Proc. Nati. Acad. Sci. USA Vol. 83, pp. 696-700, February 1986 Multiplicity of DNA sequences in Drosophila melanogaster (repeated DNA//clone instability) ALLAN R. LOHE* AND DOUGLAS L. BRUTLAGt Department of Biochemistry, Stanford University School of Medicine, Stanford, CA 94305 Communicated by David S. Hogness, September 3, 1985

ABSTRACT Three Drosophila melanogaster satellite inserts of the same size ranges as the starting material, and (1.672, 1.686, and 1.705 g/ml in CsCl), each containing therefore showing no evidence for deletion, were sequenced a simple sequence repeated in tandem, were cloned in pBR322 (see Results). as small fragments about 500 base pairs long. This precaution We investigated the instability of clones derived from the minimized deletions, since inserts of the same size as the three simple satellite DNAs by varying the size of fragments fragments used for cloning were recovered in a stable form. A used for cloning. When small DNA segments (==500 bp) were homogeneous tandem array of one sequence type usually cloned in pBR322, fragments of the same size could be extended the length of the insert. Eleven distinct repeat recovered from the plasmids although the efficiency of sequences were discovered, but only one sequence was pre- cloning was low. Clones were sequenced to verify the fidelity dominant in each satellite preparation. The remaining classes of cloning and to identify the repeating sequence. Although were minor in amount. The repeat unit lengths were restricted each of the 1.672, 1.686, and 1.705 satellites is comprised to 5, 7, or 10 base pairs, with sequences closely related. Each primarily of one simple sequence arranged in extremely sequence conforms to the expression (RRN),(RN)., where R is homogeneous, tandem arrays, an unexpected finding was A or G. The multiplicity of simple repeated sequences revealed that numerous clones contained short repeating sequences despite the small sample size suggests that numerous repeat different from the predominant type of repeat. These repeats sequences reside in heterochromatin and that particular rules differ by nucleotide changes only at certain positions of the apply to the structure of the repeating sequence. repeat sequence and not at all in others. The results indicate that the complexity of types in hetero- Approximately 20% of the genome of Drosophila melano- chromatin is much greater than suggested by simply four gaster consists of highly repeated sequences called satellite major satellite DNAs and that certain rules govern the DNA (1). This DNA falls into four distinct buoyant-density sequence permissible in a tandem array of simple repeats. classes (1.672, 1.686, 1.688, 1.705 g/ml in CsCl) and each can be isolated with high purity as satellite peaks (2, 3). The MATERIALS AND METHODS sequences are arranged in long tandem arrays and are restricted to heterochromatin, chromosomal regions defi- Cloning of Satellite DNAs. Total DNA was isolated from cient in genes (4). The sequences of three satellite DNAs nuclei ofD. melanogaster Oregon R embryos of average age (1.672, 1.686, 1.705 satellites) have been inferred by analysis 8 hr (12). Satellite DNAs were isolated in antibiotic/CsCl of pyrimidine tracts or RNA synthesized from a satellite gradients and banded as single symmetrical peaks in the DNA template (1, 3, 5, 6). This showed that the major analytical ultracentrifuge (4). A partially purified preparation repeating unit of each satellite was extremely simple, only 5 of the 1.690-g/ml satellite (1) was obtained by centrifuging or 10 base pairs (bp) long. DNA banding on the light side of an actinomycin D/CsCl Confirmation of the sequences and close examination of gradient in netropsin sulfate/CsCl. Since most of the repeat- the organization ofrepeats by current cloning and sequencing ed DNA in the 1.690 preparation is homologous to 359-bp methods have been difficult for the simple-sequence satel- repeats of the 1.688 satellite (see Results), the 1.690 satellite lites, in contrast to the 1.688 satellite (7). Most ofthis satellite appears to result from gradient tailing of the 1.688 satellite. DNA is comprised of more complex 359-bp repeating units Purified preparations of either the 1.672 or the 1.686 also arranged in tandem for many kilobases (kb) (8). Whereas satellite (5 /ug) in 