Frog Lens Research Articles

Ecology drives patterns of spectral transmission in the ocular lenses of frogs and salamanders

Abstract The spectral characteristics of vertebrate ocular lenses affect the image of the world that is projected onto the retina, and thus help shape diverse visual capabilities. Here, we tested whether amphibian lens transmission is driven by adaptation to diurnal activity (bright light) and/or scansorial habits (complex visual environments). Spectral transmission through the lenses of 79 species of frogs and six species of salamanders was measured, and data for 29 additional frog species compiled from published literature. Phylogenetic comparative methods were used to test ecological explanations of variation in lens transmission and to test for selection across traits. Lenses of diurnal (day‐active) and scansorial (climbing) frogs transmitted significantly less shortwave light than those of non‐diurnal or non‐scansorial amphibians, and evolutionary modelling suggested that these differences have resulted from differential selection. The presence of shortwave‐transparent lenses was common among the sampled amphibians, which implies that many are sensitive to shortwave light to some degree even in the absence of visual pigments maximally sensitive in the UV. This suggests that shortwave light, including UV, could play an important role in amphibian behaviour and ecology. Shortwave‐absorbing lens pigments likely provide higher visual acuity to diurnally active frogs of multiple ecologies and to nocturnally active scansorial frogs. This new mechanistic understanding of amphibian visual systems suggests that shortwave‐filtering lenses are adaptive not only in daylight conditions but also in those scotopic conditions where high acuity is advantageous. Read the free Plain Language Summary for this article on the Journal blog.

Homology modeling of ρ-crystallin from bullfrog ( Rana catesbeiana) lens

ρ-Crystallins are major protein component found in the eye lenses of frogs of the genus Rana. Structural analysis has indicated that frog ρ-crystallins belong to aldo–keto reductase superfamily (AKRs) which include aldehyde and aldose reductases, prostaglandin F synthase and several detoxification enzymes. Members of AKRs catalyze the oxidation–reduction reaction over a range of substrates using NAD(P)(H) as a cofactor. In spite of higher structural similarity with AKRs and cofactor binding affinity, the ρ-crystallins were found to be catalytically inactive. This study presents comparative or homology modeling of ρ-crystallin from bullfrog ( Rana catesbeiana) in presence and absence of cofactor NADP and a competitive inhibitor, testosterone. The predicted models are explored to examine the catalytic cleft, cofactor binding affinity characteristics and substrate binding pocket.

The 23-kDa polypeptide from Common Frog lens: isolation, characteristics and cellular localization

The major water-soluble polypeptide with molecular weight of approximately 23 kDa (the 23-kDa polypeptide) was identified in the lens of common frog Rana temporaria L. According to the gel filtration data, the peptide is a part of an oligomeric protein with molecular weight of more than 300 kDa (alpha-crystallin fraction). A highly pure fraction of the 23-kDa polypeptide was isolated by two-step ion-exchange chromatography and SDS electrophoresis and the specific antibodies were obtained. Immunohistochemistry showed the presence of the 23-kDa polypeptide in the cytoplasm of lens epithelial cells (including its central region) and in the zones neighbouring the plasma membranes in cortical fibers. The 23-kDa polypeptide was not found in the lens central zone (nucleus). It was also present in the retina (in the cells of inner nuclear layers), but not in the other tissues and organs of adult frog. Immunochemical analysis showed that the 23-kDa polypeptide was different from all known crystallins of frogs and other animals (bull, mouse, rat, and chicken). The nature of the 23-kDa polypeptide and the relation of its expression with lens cell differentiation are discussed.

