Investigative Ophthalmology & Visual Science, Vol. 31, No. 8, August 1990Copyright © Association for Research in Vision and Ophthalmology

Distribution of Carbonic AnhydraseAmong Human Photoreceptors

T. Michael Nork, Steven A. McCormick,* Gung-Mei Chao, and J. Vernon Odom

The distribution of carbonic anhydrase (CA) among human photoreceptors has been a topic of dispute.In our experiments, by modifying an established enzyme histochemical technique, reproducible stain-ing was observed. Of the cones in the peripheral retina, 91% were positive for CA. The CA-negative(CA—) cones were absent within approximately 8 arc min of the foveal center and their density peakedat 2 arc deg. No CA activity was found in the rods. Morphologic differences were noted between theCA-positive (CA+) and CA— cones. Compared to the CA+ cones, the CA— cones had longer and morenearly columnar inner segments, more nearly spherical nuclei, and more generous amounts of peri-karyal cytoplasm. In the peripheral retina, the distance between CA+ to CA+ nearest neighbors werelarger compared to CA— to CA+ nearest neighbors (P < 0.0001). The frequency distribution andmorphology of the CA— cones suggest that they are the blue-sensitive cones. As such, this studydemonstrates a biochemical similarity between blue cones and rods that may provide insight into thefunction and phylogeny of the blue cones. Invest Ophthalmol Vis Sci 31:1451-1458,1990

In recent years, there have been exciting advancesin our understanding of the biochemistry of vision,especially as it relates to phototransduction (for ex-ample, see the review by Stryer1). But these advanceshave been limited almost exclusively to rods, largelybecause most vertebrates have rod-dominated retinasthat lend themselves to mechanical isolation of theouter segments for use in traditional biochemicalanalyses. It may be erroneous to conclude that coneand rod biochemistry is similar except for their chro-mophores. After all, cones mediate not only colorvision but also photopic vision, since, in every verte-brate species except the skate, the rods saturate atfairly low light levels.2 Cones also contain greaternumbers of mitochondria in their inner segmentscompared to rods, again suggesting differences in bio-chemistry.

Investigators using immunocytochemical tech-niques have observed what may prove to be funda-mental differences in the enzyme systems of rods andcones. Specifically, most cones seem to lack the48-kD protein (also known as S-antigen and as arres-tin) that is so abundant in rods.3"7 The blue-sensitive

From the Departments of Ophthalmology and *Pathology, WestVirginia University Health Sciences Center, Morgantown, WestVirginia.

Submitted for publication: June 21, 1989; accepted August 14,1989.

Reprint requests: T. Michael Nork, MD, Department of Oph-thalmology, West Virginia University Health Sciences Center,Morgantown, WV 26506.

cones may be similar to rods in this respect because,at least in tree shrew, they also contain the 48-kDprotein.7 Cones may lack other enzymes found inrods, such as cyclic GMP and interphotoreceptorbinding protein (IRBP).6

Although cones seem to lack some of the enzymesseen in rods, they may contain their own enzymes,one of which may be carbonic anhydrase (CA). CA isan enzyme found in most organs of the body, yet notall of its physiologic functions are fully understood.In the human neural retina, CA is present in theMiiller's glial cells.8'9 Previous investigators haveplaced the frequency of CA among human conesvariously at 0%,8 10%,9 and 100%.l0 CA may" ormay not1213 be present in the cones of nonprimatevertebrates. In our earlier work using CA as a markerfor Miiller's cells, we noticed also that the cones oc-casionally were positive for this enzyme but that thepercentage of positively stained cones varied greatlybetween eyes.1415 The purpose of this study was todetermine the cause of this apparent variability in CAstaining and to ascertain whether distribution of CAamong human cones is a marker for certain spectraltypes.

