Proc. Natl. Acad. Sci. USA
`Vol. 77, No. 5, pp. 2786-2790, May 1980
`Evolution
`
`Selective amplification of variants of a complex repeating unit in
`DNA of a crustacean
`
`(evolution/amplified DNA/DNA nucleotide sequence determination/divergence measurements/crabs)
`
`NELWYN T. CHR1sTIE* AND DOROTHY M. SKINNERW
`
`‘University of Tennessee—Oak Ridge Graduate School of Biomedical Sciences; and ‘Biology Division, Oak Ridge National Laboratory,
`Oak Ridge, Tennessee 37830
`
`Communicated by Hewson Swift, January 21, 1980
`
`The nucleotide sequence of the repeating unit
`ABSTRACI‘
`of a fraction of the highly repetitive DNA of the red crab, Ger-
`yon uinquedens, is re
`rted. Treatment of total DNA with
`Hin III nuclease prodliibed an 81-base-pair monomer and
`multimers to the size of an octamer. Several of the multimers
`contained large amounts of fragments of variant sequences,
`which cannot easily be explained by random mutation alone.
`That the alterations were not random was corroborated by di-
`vergence measurements made on the distribution of Hba I nu-
`clease sites within several multimers. The analyses showed that
`a fraction of each of them is characterized by 4% divergence,
`while the amounts of dimer, tetramer, and octamer su est that
`they have undergone 2-4 times more divergence than that.
`These results, coupled with the data on se uence variants that
`are more prevalent in the dimer, indicate t at amplification of
`divergent repeating units could easily explain enhanced
`amounts of selected multimers.
`
`The structural diversity of satellites and other highly repeated
`DNAs suggests that several mechanisms may be responsible for
`their formation. The organization of both mouse satellite (1)
`and a (G+C)—rich satellite of the hermit crab (2) is compatible
`with unequal crossing-over of sister chromatids in highly re-
`peated DNAs as postulated (3). Repetitive DNAs may also arise
`by a mechanism of saltatory replication (4). Increases in genome
`size have been correlated with increases of particular fractions
`of repetitive DNAs in some organisms (5). That satellites may
`be formed by selective amplification is suggested by the pres-
`ence of specific satellite sequences in only one or a few species
`in a group of related species (6, 7). Within the Crustacea, ho-
`mology of repetitive sequences has been demonstrated for
`widely divergent species (8). Similarly, sequences homologous
`to the satellite of the mouse Mus musculus are found in related
`species, M. cervicolor and M. caroli (9). Retention of sequences
`over long evolutionary periods might be explained by frequent
`amplifications of highly repetitive DNA.
`The genome of the red crab, Geryon quinquedens, while
`lacking satellites, contains 40% highly repetitive DNA (10). In
`a subset of this repetitive DNA representing 5% of the genome,
`the distribution of HindIII restriction endonuclease sites cannot
`be explained by random mutation alone and is indicative of
`amplification of selected regions of the subset (11). We report
`here the nucleotide sequence of the repeating unit. Some se-
`quence variants occur too frequently to be explained by random
`mutation alone and may instead be derived from amplified
`regions of the genome. The complexity of the repeating unit
`[81 base pairs (bp)] is considerably greater than that of other
`crab satellite DNAs (2, 12, 13), with the exception of a
`(G+C)—rich satellite in the Bermuda land crab and a pair of
`cryptic satellites in the hermit crab (ref. 14 and unpublished
`
`The publication costs of this article were defrayed in part by page
`charge payment. This article must therefore be hereby marked "ad-
`vertisement" in accordance with 18 U. S. C. §l734 solely to indicate
`this fact.
`
`observations). The presence of complex repeats in crab DNAs
`as well as in the satellites of a number of vertebrates (15-18)
`suggests a universality in the mechanisms of origin for some
`highly repeated DNAs and possibly also in their functions. The
`range of sequence complexities observed in these DNAs, cou-
`pled with the localization of both simple and complex repetitive
`DNAs in the same centromeric regions (19), indicates that
`several mechanisms may be acting to produce a continuum of
`complexities. A plausible pathway for evolutionary change in
`the repetitive DNA of Geryon is amplification of a complex
`repeat unit, divergence from that sequence, and subsequent
`amplification of divergent subsets.