Taxon-specific ζ-Crystallin in Japanese Tree Frog (Hyla japonica) Lens

The present study demonstrated that the 38-kDa protein, instead of rho-crystallin (36 kDa), is expressed taxon specifically in the lens of Japanese tree frog (Hyla japonica). The 38-kDa protein was distinguished from rho-crystallin expressed in the lenses of bullfrog (Rana catesbeiana) and European common frog (Rana temporaria) immunochemically. Although the N terminus of the 38-kDa protein was blocked, the analyses of partial amino acid sequences showed that the protein was zeta-crystallin. Analysis of cDNA sequence encoding zeta-crystallin of the tree frog lens demonstrated that the deduced protein consisted of 329 amino acids including initial methionine and having 62.2 and 62.9% identity with zeta-crystallin of camel and guinea pig lenses, respectively. The molecular mass of the deduced structure was calculated to be 35,564 Da. zeta-Crystallin of the tree frog lens exhibited the intrinsic enzymatic activity of quinone reductase (EC, NADPH:quinone oxidoreductase). The crystallin specifically catalyzed the reduction of 9,10-phenanthrenequinone (Km, 42 microm) using NADPH (Km, 60 microm) as a cofactor. The enzymatic activity was inhibited by dicumarol, anti-coagulant drug, with IC50 of 4 microm. On gel filtration chromatography, the crystallin was recovered as 150-kDa molecular mass complex, indicating that the crystallin was homotetramer consisting of 38-kDa subunits. The crystallin gene was expressed specifically in the lens. These results show that taxon-specific crystallins such as zeta- and rho-crystallins may be available for the biochemical discrimination of Hyla- and Rana groups among frogs.

Evolution of the aldose reductase-related gecko eye lens protein rhoB-crystallin: a sheep in wolf's clothing.

rhoB-crystallin (AJ245805) is a major protein component (20%) in the eye lens of the gecko Lepidodactylus lugubris. Limited peptide sequence analysis earlier revealed that it belongs to the aldo-keto reductase superfamily, as does the frog lens rho-crystallin. We have now determined the complete cDNA sequence of rhoB-crystallin and established that it is more closely related to the aldose reductase branch of the superfamily than to frog rho-crystallin. These gecko and frog lens proteins have thus independently been recruited from the same enzyme superfamily. Aldose reductase is implicated in the development of diabetic cataract in mammals, and, if active, rhoB-crystallin might be a potential risk for the gecko lens. Apart from a replacement 298 Cys --> Tyr, rhoB-crystallin possesses all amino acid residues thought to be required for catalytic activity of the aldose reductases. However, modeling studies of the rhoB-crystallin structure indicate that substrate specificity and nicotinamide cofactor affinity might be affected. Indeed, neither recombinant rhoB-crystallin nor the reverse mutant 298 Tyr --> Cys showed noticeable activity toward aliphatic and aromatic substrates, although cofactor binding was retained. Various other oxidoreductases are known to be recruited as abundant lens proteins in many vertebrate species; rhoB-crystallin demonstrates that an aldose reductase-related enzyme also can be modified to this end.

Ascorbate Free Radical Reductase Activity in Vertebrate Lenses of Some Species

Purpose: To clarify the function of ascorbate free radical (AFR) reductase in the lens antioxidation mechanism, we investigated the difference among species in AFR reductase activity in different vertebrate lenses. Materials and Methods: Soluble and insoluble fractions were prepared from the lenses of frogs, guinea pigs, rats, rabbits, pigs, and calves. AFR reductase and diaphorase activity of each fraction was determined. Results: AFR reductase activity in the lens soluble fraction was the highest in frogs. That of guinea pigs and rabbits was at the next level; there was only a little activity in rats and pigs, and none was detected in calves. Membrane-bound AFR reductase in the lens insoluble fraction was extracted by 0.3% Triton X-100. The membrane-bound enzyme activity was almost at the same level in frogs, rats, rabbits, and calves, and a little higher in guinea pigs and pigs. However, such species-specificity of AFR reductase activity as in the soluble fraction was not observed in 0.3% Triton X-100 extracts. Diaphorase activity was 3 to 9 times as much as AFR reductase activity in the soluble fractions of frogs, guinea pigs, and rabbits, but in 0.3% Triton X-100 extracts of all vertebrate species used, it was very high, 108 to 311 times the AFR reductase activity. Conclusion: These results suggest that the lens soluble and membrane-bound AFR reductases are individual enzyme molecules and have different anti-oxidative functions. The lenses of frogs, guinea pigs, and rabbits contain a near-ultraviolet (UV) light absorbing compound, reduced pyridine nucleotide at a high concentration. Therefore, the soluble AFR reductase activity may be high in the vertebrate lenses with a near-UV light filter, and enhance the antiphotoxidation of ascorbic acid.

Isoform-specific function and distribution of Na/K pumps in the frog lens epithelium.

