volume 8 Number 201980
N u c l e i c A c i d s Research
Ordered distribution of modified bases in the DNA of a dinoQagellate
Robert E.Steele* and Peter M.M.Rae
Department of Biology, Yale University, New Haven, CT 06511, USA
Received 14 July 1980
ABSTRACT
In DNA of the dinoflagellate Crypthecodinium cohnii. 38^ of the thymine
is replaced by the modified base 5-hydroaymethyluracil, and ^ 3 ^ of the cytosine is replaced by 5-methylcytosine. Both of the modified bases are nonrandomly distributed in the DNA. Determinations of 3 1 nearest neighbors show
that HOMeU is preferentially located in the dinucleotides HOMeUpA and HOMeUpC.
Pyrimidine tract analysis shows that HOMeU is also greatly enriched in the
trinucleotide purine-HOMeU-purine. As in other eukaryotes, methylcytosine in
C. cohnii DNA occurs predominantly in the dinucleotide MeCpG. By analysis of
restriction endonuclease digestion patterns of C_. cohnii total DNA and ribosomal DNA, we have found that the central CpG dinucleotides in the sites for
the enzymes Hpa H (CCGG) and Hha I (GCGC) are extensively methylated in both
total DNA and ribosomal DNA. Results of digestion with Ava I, however, indicate that not all CpG dinucleotides in the sequence CJiCG^G are methylated in
C. cohnii DNA.
INTRODUCTION
DNAs of eukaryotes often contain modified bases.
ring of these is 5-methylcytosine.
The most widely occur-
Methylcytosine (MeC) usually constitutes
less than 5% of the total bases in DNA, although in higher plants it may comprise as much as 105? of the total bases (Shapiro, 1976).
Two other modified
bases are found in eukaryotic DNA, but they are of more limited occurrence
than methylcytosine:
DNA in several unicellular eukaryotes contains rnnnii
amounts of tr-methyladenine (see Rae and Steele, 1978, for references); the
third modified base in eukaryotic DNA, 5-hydroxymethyluracil (HOMeU), is the
most limited of the three, being found so far only in DNA of dinoflagellates
(Rae, 1973, 1976).
The amount of HOMeU in dinoflagellate DNA varies consi-
derably among different species; it constitutes 4-195? of total bases and replaces 12-70? of thymine in DNA.
DNAs of some dinoflagellate species contain
methylcytosine in addition to HOMeU, and in one species
IT-methyladenine is
found as well as HOMeU (Rae, 1976).
The only modified base in eukaryotic DNA for which the synthetic mechan-
© IRL Press Limited, 1 Falconberg Court, London W1V 5FG, U.K.
4709
Nucleic Acids Research
ism has been determined is methylcytosine, which arises by enzymatic transfer
of a methyl group from S-adenosyl methionine to cytosine residues in DNA
(Burdon and Adams, 1969; Sneider and Potter, 1969).
Cytosine in the dinuc-
leotide CpG is strongly favored as the site for modification (Sinsheimer,
1954; Grippo et al., 1968).
Neither the synthetic mechanism nor the sequence
distribution of lr-methyladenine in eukaryotic DNA is known, although in bacterial DNA this minor base is also generated by methyl transfer from S-adenosyl methionine to adenine residues in DNA (Meselson e_t al., 1972).
We have shown that ratios of thymine to hydroxymethyluracil vary in different buoyant density fractions of DNA from Crypthecodinium cohnil. indicating a non-randomness in the distribution of these two bases in DNA (Rae,
1973).
In this paper we provide additional and more direct evidence that
HOMeU is non-randomly distributed in C_. cohnii DNA. We have also found that
methylcytosine in C. cohnii DNA is predominantly in the dinucleotide MeCpG.
Using restriction endonucleases sensitive to cytosine methylation, Bird
and Taggart (1980) have examined the patterns of cytosine methylation in the
ribosomal DNAs (rDNAs) of a variety of animals, finding that rDNA is largely
unmethylated in most animals with the exception of fish and amphibia, in
which rDNA is heavily methylated.
We have found that rDNA in C_. cohnii is
extensively methylated.
Our results regarding methylcytosine in dinoflagellate DNA are of particular interest because of the unusual nature of dinoflagellate chromosomes.
These chromosomes lack histones, they remain condensed throughout the cell
cycle, and at the ultrastructural level they strongly resemble bacterial
nucleoids (reviewed by Dodge, I966).
Such peculiar chromosomal features in a
cell which is otherwise typical of a unicellular eukaryotic alga (Dodge, 1971/
have led to the proposition that dinoflagellates are evolutionarily intermediate between prokaryotes and eukaryotes (Dodge, 1965).
The results presentee
here indicate that with regard to levels and distribution of DNA methylation,
dinoflagellates are comparable to some higher eukaryotes. We have discussed
elsewhere (Rae and Steele, 1978) possible roles of hydroxymethyluracil in
dinoflagellate DNA, and the basis for our present belief that HOMeU is of
little or no contemporary functional significance in the DNA of these organisms.
MATERIALS AMD METHODS
Sources and culture of dinoflagellates.
Crypthecodinium cohnii and Pro-
rocentrum cassubicum were obtained from Dr. Luigi Provasoli (Dept. of Biology,
4710
Nucleic Acids Research
Tale University).
