Proc. Nat. Acad. Sci. USAVol. 72, No. 1, pp. 263-267, January 1975

Tay-Sachs' and Sandhoffs Diseases: The Assignment of Genes forHexosaminidase A and B to Individual Human Chromosomes

(human genetics/somatic cell genetics/lipid storage diseases/GM2 gangliosidoses)

F. GILBERT* t §, R. KUCHERLAPATI*, R. P. CREAGAN*, M. J. MURNANE*,G. J. DARLINGTON*, AND F. H. RUDDLE* t t* Department of Biology, Yale University, Kline Biology Tower, New Haven, Connecticut 06520; and t Department of HumanGenetics, Yale University School of Medicine, New Haven, Connecticut 06510

Communicated by Victor A. McKusick, June 6, 1974

ABSTRACT The techniques of somatic cell geneticshave been used to establish the linkage relationshipsof loci coding for two forms (A and B) of hexosaminidase(EC 3.2.1.30; 2-acetamido-2-deoxy-B-D-glucoside acet-amidodeoxyglucohydrolase) and to determine whether astructural relationship exists between these forms. In aseries of human-mouse hybrid cell lines, hexosaminidaseA and B segregated independently. Our results and thosereported by other investigators are used to analyze theproposed structural models for hexosaminidase. We havealso been able to establish a syntenic relationship betweenthe gene locus responsible for the expression of hexosami-nidase A and those responsible for mannosephosphateisomerase and pyruvate kinase-3 and to assign the genefor hexosaminidase B to chromosome 5 in man. Tbere isthus a linkage between specific human autosomes andenzymes implicated in the production of lipid storagediseases.

The lipid storage diseases are a family of inherited disorderscharacterized by the excessive accumulation of sphingolipidsin the body's tissues. In each, the metabolic derangement ap-pears to be the result of a deficiency of a specific lysosomalhydrolase which is involved in the catabolism of these complexlipids (1). One of these enzymes, P-N-acetylglucosaminidase(Hex; EC 3.2.1.30) is thought to be responsible for at least twolipodystrophies, Tay-Sachs' disease (TSD; GM2 gangliosidosis,type I) and Sandhoff's disease (SD; GM2 gangliosidosis, typeII). When examined electrophoretically, this enzyme is foundto exist in multiple forms, two of which (Hex A and B) havebeen well characterized biochemically (2). A third form of theenzyme (Hex C), about which relatively little is known, hasrecently been described (3). TSD is associated with a defi-ciency of Hex A and an increased activity of Hex B, and SDis associated with a deficiency of both Hex A and B (4, 5).No individual has yet been reported in whom Hex A is presentin the absence of Hex B.Biochemical, genetic, and immunological evidence suggests

that a structural relationship exists between Hex A and B.Two theories concerning this relationship have recently beenadvanced (2, 6). The first proposes that Hex A is a conversionproduct of Hex B (2). TSD would then result from the de-ficiency of a functional conversion enzyme, and SD would re-

Abbreviations: Hex, hexosaminidase; TSD, Tay-Sachs' disease;SD, Sandhoff's disease.§ Present address: Department of Human Genetics, Universityof Pennsylvania School of Medicine, Philadelphia, Pa. 19104.t To whom reprint requests should be addressed.

sult from a defect in the gene coding for the basic Hex protein.The second theory proposes that Hex A and B are eachcomposed of multiple subunits, one of which is common toboth forms (6). In this hypothesis, TSD would result from thedeficiency of the Hex A-specific subunit and SD from thedeficiency of the common subunit. It is also possible that thetwo forms of Hex are not structurally related. Hex A and Bmay be controlled by two independent genes. TSD would thenresult from an effective deficiency of the normal Hex A struc-tural gene product and SD might result from a mutation in alocus controlling expression of both enzymes or required fortheir activation.A series of human-mouse hybrid cell lines were examined for

