Download A Genetic Linkage Map for the Zebrafish

A Genetic Linkage Map for the Zebrafish
John H. Postlethwait,* Stephen L. Johnson, Clare N. Midson,
William S. Talbot, Michael Gates, Eric W. Ballinger, Dana Africa,
Rebecca Andrews, Tim Carl, Judith S. Eisen, Sally Horne,
Charles B. Kimmel, Mark Hutchinson, Michele Johnson,
Andre Rodriguez
To facilitate molecular genetic analysis of vertebrate development, haploid genetics was
used to construct a recombination map for the zebrafish Danio (Brachydanio) rerio. The
map consists of 401 random amplified polymorphic DNAs (RAPDs) and 13 simple sequence repeats spaced at an average interval of 5.8 centimorgans. Strategies that exploit
the advantages of haploid genetics and RAPD markers were developed that quickly
mapped lethal and visible mutations and that placed cloned genes on the map. This map
is useful for the position-based cloning of mutant genes, the characterization of chromosome rearrangements, and the investigation of evolution in vertebrate genomes.
The zebrafish Danio rerio (formerly Brachydanio rerio) (1) is emerging as a model
organism for the investigation of the genetic mechanisms of vertebrate development
(2). Its short 3-month life cycle and the ease
of making both haploid embryos and parthenogenetic diploid fish facilitate the identification and analysis of mutations (3). Saturating the genome with mutations that affect
various aspects of the early development of
zebrafish seems to be an attainable goal (2).
The ability to make stable lines of transgenic
zebrafish (4), as well as the ease of making
genetic mosaics (5, 6), facilitates the study
of gene interactions and gene function.
In spite of the demonstrated potential
of zebrafish genetics in understanding vertebrate development, both the genetic
and molecular investigation of zebrafish
are in their infancy. For example, only
two genes identified by mutation have
been cloned from zebrafish (7), and only
recently have any molecular genetic markers been reported (8, 9). Furthermore, no
two genetic markers in zebrafish have been
shown to be linked. Because a linkage map
is necessary to facilitate molecular genetic
analysis of zebrafish development, we constructed such a map with anonymous
DNA polymorphisms and then developed
ways to place genes known only by sequence or only by mutation on the map
with the use of the haploid genetics available in zebrafish.
For a map to be useful, it should contain
genetic markers that are abundant, evenly
distributed, highly polymorphic, and readily detected in many laboratories. A map of
J. H. Postlethwait, Institute of Neurosciences and
Institute of Molecular Biology, University of Oregon,
Eugene, OR 97403, USA.
S. L. Johnson, C. N. Midson, W. S. Talbot, M. Gates, E.
W. Ballinger, D. Africa, R. Andrews, T. Carl, J. S. Eisen,
S. Horne, C. B. Kimmel, M. Hutchinson, M. Johnson, A.
Rodriguez, Institute of Neurosciences, University of
Oregon, Eugene, OR 97403, USA.
*To whom correspondence should be addressed.
random amplified polymorphic DNA
(RAPD) markers fits these criteria (10, 1 1).
A polymerase chain reaction (PCR) with
zebrafish genomic DNA as a template and a
single, 10-nucleotide-long primer of arbitrary sequence generally amplifies 6 to 12
DNA fragments (9). Different primers amplify distinct DNA fragments. Different
strains of zebrafish differ somewhat in their
fragment patterns, presumably because of
differences in nucleotide sequence at the
primer binding sites or deletions or insertions between the primer binding sites.
These phenotypes are inherited as Mendelian genetic markers. A dominant allele
permits the amplification of a DNA fragment with a specific primer, whereas a
recessive allele results in the absence of that
fragment (9, 10).
We identified RAPD markers suitable
for mapping by performing PCRs with 134
different decamer primers (12) that each
amplified several easily scored DNA fragments specific to the DAR (Darjeeling) or
AB parental strains (neither of which is
completely inbred) (9). We followed these
genetic markers segregating among the 94
haploid offspring (13) of a single DAR/AB
female fish (called here the linkage map
cross). In haploids, recessive alleles are not
obscured by their dominant alternatives,
and thus the genotype of a haploid can be
inferred directly from its phenotype. Of 401
RAPD markers, 203 (50.6%) were from the
AB parent, 164 (40.9%) were from the
DAR parent, and 34 (8.5%) behaved as
codominant markers. Computer-assisted
linkage analysis (14) showed that the
RAPD markers fell into 29 linkage groups
(LGs) that consist of 2 to 25 loci spaced at
an average interval of 5.8 + 5.9 (SD) cM
(centimorgans) (Fig. 1). Only four scorable
RAPD loci (15AC.850, 2AG.440,
17P.710, and 15V.400) remained unlinked
to any other marker in our sample.
