Cytological Correlation ofLate X-Chromosome …Proceedings ofthe NationalAcademyofSciences Vol. 68, No. 3, pp. 544-548, March1971 Cytological andBiochemical Correlation ofLate X-ChromosomeReplication

Proceedings of the National Academy of SciencesVol. 68, No. 3, pp. 544-548, March 1971

Cytological and Biochemical Correlation of Late X-Chromosome Replicationand Gene Inactivation in the Mule

MAIMON M. COHEN AND MARIO C. RATTAZZI

Division of Human Genetics, Department of Pediatrics, School of Medicine, State University of New York at Buffalo,and Buffalo Children's Hospital, Buffalo, N.Y. 14222

Communicated by James V. Neel, November 19, 1970

ABSTRACT The correlation between late X-chromo-some replication and the quantitation of different X-linked glucose-6-phosphate dehydrogenase electrophoreticforms was studied in a natural hybrid, the female mule.In all four animals examined, a significant deviation fromthe expected 1:1 ratio of random X-chromosome inactiva-tion was observed, with the donkey X-chromosome themore frequently late-replicating one. The relative amountsof horse and donkey enzymes activities in lysates of muleskin fibroblasts and in peripheral blood were in agree-ment with this finding: the donkey enzyme was the minorcomponent. Although random expression of the enzymesfrom the two parent species was not observed, sampling,selection, or adaptation may actually be responsible forthe apparent "preferential inactivation". These studiessupport the hypothesis that late DNA replication indi-cates genetic inactivation.

The theory of dosage compensation for X-linked loci inmammals (1) has been based on the apparent relationshipbetween heterochromatin, late replication of one X-chromo-some in females, and gene inactivation (2). Although theLyon hypothesis has been generally accepted, supportiveevidence has been derived from diverse experimental systemseach yielding only a partial proof (3-11). Such investigationshave been criticized recently on several grounds by Grune-berg (12). This paper presents results of studies utilizinghybrid animals with suitable cytogenetic and enzymaticmarkers that simultaneously demonstrate, within a singlesystem, very good correlation between cytological and bio-chemical aspects of mammalian X-chromosome inactivation.

MATERIALS AND METHODSCell cultures

Fibroblast strains were established from skin biopsies ob-tained from three female mules (Equus caballus X E. asinus).The skin fragments were finely minced and placed in 60 X 15mm Petri dishes containing 5 ml of F10 nutrient medium,supplemented with 20% fetal calf serum (Grand IslandBiological Co., Grand Island, N.Y.). Incubation was at 370Cin 5% CO2 in air. When sufficient outgrowth was observed,the cells were removed by trypsinization (0.25%) or gentlescraping with a rubber policeman and transferred to 250-mltissue culture flasks. The cultured cells were subsequentlysplit 1:2 or 1:3 upon reaching confluency.

Short term cultures of peripheral lymphocytes were ex-amined from one additional female mule. Whole blood wasallowed to settle and aliquots of the leukocyte-rich buffycoat were distributed into commercially available chromosome

microculture kits (Chromosome Medium 1A, Grand IslandBiological Co.) at a final concentration of about 1 X 106lymphocytes/ml. The cultures were incubated at 370C for72 hr. Whole blood cultures of this mule were unsuccessful.

Cytogenetic techniques

Karyotypes. Chromosome analysis and karyotype construc-tion were performed on cells prepared according to a minormodification of the method of Moorhead et al. (13). Colcemide(0.05 jsg/ml) was added for the final 2 hr of incubation in bothfibroblast and lymphocyte cultures. The cells were removedfrom the culture fluid and suspended in 1% sodium citratefor 15 min at 370C, then fixed with several changes of metha-nol-glacial acetic acid 3:1. Several drops of cell suspensionwere placed on a microscope slide that had been immersed in70% methanol and passed through a flame. The preparationwas examined with phase contrast optics after staining with2% orcein in acetic acid.