0.3 M NaOAc were sonicated at low plasmid clones of the 1.688 satellite can contain inserts of up temperature (MSE Sonicator, three times for 10 sec at to 8 kb, or 33 tandem repeats, without showing evidence of maximum output) and 300- to 600-bp fragments were isolated instability (7, 9), clones of the 1.672, 1.686, and 1.705 by electrophoresis in a 1.4% agarose gel. After treatment with satellites are highly unstable (10). Restriction enzyme digests nuclease S1 (Boehringer Mannheim, 5000 units) at 370C for 30 of such clones showed initially a smear of sub-molar-insert min and the large fragment of DNA polymerase I (2.5 units) bands which could be resolved to a single fragment of about at 370C for 5 min, fragments were ligated into the Cla I site 0.5 kb after repeated rounds of single-colony isolation (10, of pBR322, which had been made blunt with large-fragment 11). Since the starting size of the purified satellite DNA in DNA polymerase I and four deoxynucleoside triphosphates these experiments was about 10 kb, more than 95% of the (25 1LM each) and dephosphorylated with bacterial alkaline satellite DNA insert had been deleted during propagation. phosphatase (Sigma, 0.2 unit, 650C for 30 min). After trans- In addition, many ofthese clones contained nonrepresenta- formation of Escherichia coli DH1, ampicillin-resistant col- tive DNA sequences and other alterations apparently intro- onies were recovered. The blunt-ending of DNA subsequent duced subsequent to cloning (unpublished observation). to size-selection resulted in some fragments being less than There were frequent interruptions in the periodicity of the 300 bp, depending on the lengths of single-stranded DNA at repeat by additions or deletions and numerous base alter- ends following sonication, and ensured that a range of ations. These aberrations were not seen when clones carrying Abbreviations: bp, base pair(s); kb, kilobase(s). The publication costs of this article were defrayed in part by page charge *Present address: Division of Entomology, CSIRO, P.O. Box 1700, payment. This article must therefore be hereby marked "advertisement" Canberra City, A.C.T. 2601, Australia. in accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom correspondence should be addressed. 696 Downloaded by guest on October 3, 2021 Genetics: Lohe and Brutlag Proc. Natl. Acad. Sci. USA 83 (1986) 697

satellite lengths were cloned. Clones of the 1.705 satellite screened by colony hybridization using purified satellite were made similarly except fragments were blunt-ended DNA as a probe. before size-selection so that fragments were greater than 300 Regardless of which of the three satellites was used as bp, and the DNA was inserted between the HindIII and starting material, only about 5% ofthe colonies were positive BamHI sites ofpBR322 by use ofHindIII and BamHI linkers. by colony hybridization. These usually retained the insert, DNA was sequenced (13, 14) using the external EcoRI site without deletions, with the size range being the same as the on the left ofthe insert or the HindIII (1.672 and 1.686 clones) starting material (300-600 bp), and no deletion was evident or BamHI (1.705 clones) site on the right of the insert. Both during large-scale growth of these plasmid DNAs. However, strands were sequenced, using 80-cm gels which allowed up most of the remaining 95% of ampicillin-resistant colonies to 600 bp to be determined from a single labeled end. had deletions, often spanning the insert and vector DNA. Satellite DNA Buoyant Density and Quantitation. The buoy- Different plasmids had deletions of varying degrees. Since ant density of satellite DNAs was determined by filter Hae III cuts the vector many times but rarely the satellite hybridization of fractions spanning a CsCl gradient of total DNA insert, the end points ofthe deletions could be mapped. DNA (300 pg). The gradient was collected in 192 fractions Deletions started from the satellite insert and continued into from the top of a VTi5O tube (Beckman). DNA from each adjacent vector DNA in a unidirectional manner, through the fraction (5 1.l) was heat-denatured, spotted onto nitrocellu- tetracycline-resistance gene and sometimes extending to the lose filters (6-mm diameter), and baked for 1.5 hr at 80'C. The origin of replication. This may result from selection against