J. Mol. Biol. (1969) 42, 441-455

Renaturation and Isolation of Single Strands from the Nuclear DNA of the Guinea pig

W. G. FLAMM?, P. M. B. WALKER AND MORA MCCALLUM

Department of Zoology, University of Edinburgh, Edinburgh, Scotland

(Received 11 July 1968, and in revised form 15 February 1969)

Preparative density-gradient centrifugation in caesium chloride was used to separate guinea pig DNA into several fractions. As in the mouse, the satellite DNA region of the gradient was shown to contain the majority of the very rapidly renaturing sequences. The observed rate of renaturation, however, was nearly twice that of mouse satellite. The specificity of the reaction was demonstrated by showing that its rate WEM appropriately enhanced by the addition of denatured homologous DNA but not by denatured heterologous DNA.

Centrifugation of satellite-enriched fractions in alkarme caesium chloride gradients revealed the presence of five distinct bands, of which four are thought to represent individual single strands of differing guanine plus thymine content. Two of these bands (a strands) were so widely separated that it was possible to isolate them directly without first preparing the satellite. Base composition, melting profiles, buoyant densities and behaviour on hydroxyaprttite crystals indicate that the two a strands are complementary to each other.

1. Introduction Recent studies on the characterization of the nuclear satellite DNA isolated from the mouse (NW musculus) have revealed some highly interesting properties which previously were not thought to be associated with DNA of higher organisms. For instance; mouse satellite can undergo rapid renaturation (Walker & McLaren, 1965a; Waring & Britten, 1966) even at very low concentrations of DNA (0.04 g/ml.) (Flamm, McCallum 6 Walker, 1967), it is highly homogeneous in base composition (Flamm, Bond, Burr & Bond, 1966; Flamm, Bond & Burr, 1966; Bond, Flamm, Burr & Bond, 1967) and undergoes visible strand separation in alkaline gradients of caesium chloride (Flamm et al., 1967). The rate at which thermally denatured satellite renatures has been interpreted to imply that mouse satellite is composed of a nucleo- tide sequence containing 300 to 400 base pairs of which there are approximately a million copies per mouse genome (Waring t Britten, 1966). Of further interest is the funding that the isolated single strands of mouse satellite are capable of interacting to a limited extent with themselves and that despite the great abundance of t,hese sequences in the mouse they are rare or perhaps non-existent in other rodents (Plamm. Walker t McCallum, 1969).

Views concerning evolution and function of the satellite have been discussed elsewhere (Britten & Kohne, 1968; Walker, Flamm & McLaren, 1968), but these depend largely upon knowing to what extent the above properties are peculiar or

t Present acldresa: Cell Biology Branch, National En~irorunontal Health Sciences Ccntst*r.. L’.O. Box 1233, Research Triangle Park, 9.C. 27700, V.S.:I.

441

442 I\:. G. FLANN, P. AI. B. if-ALKER ASD 31. JlrCALLUAI

exclusive to the mouse. For instance, if we assume all nuclear satellites of Rodentia are functionally similar, determining which properties are common to each should help us to understand better which features are obligatory to their function. Furthermore, any similarity or difference between satellites of very closely rclatcd species should provide important clues to the rate and perhaps the means by which they have evolved.

In this paper, we compare the most characteristic physical and chemical properties of guinea pig satellite to its counterpart in the mouse, though admittedly, the two species are only distantly related. An abstract of this work has appeared (Flamm, McCallum & Walker, 1966).

2. Materials and Methods (a) Isobtiolz and fractionation of guinea pig DNA

DNA of 8 to 10 million molecular weight was prepared from guinea pig livers and testes by the phenol and sodium dodecyl sulphate procedure described earlier (Walker & McLaren, 196%). DNA labelled with 3aP, prepared from embryonic cells grown in tissue culture, was isolated from CsC1 gradients following lysis of the cells in 1 o/o sodium dodecyl sulphate (Flamm, Birnstiel & Walker, 1968).

DNA was separated into 4 or more discrete fractions by equilibrium density-gradient centrifugation in CsCl (M.S.E. centrifuge, 10 x 10 aluminium angle rotor) according to the method of Flamm, Bond & Burr (1966). After the fractions were collected and read in a Beckman DB-G spectrophotometer they were diluted fourfold with 0.01 M-Tris-HCl, pH 8.6, and pelleted by centrifuging for 15 to 18 hr at full speed in either the M.S.E. 10 x 10 aluminium or titanium rotor. In this way, DNA was quantitatively recovered from solution and adjusted to whatever new concentration or ionic environment required.