Autoradiography. Tritiated thymidine (6.7 Ci/mmol) wasadded to the cell cultures (final concentration 1 MCi/ml) forthe last 6 hr of incubation, and colcemide was added for thefinal 4 hr. Slides, prepared as above, were dipped in KodakNuclear Track Emulsion (type NTB3),. stored at 40C for10 days, and developed according to the method of Schmid(14). Photomicrographs were taken of cells that showedmoderate grain counts with obvious localization of grains overa single chromosome. Silver grains were then reduced byusing a 10% solution of potassium ferricyanide (30 min)followed by 24% sodium thiosulfate (several dips), and rinsedin distilled water. Those cells previously photographed werelocated and rephotographed without the overlying silvergrains. In this way, a direct comparison could be made andthe labeled chromosomes were definitively identified.

Efforts were made to include only complete cells containing63 chromosomes. For reasons discussed below, in the fewcases in which this was not possible, only cells possessing boththe donkey X-chromosome and the largest submetacentricelement (see Fig. 1A; second line, first chromosome) wereincluded.

Biochemical determinations

Preparation of Cell Lysates. Blood was collected by vene-puncture from several donkeys, horses, and a female mule inacid-citrate-dextrose (Formula A) or heparin. White cellswere separated from the whole blood by centrifugation; the

544

Dow

nloa

ded

by g

uest

on

May

23,

202

0

Late X-Chromosome Replication in Mule 545

leukocyte-rich buffy coat was recovered. Contaminatingred cells still present were eliminated by brief suspension ofthe cells in distilled water and then in 0.9% NaCl. This pro-cedure was repeated 3 times.

Fibroblasts were obtained by gentle scraping of confluentmonolayers of cells from tissue-culture flasks with a rubberpoliceman. All cell samples were washed three times with0.9% NaCl adjusted to pH 3.8 with a few drops of 0.1 NHCl. The cells were then suspended in 25 mM H3B03-NaOHbuffer, pH 6 (1: 1.5 v/v for erythrocytes, 1: 5 v/v for leuko-cytes and fibroblasts). The buffer contained (in final concen-trations) 0.2 mM NADP, 1.0 mM #3-mercaptoethanol,1.0 mM EDTA, and 0.1 mM diisopropyl fluorophosphate.The cells were lysed by freezing and thawing three times indry-ice-acetone, and cellular debris was removed by centrif-ugation at 40,000 X g for 30 min at 40C. The supernate wasdialyzed overnight at 40C against 500 volumes of the bufferused for hemolysis.

Electrophoretic Techniques. Electrophoresis was performedon cellulose acetate gel sheets, 10 X 17 cm, 0.3 mm thick("Cellogel", Chemetron, Milano, Italy) in a Shandon Uni-versal Electrophoretic tank, MK II, by a modification of thetechnique described by Rattazzi et al. (15). The buffer usedwas 15 mM Tris-5 mM EDTA-35 mM boric acid-0.2 mMNADP at pH 7.8. Electrophoresis was performed at 40Cand, to increase the efficiency of cooling, no cover was placedover the electrophoresis tank. 450 V at 1.5 mA current wasapplied (50 V/cm for a 9-cm gap between gel supports) and

A

en}o )fl4flAAA^,nAAA/^^AA^.,,:

DONKEY " ix 4,, > .t

B

,440,*a

0

7 #

HORSE

FIG. 1. A, karyotype of the female mule. The X chromosomesare at the end of the second row; X (horse) is submetacentric;X (donkey) is acrocentric. B, autoradiographic patterns illus-trating late replication of the donkey (top) and horse (bottom)X-chromosomes.

TABLE 1. Percentages of late replication of donkey and horseX-chromosomes and of the horse contribution to glucose

6-phosphate dehydrogenase activity in mixedpopulations of mule cells

Late- Horse-typereplicating dehydro-

chromosomes genaseXd Total x2 % Xd (% of total)

Skin fibroblast 251 88 141 62.4 8.68* 59.4 + 4.2t"C "C 252 111 136 81.6 54.38* 78.4 i 5.2CC it 267 105 115 91.3 78.48* 90.8±i 3.4

Peripheral blood 30 43 69.8 6.72* 67.9 ± 1.7

Total 334 435 76.8 62.40*

* P < 0.01 for difference from 1:2 ratio for Xd:total.t Standard deviation (n = 4 for fibroblast and 5 for peripheral

blood).

the gel was allowed to equilibrate for 20 min. The samples(2-20 ul of lysate) were then applied and left to equilibratefor 30 min without the current, after which electrophoresisproceeded for 1 hr with the same current as above. Gelswere stained by the tetrazolium method (16).

