kΩ-cm): WT = 34.6 ± 32.08; KI(50/46) = 34.4 ± 22.05, KI(46/46) = 32.4 ± 9.85. The parameter values are MEAN ± SD for n lenses. WT n = 38; KI(50/46) n = 18; KI(46/46) n = 31.

Figure 4.

 

Radial R_S distribution in KI lenses. R_S is plotted as a function of normalized radial location (r/a) for (A) WT: 51 points from 23 lenses; (B) KI(50/46): 11 points from 8 lenses; and (C) KI(46/46): 51 points from 24 lenses. R_DF and R_MF (mean ± SEM) were calculated in several lenses and used to determine the fit shown. (D) A comparison of the three theory curves zoomed in at the DF-to-MF transition. As Cx46 replaced Cx50, R_DF (resistivity in KΩ cm²) increased and R_MF (resistivity in Ω cm²) decreased, producing an increase in R_S (resistance in KΩ) in DFs. The change is not so obvious for the R_S of MFs, since it is the integral of both R_DF and R_MF.

Figure 5.

 

Gap junctional coupling. Top: normalized coupling conductance in DF; middle: normalized coupling conductance in MF. Data obtained from the fitting parameters equation in 2. Bottom: MF Coupling Conductance as a function of the number of gene loci (N) expressing Cx46. The normalized (to WT G_MF) G_MF is shown for the Cx46^−/− and Cx46^+/−, WT, KI(50/46), and KI(46/46).

Table 1.

 

A Comparison of Normalized Coupling Conductances A. Coupling Conductance in KI Mice

Zone	Genotype
		WT	KI(50/46)	KI(46/46)
GDF	1.00	0.82	0.58
GMF	0.53	1.08	1.29
B. Coupling Conductance in KO Mice
		+/+	+/−	−/−
Cx46 KO	GDF	1.00	0.66	0.32
	GMF	0.39	0.20	0.00
Cx50 KO	GDF	1.00	0.72	0.44
	GMF	0.40	0.41	0.13

 Table 1B is reproduced from Baldo et al. ¹¹ The Journal of General Physiology, 2001;118:447–456 by permission of The Rockefeller University Press.

 The coupling conductance per unit area of cell-to-cell contact was calculated for DF and MF from the R_DF and R_MF values, and then for purposes of comparison, each data set was normalized to the value of conductance in the DF of WT lenses.

Figure 6.

 

Gating of fiber cell gap junction channels. Gating of fiber cell gap junction channels was assessed by their ability to close in response to cytoplasmic acidification after perfusion with a solution bubbled with 100% CO₂. (A) The series resistance (R_S) recorded in a normal lens. Our method records the series resistance of all gap junction channels between the point of voltage recording and the surface of the lens. In a normal lens, the channels in the DF are pH sensitive; hence, there is a large increase in R_S when pH drops. Conversely, when the point of voltage recording is in the MF, where the channels are pH insensitive, R_S is larger, but the increase in R_S when pH decreases is relatively small. If the channels of the MF were pH sensitive, the relative increase in R_S would be the same regardless of where the voltage was recorded. (B) and D) show this pattern of changes in R_S in WT lenses, where other studies have determined that the loss of pH sensitivity occurs at the DF-to-MF transition. ⁶ (C) and (E) show the same pattern of change in the relative values of R_S in the KI(46/46) lenses, indicating that the channels in the DFs are pH-sensitive, whereas channels in the MFs are not pH-sensitive. The observed differences in time course are not significantly different. The data shown are one experiment representative of n = 3.

The authors thank Peter Brink for thoughtful discussion of the data.