the expression of Hex activity to determine whether a struc-tural relationship does in fact exist between Hex A and B.Such interspecific hybrid cell lines, which preferentiallysegregate the chromosomes of one parent in the cross (in thisinstance, the human), have already proved useful in the as-signment of genes for specific enzymes to individual chromo-somes in man. However, their potential value as a tool for thestudy of enzyme structure has not yet been fully appreciated.We have found that, in this series of hybrid cells, human HexA and B are expressed independently.The hybrid cells were also used to establish the linkage

relationships of genes coding for Hex A and B. The humanchromosome complements and patterns of expression of aseries of isozymes with known chromosome assignments werecompared with the retention of Hex A and B activity in theseclones. On the basis of these studies we were able to assign thelocus involved in the expression of Hex B to human chromo-some 5 and to establish a syntenic relationship between genescoding for Hex A, mannosephosphate isomerase, and pyruvatekinase-3.

MATERIALS AND METHODS

Hexosaminidase Assay. The assay for hexosaminidase activ-ity is a modification of the published procedures of Okadaand O'Brien, and van Someren and van Henegouwen (4, 8).The cell homogenates were prepared as described (9). Electro-phoresis was performed (at 40) on cellulose acetate gel (Cello-gel: Chemetron, Milan, Italy) in a citrate phosphate buffer(25 mM, pH 5.6) for 3 hr at 250 V. The gel was incubatedwith the artificial substrate, 4-methyl umbelliferyl-N-acetyl-,3-glucosaminide [Pierce Chemical Co.; 1.5 mM in 0.5 M Nacitrate (pH 4.0) and 1% agarose] for 1-2 hr and then exposed

263

Proc. Nat. Acad. Sci. USA 72 (1975)

(4)

-4*- C

-_- A

-B

0

1 2 3 4 5 6 (H)

FIG. 1. Hexosaminidase activity on cellulose acetate. Cellextracts were applied to cellulose acetate and assayed as de-scribed in the text. The development time of the gel was chosento allow optimal visualization of the Hex A and B bands. Fourbands are evident-corresponding to human Hex A, Hex B, andto two mouse activities, one at the origin and a second migratingcoincidentally with human Hex C. Slot 1, hybrid with humanHex A and B; slot 2, hybrid with Hex B alone; slot 3, hybridwith Hex A alone; slot 4, hybrid with no human Hex; slot 5,

mouse parent; slot 6, human parent. The differences in intensityof Hex A and B between cell lines possibly reflect quantitativedifferences in the amounts of material applied to each slot. HexC activity (mouse and/or human) was clearly apparent in allsamples, though in some (slots 2, 3, and 6) it was faint andthus was reproduced poorly in the photograph. Some variationin the intensity of the Hex C in each cell line has been notedbetween gels. The slight difference in mobility of Hex C inslots 4 and 5 possibly reflects differences in the age of each sample.

to ammonia vapor before it was made visible under ultra-violet light.

Other Enzymes. Twenty enzymes with established mouse

and human electrophoretic differences and known humanchromosome assignments (Table 1) were also assayed bymethods detailed by Nichols and Ruddle (9).

Production of Hybrids. Series A: A number of primaryhybrid clones were established, using techniques outlined,after the fusion of a mouse cell line (A9) lacking the enzyme

hypoxanthine-guanine phosphoribosyltransferase (HGPRT)with two diploid human fibroblast lines (GM 17 and Yoder,obtained from the Institute for Medical Research, Camden,N.J. and Dr. D. Borgaonkar, Johns Hopkins UniversitySchool of Medicine, respectively) and isolation in the HATselection system (10).

Series B: This represents a number of primary and second-ary hybrid clones resulting from the fusion of several differenthuman diploid fibroblast and leukocyte lines with two HGP-RT- mouse cell lines (A9 and RAG) and established throughthe use of the methods described above. The human chromo-some complements were analyzed as the cell pellets were

prepared (R. Creagan and F. H. Ruddle, unpublished data).An average of over 25 metaphases per clone was examined.