Quantitation of Glucose-6-phosphate Dehydrogenase Activityof Bands. Upon completion of electrophoresis, the gel wasdipped in a mixture of 0.66 mM NADP, 0.33 M Tris -HClbuffer (pH 8.0), 0.33 mM MgCl2, and 1.2 mM glucose 6-phosphate. The gel was then blotted free of excess reagentand placed in a moist chamber. Development of fluorescentbands of enzyme activity was observed directly under UVlight (360 nm). The fluorescent bands were cut out of the gel,and the resulting strips (about 8 X 35 mm) were rolled with apair of forceps and inserted into the barrel of a glass tuberculinsyringe. When the syringe plunger was firmly pressed, theenzyme solution was expressed from the gel directly into thespectrophotometer cuvette, which contained all the reagentsfor quantitative determination except glucose 6-phosphate.The gel segment was squeezed twice and the syringe wasflushed twice with the assay mixture. The gel pellet was thenrotated sideways 900 in the barrel and the procedure wasrepeated. After a final rinse of the syringe, glucose 6-phosphatewas added to the cuvette and the enzyme activity was de-termined with a Gilford 2000 spectrophotometer as describedby Glock and McLean (17).

RESULTSCytogenetic studiesThe karyotype of the female mule is shown in Fig. 1A andis similar to those previously published (18, 19). The two X-chromosomes (end of second row) are distinct with regard toparental origin. The donkey X-chromosome is almost acro-centric, while the X-chromosome of the horse is submetacen-tric. Fig. 1B illustrates the late-replicating X-chromosomein the female mule (Xdonkey: Xhoree). Autoradiographic analysisof cells derived from mixed populations of either fibroblastsor lymphocytes demonstrated a significant deviation fromthe 1:1 ratio of late-replicating horse and donkey X-chromo-somes (Table 1). These deviations from the expected randomdistribution are all statistically significant when tested by thechi-square techniques (P < 0.01). In each case, an excess oflate-replicating donkey X-chromosomes was observed.

Proc. Nat. Acad. Sci. USA 68 (1971)

Dow

nloa

ded

by g

uest

on

May

23,

202

0

546 Genetics: Cohen and Rattazzi

+

FIG. 2. Cellulose acetate gel electrophoresis of cell lysates aftersaline washing, lysis at pH 6, and rapid high-voltage electro-phoresis at 40C, and selective staining for glucose 6-phosphatedehydrogenase. Left to right, horse erythrocytes, mule erythro-cytes, mule leukocytes, donkey erythrocytes. Note singlebands in horse and donkey and two distinct separable bands inthe mule.

Biochemical determinations

Cellulose acetate gel electrophoresis (15) of preparations ofhorse and donkey erythrocytes showed, for both animals, a

major and a minor band of glucose 6-phosphate dehydrogenaseactivity; the major one migrated more rapidly towards theanode. The donkey bands had a greater anodal mobility thanthe corresponding horse bands. The female mule patternrepresented a combination of the two parental types. Thisobservation of multiple bands resembles the previous findingsof others with starch gel electrophoresis (20-22). In artificialmixtures of horse and donkey erythrocyte lysates, as well as

in female mule erythrocytes, the minor (slower) donkeyband was coincident with the major (faster) horse band.Preliminary experiments with different buffer systems showedthat the relative staining intensity of the minor bands was

higher in older samples, or in fresh samples hemolyzed athigh pH, or after lengthy electrophoretic runs at room tem-perature. Thus, the variable contributions of the horse(minor) dehydrogenase band to the donkey major band madequantitation unreliable. The electrophoretic patterns forleukocyte preparations from horse, donkey, and female mulewas found, as expected, to be consistent with the X-linkedmode of inheritance demonstrated by others for the eryth-rocyte enzyme (20, 21); the female mule pattern consisted ofa superimposition of the parental bands.Leukocyte preparations (as well as fibroblasts) were found

to be relatively free of minor bands under these experimentalconditions, but the formation of the slower minor bands was

further diminished by acid-saline washings, lysis at pH 6,and rapid high-voltage electrophoresis at 4°C with efficientcooling of the gels. Under these conditions, the separationbetween horse and donkey dehydrogenase bands in mulepreparations was excellent (Fig. 2).