Materials and Methods

Suspecting a problem related to preparation of thetissue, we placed 15 adult human retinas in 4% phos-phate buffered paraformaldehyde at times rangingfrom 1 min to 24 hr after cessation of retinal bloodflow (prefixation time). Three of the eyes (with pre-fixation times of 1-5 min) were obtained as surgical

1451

1428 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Augusr 1990 Vol. 31

polymorphisms (RFLPs) for DNA markers, andprogeny from such crosses have been most useful forgenetic analysis with molecular probes in other stud-ies.12"14 The first experiment was a three-point inter-subspecific backcross designed to position the genecorresponding to zr.408 on mouse chromosome 5. Inthe second experiment, C57BL/6J mice made con-genic for the normal allele of rd, derived from thewild mouse, Mus spretus, were analyzed for the segre-gation of the putative rd gene and the segregation ofrd disease expression.

Materials and MethodsProbes

The putative rd cDNA (zr.408) was cloned as a3.45-kb fragment in the EcoRI site of the plasmidpBluescript SK(—). Because of an internal EcoRI site,two fragments were released from the plasmid afterEcoRI digestion; the sizes of the fragments were 1.65and 1.8 kb. Both fragments gave the same hybridiza-tion pattern in Northern blots of mouse retinal RNA.However, only the 1.65-kb fragment was used in theexperiments, and the 1.8-kb fragment was not testedfurther. The cDNA corresponding to the Gus gene(kindly provided by Dr. Gordon Watson, Universityof California—Berkeley, Berkeley, CA) was a 1.45-kbPstI insert in pBR322 and is referred to as pGAl.15

The cDNA corresponding to the gene Afp (obtainedfrom Dr. Shirley Tilghman, Princeton University,Princeton, NJ) was a 960-bp Hindlll insert inpBR322 and is referred to as pBR322-AFPl.16 Label-ing of all cDNAs was done with [«-32P]dCTP (3000Ci/mmole) (New England Nuclear, Boston, MA) bythe random priming method.17 All restriction en-zymes were obtained from Promega Biotec, Mad-ison, WI.

Intersubspecific Backcross

Mus musculus mice of the NFS/N strain were ob-tained from the National Institutes of Health, Divi-sion of Natural Resources, Bethesda, MD. Mice fromthe subspecies Mus musculus musculus were ob-tained from a laboratory colony derived from miceoriginally trapped in Skive, Denmark and main-tained by Dr. M. Potter (NCI contract NO1-CB2-5584) at Hazelton Laboratories, Rockville, MD.NFS/N females were mated with M. m. musculusmales and the Fl females backcrossed with M. m.musculus males to produce the experimental ani-mals.

DNAs were extracted from mouse livers by stan-dard procedures,18 cleaved with the appropriate re-striction enzyme, run on 0.4% agarose gels for 48 hrat low voltage, and transferred to nylon membranes(Amersham, Arlington Heights, IL) by the techniqueof Southern.19 The membrane blots were prehybri-

dized for 3-6 hr at 65°C in a solution containing7.0% sodium dodecyl sulfate (SDS), 0.5 M phosphatebuffer, pH 7.0, 1 mM EDTA, and 1% bovine serumalbumin. Hybridization was carried out at 60°Covernight in the same solution as that for prehybridi-zation but with the addition of labeled probe. Afterhybridization, all blots were washed for 15 min at37°C in 2X SSC (IX SSC is 150 mM NaCl and 15mM sodium citrate, pH 7.0) + 0.1% SDS. Whenprobing with zr.408, blots were washed an additionaltwo times for 20 min at 57°C in 2X SSC + 0.1% SDS,and two times for 20 min at 57°C in 0.4X SSC+ 0.1% SDS. When probing with pGAl or pBR322-AFP1, blots were washed an additional two times for20 min at 47°C in 2X SSC + 0.1% SDS, and twotimes for 20 min at 47°C in 0.2X SSC + 0.1% SDS.After washing, the blots were exposed at —80°C toKodak XAR-5 film with intensifier screens (DuPontCronex Lightning Plus) for 5-14 days.