`
`MATERIALS AND METHODS
`
`DNA was isolated as described (10). Restriction endonuclease
`fragments were isolated and sized on polyacrylamide gels, using
`fragments produced by digestion of phage ¢X174 DNA with
`Hmfl as size markers. This procedure gave a more accurate size
`of the 81-bp fragment previously sized as 75 bp by comparison
`to dye markers (10). After treatment with bacterial alkaline
`phosphatase, fragments were labeled at the 5’ end (11).
`Nucleotide sequence analyses were performed as described
`(20, 21).
`Hha I digestion products of each Hmdlll multimer were
`labeled at the 5’ end, electrophoresed on 7% polyacrylamide
`gels, frozen, sliced, and assayed for radioactivity as described
`(2).
`
`RESULTS
`
`Sequences of the Basic Repeat Unit and Organization of
`the Multimers. A multimeric series of Hindlll fragments,
`which includes sizes from an 81-bp monomer to octamer length,
`can be obtained by digestion of total Ceryon DNA (11). In Hha
`I digests of each multimer, the two major products are 47 and
`40 bp. These are designated fragments a and 1). Determining
`the sequences of these two fragments from the monomer,
`dimer, and tetramer indicated that the same-sized Hha I
`fragment from each multimer had essentially the same se-
`quence. A repeat length of 81 bp was obtained because of the
`overlap of four nucleotides for the Hin dIII site and the overlap
`of two nucleotides for the Hha I site. Fragment a was read from
`the labeled adenine of the upper strand in Fig. 1; fragment b,
`from the labeled adenine of the lower strand. The sequences
`in the basic unit are more complex than repeated DNA in sev-
`eral other cmstaceans (2, 13, 22) in that they lack short repeating
`units internal to the 81-bp repeat. Despite the considerable
`complexity of the sequence, there are several runs of pyrimi-
`dines. This DNA reassociates in the correct sequence register;
`following reassociation of dissociated DNA, fragments of
`
`Mylan v. Genentech
`Mylan V. Genentech
`Abbreviation: bp, base pairs.
`3 To whom reprint requests should be ad ressed.
`IPR2016-00710
`IPR2016-00710
`Genentech Exhibit 2061
`
`Genentech Exhibit 2061
`
`2786
`
`

`
`Evolution: Christie and Skinner
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`2787
`
`Fragment a (Hha I. 47 bp)
`20
`10
`r—————————1
`[-—-T1
`40
`30
`-lA*CCTTATCACCACCTGTAACAACTTTTTTTCTATAAGTCCCCAAGCGl—
`
`+ATAGTCGTCCACATTGTTGAAAAAAACATATTCAGGGCTTC+-
`L:___;.l
`l.__.j.|
`Hha l
`Hindlll
`
`50
`
`60
`
`80
`IT!
`I"""—'_"jl
`[T1
`-*c T T c A c c c T c c c A c A c c c c T c c T T T T c c‘c c c T c Al-
`*
`— cc GA AG'TG cc Ac cc TC TC cc GA cc AA AA cc cc cA(:T TC GA -
`Igji
`i__t:s mt:
`Hha I
`Has Ill
`Hindlll
`
`70
`
`FIG. 1. Predominant nucleotide sequence of the monomer of the Hindlll multimeric series. Sites for restriction endonuclease cleavage
`are shown. For sequence analyses performed on Hha I digestion products, sequences for each fragment were read from the indicated labeled
`"nucleotide (A* ).