C. cohnii was maintained in the M m medium of Tuttle and
Loeblich (1975) and P. cassubicum in the DV medium of Provaaoli and McLaughlin
(1963).
DNA extraction.
DNA was prepared from both dinoflagellate 3pecies es-
sentially according to Rae (1973).
C. cohnii DNA labelled in vivo with
32
P
was isolated from cells cultured for two days in M m medium containing approximately 1 yCi/ml H, TO, (New England Nuclear).
RNA and polysaccharides were
removed from DNA preparations by treatment with a combination of T^ RNase,
pancreatic RNase and a-amylase, and by CsCl gradient centrifugation.
DNA was
purified from bacteriophage SP82 (kindly provided by Dr. Gary A. Wilson, University of Rochester) by phenol extraction.
Isolation and radioactive labelling of C. cohnii ribosomal RNA. Cells
were harvested from culture medium and homogenized in liquid nitrogen with a
mortar and pestle.
The powdered cells were resuspended in buffer containing
200 mM NaCl, 5 mM Mg acetate, 20 mM Tris-HCl, pH 7.6, 0.556 Triton X-100, and
0.1% diethylpyrocarbonate.
Nuclei, mitochondria and starch granules were re-
moved by spinning the homogenate at 9,000 rpm for 15 minutes at 4° in a Sorvall HB-4 rotor.
The resulting supernatant was layered onto cushions of 33*
(w/v) sucrose in 200 mM NaCl, 5 mM Mg acetate, 20 mM Tris-HCl, pH 7.6, and
spun for 3.5 hours at 45,000 rpm and 4° in a Beckman Ti60 rotor.
The resul-
ting polyribosome pellets were resuspended in buffer containing 100 mM NaCl,
2 mM Na^EDTA, 0.5* SDS, and 20 mM Tris-HCl, pH 7.6, then precipitated overnight at -20° with two volumes of 95* ethanol.
The precipitate was recovered
by centrifugation, and RNA was extracted from it with phenol/chloroform/isoarayl alcohol (50:48:2).
17S and 24S ribosomal RNAs were purified from total
polysomal RNA by sucrose gradient centrifugation.
Ribosomal. RNAs were end-
labelled using r T'-ATP and T, polynucleotide kinase according to the procedure of Maizels (1976).
Restriction enzyme digestion and electrophoresis of DNA. Map I and Hha I
were from New England Biolabs; Hpa II was from Bethesda Research Laboratories;
Ava I was a gift from Dr. Clifford Ten. Msp I and Hpa II digestions were done
in 6 mM NaCl, 6 mM Tris-HCl, pH 7.5, 6 mM MgCl 2 , 6 mM e-mercaptoethanol, 0.1*
bovine serum albumin.
Digests with Hha I and Ava I were in 60 mM NaCl, 6.6
mM Tris-HCl, pH 7.6, 6.6 mM MgCl 2 , 6.6 mM 6-mercaptoethanol.
done at 37° with excess amounts of enzyme.
Digestions were
ELectrophoresis of DNA in agarose
gels, and staining and photography of gels, were as described by Barnett and
Rae (1979).
DNA transfer and hybridization.
DNA was transferred from agarose gels
onto nitrocellulose filters according to the method of Southern (1975). Hy4711
Nucleic Acids Research
bridization of labelled ribosomal RNA to filter-bound DNA was carried out at
37° in Seal-N-Save bags (Sears); the hybridization mixture contained 50$ formamide, 750 mM NaCI, 75 mM Na citrate, 100 mM Tris-HCl, pH 7.4, and 1-2 x 10 6
cpm of labelled RNA.
After incubation overnight, filters were washed at 37°
for several hours in hybridization buffer lacking probe, followed by washing
at room temperature in several changes of 300 mM NaCl, 30 mM Na citrate.
Filters were then blotted dry, wrapped in Saran Wrap, and appressed to X-ray
film.
Exposure was either with Kodak NS-5T No-acreen film at room tempera-
ture, or with Kodak XR-5 film and DuPont Lightning Plus intensifying screens
at -75°.
Nearest neighbor analysis.
DNA was labelled by a modification of the
nick-translation reaction (Maniatis et aL., 1975).
A reaction mixture con-
tained the following: 10 yg of C. cohnii DNA, 25-50 yCi of a single a32P-dNTP
(50-150 Ci/mmole, New England Nuclear), 50 mM Tria-HCl, pH 7.6, 5 mM MgCl 2 ,
10 mM e-mercaptoethanol, 50 pg/ml bovine serum albumin, 0.01 yg/ml DNase I,
and 4 units of E. coli DNA polymerase I (Boehringer Mannheim).
Reactions were
incubated for 1 hour at 15°. Following incubation, the reactions were phenol
extracted, approximately 100 yg of unlabelled C. cohnii was added as carrier,
and DNA was precipitated twice with ethanol to remove unincorporated dNTP.
The DNA was resu3pended in 10 mM Tris-HCl, pH 8.0.
To obtain 3•-mononucleo-
tides, an aliquot of the labelled DNA was digested sequentially with DNase TL
and spleen phosphodiesterase in buffer containing 100 mM Na acetate, pH 5.7,
and 2 mM Na2EDTA.