Epithelial cells from the anterior and equatorial surfaces of the frog lens were isolated and used the same day for studies of the Na/K ATPase. RNase protection assays showed that all cells express alpha(1)- and alpha(2)-isoforms of the Na/K pump but not the alpha(3)-isoform, however the alpha(2)-isoform dominates in anterior cells whereas the alpha(1)-isoform dominates in equatorial cells. The whole cell patch-clamp technique was used to record functional properties of the Na/K pump current (I(P)), defined as the current specifically inhibited by dihydro-ouabain (DHO). DHO-I(P) blockade data indicate the alpha(1)-isoform has a dissociation constant of 100 microm DHO whereas for the alpha(2)-isoform it is 0.75 microm DHO. Both alpha(1)- and alpha(2)-isoforms are half maximally activated at an intracellular Na(+)-concentration of 9 mm. The alpha(1)-isoform is half maximally activated at an extracellular K(+)-concentration of 3.9 mm whereas for the alpha(2)-isoform, half maximal activation occurs at 0.4 mm. Lastly, transport by the alpha(1)-isoform is inhibited by a drop in extracellular pH, which does not affect transport by the alpha(2)-isoform. Under normal physiological conditions, I(P) in equatorial cells is approximately 0.23 microA/microF, and in anterior cells it is about 0.14 microA/microF. These current densities refer to the area of cell membrane assuming a capacitance of around 1 microF/cm(2). Because cell size and geometry are different at the equatorial vs. anterior surface of the intact lens, we estimate Na/K pump current density per area of lens surface to be around 10 microA/cm(2) at the equator vs. 0.5 microA/cm(2) at the anterior pole.

Ion, water and neutral solute transport in Xenopus oocytes expressing frog lens MIP

We have expressed frog (Rana pipiens) lens major intrinsic protein (MIP) in Xenopus oocytes and observed its effect on ion conductance, water permeability and neutral solute transport. SDS-PAGE and immunoblotting demonstrated oocytes injected with MIP mRNA expressed the protein at high levels. Immunolocalization indicated the expressed MIP migrated to the plasma membrane. MIP had no effect on the slope of oocyte I-V relations in the range -50 to +10 mV, although the averaged I-V curve was shifted 10 mV positive to control. MIP increased oocyte water permeability by a factor of 1.9 +/- 0.2, whereas the permeability to sucrose, 2-deoxyglucose, inositol, sorbitol, reduced glutathione or urea was unchanged. Glycerol permeability was enhanced in oocytes expressing MIP. In contrast to control oocytes, 3H-glycerol radioactivity accumulation did not follow first order kinetics. Radioactivity continued to accumulate even after 19 h of uptake and went beyond equilibrium with the bath. The time course of MIP-mediated glycerol uptake was modeled assuming metabolic trapping with good results. Based on this model, MIP increased oocyte glycerol permeability by a factor of 2.7.

A Correlated Study of Metabolic Cell Communication and Gap Junction Distribution in the Adult Frog Lens

By using low molecular weight dye injection and laser scanning confocal microscopy it has been possible to assess the degree of dye communication in several regions (intraepithelial, epithelium-fibre cell, fibre-epithefium and fibre-fibre) of the adult frog lens. These same areas of the lens have also been examined by freeze fracture electron microscopy for the presence and organisation of gap junctions. Epithelial cell lateral membranes have numerous gap junctions which efficiently transmitted dye to their neighbours though none was detectable in the underlying fibre cells. This was correlated with an apparent absence of gap junctions at the epithelium/fibre cell interface. Dye spread between cortical fibre cells was only observed in a subset of fibres in the bow region of the lens. Neither mature cortical fibres nor immature bow fibres appeared to be dye-coupled and we detected no dye passed from fibre cells to adjacent epithelial cells at their anterior ends. This pattern of dye communication was also correlated with the apparent absence of recognisable gap junctions on the lateral membranes of either the recently differentiated bow fibres or the mature cortical fibres. Classical gap junctions were only found on the membranes of fibres between five and ten cells in from the lens bow, i.e. the subset of fibres which were dye-coupled. No gap junctions were found between deeper cortical fibres or nuclear fibres, although they were characterised by a number of square arrays. Though electrically well coupled the adult frog lens may be relatively poorly dye coupled and this could depend on the age and differentiation stage of the cells concerned. The model of a freely communicating lens clearly requires re-examination by correlated physiological and morphological studies.