`
`Fragment b (Hha I, 40 bp)
`
`monomer and dimer size are produced by Hindlll digestion
`of both second-order repetitive components but not by digestion
`of the Cot 10'5 (moles of nucleotide per liter X sec) or foldback
`fraction (10). Minor fragments were produced by Hha I di-
`gestion of the Hindlll dimer of native DNA (Fig. 2). Among
`these the 128- and 121-bp fragments are observed in digests of
`all the multimers that are clearly related to the basic repeating
`unit of 81 bp (11).
`Sequence Variants Indicating Amplification of Divergent
`Sequences. Sequence analyses of individual multimers (Fig.
`3) indicated that fragments of each size class are heterogeneous
`Differences between the monomer and dimer indicated the
`presence of one or more sequence subsets either more abundant
`or unique to the dimer. A sequence variation was found at po-
`sition 26 of the dimer (Fig. 3B). A G residue occurred in a minor
`sequence coincident with the T residue found in the major se-
`quence. Whereas there was background radioactivity in the
`G+A channel for all the Ts between positions 24 and 30, a band
`in the C channel occurred only at position 26. Although this
`sequence variant has not been detected in the sequences of the
`monomer (Fig. 3C). it might be present in small amounts. The
`same sequence variant was detected in four different sets of
`sequencing reactions performed on three different preparations
`of dimers. That background fragments contribute significantly
`to the sequences seems improbable. The base-specific reactions
`were performed on Hindlll fragments that had been further
`selected by digestion with a second enzyme and the bases we
`attribute to minor variants are distinct and few in number.
`Finally, the pattern observed in the sequencing gels near the
`a
`b
`a
`b
`A o-—x—4-o--—n—4-e
`
`B
`
`o——-u—A:——-x-—Lo
`
`47
`o—x
`
`40
`x—-to — Class 0
`
`.
`
`81
`
`C 9m._Afl..x
`
`_
`
`\ Class 1
` _/
`l62
`°—‘j?j%3 — Class 2
`
`FIG. 2. Organization of sequences in the Hindlll dimer. (A) The
`original sequence pattern is shown with restriction endonuclease sites
`given for Hindlll (O), Hha I (X), and Hae III (A). (B) A representa-
`tive of the Hindlll dimer fragments that contains all the original sites
`except the central Hindlll site is presented. (C) Hha I digestion of
`5’ end-labeled dimer yields three classes of labeled fragments. Chasm
`0, 1, and 2 have zero, one, or two altered Hha I sites.
`
`3’ end of the Hha I, 47 bp, fragment (Fig. 3B) from the dimer
`indicates the presence of other variant sequences. The presence
`of more than one strong band in all four lanes near the 3’ end
`of the molecule may indicate length heterogeneity; the observed
`pattern is compatible with three or more fragments of different
`sizes. The relative amount of DNA for each presumptive band
`of uncleaved fragments indicates that the variant sequences are
`not trivial fractions. Because of the possibility of artifacts in
`sequencing methodologies (21), our assessment of the sequence
`variants in the dimer is based on repeated observations in nu-
`merous sequencing gels utilizing separately prepared reaction
`sets.
`
`Sequence Variants Consistent with Random Mutation.
`Several minor fragments produced by cleavage at sites not
`found in the basic repeat can be explained by random mutation.
`Completely random alterations at every position in the entire
`array of repeating units would appear only as background in
`the sequence ladders of autoradiograrns. Thus, although ran-
`dom mutational alterations cannot be detected by sequence
`analyses they are detectable by the appearance of "new” re-
`striction sites within the basic repeat unit. As expected, these
`arise most often by single nucleotide changes. As long as the
`proportion of fragments containing these "new" sites is only
`a few percent of the total sequences, their presence can be
`readily explained by random mutation.