Preliminary trials showed that the "mechanical" recovery

of enzyme from cellulose acetate gels was higher than 90%without electrophoresis. However, after electrophoresis, only50% of the applied activity could be recovered. Although the

mechanism of loss of enzyme activity is unknown, it is prob-ably related to the formation of the slower minor bands.Nonetheless, this loss of activity was the same for horse,donkey, and mule preparations and does not affect the re-

sults presented in this paper. The electrophoretically separated

bands were individually quantitated and Table 1 shows thepercentage of horse G6PD. The results are presented asmeans of five determinations (peripheral blood) and of fourdeterminations (fibroblasts) from each animal. The smallstandard deviations are indicative of the high reproducibilityof this technique. In all cases, regardless of the animal in-vestigated or the cell type studied, a preponderance of horsedehydrogenase activity is observed, accounting for 60-90%of the total activity. The percentage of donkey X-chromosomeinactivation closely paralleled the relative percentage ofhorse enzyme activity (Table 1). A highly significant correla-tion was found (r > 0.95) between late replication of a givenX-chromosome and the dehydrogenase activity of that allelecarried by the euchromatic X (Fig. 3).

DISCUSSION

Previous support of the Lyon hypothesis has been based oninformation obtained from different systems possessing eithersuitable chromosomal or enzymatic markers, but not both.The use of hybrid animals with morphologically distinctX-chromosomes with regard to parental origin, coupled withunique biochemical markers paralleling these cytogeneticmarkers, allows for a more rigorous test of the hypothesis.Our results represent the first instance of simultaneous quanti-tative biochemical and cytogenetic investigations of X-chromosome inactivation within a single experimental system.The opportunity to perform such a correlative study wasafforded by modification in the electrophoretic techniqueto allow easy separation and reliable quantitation of thehorse and donkey components of glucose 6-phosphate dehy-drogenase activity in the female mule.The agreement between the percentage of late DNA replica-

tion of one X-chromosome and the level of enzyme activitycontrolled by the other X-chromosome supports the Lyonhypothesis. The highly significant correlation coefficientstrongly suggests that the late replication of the X-chromo-some indeed represents genetic inactivation. This is true, atleast for the G6PD locus of the mule.Two previous investigations can possibly be interpreted as

indicating nonrandom X-chromosome inactivation in the

100

267

90

>. 80I-

>~~~~~~~~~~~~70

*fi 60- 560

50

50 60 70 80 90 100

% LATE REPLICATING Xd

FIG. 3. Correlation between percentages of late donkey X-chromosome replication and the horse component of dehydro-genase activity. The line represents perfect (1:1) correlation.Points lie close to this line.

Proc. Nat. Acad. Sci. USA 68 (1971)

Dow

nloa

ded

by g

uest

on

May

23,

202

0

Late X-Chromosome Replication in Mule 547

female mule (22, 23). By autoradiography, Hamerton et al.(22) found that the donkey X-chromosome was late-replicat-ing in approximately 90% of the cells of three female mules.In two of the animals, both skin fibroblasts and leukocyteswere studied and considerable agreement was observed be-tween the two tissues. Hook (23) found somewhat similarresults by a qualitative biochemical approach (starch gelelectrophoresis). He examined the glucose-6-phosphate de-hydrogenase phenotype of blood and various organs of 37female mules and found that in most of the tissues (blood,pancreas, brain, cervical cord, kidney, and parotid gland)"the horse (maternal) G6PD allele is preferentially ex-pressed". In spleen and lymph node, the expression of thetwo forms was apparently equal, while in liver, the donkeyform predominated. Our results partially agree with thesefindings, since in those animals examined, deviation from theexpected random pattern of late X-chromosome replicationwas observed (Table 1). The donkey X-chromosome waspredominantly the late-replicating element in all four ani-mals. This cytological observation corresponds to the highercontribution of the horse to the dehydrogenase activity inmule cells (Table 1).Whether or not this truly represents nonrandom inactiva-