Effects of Hydrogen Peroxide Oxidation and Calcium Channel Blockers on the Equatorial Potassium Current of the Frog Lens

Hydrogen peroxide, in concentrations of 10-1000 μM, produces two major changes in the current-voltage relationships associated with the equatorial potassium current of the lens. First, the resting and reversal potentials become more negative than they were prior to treatment with hydrogen peroxide and second, the membrane resistance related to the equatorial current is decreased. The shift in the resting and reversal potentials is in the opposite direction from that produced by ouabain. Based on the Nernst equation, the shift in the reversal potential suggests that there is an increase in the concentration of potassium in the lens. The 88Rb uptake and efflux are increased. These observations suggest that hydrogen peroxide stimulates the Na,K-pump. The decrease in membrane resistance is inhibited by 100 μM of quinine, a calcium-dependent potassium channel blocker, and does not decrease in a calcium-free medium. This suggests that the decrease in resistance may be secondary to an increase in lenticular calcium. These effects of hydrogen peroxide are similar to those of p-chloromercuriphenylsulfonate (pCMPS), a nearly impermeant sulfhydryl binding agent, and suggest that permeant hydrogen peroxide may increase calcium influx by acting on sulthydryl groups on the outer surface of lens membranes. Verapamil, a calcium channel blocker, is reported to prevent cataract formation. D600, the methoxy analogue of verapamil, is a calcium channel blocker that increases the resistance associated with the equatorial current in the presence and absence of hydrogen peroxide. The gadolinium ion has a similar effect. Thus, D600 and Gd3+ partially counteract the reduction in membrane resistance produced by 50 μM hydrogen peroxide.

Molecular cloning and expression of an abundant rabbit ovarian protein with 20 alpha-hydroxysteroid dehydrogenase activity.

An abundant 37-kDa protein, which comprises up to 30% of the soluble proteins of the ovary, has been found to have 20 alpha-hydroxysteroid dehydrogenase (20 alpha HSD) activity. The steroidogenic enzyme 20 alpha HSD regulates the conversion of progesterone to 20 alpha-hydroxyprogesterone in many mammalian species. Complimentary DNA clones encoding a unique and abundant 20 alpha HSD were isolated from a mature rabbit ovary library using guinea pig antisera generated to the purified 37-kDa protein and from a 5' EcoRI fragment from the initial positive clone. A full-length cDNA clone of 1217 basepairs encoding a 323-amino acid protein with an estimated mol wt of 37 kilodaltons was obtained. Amino acid sequence data indicate a similarity to human chlordecone reductase, bovine lung prostaglandin F synthase, human aldose reductase, human aldehyde reductase, and frog lens rho-crystallin, placing rabbit ovarian 20 alpha HSD in the aldo-keto reductase family of proteins. Northern blot analysis demonstrated a 1.2-kilobase mRNA in the interstitial tissue of mature rabbit ovaries and, to a lesser extent, in corpora luteal tissue. 20 alpha HSD was expressed in bacteria as a recombinant protein and was shown to possess enzymatic activity, preferring NADP as a cofactor. These studies demonstrate that an abundant ovarian protein belonging to the superfamily of NADP-dependent aldo-keto reductases has 20 alpha HSD activity. This is the first example of an abundant crystallin-related protein with known enzymatic activity in a tissue other than the lens.

Topical Norepinephrine, Isoproterenol, Terbutaline, Forskolin and 8-Bromo Cyclic AMP Increase Lens Equatorial Current

The equatorial current of the frog lens was studied with microelectrodes and a vibrating probe. Norepinephrine, isoproterenol and terbutaline applied topically at the surface of the cornea as a 1% solution increased the equatorial current of the lens about 80%. This increase was blocked by timolol. Topical forskolin increased the equatorial current by over 80 %. This effect of forskolin was not blocked by timolol. These substances did not affect the equatorial current of the lens when the lens was suspended in a frog Ringers solution. 8-Bromo cyclic AMP increased the equatorial current when it was applied topically but not when it was added directly to the medium bathing an isolated lens. The results suggest that β-adrenergic agonists, applied topically to the cornea, increase the equatorial current of the lens by a mechanism that is mediated by cyclic AMP. The finding that 8-bromo cyclic AMP did not act directly on the lens suggests that a currently unknown substance or process is active at the level of the lens.

The cDNA sequence encoding the major intrinsic protein of frog lens

A cDNA clone encoding the frog lens major intrinsic protein (MIP) has been isolated and sequenced. The predicted protein of 28 kDa has high sequence identity and similarity to mammalian and avian lens MIP sequences. Frog lens MIP is encoded by a transcript of 4.4 kb.