`The site for Alu I is the four central nucleotides of that for
`Hindlll, and there are no additional Alu I sites located within
`the basic repeat. Nevertheless, treatment of the dimer with Alu
`I produced several fragments (Fig. 4, lane 1), which made up
`5% of the dimer DNA. The sequences of the Alu I, 45 bp,
`fragment appear identical to those of the Hha I, 47 bp, frag-
`ment. The change of a single base (C to T) at position 47 would
`produce the Alu I recognition site of (5’)A-G1C-T(3’) and yield
`a 45-bp fragment. Sequence analyses of the Alu I, 83 bp,
`fragment from the dimer indicated that this class is a mixture
`of at least two fragments. Cleavage at the Alu I site in the center
`of the dimer would produce two labeled fragments of 83 bp,
`one from each 5’ temiinus. Random mutation might easily have
`produced a fraction of dimer fragments that are missing the
`middle Hindlll site by alteration of either the first or last nu-
`cleotide in the site.
`The presence of variants of each multimer was further in-
`dicated by digestion with Hae III, whose site is at positions
`74-77 of the basic repeat. The dimer presumably contained two
`Hae III sites and should, when digested by Hae III, give frag-
`ments corresponding to the presence of either one or both sites.
`In addition to fragments of the expected sizes of 10, 75, and 91
`bp in the Hae III digests, there were also three fragments (21,
`
`

`
`2788
`
`Evolution: Christie and Skinner
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`A
`
`Hha 1,40 bp
`G
`c+T
`
`G+A
`
`B
`G+A
`
`Hha I, 47 bp
`G
`C+T
`
`C
`
`C
`
`Hae IH, 75 bp
`G G+A C+T c
`
`cqc
`
`
`
`2C
`
`(40)
`
`C
`/T
`/E‘
`;A
`G
`_/—Z
`T 2\
`#6
`—*—T(30)
`ix .
`:1; E:
`can
`DI
`(30)T
`T T
`:7 T—’’
`
`t
`
`. fl
`
`_
`T
`G’
`_
`;éC>C
`3*
`c>A
`:<_TzT(5O)
`_
`‘ 5
`egg
`4‘\
`G<A
`I M
`C>C
`C>c(4o)
`T
`
`*
`
`sar-
`
`
`
`........
`
`X
`(5o)éX
`{X
`3% "
`\ I
`2% .
`8: ‘
`G
`3
`(60)T
`
`T C
`
`G_,_/— ‘ w
`G_,_
`at
`(3
`-
`.
`A
`
`on
`
`G
`(Tom
`A
`A
`A
`
`C
`C
`6 Q Q
`
`_Z—A Tf
`2-
`!-
`' _/—c(2o)
`C
`A
`A?
`(2o)c—/_
`A_g-
`
`’
`
`._...
`due
`gun
`
`“AP ”
`
`T‘/_
`T
`6-; ‘ “‘ “"'
`c/_
`c—Z‘
`A
`<3
`(1o)<:
`A
`0
`
`.
`
`,..
`""'
`:3’
`M ,_
`f, ~
`f "‘
`
`FIG. 3. Representative autoradiograrns of DNA sequencing gels. Maxam—Gilbert analyses for fragments Hha I, 40 bp (A), Hha I, 47 bp,
`from the HindIII dimer (B), and Hae III, 75 bp, from the monomer (C) were applied to 12% polyacrylamide/urea gels.
`
`23, and 28 bp; Fig. 4, lane 2) that represented only a few percent
`of the digest. These would occur if there had been a change in
`only a single base at three positions of fragment 1) of the basic
`repeating unit. Changes to C at positions 64 and 56 would give
`21- and 28-bp fragments, respectively. A similar change at
`position 62 would give a 22—bp fragment and could account for
`the minor fragment of 23 bp. The proposed changes in the basic
`repeat unit in the three positions mentioned would give rise to
`the fragments observed, and these changes appear to be the
`most likely interpretation of the data at this time.
`Estimate of Divergence in the Multimers. The extent of
`the divergence in clames of repetitive DNAs can be determined
`by examination of the organization of restriction sites (1, 23).