Thin layer chromatography of digested DNA samples was done
as described by Rae (1973).
Radioactive spots were located by exposure of X-
ray film to the chromatogram.
England Nuclear).
Spots were excised and counted in Aquasol (New
The same amount of cellulose was excised with each spot,
and an equal-sized blank portion of the chromatogram was cut out to measure
background radioactivity.
Preparation and fractionation of pyiHmiriine oUgonucleotides.
C. cohnii
DNA labelled in vivo with 'T> was dissolved in 0.5 ml of a 10 mg/ml solution
of calf thymu3 DNA in water.
The DNA was depurinated by the method of Burton
and Petersen (I960), and pyrimidlne oligonucleotides were fractionated according to length by chromatography in DEAE cellulose (Petersen and Reeves,
1966).
Fractionation of the nonopyrimidine isostich in DEAE cellulose using
an ammonium formate gradient was done as described by fierny et al. (196S).
Radioactivity in column fractions was monitored by Cerenkov counting.
Sepa-
ration of pTp from pHOMeUp was achieved by two dimensional thin layer chromatography.
4712
Fractions from the DEAE cellulose column containing the mixture of
Nucleic Acids Research
these two components were pooled and freed of ammonium formate by repeated
drying from distilled water in a rotary evaporator.
Following a final resus-
pension in distilled water, an aliquot of the mixture was spotted near one
corner of a cellulose sheet (Eastman Chromatogram Sheet, No. 13255).
Chroma-
tography in both dimensions was in saturated ammonium sulfate-water-isopropanol, 80:18:2 (Rae, 1973).
Radioactive spots were located by exposure of X-
ray film to the chromatogram.
Spots were excised and counted in Aquasol.
RESULTS
Several methods were used to determine the distribution of hydroxymethyluracil and methylcytosine in C_. cohnii DNA.
In one of the approaches, a mo-
dified version of the nick-translation reaction of E. coli DNA polymerase I
(Maniatis et al., 1975) was used to measure frequencies of various 3' nearest
neighbor nucleotide3 of HCMedlMP and MedCMP.
The method is outlined in Fig-
ure 1, and the details are given in MATERIALS AND METHODS.
C
G
N
A
T
G
C
G
C
DNaie I
C
G
N
A
T
This procedure
Fig. 1. Scheme of the variation of the nick-translation
reaction (Maniati3 et al.,
1975) used to determine 3'
nearest neighbors of modified
nucleotides in C. cohnii DNA.
A base in either its modified
or unmodified form is represented by N, and in the example shown here the radioactively labelled dNTP is dATP.
EcoH DNA PolymeraM I
+ P-P-P-AdR
C
G
N
A
T
G
C
DNaitIl +
Spletn PhosphodletteraM
N
A
b .L
OH
+
othtr
j'-dNMP'«
TLC.cut out tpoti, count
4713
Nucleic Acids Research
involved measuring transfer of a radioactively
labelled phosphate group from
a given dNTP to both of the modified nucleotides in £. cohnii DNA.
By deter-
mining the amount of radioactivity transferred to both the unmodified (dCMP
and IMP) and modified (MedCMP and HOMedtKP) forms of a nucleotide one obtains
a measure of preferences for particular 3' nearest neighbor nucleotides of
the modified nucleotides.
The results of these experiments are expressed as
the following ratio:
ccm in modified dNMP
cpm in modified dNMP + cpm in unmodified dNMP
The ratio is used to alleviate any effects due to sequence preferences
shown by DNase I (Bernardi et al., 1973).
If the distribution of modified
bases is random, the values of HdMeU/HQMeU + T and HeC/faeC + C for every labelled dNTP will simply be their values in total C_. cohnii DNA, or 0.38 and
"^0.03, respectively (Rae, 1973; Steele, 1980; Table 2, legend).
Any deviation from these values would indicate non-random distribution
of 3' nearest neighbors of the modified nucleotides. With this approach one
makes the assumptions that the phosphodiester bond in %nrmvvi.j f^ ^ - T is cleaved
by DNase I at a frequency equal to that for the bond in I
^ ^
,-T, and that
E. coli DNA polymerase I synthesizes with equal efficiency at both types of
nick.
The results obtained by this method are given in Table 1.
It is clear
from the ratios shown in this Table that hydroixymethyluracil and methylcytosine are not randomly distributed in C_. cohnii DNA.
However, these values
reflect the non-random occurrence of HOMedlHP and MedCMP as 5' nearest neighbors of the four common nucleotides, not the actual frequencies at which these
four nucleotides occur as 31 nearest neighbors of HCMedlMP and MedCMP.
By
appropriate calculations using the results in Table 1 and other base composition information, it is possible to determine these frequencies.
The results
of such calculations are given in Table 2, and the calculations are described
in the Table legend.
The results obtained by this method indicate that any one of the nucleotides can be found as a 3' nearest neighbor of HCMedUMP, but a clear order of
preference is seen.
Deoxy-AMP is the most preferred 3' nearest neighbor, with
dCMP being somewhat less preferred.
Deoxy-GHP occurs as a 3' nearest neighbor
slightly less often than would be expected on a random basis, and TMP (or another HCMedUMP) is preferentially excluded as a 3' nearest neighbor.