Interspecific Backcross

Laboratory mice of the C57BL/6J-rd/rd strainwere used from our (MML) breeding colony. Thisline of mice is identical to the congenic C57BL/6J-rdle strain described elsewhere,3 except that the light ear(le) gene was eliminated by a series of crosses. Thus,the mice are homozygous for the rd gene (rd/rd) andhomozygous wild-type (+/+) at the le locus. M.spretus mice were obtained from the Jackson Labora-tory, Bar Harbor, ME. Over a period of several yearsthese mice were bred according to the following pat-tern: C57BL/6 J-rd/rd females were mated with M.spretus males and the Fl females (also termed Nlwhen used for repeated backcross matings) werebackcrossed to C57BX/6J-rd/rd males to produce theN2 offspring. A single eye from N2 mice was enucle-ated under ether anesthesia and was examined histo-logically3 for the presence or near-absence of photo-receptor cells to identify mice of rd/+ or rd/rd geno-types, respectively. Mice of either sex with the rd/+genotype were backcrossed with C57BL/6J-rd/rdmice to produce N3 progeny. This process was re-peated until the N8, N9, and N10 generations werereached, and these animals were used for the study.Mice from the parental lines (C57BL/6J-rd/rd andM. spretus) also were examined.

Mice were sacrificed between 17 days and 5months of age by cervical dislocation; the eyes wereenucleated and prepared for routine eye histology;3

and the brains were dissected and rapidly frozen ondry ice. Tissue DNA was extracted by a modificationof the method of Fodor and Doty.20 Individual brainswere pulverized in liquid nitrogen and the resultingpowder mixed with 3 ml of a solution containing 0.3M sucrose, 10 mM Tris buffer at pH 7.5, 10 mM

No. 8 MAPPING AND CO-SEGREGATION OF A PUTATIVE rd cDNA / Danciger er al 1429

• - • *

MgCl2 and 1% Triton X-100, and kept on ice for 10min. The suspension was then centrifuged at 5-10°Cfor 10 min at 8000 RPM in the SS-34 rotor in aSorvall RC-5B centrifuge (7649 g) and the pellet re-suspended in 3 ml of a solution containing 75 mMNaCl, 24 mM EDTA at pH 8.0, 0.5% SDS, and 600itg/m\ protease K. Incubation of the suspension wascarried out at 44°C overnight or until all particulatematter disappeared. RNAse was added to a concen-tration of 300 /ig/ml, and a second overnight incuba-tion was carried out at 37°C. The mixture was ex-tracted once with an equal volume of P/C (50% equil-ibrated phenol and 50% solution containing 24 partschloroform to 1 part isoamyl alcohol), and once withan equal volume of the chloroform-isoamyl alcoholmixture. DNA was precipitated overnight at roomtemperature in 0.1 volume of 3 M sodium acetateand 1.5 volumes of 95% ethanol. After washing in75% ethanol the centrifuged DNA pellet was dis-solved in TE buffer (10 mM Tris buffer and 1 mMEDTA) at pH 7.4.

For blot hybridization studies, the extracted mousegenomic DNAs were digested with the appropriaterestriction endonuclease, loaded at 6 jug per lane in1.2% agarose gels, and electrophoresed for 16-20 hr.Transblotting crthe DNA to nylon membranes, andprehybridization, hybridization with the zr.408probe, washing, and exposure of the membranes toX-ray film, was performed as described above.

All procedures with animals conformed to theARVO Resolution on the Use of Animals in Re-search and the guidelines of the committees on ani-mal research at our respective institutions.

Results

Intersubspecific Backcross

DNAs from the NFS/N and M. m. musculus micewere examined for RFLPs that hybridize to thezr.408 probe. When digested with Seal, DNA fromNFS/N mice showed a 12.5-kb fragment that wasabsent from M. m. musculus, and M. m, musculusDNA showed an 11.1-kb fragment that was absentfrom NFS/N (Fig. 1). DNAs were extracted from 62progeny of the backcross ([NFS/N X M. m. muscu-lus^ \ X M. m. musculus) and scored for inheritanceof the NFS/N fragment (Fig. 2). This inheritance pat-tern was compared with the inheritance patterns oftwo genes that flank the rd locus on mouse chromo-some 5, Gus and Afp. The NFS/N allele for Gus wasfollowed as a pGA1-hybridizing Hindlll RFLP of 3.7kb (the M. m. musculus RFLP was 3.9 kb), and theNFS/N allele for Afp was followed as a pBR322-AFP1-hybridizing EcoRI RFLP of 6.0 kb (the M. m.musculus RFLP was 3.8 kb).