(b) Separation and isolation of singIt? strands Single strands of DNA were isolated from alkaline gradients of CsCl following equilibrium

centrifugation in angle rotors as described by Flamm et al. (1967). However, during the course of this study we discovered that Harshaw’s optical grade of CsCl is obligatory if the published procedure is followed. Other grades of CsCl apparently contain heavy-metal contaminants which have the effect of increasing the density of alkaline DNA (pH 12.5) to a point where it will no longer band. This difficulty can be circumvented by including a chelating agent such as EDTA in the CsCl gradient. Accordingly, 1.7 ml. of 0.02 M-EDTA solution, pH 9.5, was added to 1.7 ml. of 0.01 M-Tris, pH 8.6, containing 500 pg sodium dodecyl sulphate and 200 to 600 pg DNA; 0.1 ml. of 1 M-NaOH was then added (pH 12.4 to 12.8) and the solution brought to a density of I.765 g crnM3 with Analar CsCl (B.D.H. Ltd.). When necessary, the single strands can be purified further by recycling in a second alkaline gradient without significantly reducing their molecular weight (Flamm, Birnstiel &Walker, 1968).

Because the heaviest and lightest single strands (IX strands) are so widely separated at equilibrium in the M.S.E. 10 x 10 aluminium rotor, a speed of 46,000 rev./min was selected. Equilibrium at this speed is attained within 20 hr. All density-gradient centrifugations were carried out at 25°C.

(c) Analytical density-gradient centrz@ga.tion in CsCl Analyses were conducted with an M.S.E. analytical ultracentrifuge equipped with a

standard ultraviolet absorption optical system. 10 mm thick epoxy centrepieces having a single 3” sector and fitted with alkali-resistant ‘0’ rings were used. Radial distances of the cells and reference lines are identical to those of the Spinco An-D rotor. Centrifugation was carried out at 25°C for 20 to 24 hr at 48,000 rev./mm Photographic records were taken on 35 mm spectroscopic safety film (Kodak type 103-O) and the exposures analysed with a Joyce-Loebl densitometer. Initial solution densities were determined refracto- metrically (Ifft, Voet & Vinograd, 1961) except when the Na+ concentration exceeded 0.05 M and these were determined pycnometrically. For neutral gradients, either mouse

NUCLEAR DNA OF THE GUINEA PIG 443

satellite DNA at a density scale of l-691 (Chun & Littlefield, 1963), or Microc~~~us Zysodeikticw at 1.731 (Schildkraut, Marmur & Doty, 1962) or both were used aa a reference. For alkaline gradients, buoyant densities were determined from root-mean-square positions and are expreesed in termed of the initial density of the solution at atmospheric pressure (Vinograd & Hearst, 1962).

(d) Hydroxyapatite fractionution

Hydroxyapatite fractionations were carried out according to McCallum & Walker (1967) except that the more convenient centrifugation procedure was followed (Flamm et al., 1969) when renaturation rates were investigated. In all ceaes, the DNA wa8 sonically disrupted to lengtha of about 600 base pairs (Walker & McLaren, 1965b) and heat denatured (10 mm at 100°C) before incubation at 60°C in 0.12 ~-sodium phosphate, pH 6-8, and fractionation at 70°C.

(e) Melting profdes The 260 mp optical absorbance of DNA in 0.06 M-sodium phosphate, pH 6.8, was

monitored during continuous temperature change in l-cm closed, thermostatically con- trolled quartz cells. Temperatures and absorbancies were recorded simultaneously from a Zeiss PM& II spectrophotometer. All curves have been corrected for thermal expansion.

3. Results When DNA is extracted from either whole cells or isolated nuclei of guinea pig

tissue and centrifuged to equilibrium in a density-gradient of C&l, two DNA bands appear, the smaller of which is referred to as a satellite band (insert, Fig. 1). Superficially, the guinea pig satellite differs in two respects from that found in the mouse. It has a substantially higher buoyant density (l-705 as compared to

I.700 1.705 4

4

:;;

Fraction no.

FIG. 1. Typical band-profile of guinea pig DNA centrifuged to equilibrium in a preparative C&l density gradient.

50 pg DNA from guinea pig liver was centrifuged at 35,006 rev./min for 60 hr at 25% in an M.S.E. 10 x 10 aluminium rotor. Initial solution density was 1.715 g crne3 at 25°C. Bar graph indicates fractions taken for subsequent analysis. Insert represents a microdensitometer tracing of an ultraviolet photograph of the same DNA (2 pg) centrifuged to equilibrium at 45,000 rev./min in an M.S.E. analytical ultracentrifuge.

444 \v. c. E’LAMM, P. 1\1. u. \VALKEK AND 51. ,\lC(‘i~Ll,U~1

1.691 g cm -3 for mouse satellite) and is less well resolved from the main band of DNA. If we assume the width of the main band is nearly equivalent to that determined for many other rodents of roughly the same molecular weight (Kit, 1961,1962), it is unlikely that guinea pig satellite could ever be completely separated from main band fragments by CsCl centrifugation. For this reason the term satellite is in- appropriate though we continue to use it here since the term appears in the literature and the only alternative is to call it fraction II.