TABLE 1. Enzymes studied and their human chromosomeasszgnments*

HumanEnzyme chromosome

Dipeptidase-1 (PEP C) 1Malate dehydrogenase (NAD +) (MDH-1) 2Isocitrate dehydrogenase (NADP+) (IDH-1) 2Malic enzyme (NADP+) (MOD) 6Indophenol oxidase (SOD-2) 6Mannosephosphate isomerase (MPI) 15Pyruvate kinase (PK) 15Glutamate-oxaloacetate transaminase (GOT) 10Esterase A4 (Es-A4) 11Lactate dehydrogenase-A (LDH-A) 11Lactate dehydrogenase-B (LDH-B) 12Tripeptidase-1 (PEP B) 12Nucleoside phosphorylase (NP) 14Adenine phosphoribosyltransferase (APRT) 16Dipeptidase-2 (PEP A) 18Glucosephosphate isomerase (GPI) 19Adenosine deaminase (ADA) 20Indophenol oxidase (SOD-1) 21Phosphoglycerate kinase (PGK) XGlucose 6-phosphate dehydrogenase (G6PD) X

* See ref. 9.

Heat Inactivation. Cell homogenates of the human andmouse fibroblast parents and of hybrid lines from both seriesA and B were heated for three hours at 500. The heat-in-activated samples as well as untreated controls were thenanalyzed for hexosaminidase activity by cellulose acetate gelelectrophoresis. Hex A activity has been reported to be thermo-labile while Hex B activity is thermostable.

Toxin Treatment. Human, mouse, and hybrid somatic cellswere subjected to various concentrations of diphtheria toxin asdescribed (R. P. Creagan, S. Chen, and F. H. Ruddle, manu-script in preparation).

RESULTSThe electrophoretic patterns of the multiple forms of Hex inthe human and mouse parental cell lines and in selectedhybrid clones are illustrated in Fig. 1. The human cell linesdemonstrate the previously described three-band pattern cor-responding to Hex A, B, and C. The mouse cell line has onedistinct band which migrates to a position coincident with thehuman Hex C and a second band at the origin. A mixture ofhomogenates of the parent lines (human and mouse), whenassayed for Hex, demonstrates all four bands.Two series of human-mouse hybrid clones (series A and B)

were examined for the presence of Hex A and B. All of theclones retained enzymatic activity in the Hex C region thatcould be of human and/or mouse origin. Cell homogenates ofthe mouse and human fibroblast parents and of several hybridclones from both series A and B (three of which were HexA+/B+, one that was Hex-/B+, and three that were HexA+/B+) were subjected to heat inactivation and analyzed forhexosaminidase activity by gel assay. Heat inactivation re-

sulted in the loss of Hex A activity in all of the lines in which itwas present prior to treatment. Hex B activity was unaffectedby the heat treatment. The distribution of clones retainingboth Hex A and B, either Hex A or B alone, or neither, is

264 Genetics: Gilbert et al.

Tay-Sachs' and Sandhoff's Diseases 265

TABLE 2. Hexosaminidase activity in human-mousehybrid clones

Hexosaminidase activity (no. of clones)

Hex A Hex A Hex Band B alone alone Neither

Series A 12 2 8 7Series B 0 4 6 5

Two series of human-mouse hybrid clones were established asdescribed in the text and analyzed for expression of Hex activity.

given in Table 2. The results indicate that human Hex Aand B segregate independently.The hybrid clones in series A were also analyzed for twenty

other enzymes whose human chromosome assignments havealready been established (see Table 1). The only concordancenoted between the multiple forms of Hex and the other en-zymes was between Hex A and the human forms of mannose-phosphate isomerase (EC 5.3.1.8) and pyruvate kinase-3 (EC2.7.1.40) (Table 3). There was no positive correlation evidentbetween the expression of Hex B activity and the otherenzymes studied. Chromosome analysis of hybrid clones inseries B revealed that Hex B activity was correlated with thepresence of chromosome 5 (Table 4).