1 2

t HI 12.5 kb.1 kb

Fig. 1. Autoradiogram of a Southern blot of DNAs from inter-species backcross parental controls M. m. musculus (lane I) andNFS/N (lane 2) hybridized with the zr.408 cDNA; each lane has 6Mg DNA. The DNAs were digested with Seal restriction endonucle-ase. The arrows point to t*ie 12.5-kb NFS/N fragment and the11.1 -kb M. m. musculus fragment.

In the 62 mice for which all three genes werescored, four recombinants were found to contain theNFS/N Afp aJlele but not the NFS/N fragment cross-reactive with zr.408. This suggests a map distance of 6± 3.1 cM (recombination = 4/62 = 6 ± SE 21). Forzr.408 and Gus, the corresponding map distance was21 ±5.2 cM, and for Afp and Gus, 27 ± 5.7 cM(Table 1). Considering the three genes taken together,all recombinants were single recombinants. That is tosay, whenever there was a recombinant between Afpand zr.408, the allelic pattern of Gus was the same asthat of zr.408, and whenever there was a recombinantbetween Gus and zr.408, the allelic pattern of Afp wasthe same as that of zr.408. There was no instance inwhich zr.408 was of one allelic pattern and both Afpand Gus were of the other (double recombinant).Therefore, the most likely gene order is Afp—zr.408—Gus. The corresponding distances in the standardmouse linkage maps of Lyon9 and Davisson et al10 are

1430 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / Augusr 1990 Vol. 31

1 2 3 4 5

I

* * •

Fig. 2. Autoradiogram of a Southern blot of interspecies back-cross progeny DNAs of the cross (NFS/N X M, m. musculus)F\X M. m. muscttlus. DNAs were digested with Seal and hybridizedwith the zr.408 probe; each lane has 10 tig DNA. Five representa-tive backcross progeny are in tanes I-5. The lower arrow marks the11. l-kb M. m. musculus fragment present in all backcross progeny,and the upper arrow marks the 12.5-kb NFS/N fragment presentonly in progeny heterozygous for the NFS/N putative rd allele (ie,lanes 3 and 5).

smaller in this region of chromosome 5 than are ourrecombination distances. However, the distances re-ported here between Afp and zr.408 and betweenzr.408 and Gus are not statistically different fromexpected distances between Afp and rd and betweenrd and Gus in this sample size (x2 = 2.06 and x2

= 3.59, respectively; P < 0.05 for both). Furthermore,our recombination distances fall within the range ofrecombination distances reported for rd, Gus, andAfp in earlier studies.22'23 Therefore, the location ofzr.408 between Afp and Gus and its position relativeto the two genes strongly suggests that zr.408 corre-sponds to the rd gene.

Analysis of Segregation of rd and zr.408

DNAs from C51BL/6J-rd/rd and M. spretus micewere examined for RFLPs that hybridized with the

zr.408 probe. BamHI digestion revealed two promi-nent fragments of 12.5 and 7.5 kb present in M.spretus and absent from the C51BL/6J-rd/rd mice(Fig. 3, lanes I and 2). Histologically, M. spretus micehad normal retinas and C57BL/'6J-rd/rd mice hadretinas nearly devoid of photoreceptors, as expected.3

DNAs were then extracted from mice from the N8,N9, and NIO generations and examined for the twoM. spretus RFLPs, ie, the M. spretus allele (Fig. 3,lanes 3-9). These results were compared to the histo-logic results demonstrating the presence (rd/rd) orabsence (rd/+) of retinal degeneration. Of 72 micetested by blot hybridization with the zr.408 probe, 45were heterozygous (contained the M. spretus allele)and 27 were homozygous (showed only the C57BL/6J hybridization pattern). Each of the 45 heterozy-gotes had normal retinas and each of the 27 homozy-gotes had degenerative retinal disease. Thus, no re-combinants were detected between the putative rdgene and the appearance of retinal degeneration.