In any case, we want to know whether this region of the gradient contains the most frequently repeated nucleotide sequences of the nuclear genome as does its counterpart, the mouse satellite. Such evidence has been provided (Flamm, Birnstiel & Walker, 1968) and is consistent with Figure 2 where the ability of individual fractions to re-form DNA duplexes after thermal denaturation and a short incubat’ion has been examined using the hydroxyapatite procedure. Clearly, the region of the gradient containing the greatest proportion of abundant, and hence rapidly renaturing, sequences is coincident with the satellite shoulder.

r

j t

Fraction no.

FIQ. 2. Ultraviolet absorption (-O-O-) and renaturetion profile of guinea pig DNA. 60 M of 3aP-labelled DNA (2000 cts/min/pg) from embryonic cells grown in tissue culture w&e centrifuged in C&l 8s described in Fig. 1.

-@-a-, Percentage renaturation of each frection after & lo-min incubation at 6O”C, 8t & DNA concentration of 1 pg/ml. in 0.12 M-phosphate. The hydroxyapatite procedure described by McCaJlum & Walker (1967) was used to 8we.w the extent of renaturation (see Materials and Methods).

Since some of the sequences with a lower degree of repetition are as dense as, or more dense than, the satellite region of the gradient, methods other than neutral CsCi centrifugation are needed if guinea pig’s most highly repetitive sequences are to be purified properly. The logical choice was the hydroxyapatite procedure with which highly purified mouse satellite DNA has been prepared. The method is based on the ability of hydroxyapatite to separate thermally denatured DNA from those DNA molecules which have re-formed duplex structures (Walker & McLaren, 1965a; McCallum & Walker, 1967).

Figure 3(a) shows the banding profile of such a fract’ion. The narrow band width

NUCLEAR DNA OF THE GUTSEA PIG 445

(a)

I I

- (b)

1.3 -

-2 z I.2 - N

,” 6

-E 0 -

2 -% I,I-

40 60 80

Temperature (“Cl

FIG. 3. Densitometer tracing (a) and melting profile (b) of hydroxyapatite-prepared scttellite. DNA from guinea pig liver (100 &ml.) was sonic&ted, heat denatured (lOO’C, 5 min) and

incubated at 60°C for 5 rnb in O-12 M-phosphate then fractionated on hydroxyapetite as described by McCallum & Walker, 1967. The 0.3 M-phosphate fraction (renetured fraction) represented 9.3% of the total DNA and was wed in the analyses shown above. A midpoint in the tempera- ture melting curve of 70°C is indicated.

shows it is of high molecular weight after renaturation. This can be accounted for by networks which form during renaturation and are apparently caused by an inter- molecular joining of short single-stranded regions of partially renatured DNA to give high molecular weight polymers held together by hydrogen bonding (B&ten & Waring, 1965). The rapid formation of networks, as evidenced by Figure 3(a), is characteristic of highly repetitive DNA’s such as mouse or guinea pig satellite.

The midpoint temperature and sharpness of the melting curve (Fig. 3(b)) show the “rapidly reassociating sequences” possess a highly ordered secondary structure similar to the original duplexed molecule.

One of the important questions concerning the properties of guinea pig satellite is how closely its rate of renaturation compares with that of mouse satellite (Waring & B&ten, 1966; Flamm et al., 1967). We use the hydroxyapatite fraction shown in Figure 3 which occurs predominantly in the satellite region of the gradient. It is by

446 W. 0. FLAMM, P. M. B. WALKER AND M. McCALLLTM

measuring this rate, k,, that we can arrive at an estimate of the complexity of DNA (Britten & Waring, 1965; Waring & Britten, 1966; Britten & Kohne, 1966; Wetmur & Davidson, 1968). For in fact, the rate-constant for rcnaturation (IL,) is inversely proportional to the complexity of DNA (Britten & Waring, 1965) and providing that all sequences are unique, the genome size of an organism can be estimated on the basis of renaturation data using the formula shown below (W’rtmur C% Davidson, 1968).

From the data presented in Figure 4 we have calculated k2 values of 8.5 and 15 x lo3 litres/mole second for mouse and guinea pig, respectively. Since hydroxy-

3000 6000

Incubation time (set)

Fm. 4. The second-order rate plot for the renaturation of guinea pig hydroxyapatite-prepared satellite and mouse satellite DNA. Both samples were heat-denatured (lOO”C, 5 min) and incubated at 0.04 pg/ml. in 0.12 M-phosphate for the times shown. The extent of renaturation at any given time was determined by the hydroxyapatite procedure (Flamm et al., 1969). The total extent of renaturation (95%, mouse; 87%, guinea pig) wae determined by extrapolation of a double reciprocal plot of time against extent of reaction and was used to calculate the size of the un- renatured fraotion. Hydroxyapatite-prepared satellite (-@-a-) was obtained from the same preparation used in Fig. 2 and prepared according to Fig. 3. Specific activity of mouse satellite (-O--O-) was 30,000 cts/min/pg. The unrenatured fraction is expressed in moles nucleotide/l.