The re-
sults in Table 2 for methylcytosine distribution in C_. cohnii DNA reflect the
situation in other eukaryotes, in that dGMP is strongly preferred as the 3'
nearest neighbor.
The second approach used to measure HOMeU distribution in C. cohnii DNA
4714
Table 1
Frequencies of Transfer of Radioactivity from
a32p-dNTPs to Modified Nucleotides in C. cohnli
DNA by Procedures Outlined in Figure 1
Ratios
dNTP
Substrate
R a d i o a c t i v i t y i n cpm Transferred t o :
HCMedUMP
TMP
dCMP
MedCMP
HCMeU
HOMeU + T
MeC
MeC + C
0.013
dATP
18,692
16,866
21,263
287
0.526
TTP
9,927
19,612
95,074
162,573
66,810
104,386
346
1,032
0.095
0.108
ave. = 0.101
0.005
0.010
ave.
0.007
dGTP
5,901
6,115
12,311
12,493
4,273
5,961
754
593
0.324
0.329
ave. = 0.326
0.150
0.090
ave.
0.120
dCTP
29,157
45,035
23,846
30,362
68,784
100,842
0
0
0.550
0.597
ave. = 0.574
0
0
Data from experiments of the type shown in Figure 1 are given for each of the four dNTPs. Since DNA
polymerase I will substitute dCTP for MedCTP, and TTP for HOMedUTP, values for the experiments in which
dCTP and TTP were used as substrates could include cases in which MedCMP and HCMedUMP are 3 1 nearest
neighbors of modified nucleotidea.
Nucleic Acids Research
Table 2
Calculation of the Frequency Distribution of 3' Nearest
Neighbors of Modified Nucleotides in C. cohnil DNA
HCMeU
Nucleotide
Contribution
HOHeU + T
Corrected Yield
to Total DNA from T a b l e 1 of 5'-HOMeUuX-3'
Percentage
of 31 Nearest
Neighbors of HCMeU
29.6*
29.2
0.526
15.57
41.59
0.101
2.95
7.88
dCMP
20.9
0.326
6.81
18.19
dCMP
(orMedCMP)
21.1
0.574
12.11
32.34
dAMP
WP
(orHOMedUMP)
37.44
MeC
Nucleotidg
dAMP
TMP
(orHOMedlMP)
dfflP
dCMP
(orMedCMP)
Contribution
MeC + C
to Total DNA from Table 1
Percentage
Corrected Tield of 3' Nearest
of 5'-MeCpX-3' Neighbors of MeC
29.6*
0.013
0.38
12.30
-a
2
0.007
0.20
6.47
20.9
0.120
2.51
81.23
21.1
0
0
0
y
3.09
The nucleotide composition of C. cohnii DNA is 29.6* dAMP, 18.1* TMP, 11.1*
HOMedUMP, 20.9* dCMP, 20.4* dCMP and «0.7* MedCMP (42* G+C); the HOMeU/HCMeU
+ T ratio is 0.38, and the MeC/fceC + C ratio is -vO.03 (Rae, 1973; Steele,
1980). If all bases were distributed randomly in the DNA, the relative yield
of a dinucleotide of the composition 5'-HOMeUpX-3' would simply be the percentage of X in total DNA. Were X to be A, for example, the relative yield of
HOMeUpA would be 29.6*. It is evident from Table 1 that HOMeU and MeC are not
randomly distributed, and the values in Table 1 can be used to adjust the relative yields of the various 5'-HOMeUpX-3' dinucleotides. For the example X
= A, the corrected yield would be (29.6)(0.526) =15.57. By summing the corrected yields for all HCMeUpX and determining the percentage of this total
contributed by each, the frequency at which each nucleotide occurs as the 3 1
nearest neighbor of HOMeU is obtained. The same method is used to calculate
frequencies of 31 nearest neighbors of MeC.
involved determining amounts of hydroxymethyluracil and thymine in the monopyrimidine fraction derived by depurination of in vivo
T> labelled C. cohnii
DNA. Monopyrimidines were prepared and fractionated as described in MATERIALS
AND METHODS, and results of the various 3teps are shown in Figure 2.
4716
The a-
Nucleic Acids Research
5 p,
i
n
m
I
3Q vn
4
3
2
I
b 0
A
10
20
30
40
50
pTp
pCp
1.5
pHp
1
1.0
0.5
jl
-
/I
0.0
10
20
i
.
30
froction
40
50
Fig. 2. Fractionation of pyrimidine oligonucleotides from ^
^
cohnii DNA. The elution profile of pyrimidine isostichs from a DEAE cellulose
column developed with a sodium chloride gradient is shown in the upper left
panel. 32p_iaDeliecl inorganic phosphate eluted first, followed by pyrimidine
isostichs of increasing length. The monopyrimidine fraction (peak I) was separated according to base composition in a DEAE cellulose column developed
with an ammonium formate gradient (lower left panel). In two completely separate runs, the monopyrimidine fraction contained 36.2% and 37.75? of all the
(Py)Pn+l, for an average of 37.C#. Of this, the pTp + pHOMeUp fraction comprised 58.35? and 54.556 of all pPyp (ave. 56.W?). Separation of pHOMeUp from
pTp was achieved by two dimensional thin layer chromatography as shown in the
autoradiogram at the right. The origin was at the black dot, and arrows indicate the directions of the first and second dimensions of chromatography.