These data can be used to estimate the maximumpossible distance between rd and the zr.408-hybridiz-ing sequences by the formula P = (I - R)n 2t, where R= recombination distance between zr.408-hybridiz-ing genomic sequences and the gene responsible forthe observed retinal degeneration; n = the number ofmice; and P = probability. The expression (I — R)n

represents the probability P that two genes will segre-gate together in a mating of n mice with zero recom-binants. If P < 0.05, then the two genes are signifi-cantly linked within a distance of R. At the 95% con-fidence limit Pis 0.05. Since I mouse can be added tothe total studied for each N backcross analyzed, a

Table I. Segregation of a zr.408 hybridizingfragment (putative rd gene) with Afp and Gusamong 62 backcross progeny between NSF/Nand Mus musculus musculus (Skive)*

Inheritance of NFS/N allele

Mice Afp-1Putative

rd GusNumber

of progeny

Nonrecombinants(parentalgenotypes)

Singlerecombinants

Recombination (Afp, putative rd)(Afp, Gus)(putative rd, Gus)

= 4/62= 17/62= 13/62

2124

406

= 6 ± 3 . I f= 27 ±5 .7= 21 ± 5.2

Gene order: centromere—Afp—putative rd—Gus

* Distances in cM between hybridizing fragments and standard errorswere calculated according to Green21 from the number of recombinants.

t All recombination values are significant to the 0.05 level.

No. 8 MAPPING AND CO-SEGREGATION OF A PUTATIVE rd cDNA / Donciger er ol 1431

1 2 3 4 5 6 7 8 9

23.1kb»

9.4 kb-

6.6 kb-

4.4 kb-

mi

Gus in our three-point cross was comparable to theposition of the rd gene in standard mouse gene maps.

The second study demonstrated linkage betweenthe putative rd gene and rd expression (in this case,photoreceptor degeneration). With zero recombi-nants in the equivalent of 80 mice, we were able toestablish with a >95% probability that the zr.408cDNA hybridized to DNA sequences that are within4 cM of the rd gene.

Taken together, the data from the two studies posi-tion the genomic sequences corresponding to zr.408to a site on chromosome 5 at or near the rd mutation,and demonstrate its co-segregation with retinal de-generation in genetic crosses. Thus, the data are con-sistent with our characterization of zr.408 as the nor-mal product of the rd locus.

Key words: rd mouse, rd gene, interspecies backcross, co-segregation, fine mapping

2.3 kb»

Fig. 3. Autoradiogram of a Southern blot of progeny of the N8and N9 generation of crosses (described in the text) that were de-rived from the original backcross (C57BL/6i-rd/rd X M.spretus)Fl X C51BL/6J-rd/rd. DNAs were digested with Bam HIand hybridized with the zr.408 probe; each lane has 6 fig DNA. TheC57BL/6J-rdfrd mouse control is in lane 1 and the M. spretuscontrol is in lane 2. Representative progeny from the N8 and N9generations are in lanes 3-9. The arrows on the right mark the 12.5-and 7.5-kb RFLPs found in the M. spretus control (lane 2) and insome of the backcross progeny (lanes 4, 5, and 7-9). The backcrossprogeny in lanes 3 and 6 show only the C57BL/6J-rd/rd pattern.The numbers on the left (in kilobase pairs) mark the positions of XDNA fragments produced by digestion with Hindlll.

minimum of 8 can be added to the 72 mice tested,such that n = 80. Then,

(1 - R ) 8 0 = 0.05,

and

R = 0.0368.

Therefore, there is a 95% probability that the mousegenomic DNA sequences to which zr.408 hybridizesare within 3.68 cM of the known raf gene.

Discussion

We tested a putative rd cDNA, zr.408 (recentlycloned in our laboratory) by two methods, to verifythat this cDNA corresponds to the rd gene.

With a three-point intersubspecific backcross weshowed that zr.408 hybridizes to mouse DNA se-quences that lie between the genes Afp and Gus. Theposition of the putative rd gene relative to Afp and

Acknowledgments

The authors thank Drs. Robert Sparkes and BronwynBateman for helpful discussions; Dr. Lawrence Pinto forthe initial suggestion to produce the interspecific crosseswith M. spretus; and Edwin Mar, Gregg Gorrin, andDouglas Yasumura for expert technical assistance.

References

1. Lasansky A and DeRobcrtis E: Submicroscopic analysis of thegenetic dystrophy of visual cells in C3H mice. Journal of Bio-physical and Biochemical Cytology 7:679, 1960.