apatite reassociation rates are about twice as fast as those measured optically (Britten & Kohne, 1966), we introduce an extra factor of two into the Wetmur & Davidson formula ( 1968),

2 N 6.5 x 108 (s;;,.l,“)“s” D

=

ka >

where SX,“,‘w” is the corrected sedimentation coefficient of DNA in alkaline solution (determined according to Studier, 1965, and taken as equal to 4.0) and correcting for Na + concentration, the complexity of the satellites in molecular weight units (N,) is

NUCLEAR DNA OF THE GUINEA PIG 447

1.1 x lo5 daltons for mouse and 6-O x lo4 daltons for guinea pig. Dividing by the molecular weight of a nucleotide pair, we estimate that mouse and guinea pig satellite possess a basic repeated length of 180 and 100 nucleotides, respectively.

The calculation is subject to the following criticism: the repeated element is slightly shorter than the reactant, thereby permitting more than one registration between pairs of complementary fragments; no correction has been made for base composition though it is thought to have some influence on the renaturation rate (Wetmur & Davidson, 1968); the conversion factor used to convert kinetic data determined by the hydroxyapatite method to its equivalent measured optically is not precisely known. In fact, the final values may be in error by as much as a factor of two and should be so regarded. We do not, therefore, claim to differ significantly from Waring & Britten (1966) who report a value of 350 nucleotides for the repeated length of thr mouse satellite sequence.

In any case, since each satellite constitutes approximately 9% of the mass of nuclear DNA, they must each represent a total of 5 x lOa nucleotide pairs per genome. This means that each basic sequence length of 180 to 100 nucleotide pairs is repeated 3 to 5 x lo6 times per cell. Whether this means that all sequences are identical or that they merely resemble each other we are not able to say.

It seems, however, that the renaturation reaction involves a reasonably specific reassociation of complementary strands. In Table 1, a times 100 excess of either unlabelled homologous DNA (guinea pig) or unlabelled heterologous DNA is added and denatured with the 32P-labelled fraction of guinea pig DNA shown in Figure 3.

TABLE 1

l3ffect of a loo-fold excess of unlubelled DNA upon the renaturation half-life of hydroxyapatite-satellite

Unlabelled DNA Half-life

at 0.01 pg/ml. at 0.04 pg/ml.

.

None 5 s 12.5 Guinea pig 7 1 ,.-i

Escherichia coli 70 12.0 Mouse 60 17.0

Peromyacu.8 poli 70 12.0 Apodemus sylvaticus - 12.0

Rat 70 20.0

Unlabellod DNA was heat-denatured and incubated at 60°C with 32P-labelled hydroxyapatite- satellite DNA. After 5, 10, 30 and 100 min, the extent of renaturation was determined by the hydroxyapatite procedure (Flamm et al., 1969).

If the heterologous DNAs contained sequences like the guinea pig fraction, or if the reaction were non-specific, addition of the loo-fold excess would drive the reaction faster to the right and shorten the reaction half-life (Walker & McCallum, 1966; Walker et al., 1968). Clearly, this does not happen. An enhancement of the rate of renaturation occurs only when guinea pig DNA is itself added. Hence, the reaction is specific and guinea pig DNA is again similar to mouse in the sense that its most

30

448 W. G. FLAMM, P. M. B. WALKER AND M. McCALLUM

highly repetitive sequences are not abundant in the other rodents examined (Tl’lamm et al., 1969).

If guinea pig satellite actually consisted of a single nucleotide sequence containing only 100 base pairs repeated millions of times in each genome, then it is certainly homogeneous enough to undergo visible strand separation in an alkaline gradient of CsCl providing there exists an interstrand compositional bias. As Vinograd, Morris, Davidson & Dove (1963) and Baldwin $ Shooter (1963) have shown, the buoyant density of native DNA is insensitive to changes of pH above neutralny until a critical pH greater than 11 is reached, whereupon the DNA rapidly denatures (Davison, 1966) and increases in density. The increase is attributable to: (a) loss of secondary structure; and (b) titration of the N-H protons in the guanine and t’hymine residues resulting in the addition of one caesium ion per residue titrated.

Obviously, if the molar proportion of G plus T differed between the two st’rands of a double-stranded duplex, the (G + T)-rich strand would acquire a higher buoyant density than its complement following titration. The most extreme example is that of poly d(T-G).d(A-CC) which splits into two widely separate bands following equilibrium centrifugation in alkaline CsCl (Doerfler & Hogness, 196Sa; Wells & Blair, 1967). The poly d(T-G) strand becomes 0.14 gcmm3 denser than its poly d(A-C) complement. Certain bacteriophage DNA’s (h, SP84) also exhibit two bands in alkaline CsCl although with considerably less*separation (Doerfler & Hogness, 1965b; Saunders & Campbell, 1965).