In order to further test the independent expression of Hex Aand B and to confirm our assignment of the gene for Hex B tochromosome 5, we treated three cell lines, AIM-15a, JFA-14a-13, and JFA-16a-8, with various concentrations ofdiphtheria toxin. Mouse cells are resistant to the toxinwhereas human cells are sensitive. Hybrid cells that retainhuman chromosome 5 are as sensitive as human cells, whereasother human chromosomes leave them toxin-resistant (R. P.Creagan, S. Chen, and F. H. Ruddle, manuscript in prepara-tion). The results of toxin treatment, in terms of their Hexexpression, are presented in Table 5. These results confirmour assignment of Hex B to chromosome 5 and the inde-pendence of Hex A and B expression.

DISCUSSIONThe lysosomal enzyme, Hex, implicated in the production ofthe GM2 gangliosidoses, is found to exist in at least two elec-trophoretic forms, A and B, between which a structural rela-tionship has been presumed to exist. We have used the tech-niques of somatic cell genetics to gain insight into this relation-ship and to establish the chromosomal assignments of genesinvolved in the expression of this enzyme.

TABLE 3. Segregation ofHex with mannosephosphate isomeraseand pyruvate kinase

MPI PK

+ - + -+ 12 1 + 5 2

Hex A Hex A- 1 15 - 0 10

Human-mouse hybrid clones (series A) were analyzed for Hexactivity and for the enzymes listed in Table 1. Synteny wasevident only between Hex A and mannosephosphate isomerase(MPI) and pyruvate kinase (PK).

TABLE 4. Segregation of Hex B with specific humanchromosome

Chromosome 5

+ 7 0Hex B

- 0 9

Human-mouse hybrid clones (series B) whose human karyo-types are known were analyzed for Hex activity. Synteny wasevident between Hex B and chromosome 5. In clones positive forHex B, chromosome 5 was found in 30-90% of the metaphasesexamined. In clones negative for Hex B, this specific chromosomewas absent in all of the metaphases examined.

In a series of human-mouse somatic cell hybrid clones, HexA was found to segregate concordantly with the human formsof mannosephosphate isomerase and pyruvate kinase-3.This synteny has been independently confirmed by anothergroup of investigators (13, 16). The gene for mannosephos-phate isomerase has been assigned to chromosome 7(11).Van Heyningen et al. (17) in a recent study assigned the genesfor mannosephosphate isomerase and pyruvate kinase-3 tochromosome 15. Ruddle and McMorris (18), on the basis ofan extensive study, retracted their original assignment ofmannosephosphate isomerase to chromosome 7. It is thereforepresumed that a genetic locus, most probably the structurallocus, for Hex A can be assigned to chromosome 15 in man.Hex B was found to segregate independently of Hex A. On thebasis of an analysis of a series of hybrid clones with definedhuman chromosome complements, we were able to assign alocus, again probably the structural locus, for Hex B to chro-mosome 5. The independent segregation of Hex A and B andthe assignment of Hex B to chromosome 5 have been con-firmed by our studies on the effects of diphtheria toxin treat-ment of three hybrid cell lines.Hypotheses have been advanced that seek to explain mecha-

nisms governing the expression of Hex A and B in man. Thesehypotheses are divisible into two major categories: (A) modelsbased on structural interrelationships between Hex A and B,and (B) a model that proposes complete structural indepen-dence of the Hex A and Hex B gene products. The two majortheories under consideration in category A are: (1) the enzymeconversion model and (2) the common subunit model. Thefirst would require there be at least two loci involved in Hexexpression, one coding for the basic structural protein, pre-sumed to be Hex B, and a second coding for the enzyme re-sponsible for the conversion of this protein to Hex A. Thesecond hypothesis, advanced by Beutler and illustrated in

TABLE 5. Expression of hexosaminidase in three hybrid celllines before and after treatment of the cells with diphtheria toxin

Hex expression Chromosome 5

Cell line No toxin Toxin No toxin Toxin

AIM-15a A+B+ A+B- + n.d.JFA-14a-13 A-B+ A-B- +JFA-16a-8 A-B- A-B-

n.d. = not done.