filter-bound DNA samples were hybridized in 450 mM deletions extending leftward into the ampicillin-resistance NaCl/45 mM sodium citrate/50% (vol/vol) formamide in gene used to select colonies. The deletion event appeared to separate vials at room temperature for 16 hr. The probe was occur early in plasmid replication, possibly during the first satellite insert DNA, excised with EcoRI and HindIII or replication of the transformed DNA, because multiple bands BamHI and labeled at the termini to 5000 cpm/ng with characteristic of continued deletion were not observed. Rapid [a-32P]dTTP (3000 Ci/mmol; 1 Ci = 37 GBq) and unlabeled deletion was due to the repeated DNA originating from D. deoxynucleoside triphosphates, using large-fragment DNA melanogaster rather than E. coli and was independent of the polymerase I. After hybridization, filters were washed, particular sequence type or highly reiterated nature of the heated at 10'C below the Tm (Tm is mean melting temperature; simple satellite DNAs. Excision ofthe 357-bp fragment in clone Tms are sequence-specific, unpublished observations) and 1.705-42 (Table 1) and recloning in pBR322 gave 95% positive assayed for radioactivity in a scintillation counter. The colonies by colony hybridization, in contrast to 5% positive sequence comprising the bulk of each of the 1.672, 1.686, and observed when D. melanogaster 1.705 satellite was used for 1.705 satellites (Table 2) was used as a marker, and the cloning. D. melanogaster DNA containing complex tandem positions of these sequences in the gradient plotted against repeats or unique sequences did not behave in this way (see buoyant density fell on a straight line. below) and could be cloned with much higher efficiency. Satellite DNA sequences were quantified using the dot-blot Although the satellite segments are relatively stable once procedure (15), except that a scintillation counter was used cloned, detailed examination did reveal minor deletions as instead of autoradiography. The probe and loading and faint bands flanking the major insert in overloaded gels. In hybridization of filters were the same as for gradient hybrid- addition, end-labeled insert DNA analyzed on a sequencing ization except that excess pBR322 DNA, linearized with Hae gel contained a "ladder" of shadow bands surrounding the III, was added to the hybridization mixture as a competitor major band and these were always at intervals of the repeat to eliminate hybridization by trace amounts of labeled unit, whether 5, 7, or 10 bp. The heterogeneity was observed pBR322 DNA. The reference series used satellite insert DNA even with a 109-bp fragment (22 tandem repeats) and a 125-bp (0.1-0.8 ng per filter) loaded as Hae III-linearized, total fragment (12.5 repeats). Clones with the repeat sequence 5' plasmid DNA. D. melanogaster total DNA (10-50 ng, except AACAA 3' were particularly susceptible to such deletions for 5' AACAA 3' quantitation, for which 150-900 ng was and DNA preparations of these clones contained from 50 to used) was loaded in a parallel series and each filter was 100% plasmid dimers. hybridized separately, with the same amount and concentra- Occasional major deletions were also observed. Of 31 tion of x probe (about 4 105 cpm/ml). Since the probe is in colonies containing plasmid 1.705-42 and analyzed by the excess, hybridization increases linearly with increasing alkaline-extract method (17), 30 showed the major 357-bp amounts of filter-bound DNA. Linear regression of cpm hybridized against pg of cloned satellite DNA loaded gave a calibration slope taken to represent 100% repeat sequence, Table 1. Satellite DNA sequences in D. melanogaster and genomic amounts were calculated by dividing the cor- Sequence of responding D. melanogaster slope by the calibration slope. repeat unit, Alterations in DNA was quantified by A260 and diphenylamine assays (16), Clone 5'1-* 3' Length, bp repeat units and the amount of satellite insert was calculated from the 1.672-38 AATAT 335 1 proportion of insert to plasmid DNA as shown by DNA 1.672-349 AATAG 235 1 sequencing. The results of this method are in good agreement 1.672-614 AATAC* 205 2 with other estimates of satellite DNA content (1) and were 1.686-198 AAGAC 427 1 more rapid and sensitive. 