Neither the DNA of higher organisms nor bacterial DNA can be expected to show such interstrand bias since their greater complexity should cause any differences t’o be averaged out. Not so, of course, with simple satellites as shown in the mouse by t,he well-resolved bimodal distribution in alkaine CsCI. Its light (42 %, G + T) and heavy (57%, G + T) bands are separated by O-025 gcme3 (Flamm et aE., 1967).

With guinea pig DNA we have a more difficult problem. The hydroxyapatite- prepared fraction (of Fig. 3) was sheared and is too small in molecular weight to band sharply in alkaline CsCl while the isopycnically prepared satellite (fraction II) is contaminated by main band sequences. Nevertheless, as Figure 5(b) illustrates, fraction II undergoes visible strand separation in alkaline CsCI, but unlike mouse satellite exhibits five distinct modes. The broad band found at a density of 1.76 g cm - 3 is undoubtedly attributable to main band contamination, since its density increment (native to alkaline DNA, O-055 to O-060 gcmA3) is consistent with that of a DNA containing 50% G + T on either strand (Vinograd, Lebowitz & Watson, 1968).

The two relatively sharp bands immediately adjacent to and a,ppearing on either side of the broad band will be referred to as p bands. They are separated by 0.03 g cmA3 and are thought to represent the complementary strands of a duplex possessing an interstrand compositional bias similar to that of mouse satellite.

Of greater practical interest are the two extreme bands (a bands) which are separated by a density difference of 0.09 g cme3. By comparison with the poly d(T-G).d(A-C) system, we would judge them to be complementary to each other in base composition with the heavy strand (Ha) containing 60% more G plus T than the light strand (LX). This has been confirmed by chemical analysis of their base composition (Table 2).

So great is the separation between a bands that either strand can be isolated directly from alkaline gradients of whole nuclear DNA (Fig. 5(a)). In fact, all sub- sequent experiments have been performed on a strands isolated from whole nuclear

NUCLEAR DNA OF THE GUINEA PIG

TABLE 2

Baee composition of a strands

449

Cytosine A&nine Guanine Thymine

LCr 37.8 f 0.4t 39.6 f 3.2 4.4 f 0.0 18.3 f 2.7

Ha 5.2 j, 1.2 20.9 f 0.4 30.6 f 1.8 43.3 f 3.3

t Preliminary results (Edwin Southern, personal communication) indicate the presence of at least two other components in addition to cytosine.

I I I I I ,

I.714 I.76 I.80 Density (g/cm3)

FIQ. 5. Densitometer tracing of nuclear DNA (a) and fraction II (b) centrifuged to equilibrium in an alkaline gradient of CsCl.

DNA was extracted from nuclei of guinea pig liver isolated according to Dejardins, Smetana, Grogan, Higashi & Bush, 1966. Is order to emphasize the presence of minor components in (a), the gradient wae overloaded with 30 pg of DNA and the tracing of the major band left open. Approximately 12 c(g DNA was used in (b). From preparative banding profiles of nuclear DN4 the minor components at 1.71 and 1.80 g cmW3 were each estimated to constitute 2.4% of the total DNA maas. OS this baais, the combined mass of fi bands appears to constitute another 5qh. Centrifugation was carried out aa described in Materials and Methods.

DNA centrifuged to equilibrium in preparative gradients of alkaline CsCl. It is clear, however, from analysis of all four fractions in Figure 1, that only those fractions which contain the satellite exhibit a bands in alkaline CsCl.

Further evidence suggesting that Ha and La are complementary single strands is provided by rebanding experiments in neutral CsCl (Fig. 6). Both Ha and La display unimodal bands with densities of 1.754 and l-706 g cme3, respectively. Neither density is affected by heat treatment (1OO’C for 5 min) or by subsequent incubation at 60°C (in 0.12 M-SOdkIm phosphate, pH 6%). However, when equimolar quantities of Ha and La are mixed together at room temperature in 0.12 M-sodium phosphate they form a single hybrid band at a density (l-721 g cmV3) intermediate between La and Ha (Fig. 6(c)). Heat denaturation and incubation for one or two minutes at 60°C at 10 pg DNA/ml. reduces the density of the hybrid to 1.715 g cme3, presumably by ensuring that only properly matched duplexes remain paired.

450 W. G. FLAMM, P. M. B. WALKER AND M. McCALLUM

I.691 I.706 I.721 1.754

Density (g/cm3)

FIG. 6. Densitometer tracing of (a) isolated La, (b) isolated Ha and (c) an equimolar mixture of LX and Ha centrifuged to equilibrium in IX neutral CsCl gradient as described in Materials and Methods. Native mouse satellite served as a density marker (1.691 g cmS3).