tion remains a moot question. Cytogenetically, chromosomeidentification may prove a minor complicating problem in themule since the largest submetacentric element is also late-replicating and might be mistaken for the donkey X-chromo-some (Fig. 1A); a spurious increase in the frequency of late-replicating donkey X-chromosomes would result. This prob-lem can be overcome by using only cells in which both chro-mosomes are unequivocally identified.More importantly, the effects of sampling, selection, or

adaptation on the apparent "preferential inactivation" mustnot be overlooked. Cells from adult animals are obtained longafter the time of supposed X-chromosome inactivation inembryogenesis. Therefore, skin biopsies may have been ob-tained from a rather large "clonal" area and yield a biasedimpression of late replication or gene inactivation of only oneor the other X-chromosome. Linder and Gartler have studiedthe expression of two G6PD alleles (A and B) with regard toaverage "patch size" (areas of pure A or B cells) in humanfemale heterozygotes (24). In most cases, "patch sizes" of0.3 cm2 were large enough to ensure expression of both alleles.However, in one case, it was estimated that a patch size of1.0 cm2 represented the expression of a single allele. Shouldthe average patch size in the mule epithelium be large, de-viations from a 1:1 distribution of the active parental X-chromosomes would be expected in the original sample (0.3cm2) and would be expressed in the resultant fibroblasts.However, the migration of cells during development, afterX-chromosome inactivation, may result in the absence oflarge pure patches (expressing only one allele).The concept of different selection pressures on X-linked

genes in given organs is suggested by Hook's data (23), whichshow different patterns of enzymatic activity in variousorgans. Although the apparent preponderance of one enzymicform was uniform within a given organ of all the animalsstudied, this form was not active to the same degree in allorgans. Similarly, adaptive pressures of a specific tissue cul-ture environment on the fibroblasts may result in the selectionof cells expressing the same functional X-chromosome. This

enzyme activities ascribable to either the horse or the donkeycomponent in the derived fibroblasts with that of the originalbiopsy material. The possible in vivo and in vitro selectionpressures which may influence the relative proportions of thetwo G6PD alleles in heterozygotes have been discussed byGartler and Linder (25). Additionally, Nyhan et al. (26) haveshown intrachromosomal (gene) effects on the nonrandomexpression of the glucose-6-phosphate dehydrogenase pheno-type in human females, heterozygous for both the A and Balleles of glucose-6-phosphate dehydrogenase, and hypoxan-thine: guanine phosphoribosyl transferase (HGPRT) de-ficiency. The observed clonal phenomenon was explained bythese authors as selection against the HGPRT- cells afterrandom inactivation of the X-chromosome.Although the donkey X-chromosome was inactivated in a

majority of the cells (Table 1), we do not interpret this findingas supporting the "preferential inactivation" suggested byHamerton et al. (22). Even though the number of animalsstudied is small, the percentage of late replication of donkeyX-chromosomes in the female mule cells (and likewise of theactivity of the horse component of the dehydrogenase)varied between 60 and 90%. Furthermore, Mukherjee et al.(27) have demonstrated an almost 1:1 ratio of Xd:Xh chro-mosome late replication in fibroblasts of a female mule plusthe lymphocytes of a second animal (18). This, in essence,covers one-half of the entire range of possibilities. Perhapsa larger sample may include animals whose horse X-chromo-some is predominantly inactivated, representing the otherhalf of a continuous distribution. Nance (28) has shown thatin a population of proven heterozygous human females, theexpression of the two G6PD alleles (A and B) follows aGaussian distribution, supporting the concept of randominactivation of the two X-chromosomes. Thus, the apparent"nonrandom inactivation" of the X-chromosomes in themule (22) may reflect a sampling bias in obtaining only oneextreme of a continuous distribution of "mosaic" animals.