`This approach indicates the fraction of altered sites in an array
`of sites originally spaced uniformly. From the fraction of sites
`that has been altered, the divergence in the entire sequence can
`
`l
`
`2
`
`-152
`
`-162
`
`-115
`
`1- as
`
`_ 9‘
`-75
`
`1 — 45
`
`913,
`fl
`
`FIG. 4. Autoradiogram of digestion
`products of the HindIII dimer. The dimer
`~ 4 — 28
`N. _ 23 was labeled at the 5’ end, digested with Alu
`._, _ 21
`I (lane 1) or Hae III (lane 2), and electro-
`phoresed on a 7% polyacrylamide gel. In
`lane 2, the 10-bp fragment is not shown.
`Sizes of fragments are given in bp.
`
`be calculated. Assessment of a set of restriction sites in a re-
`
`petitive DNA is equivalent to selecting a small population of
`nucleotides from which to predict the amount of change that
`has occurred in the entire population. As divergence within the
`sequences increases, the number of sites altered at consecutive
`positions will increase. As a result, a multimeric series of frag-
`ments will be produced from the original population of
`monomers. Regardless of the overall extent of divergence, the
`relative amount of each successively larger multimer should
`be less than the preceding multimer. Other multimeric series
`digested from either satellites (1, 24, 25) or total DNA (26, 27)
`were compatible with formation by random mutation alone.
`For the HindIII multimeric series of Ceryon total DNA, the
`plot of the logarithm of the relative amount of each multimer
`versus (n — 1), in which n = 1 for a monomer, deviates signif-
`icantly from the straight line typically observed ( 1, 11). In
`particular, there are enhanced amounts of dimer, tetramer, and
`octamer. These results, coupled with specific indications of
`amplified divergent sequences in the dimer and in the octamer
`(see below), suggest that random mutation alone cannot explain
`the observed distribution of restriction sites.
`After the observation of a nonrandom spacing of HindIII sites
`in the Geryon genome, the distribution of Hha I sites within
`specific HindIII fragments was investigated. Hm dIII fragments
`were digested with Hha I. Fragments were divided into classes§
`according to the number of altered Hha I sites. For each mul-
`timer, class 0 fragments are those with no altered Hha I sites
`(always the 47- and 40-bp fragments), while classes 1-4 have
`between one and four consecutive altered sites. In Hha I digests
`of the monomer, class 1 will be the enzyme-resistant fraction,
`whereas for the dimer it will be fragments of 121 and 128 bp
`(Fig. 2). The relative amounts of each class are plotted in Fig.
`5 for the first four multimers. On each graph there are two
`broken lines representing two different levels of divergence,
`
`§ The fragments in these classes are clearly related to the basic repeat
`as determined by the positions of restriction sites.
`
`

`
`Evolution: Christie and Skinner
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`2789
`
`
`
`FIG. 5. Distribution of Hha I sites in HindIII multimers. The
`relative amounts of each Hha I digestion product of the monomer (A),
`dimer (B), trimer (C), and tetramer (D) were quantified and plotted
`(solid line) according to the arrangement of Hha I sites (see Fig. 2).
`Classes of fragments are produced in which cleavage occurred at the
`first through the fourth site from each labeled 5’ terminus, depending
`on the number of consecutive altered sites. The plot of log a,., the
`fractional amount of each Hha I digestion product, versus s, the
`number of consecutive altered sites per fragment, should yield a
`straight line if the distribution of Hha I sites were due entirely to
`random mutation of a previously uniform array of sites (1, 11). The
`broken lines are plots of log c,, versus s for divergence values of 3.7%
`(a) and 15.7% (b), in which c,. is the number of copies of each multimer
`relative tothe monomer; i.e., c,, = 1 when n = 1 (11). Uncut refers to
`the enzyme-resistant fraction. For the tetramer, class 3 fragments
`were not detected.
`
`(a) 3.7% and (b) 15.7%. These two lines are included to allow
`a comparison to the distribution of HindIII sites in the total
`genome (11). For the HindIII multimeric series, the relative
`amounts of trimer, pentamer, hexamer, and heptamer best fit
`line a, whereas the dimer and octamer best fit line b. The tet-
`ramer value is intermediate. The dimer, tetramer, and octamer
`compose 0.25, 0.5, and 3.0% of the genome, respectively
`(1 1).