The melting curve of the re-annealed hybrid, prepared by incubating equimolar proportions of Ha and La for five minutes at 60°C shows it to contain duplexed regions. Isolated Ho: or La, however, melt in a manner characteristic of single- stranded polynucleotides (Fig. 7). Although it is by no means certain, available information suggests that the great difference between the hyperchromicities given by Ha and Lu might be related to their respective molar contents of A+ C. In fact, the following percentage hyperchromicities have been reported for different mole fractions of A+C: 33%, O-77; 29%, 0.59; 24%, 0.54; 20%, O-48; 15%, 0.43 and 9%, 0.26 (Saunders & Campbell, 1965; Flamm et al., 1967; Table 2).

The complementarity between u strands is further supported by hydroxyapatite experiments (Fig. 8). Equimolar mixtures of the two u strands elute predominantly at high salt concentrations where renatured or native DNA’s are known to chromato- graph, whereas Lu and Hu are completely eluted from hydroxyapatite at low salt concentrations as expected of single-stranded polynucleotides. They do, however, differ in this respect from the L and H strands of mouse satellite which possess a small fraction (10 to 15%) eluting at 0.3 M (Flamm et al., 1967). This fraction has been attributed to the presence of a short sequence which is periodically reversed between the L and H strands of mouse satellite thereby permitting homologous strands to interact to a limited extent with themselves (Flamm et aE., 1969).

The specificity of the reaction of CL strands is indicated by the failure of both Ha and La to associate with either the L or H strands of mouse satellite (Table 3) under the same conditions as those used in Figure 8. In fact, the data of Tables 1 and 3 and that reported by Flamm et al., 1969, show that mouse and guinea pig satellites are composed of different sequences since neither interacts with the other.

NUCLEAR DNA OF THE GUINEA PIG 451

TABLE 3

Lack of complementarity between guinea pig M. strands ad the single strands of mouse satellite

32P-labelled *owe

satellite strand

U&belled strand

pg unlabelled

pg labelled

0, /O rcnstured

- L mouFIe

L G[ guinea pig L G( guinea pig

- H mouse

H 0: guinea pig H a guinea pig

- 1.0

10.0 10.0 - 1.0

10.0 10.0

0.0 80.0

0.4 0.4 1.2

75.0 1.5 2.0

32P-labelled single strands were denatured (lOO’C, 5 min) and incubrtted with unlabelled single strands at 60°C in 0.12 ix-phosphate for 200 min. Concentration of the labelled strand was 0.02 pg/ml. Percentage renaturation was determined by the hydroxyapatite procedure (Flamm et al., 1969).

IO 20 30 40 50 60 70 60 90 Temperature ("C)

FIG. 7. Melting curve of c( fraction in 0.06 M-phosphate. (0) Isolated La strand; (A) isolated Ha strand; (0) en equimolar mixture of La and Ha- heat treated (lOO”C, 5 min) and incubated for 5 min at 60°C before cooling and analysis. Measurements were as specified in Materials and Methods.

W. G. FLAMM, P. M. B. WALKER AND M. McCALLlT&\I

2000

IO00

Y x

1500

900

3 15 Fraction no.

FIG. 8. Stepwise elution of 32P-labelled DNA (specific activity, 10,000 cts/min/pg) from hydroxyapatite crystals after heat denaturation and a 30-mm incubation of 0.5 pg DNA/ml. in 0.12 M-phosphate. Incubation temperature was 60°C with fractionation at 70°C. (a) -O-O-, Isolated La strand; -a-+--, isolated Ha strand. (b) Equimolar mixture of La and HG( strands. The rate (ka) by which the a strands m-associate to form complexes eluting at 0.3M was determined at 0.02 pg single strand/mI. to be 8.4 x IO3 I./moIe sec.

4. Discussion With the exception of possessing a higher buoyant density and exhibiting twice

as many bands in alkaline CsCl, the satellite DNA of guinea pig appears to be basically similar to its counterpart in the mouse.

Neither satellite is completely homogeneous in terms of being composed of only one type of nucleotide sequence, Guinea pig satellite contains u and /3 duplexes while mouse satellite consists of at least two hinds of sequences on either strand. One type, the single-strand interaclting portion, appears as a very short sequence which is periodically reversed between the H and L strands of mouse satellite (Flamm et al.,

1969). The second type, representing the bulk of the satellite fraction, is a sequence of 180 nucleotide pairs (or 350 as reported by Waring & Britten, 1966) organized in such a way that its complement is always restricted to the opposite L or H strand (Flamm et al., 1967,1969).

Perhaps, the a and #I sequences of the guinea pig bear some functional or evolutionary relationship to the two types of sequences found in mouse satellite. There are, however, notable differences between them.