1.705-42 AAGAG 357 0 1.705-1232 GAGAG* 231 0 RESULTS 1.672-233 AACAA 297 0 Cloning of 1.672, 1.686, and 1.705 Satellite DNAs. It has 1.672-453 AATAAAC* 479 0 been shown previously that cloned 10-kb fragments of the 1.672-563 AATAGAC 568 0 1.672, 1.686, and 1.705 satellites are highly unstable but that 1.705-34 AAGAGAG 309 0 deleted derivatives of less than 1 kb can be recovered as 1.686-171 AATAACATAG 125 0 stable plasmids after repeated single-colony isolations (10). 1.690-1 353 bp 470 1 Therefore, the effect of cloning short DNA fragments on aDm23-24 359 bp 359 stability was examined. Highly purified preparations of these pDm688-169 254 bp 8.4 x 103 satellites were sonicated and 300- to 600-bp fragments were *A tandem array ofthis sequence as unit repeat was linked to another cloned in pBR322. Ampicillin-resistant colonies were tandem array in the clone (unpublished observations). Downloaded by guest on October 3, 2021 698 Genetics: Lohe and Brutlag Proc. Natl. Acad. Sci. USA 83 (1986) insert fragment and one a major deletion. In four of these Table 2. Buoyant density and genomic amount of satellite DNA colonies, a deleted fragment of 60-120 bp appeared in addition to the 357-bp fragment. A similar low level of cscl instability has been observed with 12 tandem copies of the buoyant 40-bp E. coli lactose operator (18). Sequence density, 1.672 1.686 Sequences of the Satellite DNAs. The insert DNA of clones 5AA3T g/ml % genome* satellite satellite showing no deletion into the vector was sequenced. Most AATAT 1.672 3.1 ± 0.4 85 clones contained one sequence repeated in tandem for the AATAG 1.693 0.23 ± 0.02 5 entire length. A single repeat class did not comprise the AATAC 1.680 0.52 ± 0.04 13 6 majority of the clones from any one satellite as would be AAGAC 1.689, 1.701 2.4 ± 0.2 18 expected from other data (2, 3, 5). Further, the same repeat AAGAG 1.705 5.6 ± 0.4 sequence could sometimes be found in clones from prepara- AACAA 1.663 0.06 ± 0.01 0.5 tions of different satellite DNAs. Such cross-contamination AATAAAC 1.669 0.23 ± 0.05 5 despite often distinct buoyant-density differences and exten- AATAGAC 1.688 0.24 ± 0.02 6 sive purification procedures may result from physical linkage AAGAGAG ND 1.5 ± 0.2 to the major repeating sequence class (12, 19). The 16 clones AATAACATAG 1.686 2.1 ± 0.2 73 of 1.672 satellite that were examined fell into six sequence 359 bp 1.688 5.1 ± 0.8 classes. Five sequence classes were found in 1.686 satellite ND, not done. clones, two being different from any found in the 1.672 *Mean ± SEM (n - 4-10). clones; and three sequences were found in 1.705 clones (Table 1). In total, 11 repeated sequences were found in the three satellite DNAs and, of these, 7 were 5-bp repeats. Only Together the sequences amount to 21% of the genome, in repeat units of 5, 7, or 10 bp were found and each of the good agreement with the value of 20-22% estimated for the repeats showed a similar arrangement of their nucleotides. highly repeated DNA content in D. melanogaster (2). For example, a 5-bp sequence is related to another 5-bp A 353-bp Repeating Sequence. Detailed analysis of sequence by one base change. With the exception of clone antibiotic/CsCl gradients detected a fifth satellite DNA ofD. 1.705-1232, the expression (AAN)m(AN)n describes each melanogaster banding at 1.690 g/ml (4). A preparation of this sequence, in agreement with Endow et al. (3). The exception, satellite, purified to about 30% homogeneity, was cloned, and 5' GGAGA 3', suggests that a more general formula may be 34% of colonies hybridized to repeated sequences in the 1.690 (RRN)m(RN)n, where R represents a purine (A or G) and N satellite preparation. However, these same clones were also represents A, G, T, or C. homologous to the 359-bp repeat ofthe 1.688 satellite (8). One Homogeneity of Sequences. Clones appeared to fall into two clone, 1.690-1, hybridized to the 359-bp repeat with thermal classes with respect to homogeneity of repeating units. In 14 stability reduced by 9.50C compared with homologous hy- clones, each repeat unit was identical, and 6 other clones brids. It contained a variant 353-bp repeating sequence, 80% contained a single altered repeat unit. Five of these changes homologous to both the 359-bp repeat [aDm23-24, (8)] and were single-base-pair alterations with transitions or transver- 254-bp repeat [pDm688-169, (9)] of the 1.688 satellite (Fig. 1). sions occurring at any position in the repeat, and one change Most differences between the 353-bp and 359-bp repeats are was a single-base-pair insertion or 4-bp deletion. The amount scattered single-base changes, but 353-bp sequences contain of sequenced DNA in these clones was 7565 bp, so that a 6-bp deletion of a G+C-rich stretch compared with the sequence alterations amounted to about 1 bp per 1000 bp. We 359-bp sequence. do not know whether these changes are regular and reflect a Both Hae III and Hinf resolve the 1.688 satellite into two higher order arrangement or whether they are mutations classes, a major 359-bp monomer class, including dimer and scattered at random along the satellite. In contrast to the 20 trimer bands, and a minor class of resistant DNA (7, 8). The homogeneous clones, 5 clones showed a much higher number 254-bp variant repeat is a minor sequence class deriving from of alterations (unpublished data). Homogeneous clones could Hae III-resistant 1.688 satellite, but a second class of variant be found for each repeated sequence class whether major or DNA containing repeat units slightly shorter than 359 bp was minor in the genome. Five clones containing the repeat 5' also found (7). Since the sequenced 353-bp repeat contains no AACAA 3' were examined in detail because this satellite is Hae III or Hinf sites (Fig. 1), it should be located in Hae III- particularly low in abundance (0.06%, or 100 kb of the or Hinf-resistant 1.688 satellite DNA. This expectation was genome, see next section). The clones contained only this confirmed by Southern hybridization (20) to total DNA, using sequence, repeated serially. There were two sequence alter- the 353-bp and 359-bp repeats as probes. At low stringency, ations in a total of 1567 bp, a proportion similar to that found 25°C below the Tm value, both repeats cross-hybridize and an in the more abundant repeat classes. almost identical hybridization pattern was seen with six Quantitation of Sequences. One explanation for the large restriction enzymes. However, heating the filter to 50°C number of sequence classes found in only three satellite (10°C above the Tm of the cross-hybrids) showed few if any DNAs is that the cloning system selects against the major 359-bp sequences to be located in Hae III-resistant DNA repeat type. To determine whether the recovered satellite migrating at the limiting mobility. Almost all 353-bp repeats clones are representative ofthe satellite DNA, the proportion were found in this region (Fig. 2). A similar case was seen of a sequence in the satellite preparation and genomic with Hinf, except that most 353-bp sequences formed an amounts were determined for 11 repeated sequences, using a irregular array extending from 0.7-kb (dimers) to 7 kb. modification of the dot-blot hybridization procedure (Table Differences between the repeat types were also seen for the 2). The 1.672 satellite consists mainly of the repeated se- remaining enzymes tested. Thus, susceptibility to Hae III or quence 5' AATAT 3'; the 1.686 satellite, 5' AATAACATAG Hinf distinguishes between 353-bp and 359-bp domains, 3'; and the 1.705 satellite, 5' AAGAG 3', in accord with which comprise the 1.688 satellite, and shows that these previous findings (2, 3, 5, 6). Thus, each repeat class repeat types are not interspersed. regardless of sequence can be propagated stably and without selection in E. coli, although the abundant repeat types DISCUSSION appear to be underrepresented in the clone bank. These data also show that the majority of highly repeated sequences in We have shown that the simple repeated sequences of D. melanogaster have been isolated in these plasmid clones. Drosophila satellites can be cloned successfully, provided Downloaded by guest on October 3, 2021 Genetics: Lohe and Brutlag Proc. Natl. Acad. Sci. USA 83 (1986) 699

10 20 30 40 50 60 70 80 90 353bp GTAAATATCA ACTTTTTGGC AAAATCCGTT TTTCCAAATT TCGGTCATCA AATAATCAGT GTTTTCTGCT ACAACTTTAA AAACAATTGT 359bp ------C ------T-- -G-G--T------*-T-A--T---C-----A-A-- ---T------254bp ------A------T-*A --C--T ------T------T---C ------TT------

100 110 120 130 140 150 160 170 180 353bp CTGAATATGG AAACTCATAC GTCGCTGAGC TCGTAATTAA ATTTCCAATC AAACTGTGTT CAAAAATGGA AATTAAATTT CTTTGACATA 359bp ------TG-----T C--A------A ------T--G-C--C- 254bp ------TG------C----NN--- -N------G------C----A------C---- *--C-G--N-