Quantitation of such separations is given in Table 3.
mount of radioactivity in pHOMeUp versus the amount in pTp from four determinations is given in Table 3.
These results show that there is a preference
for HOMeU over T in the monopyrimidine fraction.
From the radioactivity dis-
tributions in the column elution profiles in Figure 2, the values for the
amounts of thymine and hydroxymethyluracil in total C. cohnii DNA (legend to
Table 2 ) , and the results in Table 3, one can calculate that 30.65? of the total thymine in the DNA is in the monopyrimidine fraction, while UU-8% of the
total hydroxymethyluracil in the DNA is in this fraction.
These values are
obtained in the following way: (i) from the legend to Figure 2, 37.05? of all
pyrimidines are in the monopyrimidine fraction, and 56.^8 of this fraction is
4717
Nucleic Acids Research
Table 3
Relative Amounts of pTp and pHOMeUp in the
Monopyrimidine Fraction from C. conni i DNA
Determination
Percentage of
Radioactivity in pTp*
Percentage of
Radioactivity in pHOMeUp*
1
51.1 (1387)
2
58.4
(272)
48.9 (1329)
41.6 (201)
3
4
51.6
(524)
48.4 (492)
50.3
(275)
average
52.8
49.7 (271)
47.2
expected for
62.0
random
37.0
distribution
* Values in parentheses are radioactivity in counts per minute.
+ From the nucleotlde composition of C_. cohnii DNA given in Table 2, IMP +
HOMedlMP comprise 29.2^ of all micleotides. If these were distributed randomly in DNA, pTp would amount to (18.1/29.2)100 = 62.0^ of pTp + pHOMeUp, and
pHOMeUp would constitute the balance.
pTp + pHOMeUp, so that (37.0)(.564) = 20.9^ of all pyrinrldines are pTp or
pHOMeUp monopyrimidines; (il) data in Table 3 indicate that 52.8^ of the pTp +
pHOMeUp fraction is pTp, and 47.2^ is pHOMeUp, so that II.O56 and 9.956 of all
pyrimid±nes are, respectively, the monopyrimidines pTp and pHOMeUp; (iii) from
the nucleotide composition of C. cohnii DNA (Table 2, legend), TMP comprises
(18.1/18.1 + 11.1 + 21.1)100 = 36.C# of all pyrimidines, and HOMedlMP comprises
22.13; (iv) thus, (11.0/36.0)100 = 30.6* of all T is flanked by purines, as is
(9.9/22.1)100 = 44.856 of all HOMeU.
If all nucleotides were randomly distri-
buted in the DNA, only 25^ of each of T and HOMeU would be found in this fraction.
Thus one finds somewhat more of the total thymine in this fraction than
expected from a random distribution, but nearly twice as much hydroxymethyluracil.
Another approach used to examine the distribution of methylcytosine in
C. cohnii DNA involved analyzing cleavage patterns of the DNA with various
restriction endonucleases whose activities are inhibited by the presence of
methylcytosine in their recognition sequences.
The enzyme Hpa II recognizes
and cleaves the sequence C+CGG (Garfin and Goodman, 1974).
If, however, the
internal cytosine residue in the sequence is methylated, cleavage is prevented
(Mann and Smith, 1977).
Another enzyme which cleaves the same sequence, M3p
I, is not affected by the presence of the modified cytosine (Waalwijk and
Flavell, 1978).
Together, these two enzymes provide a very useful tool for
studying distribution of methylcytosine in eukaryotic DNA, in which most of
4718
Nucleic Acids Research
the methylcytosine occurs in the dinucleotide HeCpG.
The enzymes Hha I,
which cleaves GCG+C (Roberts et al., 1976), and Ava I, which cleaves
^
(Hughes and Murray, 1980), also fail to cleave DNA when the internal cytosine
in the recognition site is methylated (Bird and Southern, 1978; Mann and
Smith, 1979)> making these enzymes useful probes as well for studying methylation of DNA.
To rule out interference with these enzymes by HDMeU, each wa3 tested on
DNA from the bacteriophage SP82, a Bacillus subtilis phage in which thymine
is entirely replaced by hydroxymethyluracil (Hemphill and Whiteley, 1975).
The results of these experiments are shown in Figure 3. Msp I and Hpa II both
cleaved SP82 DNA extensively to give identical fragment patterns (lanes a and
b).
Hha I also cleaved SP82
SP82 DNA
DNA
extensively (lane c ) . Ava I did not cleave
(not shown), but since this enzyme cleaves C,. cohnii DNA (Figure 4,
lane h) it is evidently not inhibited by
HCMeU, and its inability to cleave
SP82 DNA seems likely to be due to an absence of sites for this enzyme in the
DNA.
When G_. cohnii DNA was treated with Hpa H , little cleavage was detected
(Fig. 4, lane b ) ; however, Msp I cleaved the DNA extensively (lane a ) . The
presence of interfering substances in the Hpa II reaction mixture was ruled
out by determining that bacteriophage \ DNA included in the mix was digested
Fig. 3. DNA of bacteriophage SP82 treated with Msp I
(a), Hpa II (b) or Hha I (c), electrophoresed in X% (a,
b) or 1.1& (c) agarose, and stained with ethidium bromide.