Proc. Nat. Acad. Sci. USA 72 (1975)

Proc. Nat. Acad. Sci. USA 72 (1976)

HexA: AB HexA: AXHexB: BB HexB: BX

I II

FIG. 2. Models of subunit structure of hexosaminidase A andB dimers. (I) Two-subunit model: B = common subunit, A =Hex A-specific subunit. (II) Three-subunit model: X = commonsubunit, A = Hex A-specific subunit, B = Hex B-specific subunit.

Fig. 2, would require two or three loci, each coding for a dis-tinct subunit type (6). In the two-subunit model, each Hex Adimer would be formed from two subunit types, one of whichis common to both forms and one of which is Hex A-specific.Each Hex B dimer would consist exclusively of the commonsubunits. The three-subunit model would require three loci,one coding for the subunit common to both forms and twoothers coding for the Hex A- and Hex B-specific subunits.The Hex A and B dimers would each be composed of onecommon and one specific subunit. It has recently been pro-posed that Hex C may be the Hex A-specific subunit (14).

In a study of 60 human-mouse hybrid clones analyzed forHex activity and for a number of other enzymes, Lalley et al.reported no clones in which Hex A was expressed in the ab-sence of Hex B (13, 16). They concluded that the expressionof Hex B is a necessary prerequisite for the expression of HexA. If this were so, and one assumed random segregation ofhuman chromosomes in the hybrid clones, it would be expectedthat a significant number of clones that retained mannose-phosphate isomerase and pyruvate kinase-3 activity would notexpress Hex A since they have lost the chromosome responsi-ble for Hex B (chromosome 5). They found, however, onlyone of 35 clones in which mannosephosphate isomerase waspresent and Hex A was absent. Thus, their data, while con-sistent with a model requiring the presence of Hex B forexpression of Hex A, do not provide conclusive proof for sucha scheme.

In our series of hybrid clones we found all four possible com-binations of Hex A and B-those with both, those withneither, and those with Hex A or B alone. The studies by vanSomeren and van Henegouwen (8) of 105 Chinese hamster-human hybrids also show that the two enzymic forms segre-gate independently. The finding that human Hex A activitycan exist in the absence of demonstrable Hex B in these studiesmakes it doubtful that the former is a conversion product ofthe latter.

If it were possible, in the hybrid situation, for a hypotheticalmouse conversion enzyme to substitute for the correspondinghuman form and change Hex B to Hex A, one would expectto find hybrid clones that have chromosome 5 and thus HexB as well as Hex A activity but that lack mannosephosphateisomerase and pyruvate kinase-3 activity as a consequenceof the loss of human Hex A, mannosephosphate isomerase,and pyruvate kinase-3 linkage group. Such clones were notobserved. It is also conceivable, though unlikely, that ahuman conversion enzyme, present in the hybrids which re-tained mannosephosphate isomerase expression but notchromosome 5, could alter the mouse structural protein suchthat it migrates coincidentally with human Hex A. However,it would seem unlikelv that such a conversion product wouldhave an electrophoretic mobility precisely that of humanHex A.Based on immunochemical data, Beutler (6) proposed a

unit. This would require two or three genetic loci, each codingfor a different subunit. If the three-subunit model were cor-rect, one would expect to correlate the presence of Hex A andHex B with two chromosomes each. Our karyotypic data in-dicate that only one chromosome each is required for the ex-pression of these enzymes. The two-subunit model predictsdependent segregation of Hex A or Hex B, depending uponwhich of these enzymes is the heteropolymer. If we assumethat there is no interaction between the mouse and humanforms of the enzyme, our results do not support this hy-pothesis.In clones that express Hex A alone, a situation that has not

yet been reported in vivo, the possibility that a heteropolymermay have been formed between human HexA or B componentand a component contributed by the mouse cell has to be con-sidered. This heteropolymer then could migrate to a positionsimilar to that of normal human Hex A. Heteropolymers havebeen reported to occur in somatic cell hybrids (see, for ex-ample, ref. 15). If we assume that this hypothesis is correct,our results might be considered to be consistent with Beutler'stwo-subunit model. However, one might expect to see somedifferences in the mobilities of the heteropolymer and thenormal human Hex A. On repeated enzyme analyses we foundthat Hex A migration in all hybrid clones was identical andthat there was no reduction in the intensity of mouse or humanbands. This observation makes it less likely that the two-sub-unit model of Hex structure is correct, but does not exclude itentirely.Our data strongly support the hypothesis that there is no