Molecular cloning of rat liver 3α-hydroxysteroid dehydrogenase and identification of structurally related proteins from rat lung and kidney

3α-Hydroxysteroid dehydrogenase and related enzymes play important roles in the metabolism of endogenous compounds including androgens, corticosteroid, prostaglandins and bile acids, as well as drugs and xenobiotics such as benzo( a)pyrene. Complementary DNA clones encoding 3α-hydroxysteroid dehydrogenase were isolated from a rat liver cDNA lambda gt11 expression library using monoclonal antibodies as probes. A full-length cDNA clone of 1286 base pairs contained an open reading frame encoding a protein of 322 amino acids with an estimated M w of 37 kD. When expressed in E. coli, the encoded protein migrated to the same position on SDS-polycrylamide gels as the enzyme in rat liver cytosols. The protein expressed in bacteria was highly active in androsterone oxidation in the presence of NAD as cofactor and this activity was inhibited by indomethacin, a potent inhibitor of 3α-hydroxysteroid dehydrogenase. The predicted amino acid sequence of 3α-hydroxysteroid d dehydrogenase was related to sequences of several other aldo-keto reductases such as bovine prostaglandin F synthase, human chlordecone reductase, human aldose reductase, human aldehyde reductase and frog lens epsilon-crystallin, suggesting that these proteins belong to the same gene family. Recently, we have found that monoclonal antibodies against 3α-hydroxysteroid dehydrogenase also recognized multiple antigenically related proteins in rat lung, kidney and testis. Further screening of liver, lung and kidney cDNA libraries using these monoclonal antibodies as probes resulted in the isolation of additional five different cDNAs encoding proteins with high degree of structural homology to rat liver 3α-hydroxysteroid dehydrogenase.

Potassium currents from isolated frog lens epithelial cells

Using the perforated patch version of whole-cell recording, we have measurd currents from isolated frog lens epithelial cells. Three types of currents were seen. A time-independent outwardly rectifying potassium current was identified that sets the resting voltage. This potassium current differs significantly from any of the potassium currents recorded with the whole-cell technique in mammalian lens epithelial cells. In addition to the potassium current, the two other currents present were both outwardly rectifying: one was time-independent while the other showed distinct activation.

LEVELS OF REDUCED PYRIDINE NUCLEOTIDES AND LENS PHOTODAMAGE

Since most of the known factors that are associated with cataract formation are oxidative in nature, one would expect that a highly reductive environment might arrest or retard the progress of cataract formation. Reduced nucleotides, both NADH and NADPH, are potent reductants with a large negative redox potential of -320 mV. Lenses of certain species contain high levels of these nucleotides, presumably due to the presence of taxon specific crystallins. We have utilized this situation to investigate whether the levels of reduced pyridine nucleotides modulate photo-oxidative damage to the lens. We have monitored the time dependent loss of tryptophan fluorescence upon photodamage for lenses from guinea pig, rabbit and frog (Rana) that contain high levels of pyridine nucleotides and compared with the lenses from rat, Xenopus and a mutant strain of guinea pig that contain significantly lower amounts of these nucleotides. About 75% and 90% of the initial fluorescence intensity is lost in the case of rat and Xenopus lenses, respectively, after a total of 35 min exposure. Rabbit, guinea pig and frog lenses, under identical conditions, show only about 35-40% loss of the initial fluorescence. It appears that the lenses that contain high levels of reduced nucleotides are less susceptible to photodamage. The observed anti-oxidative role of reduced nucleotides in the lenses indicates the possibility of testing reductants (NADPH, NADH and their functional analogues) as potential candidates to therapeutically intervene in the process of cataractogenesis.

Divalent cation effects on lens conductance and stretch-activated cation channels

In patch clamp studies of apical membrane from frog lens epithelium, the most frequently observed channel is ‘stretch-activated’, highly selective for cations over anions but showing little selectivity for Na − vs. K +. In normal physiological saline, the open channel conductance is 25–30 pS and quite linear over ±100 mV. In the absence of extracellular divalent ions, the open channel conductance for inward current flow increases to about 50 pS at the normal lens resting voltage of −75 mV, whereas the conductance for outward current flow is unaffected. In the intact lens, removal of extracellular divalents causes the input conductance approximately to double and the intracellular voltage to depolarize from −74 to −58 mV. A variety of divalent ions block this change in whole lens conductance and voltage in the same order in which they block the ‘stretch channels’. Single voltage-clamped epithelial cells also increase their conductance when Ca 2+ is removed from their bathing medium. There are, therefore, some striking parallels between the open channel properties of the ‘stretch-activated’ cation channel and the response of the whole lens or single lens cells to removal of extracellular Ca 2+. There are also inconsistencies. This channel is apparently not open in the normal resting lens so removal of extracellular Ca 2+ must cause it to open if it is indeed responsible for the increase in lens conductance. However, we have not been able to demonstrate convincingly an increase in open probability at the single-channel level when external divalents are removed.