4719
Nucleic Acids Research
Fig. 4- Lanes a, b, e and h show ethidium bromide staining patterns of gelfractionated C_. cohnii DNA which was treated with Msp I, Hpa H , Hha I and
Ava I, respectively. Lanes c and d show autoradiograms obtained when DNA in
lanes a and b was transferred to a nitrocellulose filter and hybridized with
a mixture of end-labelled 17S and 24S rRNAs. Lanes f and g show the hybridization patterns obtained when DNA treated as in lane e was transferred to nitrocellulose filters and probed with end-labelled 17S (f) or 24S (g) rRNA.
Lane i shows the hybridization pattern obtained when DNA in lane h was transferred to a nitrocellulose filter and probed with a mixture of end-labelled
17S and 2AS rRNAs. Lanes j, k and 1 3how ethidium bromide staining patterns
of undigested, Hpa II-treated, and Msp I-treated P. cassubicum DNA, respectively. All of the gels shown in this Figure are 1? agarose. The arrowhead in
lane e is explained in the text.
to completion (not shown).
It was of interest to determine if a particular
sequence in the DNA of C_. cohnii was always methylated or not, and for this
the repeated genes for ribosomal RNA were chosen. When C. cohnii DNA was
digested with Msp I, electrophoresed in an agarose gel, blotted onto a nitrocellulose filter, and hybridized with rRNA, fragments varying in size from
10.2 kb to 1.4 kb were labelled (Fig. k, lane c ) .
No hybridization to DNA at
the limiting mobility position was seen. On the other hand, when DNA incubated with Hpa II was hybridized with rRNA, labelling was to DNA at the limiting
mobility position (lane d ) .
This result indicates that the sites detected by
Msp I are modified in most or all copies of the ribosomal RNA genes.
Treatment of C. cohnii DNA with Hha I had little effect on DNA size (Fig.
4720
Nucleic Acids Research
4, lane e ) . If bases are taken to be randomly distributed in C_. cohni 1 DNA,
it is possible to construct a theoretical distribution curve for Hha I fragment sizes as described by Botchan e_t &L. (1974)-
In such a distribution only
135^ of the DNA should be present as fragments greater in size than 2 kb.
The
position at which a fragment of this size would ml grate in the gel shown in
lane e is indicated by the white arrowhead.
Nearly all of the DNA is well
above the arrowhead, indicating that a large fraction of Hha I sites are modified.
The possibility that impurities in the DNA sample prevented Hha I
cleavage was eliminated by the finding that bacteriophage A DNA was digested
to completion by Hha I in a mixture of C_. cohnii DNA and \ DNA (not shown).
As the results in lanes f and g of Figure 4 show, when either 17S (lane f) or
24S rRNA (lane g) is hybridized to Hha I-treated DNA, a complex distribution
of high molecular weight fragments and DNA which did not separate in the gel
are labelled.
This result suggests that some Hha I sites in rDNA may be me-
thylated in a n repeats while others are unmethylated in » n repeats, or that
a given Hha I site may be methylated in some rDNA repeats but not in others.
Ava I cleaves C_. cohnii DNA (Fig. 4, lane h ) , suggesting that a significant fraction of the Ava I sites do not contain methylated internal cytosine
residues.
As the results of hybridization with labelled rRNA show (lane i ) ,
rDNA is cleaved by Ava I as well.
Some Ava I sites (those with the sequence
CCCGGG) are a subset of Hpa Il/Hsp I sites.
These are also Sma i/Xma I sites
(Roberts, 1980), and both of these latter enzymes do not cleave C.. cohnii DNA
(not shown).
Their lack of activity is presumably due to the presence of me-
thylcytosine in the recognition sequence (see also Gautier e_t al., 1976).
These results suggest that Ava I cleavage of C_. cohnii DNA is restricted to
CTCG33 sites, and that these sites are largely unmethylated.
DNA from Prorocentrum cassubicum. another dinoflagellate species whose
DNA contains methylcytosine (Rae, 1976), was also tested for its sensitivity
to cleavage by Hpa II and Msp I.
Figure 4.
The results are shown in lanes k and 1 of
Hpa II does not give significant cleavage of the DNA, whereas Msp I
cleaves the DNA extensively enough to allow a significant fraction of it to
migrate past the limiting mobility position in the gel.
Thus in this dino-
flagellate species as well, CpG dinucleotides are methylated.
DISCUSSION
The functional or structural significance of modified bases in eukaryotic
DNAs is unknown.
The most widely occurring of these bases, methylcytoaine,
appears to be absent from some eukaryotic DNAs (Rae and Steele, 1979; Bird and
Taggart, 1980), indicating that it is not fundamental to eukaryotic genome
4721
Nucleic Acids Research
structure or function.
To extend the observations on modified bases in eu-
karyotic DNA, we have examined the diatribution of methylcytosine and hydroxymethyluracil in the DNA. of a dinoflagellate, Crypthecodinium cohnii.