190 200 210 220 230 240 250 260 270 353bp GTGTGCAAAT TTTGATGA** *A--'*TrGTTA CAAAATATGT GAAAATTTGC CAAAAAATTG ATTTCTCTAA ATCCTTGAAA AAGTAATAGG 359bp T-T------CC CCCC-CC------A---C ------GAT -C------A -----C------C------FIG. 1. Sequence compari- 254bp *-T--- A-----A----AC CCCC**C--* -----A---C ------AAC GC------C------C--- -- son of the 353-, 359-, and 254-bp repeats ofthe 1.688 satellite. On- ly differences in the sequences are indicated, and deletions are 280 290 300 310 320 330 340 350 360 353bp GATCGTCAGC ACTGGTAATT AACTGCTCAA AACAGTTTTT CATGCATCTA TATGACCCTT TTTAGCCAAG TTATGACAAA AATTTCGTTT shown by asterisks. The 359-bp 359bp ------T------G ------A-A-- -G-A ------G -----A------A--G ------sequence is from ref. 8, and the * 254bp ----*,* * * * * ** e -- -^- * n11 ------? ^, 2** 254-bp sequence, from ref. 9. short DNA fragments of about 0.5 kb are used as starting from much longer satellite fragments (unpublished results). material. This precaution reduced markedly the instability In our clones, a single repeat sequence usually extended the evident when fragments of 10 kb were cloned in plasmid length of the insert, in agreement with other results showing vectors (10, 11) and enabled the sequencing of the entire that a given satellite sequence occupies long regions of the satellite insert. The sequence data showed that satellites had chromosomal DNA (1, 12). Further, the extreme sequence been cloned with fidelity, in contrast to those clones derived homogeneity ofless than one base-pair alteration per kilobase shown by these clones agrees well with estimates from 359 bp 353 bp renaturation and melting studies using gradient-purified sat- ellite DNAs (3, 12) or from the average length of 750 bp for 1 2 3 4 5 6 1 2 3 4 5 6 pyrimidine tracts in the 1.705 satellite (21). Two unexpected findings emerge from our results. First, although each ofthe major satellites (1.672, 1.686, 1.688, and 1.705) is comprised primarily of one type of repeating

23.1- sequence, there exist at least 14 families of highly repeated 9.4 - satellite DNAs in D. melanogaster (Table 1). Extensive 6.7 - 4.4 - characterization of clones derived from the 1.672 satellite revealed six classes of tandemly repeated sequences present u in this satellite alone. One class was predominant (5' AATAT 2.3 - tI 3' repeats) with the remainder being minor in amount. The ~~~~~~~~~~~~~4 . 2.0 - resolution afforded by molecular cloning has altered the view L~~~~~~~~~~~~~~~~~~~~wWW of simplicity, suggested by just four satellite DNAs in D. 1.44 - melanogaster, into one of complexity. Apart from abun- dance, minor satellite DNAs were identical to the major 1.08- satellite sequences in banding as satellites in CsCl gradients .. (Table 2) and showing extreme homogeneity of their repeat units. 0.718- '4 - The second feature of our results is the striking restriction on nucleotide changes apparent when the sequences of the different repeats are compared. Each of the 11 simple sequences has a common pattern and can be written in the __ N general form (RRN)m(RN)n. Therefore, the relationships 0.359 - between these satellites are not consistent with divergence _~~~4 and amplification in a neutral manner, since the repeating sequence is highly nonrandom. If sequences violating the expression arose from an ancestral satellite sequence, they presumably must be selected against and would not be amplified into a major satellite DNA. A similar nonrandom FIG. 2. The 359-bp and 353-bp variant repeats of the 1.688 choice of repeat sequence has been reported for three satellite are located in nonoverlapping arrays. Total D. melanogaster satellite DNAs in D. virilis (22). Each is a 7-bp repeat and DNA (5 ,ug), digested to completion using excess restriction enzyme, conforms to the formula (AAN)m(NA)n, itself related to the was divided into two aliquots and electrophoresed in a 1.4% agarose D. melanogaster sequence type. However, a fourth satellite gel. After transfer to nitrocellulose (20), the filter was cut in two and sequence, 5' AATATAG 3', fits the D. melanogaster pattern the filter-bound DNA was hybridized with either the 359-bp repeat (23). One exception to the sequence rule could be the repeat [aDm23-24 (8)] or the 353-bp repeat (1.690-1) at 25°C in 450 mM 