`Examination of the distribution of Hha I sites reveals two
`points. First, there is always a large amount of enzyme-resistant
`fragments (indicated as uncut in the figure). Although a frac-
`tion of this DNA may be unrelated or “background fragments,"
`a similar result was observed for the Drosophila melanogaster
`1.688 g/cm3 satellite (28). Random inactivation of restriction
`sites is insufficient to explain completely this organization of
`restriction sites. Such resistant fractions may well be the result
`of amplifications of DNA segments in which the original
`spacing of restriction sites was lost.
`The second and more easily interpretable result obtained
`from these analyses is the indication of the extent of divergence
`of Hha I sites within the HindIII sequences. A precise deter-
`mination of divergence of the Hha I sites could not be made
`because of the small number of sites within the first four mul-
`timers for which quantitation was performed. Multimers of five,
`six, and seven monomer lengths were present in too few copies
`to be used for quantitation. The octamer was not included in
`this analysis because it contained enhanced amounts of class 3
`fragments, which will be discussed later. Only the dimer, tn‘-
`mer, and tetramer will be considered for an estimate of diver-
`gence in Hha I sites, because an estimate from the monomer
`would depend on the enzyme-resistant fraction. The relative
`amounts of each class of fragments for the dimer, trimer, and
`tetramer are most compatible with"the lower divergence of line
`a. These results support the idea that the divergence in at least
`one subset of the HindIII fragments is about 4% and that the
`large amounts of dimer, tetramer, and octamer may be due to
`selective amplification.
`The nature of at least one amplification event is indicated
`
`
`
`FIG. 6. Distribution of Hha I sites in the HindIII octamer. The
`relative amount a,. of each class of Hha I digestion products from the
`octamer is plotted versus s, the number of consecutive altered sites,
`as in Fig. 5. Fragments missing six or seven sites were not detected.
`
`from a quantification of the Hha I products of the octamer (Fig.
`6). Over 50% of the digest is composed of class 3 fragments,
`which have presumably lost three consecutive Hha I sites.
`Amplification of a DNA segment containing a divergent oc-
`tamer would yield this result. Clearly there are also some less
`divergent octamers, because classes 0, 1, and 2 are also present.
`From the relative amounts of these latter classes we conclude
`that the divergence due to random mutation in a fraction of this
`DNA is only 4%. In addition, the presence of unusually large
`amounts of particular HmdIII multimers suggests amplification
`of a DNA segment containing divergent multimers. It is also
`possible that the large amounts of Hha I-resistant fractions in
`each multimer resulted from amplification of divergent se-
`quences.
`
`DISCUSSION
`
`The relative proportions of Hindlll multimers in total Ceryon
`DNA cannot be explained solely by the random accumulation
`of mutations in a homogeneous array of monomers (11). The
`amounts of the trimer, pentamer, hexamer, and possibly the
`heptamer could result from approximately 4% divergence,
`whereas the amounts of the dimer, tetramer, and octamer
`would require 2-4 times that divergence to have been produced
`by random mutation. The distribution of Hha I sites was ex-
`amined to give an independent assessment of divergence. Their
`distribution best fits a divergence of 4%, easily distinguishable
`from a divergence of 16%, and much too low to explain the
`large amounts of particular multimers observed. Mutation may
`produce specific changes in the sequence if access to the DNA
`is limited by chromatin structure (29). In addition, random base
`changes may produce "hot spots” for further mutation in only
`a few multimers. Single-nucleotide changes have produced
`altered mutation rates in the III locus of phage T4 (30). We
`conclude that while selective mutation may occur to some ex-
`tent, selective amplification seems necessary to explain the
`observed distributions of restriction sites.