It was
of interest to determine how modified bases are distributed in such DNA because of the unique position that dinoflagellates occupy among eukaryotes by
virtue of their unusual chromosome structure (see INTRODUCTION).
Nearest neighbor analysis showed that methylcytosine in C_. cohni i DNA is
largely restricted to the dinucleotide MeCpG.
As do other features of dino-
flagellate DNA (Rae and Steele, 1978), the fact that the amount and distribution of methylcytosine in C_. cohnii DNA are similar to those in some higher
eukaryotes argues against the suggestion of Dodge (1965) that dinoflagellates
are intermediate between prokaryotes and eukaryotes.
The results obtained
when DNA from Prorocentrum cassubicum was treated with Hpa II and Msp I (Fig.
k) indicate that in this dinoflagellate as well, the CpG dinucleotide3 are
extensively methylated.
In light of the proposal by Loeblich (1976) that
dinoflagellates in the order Prorocentrales are primitive species, this result
suggests that considerable CpG cytosine methylation is an evolutionarily old
feature of dinoflagellate DNA.
It appears from the results of digests of C. cohnii DNA with various restriction endonucleases sensitive to cytosine methylation that CpG dinucleotides are not all equally susceptible to methylation.
In particular, a sub-
set of the sites recognized by Ava I, those with the sequence C T C G Q G , are not
extensively methylated.
This result suggests that sequences flanking the CpG
dinucleotides may influence the binding of the methylase to DNA.
The extensive methylation of C. cohnii rDNA at Hha I and Hpa Il/Msp I
sites is interesting in comparison with the results of Bird and Taggart (198O),
which show that rDNA in animals is heavily methylated only in amphibia and
fish.
Because no studies on rDNA methylation in plants have been published,
we cannot state if dinoflagellate DNA is typical in this regard.
However,
the large amounts of methylcytosine in DNA of higher plants make it likely
that their rDNA is methylated.
The synthetic mechanism for HOMeU in dinoflagellate DNA is unknown.
The
most straightforward mechanism would be modification of thymLne at the DNA
polymer level.
By analogy with the systems responsible for modification of
bacterial DNA and methylation of cytosine in eukaryotic DNA, one might expect
that modification of thymine in dinoflagellate DNA to yield hydroxymethyluracil would be catalyzed by an enzyme which acts on thymines at particular sites
in the DNA.
4722
An obvious result of such a mechanism would be non-random distri-
Nucleic Acids Research
bution of HOMeU In the DNA. We showed earlier that the percentage of HOMeU replacing T varies in different buoyant density fractions of C_. cohnii DNA (Rae,
1973).
A more direct demonstration of non-random distribution of HOHeU in C.
cohnii DNA is provided by the nearest neighbor nucleotide analysis and pyrimidine distribution analysis in this paper.
The occurrence of more than one
3' nearest neighbor was to be expected in light of the fact that nearly l£ff>
of the expected thymine is replaced by hydroxymethyluracil.
A stringent re-
quirement for a single nucleotide as a 3' nearest neighbor would reduce the
number of available sites for HQMeU to a level insufficient to achieve such a
high degree of substitution (aside from TpT, the most frequent TpN in a DNA
having about l£fj> G+C is TpA; on a random basis, this would comprise 95^ of all
dinucleotides, and 155t of all T).
It is somewhat surprising, however, that
of the two most preferred nearest neighbors (dAMP and dCMP), one is a purine
nucleotide while the other is a pyrimidine nucleotide.
Furthermore, it is
interesting that the most preferred neighbor is dAMP, which gives the symmetrical sequence TpA/ApT when part of a dinucleotide with IMP.
Sequences
recognized by bacterial modification enzymes are symmetrical, and in eukaryotes methylation of cytosine residues is preferentially the symmetrical sequence CpG/GpC.
The results on distribution of HCMeU in C. cohnii DNA are interpreted as
favoring a mechanism for HCMeU production Involving modification of thymine
residues in DNA.
results.
However, two other possibilities are not excluded by these
One is that HQMeU is incorporated into DNA via HOMedUTP as it is in
SP phage DNAs (Roscoe and Tucker, 1966).
In such a scheme, both TTP and
HCHedUTP would be present as precursors, and some mechanism would operate to
give rise to the non-random distribution of HQMeU in DNA.
There is no evi-
dence that DNA polymerase is capable of directed incorporation of alternative
precursors into DNA, so synthesis involving both HOMedUTP and TTP is considered an unlikely mechanism for the appearance of HCMeU in dinoflagellate DNA.
The other possibility is that TTP is not synthesized by dinoflagellates, and
the DNA precursor is HCHedUTP.
The presence of thymine in DNA would then be
the result of dehydroxylation of hydrcotymethyluracil at certain sites in DNA
to produce thymine.
Incorporation of HOMedUTP into DNA and subsequent conver-
sion of the base to thymine has been shown for DNA of +W-14, a phage which
infects Pseudomonas acldovorana (Neuhard et aL., 1980).
An understanding of
the mechanism responsible for HCMeU synthesis in dinoflagellates must await
studies of nucleotide precursor pools and characterization of nucleic acid
metabolizing enzymes in these organisms.