First, tc and /3 sequences are nearly of equal mass in the satellite fraction whereas the reversed sequences of mouse constitute no more than 10% of the total satellite mass

NUCLEAR DNA OF THE GUINEA PIG 453

(Flamm et al., 1969). Second, reversed sequences are interspersed between long stretches of DNA which contain the other mouse satellite sequence, while both 0: and /3 sequences are organized into separate and relatively longer blocks of DNA. From molecular weight estimates of single a or ,9 strands (1 to 2 x lo6 daltons, calculated from equilibrium band-widths, see Vinograd & Hearst, 1962), at least 50 repeated a

sequences or 50 repeated /I sequences must exist within a single stretch of DNA. We do not know whether blocks of a sequences are adjacent to blocks of /? sequences within the genome, but clearly they are segregated in DNA of the size we isolate. The main evidence suggesting that /3 sequences renature rapidly is based on the observa- tion that the c( plus /3 strands constitute one-half the mass of fraction II and that approximately one-half this mass undergoes rapid renaturation (Flamm, Birnstiel & Walker, 1968). In further support of this view is the recent observation (unpublished) that only 25% of fraction II renatures following the removal of a bands. Also, the pro- portions of a and /3 sequences (4.8 and 5.0% of total DNA, respectively) are in agrec- ment with the size of the very rapidly renaturing fraction (Fig. 3).

Although u sequences do not appear to possess single-strand interacting regions as do the L and H strands of mouse satellite (reversed sequences), it seems probable they contain some other form of microheterogeneity. The failure of either the satellite or a strands to rcnature completely as judged by buoyant density and spectrophotometric measurements, contrasts sharply with the behaviour of mitochondrial DNA, which renatures more slowly but forms a more nearly native product following thermal or alkaline denaturation (Corneo, Moore, Sandi, Grossman & Marmur, 1966; Borst & Ruttenberg, 1966).

One possible explanation might involve the problem of microheterogeneity. If, for example, a short sequence (heavy lines, Fig. 9(a)) were imposed between every three or four basic repeated sequences (light lines, Fig. 9(a)) which govern the rate of re- naturation (k,), it could prevent fully contiguous base-pairing and so influence the nature of the product as shown by Figure 9(a). Reduction of the molecular weight either by sonication or heat treatment should “release” the microheterogeneity from the repeated sequence and thereby facilitate fully contiguous base-pairing and bhe formation of a more nearly native product (Fig, 9(b).

Support for this view has been provided by Bond et al. (1967) in the case of mouse satellite and to some extent by the studies reported here with guinea pig satellite. Corneo, Ginelli & Polli, 1968, who observed a similar phenomenon, reported that even though the complementary strands of human satellite DNA very rapidly re-join, not all the base pairs enter into register. The amount of base pairing, however, is increased by long incubations (5 hr) at 65”C, which presumably reduce the molecular weight and give rise to the type of products depicted in Figure 9(b).

The type of heterogeneity proposed above represents only one possible class. Another type could include the sequences we have referred to as repeated. They might, for instance, all vary slightly one from another but retain certain basic similarities in base composition and nucleotide sequence, The best example of this kind of heterogeneity is provided by crab poly d(A-T), which unlike the in vitro polymer, renatures with a temperature at the midpoint of the melting curve 5 deg.C lowrr than the native material (Davidson et aE., 1965). This is presumably due to the small percentage of cytosine and guanine residues scattered through the molecule (Swartz, Trautner & Kornberg, 1962). The only argument in favour of the so-called repeated sequences being highly similar is based on t’he species specificity of their

454 W. G. FLAMM, P. M. B. WALKER AND 114. McCALLUM

(a)

H L-H

L H L-H

FIG. 9. Model for tho m-association of complementary single strands of high (a) and low (1~) molecular weight. For the purpose of simplicity, the stretches of microheterogeneity (- ) and the basic repeated sequence (- ) are assumed to be of identical length.

renaturation. Direct chemical measurements, however, are needed to establish the exact nature and extent of this kind of heterogeneity.

The central question concerning the biological purpose served by so many similar or identical sequences remains unanswered. It is clear, however, that the blocks of DNA which contain the satellite sequences are more complex than originally pro- posed. Possibly a better understanding of the nature of this heterogeneity will provide a key to the question of function.

Note added in proof. Since this paper was submitted, a report by G. Corneo, E. Ginelli, C. Soave & G. Bernardi (1968, Biochemistry, 7, 4373) has appeared in which the guinea pig satellite was isolated by a Ag + - Cs,S04 density-gradient technique. Tho strands referred to in our paper as Lee and Her strands were observed by them and found to have properties similar to those reported by us. They did not, howcvor, observe the ,6 bands which arc oxhibited in Figure 5 of this report.