`Nonrandom distribution of restriction sites has been observed
`for mouse satellite DNA, although only a small fraction, ap-
`proximately 5%, of the basic repeat units contain Hae III sites
`(1). For Geryon, sequences characterized by nonrandom dis-
`tribution of restriction sites represent at least 20% of the HindIII
`sequences, using the assumption that half of each of the current
`amounts of dimer, tetramer, and of one of the octamers was
`produced by amplification of divergent sequences. Although
`it is likely that the same general mechanisms for amplification
`occur in all or most eukaryotes, differences within particular
`
`

`
`2790
`
`Evolution: Christie and Skinner
`
`Proc. Natl. Acad. Sci. USA 77 (1980)
`
`taxonomic groups are likely. Highly repetitive DNA composes
`between 25 and 50% of the genomes of several crabs (10, 31,
`32). Other arthropods, specifically certain insects, also have
`large amounts of highly repetitive DNA (7, 33). Although such
`examples are not considered indicative of a requirement for
`large amounts of highly repetitive DNAs, they might well
`suggest that there are active mechanisms for their formation.
`A rodent, the kangaroo rat, Dipodomys ordii, contains over 60%
`highly repetitive DNA (34). By contrast the genomes of six
`primates, including humans, contain less than 10% highly re-
`petitive DNA (6).
`The homogeneity of repeating units in certain arthropod
`satellite DNAs should be contrasted with the heterogeneity in
`rodent satellite DNAs. For guinea pig at-satellite or mouse
`satellite the predominant sequence is representative of only 50%
`of each DNA (35, 36). For crab poly(dA—dT) (12, 37) and two
`satellites in the hermit crab (2, 13), the predominant sequence
`represents more than 90% of the total sequences. This feature
`may again imply that an unusually high rate of amplification
`occurs in some repetitive DNAs of crabs. Multiple rounds of
`amplification at frequent time intervals would have the effect
`of maintaining a closely related set of sequences. For example,
`crab poly(dA—dT) appears in widely divergent species (14, 37).
`Clearly, the inclusion of a divergent multimer in the segment
`for amplification would increase the magnitude of the diver-
`gent sequences. Additional support for this idea is derived from
`the low divergence observed in a fraction of three Hindlll
`multimers of Ceryon. Amplification of both divergent and
`nondivergent multimers would effectively maintain a portion
`of the original subset and simultaneously magnify a particular
`divergent subset. "Divergent" and "nondivergent" refer only
`to the set of restriction sites being tested; other modifications
`would be undetected. Because each of three multimers contains
`fractions showing 4% divergence, it is possible that a single
`amplification contained a dimer, trimer, and tetramer that had
`retained the original spacing of Hha Isites. This would indicate
`that the minimum amount of amplified DNA was the com-
`bined length of these three multimers (729 bp). The data do not
`require that these three multimers be contiguous for amplifi-
`cation. Similarly, from the presence of large amounts of a di-
`vergent octamer the amplification unit could be at least 648 bp.
`Whether this latter amplification occurred at the same time as
`that of the smaller multimers cannot be determined. These
`estimates of the length for an amplification unit are minimal,
`because a length of several thousand nucleotides would still be
`small enough so that the distribution of restriction sites in the
`amplified DNA might not reflect the distribution in the entire
`set of Hindlll sequences, approximately 3 X 103 bp.
`
`This research was supported by the Office of Health and Environ-
`mental Research, U.S. Department of Energy, under contract W-
`7405-eng-26 with the Union Carbide Corporation. N.T.C. was sup-
`ported as a predoctoral fellow by Grant GM 1974 from the National
`Institute of General Medical Sciences.
`
`1. Southern, E. M. (1975) ]. Mol. Biol. 94, 51-69.
`2. Chambers, C. A., Schell, M. P. & Skinner, D. M. (1978) Cell 13,
`97-110.
`
`Smith, G. (1973) Cold Spring Harbor Symp. Quant. Biol. 38,
`507-513.
`
`Britten, R. J. & Davidson, E. H. (1971) Q. Rev. Biol. 56, 111-
`138.
`
`Bozzoni, J. 81 Beccari, E. (1978) Biochim. Biophys. Acta 520,
`245-252.