4723
Nucleic Acids Research
ACKNCWLEDGIMENTS
We thank Dr. Elizabeth Blackburn for suggesting the use of the nicktranslation reaction for nearest neighbor analysis. This work was supported
by NSF and NIH grants to PMMR. RES was supported by an NSF Graduate Fellowship and a USPHS Cell and Developmental Biology Training Grant. A portion of
this work is from a dissertation submitted by RES in partial fulfillment of
the requirements for the Ph.D. degree at Yale University.
*
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
4724
Present address: Fred Hutchinson Cancer Research Center, 11^4 Columbia
Street, Seattle, Washington 98104, USA
Barnett, T. and Rae, P.M.M. (1979) Cell 16, 763-775.
Bernard!, G., Ehrlich, S.D. and Thiery, J.-P. (1973) Nature New Biol.
246, 36-40.
Bird, A.P. and Southern, E.M. (1978) J. Mol. Biol. 118, 27-47.
Bird, A.P. and Taggart, M.H. (1980) Nucleic Acids Res. 8, 1485-1497.
Botchan, M., McKenna, G. and Sharp, P.A. (1974) Cold Spring Harbor Symp.
Quant. Biol. 38, 383-395.
Burdon, R.H. and Adams, R.L.P. (1969) Biochim. Biophys. Acta 174, 322329.
Burton, X. and Petersen, G.B. (I960) Biochem. J. 75, 17-27.
Serny, R., Mushynski, W.E. and Spencer, J.H. (1968) Biochim. Biophys.
Acta 169, 439-450.
Dodge, J.D. (1965) Excerpta Med. Int. Congr. Ser. 91, 264-265.
Dodge, J.D. (1966),in "The Chromosomes of the Algae", Godward, M.B.E.,
ed., St. Martin's Press, New York, pp. 96-115.
Dodge, J.D. (1971) Protoplasma 73, 145-157.
Garfin, D.E. and Goodman, H.M. (1974) Biochem. Biophys. Res. Commun. 59,
108-116.
Gautier, F., Mayer, H. and Goebel, W. (1976) Mol. Gen. Genet. 149, 23-31.
Grippo, P., Iaccarino, M., Parisi, E. and Scarano, E. (1968) J. Mol.
Biol. 36, 195-208.
Hemphill, H.E. and Whiteley, H.R. (1975) Bact. Rev. 39, 257-315.
Hughes, S.G. and Murray, K. (1980) Biochem. J. 185, 65-75.
Loeblich I H , A.R. (1976) J. Protozool. 23, 13-28.
Maizels, N. (1976) Cell 9, 431-438.
Maniatis, T., Jeffrey, A. and Kleid, D.G. (1975) Proc. Natl. Acad. Sci.
USA 72, 1184-1188.
Mann, M.B. and Smith, H.O. (1977) Nucleic Acids Res. 4, 4211-4221.
Mann, M.B. and Smith, H.O. (1979) in "Transmethylation". Developments in
Neuroscience, Vol. 5, Usdin, E., Borchardt, R.T. and Creveling, C.R.,
eds., ELsevier/North Holland, pp. 483-492.
Meselson, M., Yuan, R. and Heywood, J. (1972) Ann. Rev. Biochem. 41, 447466.
Neuhard, J., Maltman, K.L. and Warren, R.A.J. (1980) J. Virol. 34, 347353.
Petersen, G.B. and Reeves, J.M. (1966) Biochim. Biophys. Acta 129, 438440.
Provasoli, L. and McLaughlin, J.J.A. (1963) in "Symposium on Marine
Microbiology", Oppenheimer, C.H., ed., C.C. Thomas, Springfield, Illinois, pp. 105-114Rae, P.M.M. (1973) Proc. Natl. Acad. Sci. USA 70, 1141-1145.
Rae, P.M.M. (1976) Science 194, IO62-IO64.
Nucleic Acids Research
28
29
30
31
32
33
34
35
36
37
38
39
Rae, P.M.M. and Steele, R.E. (1978) Biosystems 10, 37-53.
Rae, P.M.M. and Steele, R.E. (1979) Nucleic Acids Res. 6, 2987-2995.
Roberts, R.J. (1980) Nucleic Acids Res. 8, r63-r8O.
Roberts, R.J., Meyers, P.A., Morrison, A. and Murray, K. (1976) J. Mol.
RLol. 103, 199-208.
Roscoe, D.H. and Tucker, R.G. (1966) Virology 29, 157-166.
Shapiro, H.S. (1976) Handbook of Biochemistry and Molecular Biology,
Nucleic Acids, Vol. 2, 3rd edit., Faaman, G.D., ed., CRC Press, Cleveland, pp. 241-281.
Sinsheimer, R.L. (1954) J. Biol. Chem. 208, 445-459.
Sneider, T.W. and Potter, V.R. (1969) J. Mol. Biol. 42, 271-284.
Southern, E.M. (1975) J. Mol. Biol. 98, 503-517.
Steele, R.E. (1980) Ph.D. Thesis, Tale University.
Tuttle, R.C. and Loeblich I H , A.R. (1975) Phycologia 14, 1-8.
Waalwijk, C. and Flavell, R.A. (1978) Nucleic Acids Res. 5, 4631-4641.
4725
Nucleic Acids Research
4726