We are grateful to Dr AM McLaren for much helpful discussion and to David Richmond for invaluable technical assistance. We also thank the Nuffield Foundation for support in the early stages of the project and the Medical Research Council for the support of a Research Group on the Mammalian Genome. One of us (W. G. F.) was on foreign assign- ment from the National Cancer Institute, Bethesda, Maryland, U.S.A.

REFERENCES

Baldwin, R. L. & Shooter, E. M. (1963). J. Mol. Biol. 7, 511. Bond, H. E., Flamm, W. G., Burr, H. E. & Bond, S. (1967). J. &ZoZ. Biol. 27, 289. Borst, P. & Ruttenberg, G. (1966). Biochim. biophys. Acta, 114, 647. Britten, R. J. & Kohne, D. E. (1966). Yearb. Carnegie Inst. 1965, p. 78. Britten, R. J. & Kohne, D. E. (1968). In Handbook of illolecular Cytology, ed. by A. Lima-

de Faria. Amsterdam: North Holland Press, in the press.

NUCLEAR DNA OF THE GUINEA PIG 4.x

Brit,ten, R. J. & Waring, M. J. (1965). Yearb. Carnegie Inst. 1964, p. 313. Chun, E. H. & Littlefield, J. W. (1963). J. Mol. Biol. 7, 245. Corneo, G., Ginelli, E. & Polli, E. (1968). J. Mol. Bill. 33, 331. Cornea, G., Moore, C., Sandi, D. R., Grossman, L. J. & Marmur, J. (1966). Science, 151,

687. Davidson, N., Widholm, J., Na;ndi, U. S., Jensen, R., Olwerg, C. & Wang, J. C. (1965).

Proc. Nat. Acad. Sci., Wash. 53, 111. Davison, P. F. (1966). J. Mol. Biol. 22, 97. Dejarclins, R., Smetana, K., Grogan, D., Higashi, K. & Bush, H. (1966). Cancer Res. 26,

97. Doorflor, W. & Hogness, D. S. (1965a). J. Mol. BioZ. 14, 237. Doerflor, W. & Hogness, D. S. (19653). Fed. Proc. 24, 226. Flamm, W. G., Birnsteil, M. L. & Walker, P. M. B. (1968). In Subcellular Components, od,

by G. D. Birnie, p. 125. London: Butterworths. Flamm, W. G., Bond, H. E. & Burr, H. E. (1966). Biochim. biophys. Acta, 129, 310. Flamm, W. G., Bond, H. E., Burr, H. E. & Bond, S. (1966). Biochim. biophys. Acta, 123,

652. Flamm, W. G., McCallum, M. &Walker, P. M. B. (1967). Proc. Nut. Acad. Sci., Wash. 57,

1749. Flamm, W. G., McCallum, M. & Walker, P. M. B. (1968). Biochem. J. 108, 42. Flamm, W. G., Walker, P. M. B. & McCallum, M. (1969). J. Mol. BioZ. 40, 423. Ifft, J. B., Voet, D. H. & Vinograd, J. (1961). ,J. Phys. Chem. 65, 1138. Kit, S. (1961). ,J. Mol. BioZ. 3, 711. Kit, S. (1962). Nature, 193, 274. McCttllum, M. 8: Walker, P. M. B. (1967). Biochem. J. 105, 163. Saunders, G. & Campbell, L. (1965). Biochemistry, 4, 2836. Schildkraut, C., Marmur, J. & Doty, P. (1962). J. MOE. BioZ. 4, 430. Studier, F. W. (1965). J. Mol. BioZ. 11, 373. Swartz, M. N., Trautner, T. A. & Kornberg, A. (1962). J. BioZ. Chem. 237, 1961. Vinograd, J. & Hearst, J. E. (1962). Fortechr. Chem. org. Naturatoffe, 20, 373. Vinograd, J., Lebowitz, J. & Watson, R. (1968). J. Mol. BioZ. 33, 173. Vinograd, J., Morris, J., Davidson, N. & Dove, W. F., Jr. (1963). Proc. Nat. Acad. Sci.,

Wash. 49, 12. Walker, P. M. B., Flamm, W. G. & McLaren, A. (1968). In Handbook of Molecular Cytology,

nd. by A. Lima-de-Faria. Amsterdam: North Holland Press, in the press. Walker, I?. M. 13. & McCallum, M. (1966). J. Mol. BioZ. 18, 215. Walker, P. M. B. & McLaren, A. (1965a). Nature, 208, 1175. Walkar, P. M. B. & McLaren, A. (1965b). J. Mol. BioZ. 12, 394. Waring, M. & Britten, R. J. (1966). Science, 154, 791. \Vr:lls, W. D. & Blair, J. E. (1967). J. Mol. BioZ. 27, 273. \I’ot’mur, J. G. & Davidson, N. (1968). J. Mol. BioZ. 31, 349.