We thank Dr. A. Hadani of the Bet Dagon Veterinary Insti-tute, Israel, and the staff of the Buffalo Zoological Gardens forproviding specimens of horse, donkey, and mule cells. The in-vestigation was supported in part by grants from the UnitedStates Children's Bureau (Project no. 417) and NIH GeneralResearch Support Branch (RR-05493). One of us (MCR) is arecipient of a Dr. Henry Buswell and Bertha H. Buswell Re-search Fellowship. We are deeply indebted to Miss ClaudiaGray, Mrs. Clara Lockwood, and Mrs. Martha Molnar for ex-cellent technical assistance, and to Dr. R. G. Davidson for hishelpful discussion during the course of these studies and criticalreading of the manuscript.

1. Lyon, M. F., Nature, 190, 372 (1961).2. Ohno, S., and B. M. Cattanach, Cytogenetics, 1, 129

(1962).3. Russell, L. B., Science, 133, 1795 (1961).4. Beutler, E., M. Yeh, and V. F. Fairbanks, Proc. Nat.

Acad. Sci. USA, 48, 9 (1963).5. Davidson, R. G., H. M. Nitowsky, and B. Childs, Proc.

Nat. Acad. Sci. USA, 50, 481 (1963).6. Danes, B. S., and A. G. Bearn, J. Exp. Med., 123, 1

(1966).7. Migeon, B. L., Science, 160, 425 (1968).8. Taylor, J. H., J. Biophys. Biochem. Cytol., 7, 455 (1960).9. Mukherjee, B. B., and A. K. Sinha, Can. J. Genet. Cytol.,

5, 490 (1963).10. Mukherjee, B. B., and A. K. Sinha, Proc. Nat. Acad. Sci.

USA, 51, 252 (1964).11. Gartler, S. M., and B. Burt, Cytogenetics, 3, 125 (1967).

question, however, is easily tested by comparing the relative

Proc. Nat. Acad. Sci. USA 68 (1971)

12. Gruncb.erg, H.', Ann. Human Genet., 30, 230 (1967).

Dow

nloa

ded

by g

uest

on

May

23,

202

0

548 Genetics: Cohen and Rattazzi

13. Moorhead, P. S., P. C. Nowell, W. J. Meilman, D. M.Battips, and D. A. Hungerford, Exp. Cell Re8., 20, 613 (1960).

14. Schmid, W., Cytogenetic8, 2, 175 (1963).15. Rattazzi, M. C., L. F. Bernini, G. Fiorelli, and D. M.

Mannucci, Nature 213, 79 (1967).16. Meera Khan, P., and M. C. Rattazzi, Biochem. Genet., 2,

213 (1968).17. Glock, G. E., and P. McLean, Biochem. J., 55, 440 (1953).18. Benirschke, K., L. E. Brownhill, and M. M. Beath,

J. Reprod. Pert., 4, 319 (1962).19. Trujillo, J. M., C. Stenius, L. C. Christian, and S. Ohno,

Chromosnoma, 13, 243 (1962).20. Trujillo, J. M., B. Walden, P. O'Neil, and H. B. Anstall,

Science, 148, 1603 (1965).

Proc. Nat. Acad. Sci. USA 68 (1971)

21. Mathai, C. K., S. Ohno, and E. Beutler, Nature, 210, 115(1966).

22. Hamerton, J., F. Giannelli, F. Collins, J. Hallett, A. Fryer,V. M. McGuire, and R. V. Short, Nature, 222, 1277 (1969).

23. Hook, E., Genetics, 2, s30 (1970) (abstract).24. Linder, D., and S. M. Gartler, Amer. J. Human Genet.,

17, 212 (1965).25. Gartler, S. M., and D. Linder, Cold Spring Harbor Symp.

Quant. Biol., 29, 253 (1964).26. Nyhan, W. L., B. Bakay, J. D. Connor, J. F. Marks, and

D. K. Keele, Proc. Nat. Acad. Sci. USA, 65, 214 (1970).27. Mukherjee, B. B., A. B. Mukherjee, and A. B. Mukher-

jee, Nature, 228, 1321 (1970).28. Nance, W. E., Cold Spring Harbor Symp. Quant. Biol.,

29, 415 (1964).

Dow

nloa

ded

by g

uest

on

May

23,

202

0