structural relationship between the two human specific en-zymes. The independent segregation of the two enzymic formsin hybrid cell lines reported here and by van Someren and vanHenegouwen (8) are consistent with this model. On this basis,TSD can be explained by a mutation at a locus responsible forHex A expression (possibly on chromosome 15) and SD to bethe result of either two mutations involving Hex A and Bstructural proteins (involving chromosome 5 in addition tochromosome 15), or more likely a single mutation at a locusthat regulates the expression of these genes, or is necessary forthe proper packaging of the enzymes within the lysosome, oris required for enzyme activation.We have thus been able to assign the locus involved in the

expression of Hex B to chromosome 5 and establish syntenicrelationship between the genes for Hex A, mannosephosphateisomerase, and pyruvate kinase-3.

We are grateful to Ms. Elizabeth A. Nichols, Ms. Susie Chen,and Mrs. Mae Reger for their expert assistance. This investigationwas supported by N.I.H. Grant USPHS 5-RO1-GM-09966, andNSF-GB 34303. R.K. is a Damon Runyon Memorial CancerResearch Fellow.

1. Brady, R. 0. (1973) Fed. Proc. 32, 1660-1667.2. Robinson, D. & Stirling, J. L. (1968) Biochem. J. 107,

321-327.3. Hooghwinkel, G. J. M., Veltkamp, W. A., Overdijlc, B. &

Lisman, J. W. (1972) Z. Physiol. Chem. 353, 839-841.4. Okada, S. & O'Brien, J. S. (1969) Science 165, 698-700.5. Sandhoff, K., Andrae, U. & Jatzkewitz, H. (1968) Life Sci.

7, 283-288.6. Srivastava, S. K. & Beutler, E. (1973) Nature 241, 463.7. Ruddle, F. H. (1973) Nature 242, 165-169.8. van Someren, H. & van Henegouwen, H. (1973) Human-

genetik 18, 171-174.9. Nichols, E. A. & Ruddle, F. H. (1973) J. Histochem. Cyto-

model in which each of the enzymic forms has a common sub-

266 Genetics: -Gilbertet al.

chem. 21, 1066-1081.

Proc. Nat. Acad. Sci. USA 72 (1975)

10. Klebe, R. J., Chen, T. R. & Ruddle, F. H. (1970) J. CellBiol. 45, 74-82.

11. McMorris, F. A., Chen, T. R., Ricciuti, F., Tischfield, J.,Creagan, R. & Ruddle, F. H. (1973) Science 179, 1129-1131.

12. Shows, T. B. (1972) Amer. J. Hum. Genet. 24, 13 abstr.13. Lalley, P. A., Rattazzi, M. C. &' Shows, T. B. (1973) New

Haven Conference, First International Workshop on HumanGene Mapping. Birth Defects: Original Article Series, ed.

Bergsma, D., Vol. X, no. 3.

14. Ropers, H. H. & Schwantes, U. (1973) Humangenetik 20,167-170.

Tay-Sachs' and Sandhoff's Diseases 267

15. Shows, T. B. & Ruddle, F. H. (1968) Science 160,1356-1357.16. Lalley, P. A., Rattazzi, M. C. & Shows, T. B. (1974) Proc.

Nat. Acad. Sci. USA 71, 1569-1573.17. van Heyningen, V., Bobrow, M. & Bodmer, W. F. (1974).

In Rotterdam Conference (1974). Second InternationalWorkshop on Humran Gene Mapping. Birth Defects: OriginalArticle Series, in press.

18. Ruddle, F. H. & McMorris, F. A. (1974). In Rotterdam Con-ference (1974). Second International Workshop on HumanGene Mapping. Birth Defects: Original Article Series, inpress.