5' GATA 3', reported to be present as a minor repeat class in NaCl/45 mM sodium citrate/50%o formamide. The filter then was D. melanogaster (24). In contrast to the simple-sequence heated at 50°C in the hybridization solution and exposed to x-ray film for 2 days. Restriction enzymes used were Hae III (lanes 1), Hinf satellites, clones of this repeat did not show sequence (lanes 2), Sau3A (lanes 3), Alu I (lanes 4), Taq I (lanes 5), and Mnl homogeneity and the 4-bp repeat is also transcribed. I (lanes 6). Markers at left (in kb) represent X DNA digested with The restriction on nucleotide sequence repeat suggests that HindIlI; the positions of 1-4 repeat units ofthe 1.688 satellite are also at least for the simple-sequence satellite DNAs, the sequence included (0.359, 0.718, 1.08, and 1.44 kb). of the repeating unit may be important for some cellular Downloaded by guest on October 3, 2021 700 Genetics: Lohe and Brutlag Proc. Natl. Acad. Sci. USA 83 (1986) functions. For example, these could involve particular con- recombination, are undetectable in satellite DNA prepara- formations of the DNA helix or satellite-specific DNA- tions and no unusual structures are seen in electron micro- binding proteins. Unlike the simple satellites, however, scopic examination of satellite DNA (unpublished results). constraints on sequence changes do not seem to apply Sequences arranged as palindromes are deleted rapidly when between the complex repeats of the three repeat families in cloned (27), but there is no evidence for such repeat organi- the 1.688 satellite. Multiple changes are scattered throughout zation in the Drosophila satellite DNAs (2, 21). the repeating units such that each of the 254-bp, 353-bp, and Although the cloned satellite DNAs account for the ma- 359-bp repeats are about 80% homologous. Like the simple jority of the D. melanogaster highly repeated DNAs in a repeats, adjacent 353-bp repeats can share a high degree of quantitative sense, they may not be fully representative ofthe homology. Clone 1.690-1 contained 470 bp, or a complete types of highly repeated sequences in this species. The six repeat, plus 117 bp ofthe next repeat, and this adjacent DNA major repeated sequences amount to 20% ofthe genome, and contained a single base different from the first repeat. That the collection of cloned satellite DNAs may be biased by most changes are random when 353-bp and 359-bp repeats are these abundant classes. Five minor satellite sequences compared, yet adjacent 353-bp repeats and adjacent 359-bp amount to only 1.3% and would not have been detected repeats appear to be highly conserved, could be explained by optically as satellite peaks. Nevertheless, they band as adjacent copies in a tandem array being evolutionarily young, distinct satellite DNAs and must be located in tandem arrays possibly deriving from the amplification ofa single copy. This separate from the major satellite repeating units. Since we is supported by Southern hybridization, which showed that characterized only a small number of clones but found 11 353-bp and 359-bp repeats are not interspersed but arranged simple sequence classes, more extensive screening might in separate arrays for many kilobases of DNA. have revealed other, minor repeat classes. A puzzling aspect of the 353-bp hybridization data is the irregular series of bands seen with Hinf digestion, unlike the We thank Liz Dennis for comments on the manuscript. The work 359-bp repeats which are cut into a monomer, dimer, and was supported by Grant GM21498 from the National Institutes of trimer series by this enzyme (Fig. 2). However, 353-bp Health. The DNA sequence analyses in this work were performed repeats are still arranged in tandem arrays, because Sau3A or using National Institutes of Health BIONET computer resource. Alu I digestion results predominantly in monomer fragments. 1. Peacock, W. J., Lohe, A. R., Gerlach, W. L., Dunsmuir, P., The sequenced 353-bp repeat is representative of 353-bp Dennis, E. S. & Appels, R. (1977) Cold Spring Harbor Symp. repeats generally because it contains one Sau3A and one Alu Quant. Biol. 42, 1121-1135. I site but no Hinf sites. It also contains nine sites where a 2. Peacock, W. J., Brutlag, D., Goldring, E., Appels, R., Hinton, single nucleotide change would generate a Hinf site. Frag- C. W. & Lindsley, D. L. (1973) Cold Spring Harbor Symp. Quant. ments of variable length could be formed depending on the Biol. 38, 405-416. 