`
`Iones, K. W. (1977) in Molecular Structure of Human Chro-
`mosomes, ed. Yunis, I. J. (Academic, New York), pp. 295-326.
`Barnes, S. R., Webb, D. A. & Dover, G. (1978) Chromosoma 67,
`341-363.
`Graham, D. E. & Skinner, D. M. (1973) Chromosoma 40,
`135-152.
`
`Rice, N. & Strauss, N. (1972) Carnegie Inst. Washington Yearb.
`71, 264-269.
`Christie, N. T. & Skinner, D. M. (1979) Nucleic Acids Res. 6,
`781-796.
`
`Christie, N. T. 81 Skinner, D. M. (1980) Nucleic Acids Res. 8,
`279-298.
`
`Skinner, D. M. (1967) Proc. Natl. Acad. Sci. USA 58, 103-
`110.
`Skinner, D. M., Beattie, W. G., Blattner, F. R., Stark, B. P. &
`Dahlberg, J. E. (1974) Biochemistry 13, 3930-3937.
`Beattie, W. G. & Skinner, D. M. (1972) Biochim. Biophys. Acta
`281, 169-178.
`Maio, J. 1., Brown, F. L. & Musich, P. 11. (1977) 1. Mol. Biol. 117,
`637-655.
`
`Rosenberg, H., Singer, M. & Rosenberg, M. (1978) Science 200,
`394-402.
`Manuelidis, L. 81 Wu, J. (1978) Nature (London) 276, 92-94.
`Kopecka, H., Macaya, G., Cortadas, ]., Thiéry, I.-P. & Bernardi,
`C. (1978) Eur.
`Biochem. 84, 189-195.
`Manuelidis, L. (1978) Chromosoma 66, 23-32.
`Maxam, A. & Gilbert, W. (1977) Proc. Natl. Acad. Sci. USA 74,
`560-564.
`
`Maxam, A. & Gilbert, W. (1980) Methods Enzymol. 65, in
`press.
`Skinner, D. M. (1977) Bioscierice 27, 790-796.
`Slack, ]. (1974) Biopolymers 13, 2241-2264.
`Altenberger, W., Horz, W. & Zachau, H. G. (1977) Eur. ]. Bio-
`chem. 73, 393-400.
`Fittler, F. (1977) Eur. J. Biochem. 74, 343-352.
`Manteuil, S., Hamer, D. 81 Thomas, C. A., Jr. (1975) Cell 5,
`413-422.
`
`(1975) Mol. Biol. 94, 87-99.
`Cooke, H.
`Carlson, M. & Brutlag, D. (1977) Cell 11, 371-381.
`Musich, 1’. 11., Maio, ]. I. & Brown, F. L. (1977) Mol. Biol. 117,
`657-677.
`Koch, R. E. (1971) Proc. Natl. Acad. Sci. USA 68, 773-776.
`Vaughn, J. C. (1975) Chromosoma 50, 243-257.
`Holland, C. A. & Skinner, D. M. (1977) Chromosoma 63, 223-
`240.
`
`(1977)
`
`Brutlag, D., Appels, R., Dennis, E. S. 61 Peacock, W.
`Mol. Biol. 112, 31-47.
`Hatch, F. T. & Mazrimas, J. A. (1970) Biochim. Biophys. Acta
`224, 291-294.
`I
`227, 794-798.
`Southern, E. M. (1970) Nature
`Biro, P. A., Carr-Brown, A., Southern, E. M. & Walker, P. M. B.
`(1975) Mol. Biol. 94, 71-86.
`Skinner, D. M. & Kerr, M. S. (1971) Biochemistry 10, 1864-
`1872.
`
`10.
`
`12.
`
`13.
`
`14.
`
`15.
`
`16.
`
`17.
`18.
`
`19.
`20.
`
`21.
`
`22.
`
`24.
`
`26.
`
`27.
`28.
`29.
`
`31.
`32.
`
`36.
`
`37.