Cell to cell communication and pH in the frog lens.

Fiber cells of the lens are electrically and diffusionally interconnected through extensive gap junctions. These junctions allow fluxes of small solutes to move between inner cells and peripheral cells, where the majority of transmembrane transport takes place. We describe here a method utilizing two intracellular microelectrodes to measure the cell to cell resistance between fiber cells at any given distance into the intact lens. We also use ion-sensitive microelectrodes to record intracellular pH at various depths in the intact lens. We find that gap junctions connecting inner fiber cells differ in pH sensitivity as well as normal coupling resistance from those connecting peripheral cells. The transition occurs in a zone between 500 and 650 microns into the lens. Fiber cells peripheral to this zone have a specific coupling resistance of 1.1 omega cm2, whereas those inside have a specific coupling resistance of 2.7 omega cm2. However, when the cytoplasm of fiber cells is acidified by bubbling with CO2, peripheral cells uncouple and the cell to cell resistance goes up more than 40-fold, whereas junctions inside this zone are essentially unaffected by changes in intracellular pH. In a normal frog lens, the intracellular pH in fiber cells near the lens surface is 7.02, a value significantly alkaline to electrochemical equilibrium. Our data suggest that Na/H exchange and perhaps other Na gradient-dependent mechanisms in the peripheral cells maintain this transmembrane gradient. Deep in the lens, the fiber cell cytoplasm is significantly more acidic (pHi 6.81) due to influx of hydrogen across the inner fiber cell membranes and production of H+ by the inner fiber cells. Because of the normally acid cytoplasm of interior fiber cells, their loss of gap junctional sensitivity to pH may be essential to lens survival.

Sodium channels in ocular epithelia.

Voltage-gated, tetrodotoxin(TTX)-blockable sodium channels are found in most excitable cells and are the primary contributors to action potentials generated by many of these cells. To date, there has only been one report of a non-cultured vertebrate epithelial cell type containing TTX-blockable Na+ channels: rabbit non-pigmented ciliary body epithelial cells [Cilluffo MC et al. (1991) Invest Opthalmol Vis Sci 32: 1619-1629], and three reports of cultured epithelial cells containing TTX-blockable Na+ channels: rabbit non-pigmented and pigmented ciliary body epithelium [Ciluffo MC et al. (1991) Invest Opthalmol Vis Sci 32: 1619-1629; Fain GL, Farahbakhsh (1989) J Physiol (Lond) 417: 83-103] and human lens epithelium [Cooper K et al. (1990) J Membr Biol 117: 285-298]. We report here the presence of sodium currents in two different non-cultured, freshly dissociated transporting epithelial cell types: the rabbit corneal endothelium and the frog lens epithelium. We also report the occurrence of sodium currents in six additional cultured ocular epithelial cell types from three different species. These currents have a current/voltage (I/V) relationship consistent with traditional voltage-gated Na+ currents, are quinidine- and TTX-blockable (of the low-affinity TTX-sensitive type), and disappear following bath substitution of Na+ with Cs+ or K+.

Microtubules rich in post-translationally modified α-tubulin form distinct arrays in frog lens epithelial cells

Isolated frog lens epithelia were stained with antibodies against tyrosinated, detyrosinated or acetylated α-tubulin and observed by several means including a scanning confocal microscope. The most prominent feature of Rana pipiens lens cells was a primary cilium close to the apical surface of the cells above the centrosome. This structure was associated with microtubules rich in modified α-tubulin. The cilium was less pronounced but still discernible in the cells of another species R. ridibunda. In both species, the modified (acetylated or detyrosinated) microtubules formed arrays spatially distinct from the unmodified (tyrosinated) microtubules. The modified microtubules formed a basket of microtubules with a curly distribution around the nucleus while the tyrosinated array consisted predominantly of rather straighter microtubules running from the apical centrosome to the cell periphery, down the lateral sides of the cells and across the basal surface adjacent to the lens capsule and basement membrane. It is concluded that the organization of modified microtubules previously described for several types of cultured cells may represent a remnant of the three-dimensional perinuclear array of such microtubules described here for the cells of an intact epithelium.