3. Endow, S. A., Polan, M. L. & Gall, J. G. (1975) J. Mol. Biol. 96, distance between consecutive mutations generating such 665-692. Hinf sites. This degree of heterogeneity is masked in the 4. Peacock, W. J., Appels, R., Dunsmuir, P., Lohe, A. R. & Gerlach, 359-bp repeats because the repeating unit already contains a W. L. (1977) in International Cell Biology, eds. Brinkley, B. L. & Hinf site. Thus, sequence variability can generate two pat- Porter, K. R. (Rockefeller Univ. Press, New York, NY), pp. terns in complex satellite DNAs: a dimer, trimer, ..., n-mer 494-506. series, resulting from single mutations inactivating a restric- 5. Brutlag, D. L. & Peacock, W. J. (1975) in The Chromo- some, eds. Peacock, W. J. & Brock, R. D. (Australian National tion enzyme site normally present once in the repeating unit, Univ., Canberra, Australia), pp. 35-45. and an irregular series of bands, deriving from single muta- 6. Sederoff, R., Lowenstein, L. & Birnboim, H. C. (1975) Cell 5, tions generating a site usually absent in the repeating se- 183-194. quence. 7. Carlson, M. & Brutlag, D. L. (1977) Cell 11, 371-381. The instability observed for long satellite DNA fragments 8. Hsieh, T. & Brutlag, D. (1979) J. Mol. Biol. 135, 465-481. inserted into plasmids may have two components. 9. Carlson, M. & Brutlag, D. (1979) J. Mol. Biol. 135, 483-500. Instability 10. Brutlag, D., Fry, K., Nelson, T. & Hung, P. (1977) Cell 10, does depend on fragment length, since cloned fragments of 509-519. about 10 kb are deleted rapidly (10), whereas this work 11. Brutlag, D., Carlson, M., Fry, K. & Hsieh, T. S. (1977) Cold showed inserts of about 0.5 kb to be relatively stable once the Spring Harbor Symp. Quant. Biol. 42, 1137-1146. plasmids were established in E. coli. We did not clone 12. Brutlag, D. L., Appels, R., Dennis, E. S. & Peacock, W. J. (1977) fragments longer than 0.6 kb to determine the upper limit in J. Mol. Biol. 112, 31-47. 13. Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol. 65, fragment size conferring stability. Even when short frag- 499-560. ments were cloned, most inserts appeared to be deleted 14. Smith, D. R. & Calvo, J. M. (1980) Nucleic Acids Res. 8, spontaneously, so that they were negative by colony hybrid- 2255-2274. ization. This phenomenon of rapid deletion was restricted to 15. Kafatos, F. C., Jones, W. C. & Efstratiadis, A. (1979) Nucleic the 5-to 10-bp repeat sequences in the 1.672, 1.686, and 1.705 Acids Res. 7, 1541-1552. satellites and did not occur with more 16. Burton, J. (1956) Biochem. J. 62, 315-323. complex repeating 17. Birnboim, H. C. & Doly, J. (1979) Nucleic Acids Res. 7, DNA. More than 30% of ampicillin-resistant colonies in a 1513-1523. partially purified 1.688 satellite preparation contained 359-bp 18. Sadler, J. R., Tecklenburg, M. & Betz, J. L. (1980) Gene 8, inserts, a proportion similar to the amount of 1.688 satellite 279-300. DNA in the preparation as seen by analytical ultracentrifuga- 19. Steffensen, D. M., Appels, R. & Peacock, W. J. (1981) tion. Also, rapid deletion appeared to be a property of Chromosoma 82, 525-541. satellite DNA deriving from Drosophila, because once pas- 20. Southern, E. M. (1975) J. Mol. Biol. 98, 503-517. saged through E. coli, it could be recloned at high efficiency. 21. Birnboim, H. C. & Sederoff, R. (1975) Cell S, 173-181. The source of rapid deletion is unclear, particularly since it 22. Gall, J. G. & Atherton, D. (1974) J. Mol. Biol. 85, 633-664. applies only to 23. Mullins, J. I. & Blumenfeld, M. (1979) Cell 17, 615-621. Drosophila DNA with simple sequences 24. Singh, L. S., Phillips, C. & Jones, K. W. (1984) Cell 36, 111-120. arranged serially and not to Drosophila DNA in general. Base 25. Urieli-Shoval, S., Gruenbaum, Y., Sedat, J. & Razin, A. (1982) modification by methylation is unlikely because N6- FEBS Lett. 146, 148-152. methyladenine, 5-methylcytosine, or other methylated bases 26. Achwal, C. W., Ganguly, P. & Sharat Chandra, H. (1984) EMBOJ. occur rarely, if at all, in embryonic DNA of Drosophila (25, 3, 263-266. 26). Single-stranded gaps, regions which might enhance 27. Collins, J., Volckaert, G. & Nevers, P. (1982) Gene 19, 139-146. Downloaded by guest on October 3, 2021