2. Shiosi Y and Sonohara O: Studies on retinitis pigmentosa:XXVI. Electron microscopic aspects of the early retinalchanges in inherited dystrophic mice. Jpn J Ophthalmol72:299, 1969.

3. LaVail MM and Sidman RL: C57BL/6J mice with inheritedretinal degeneration. Arch Ophthalmol 91:394, 1974.

4. Carter-Dawson LD, LaVail MM, and Sidman RL: Differentialeffect of the rd mutation on rods and cones in the mouse retina.Invest Ophthalmol Vis Sci 17:489, 1978.

5. Farber DB and Lolley RN: Cyclic guanosine monophosphate:Elevation in degenerating photoreceptor cells of C3H mouseretina. Science 186:449, 1974.

6. Farber DB and Lolley RN: Enzymatic basis for cyclic GMPaccumulation in degenerative photoreceptor cells of mouseretina. Journal of Cyclic Nucleotide Research 2:139, 1976.

7. Lolley RN, Farber DB, Rayborn ME, and Hollyfield JG: Cy-clic GMP accumulation causes degeneration of photoreceptorcells: Simulation of an inherited disease. Science 196:664,1977.

8. Sidman RL and Green MC: Retinal degeneration in themouse. J Hered 56:23, 1965.

9. Lyon MF: Mouse chromosome atlas. Mouse Newsletter 81:20,1988.

10. Davisson MT, Roderick TH, Hillyard AL, and Doolittle DP:Locus map of the mouse. Mouse Newsletter 84:15, 1989.

11. Bowes C, Danciger M, Kozak CA, and Farber DB: Isolation ofa candidate cDNA for the gene causing retinal degeneration inthe rd mouse. Proc Natl Acad Sci USA 86:9722, 1989.

12. Robert B, Barton P, Minty A, Daubas P, Weydert A, Bon-

1432 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / August 1990 Vol. 31

homme F, Catalan J, Chazottes D, Guenet J, and BuckinghamM: Investigation of genetic linkage between myosin and actingenes using an interspecific mouse backcross. Nature 314:181,1985.

13. Amar L, Arnaud D, Cambrou J, Guenet J, and Avner P: Map-ping of the mouse X chromosome using random genomicprobes and an interspecific mouse cross. EMBO J 4:3695,1985.

14. BrocdorfFN, Cross G, Cavanna J, Fisher E, Lyon M, Da vies K,and Brown S: The mapping of a cDNA from the human X-linked Duchenne muscular dystrophy gene to the mouse Xchromosome. Nature 328:166, 1987.

15. Watson G, Felder M, Rabinow L, Moore K, LaBarca C, TietzeC, Vander Molen G, Bracey L, Brabant M, Cai J, and PaigenK: Properties of rat and mouse /3-glucuronidase mRNA andcDNA, including evidence for sequence polymorphism andgenetic regulation of mRNA levels. Gene 36:15, 1985.

16. D'Eustachio P, Ingram RS, Tilghman SM, and Ruddle FH:Murine a-foetoprotein and albumin: II. Evolutionary linked

proteins encoded on the same mouse chromosome. Somat CellGenet 7:289, 1981.

17. Feinberg AP and Vogelstein B: A technique for radiolabelingrestriction endonuclease fragments to high specific activity.AnalBiochem 132:6, 1983.

18. Hoggan MD, Halden NF, Buckler CE, and Kozak CA: Geneticmapping of the mouse c-frns proto-oncogene to chromosome18. J Virol 62:1055, 1988.

19. Southern EM: Detection of specific sequences among DNAfragments separated by gel electrophoresis. J Mol Biol 98:503,1975.

20. Fodor EJB and Doty P: Highly specific transcription of globinsequences in isolated reticulocyte nuclei. Biochem BiophysResCommun 77:1478, 1977.

21. Green EL: Genetics and Probability in Animal Breeding Ex-periments. New York, MacMillan, 1981.

22. Chapman VM, Noell WK, and Adler D: Note. In MouseNewsletter 53:61, 1975.

23. Paigen K. and Noell WK: Two linked genes showing a similartiming of expression in mice. Nature 190:148, 1961.