In addition to RAPD markers, we also
SCIENCE
*
VOL. 264
*
29 APRIL 1994
mapped 13 of the 16 published simple
sequence repeats (SSRs) by implementing a
PCR on DNA from haploid embryos of the
linkage map cross using published primers
(8). Of the three unmapped SSRs, SSR 22
was unlinked to any other marker, SSR 17
did not segregate in the linkage map cross,
and SSR 27 did not reliably amplify a
product.
Because the mapping of cloned genes
facilitates the identification of candidate
genes for mutations, we developed a strategy to map genes known by DNA sequence. In this approach, DNA from each
haploid of the linkage map cross serves as
a template for a PCR with the use of a pair
of primers that amplify a fragment containing either the target gene's 3' untranslated region (UTR) or one of its introns.
In some cases, this results in a fragment
size polymorphism. For example, primers
spanning intron 1 in the major histocompatibility complex (MHC) class II chain
locus (15) amplified fragments of 520 and
580 base pairs (bp) that segregated as
codominant alleles in the linkage map
cross and provided a location on LG VIII
(Fig. 1). Likewise, snail) (16) was mapped
to LG XI on the basis of codominant
length variants of 210 and 250 bp arising
from the CA repeat contained in its 3'
UTR. Although primers amplifying the 3'
UTR of msxB (17) yielded fragments of
uniform size from all 94 haploids of the
linkage map cross, the restriction enzyme
Msp I cut the DAR allele but not the AB
allele, allowing msxB to be localized to LG
I near leopard (Fig. 1).
The utility of a genetic map is related to
its degree of completeness. The current
map has four more LGs than the number of
haploid chromosomes (18), and 4 of the
405 RAPD markers and 1 of the 14 SSR
markers remain unlinked to any other
marker; thus, a minimum of nine gaps
remain to be filled. Assuming that markers
are randomly distributed, about 99% of loci
are estimated to be located within 20 cM of
a marker on the map (19). The fact that
only 5 of the 419 DNA polymorphisms
studied remain unlinked to any other marker supports this estimate.
A parameter to consider when planning
map-based cloning of mutated loci is the
relation of physical to genetic distance. To
estimate the average length of 1 cM in
kilobase pairs (kbp) of DNA, one must
compare the size of the zebrafish physical
map to the length of its genetic map. The
size of the linkage map shown is 2317 cM
(Fig. 1). To estimate the size of the complete map, one must account for the distance from the end markers on each chromosome to their adjacent telomeres and
any unfilled gaps. This would add 145 cM
to account for the telomeres (25 chromo699
Downloaded from www.sciencemag.org on September 20, 2011
M
this estimate gives an average of about 625
kbp cM- , although the relation of physical
and genetic distance may vary from one
region of the genome to another (22). In
contrast, the mouse genome has on average
1800 kbp cM- (23).
A strategy that exploits the haploid genetics of zebrafish and the advantages of
RAPD markers allowed us to quickly identify loci closely linked to a mutation. This
in turn allowed us to place the mutation on
the map, thus serving as a prelude to molecular isolation of the gene (24). This
-211.800A)
11000(D
-
18O317.0
t.3
-
-1418100(A)
16&6(D)
5.7 -1701000(
-MB.1300(
0.0
r)
1.1
MG
-
CISOQ
C150225(D
Z1650(D)
1.1
1u6.1450A
SA
1.1
780(D)O
6.4
62
12
14080(A
20425D
-
0.0
14Y.112A)
7.7
5.1
22
IV.9750)
17X~)
17AE.14M0()
1.6
3.8
14C52~1D)
9.2
6.6
0.0
32
4.1
9.7
2
5.3
132
38.6550(
17X~
4.6-
1.0>
4.3
11316L3
-4X25(
14Y.72RA)
80
a1- \ 2n
Y 2~1
11L1
..
-
-
17A01O(D)
1.17
UP)~
12R60A0)
4.4
SW.3A
-E
74
28.0
-4C.1~D)
17P.(D)
+18
IO1N.6tED)
451120(D)
a 14
8BR14(ND)
r
-14C.620(A
20K(1480(A
11~
I\A
W1Z
167.41(D)
13.725)
11.1
)
15V.525()
21.0 -
11010(D)
13B. 112W
6.7
23.4
-
11.5
-
0(A
150.575W
511H.350(A
10.6 - 0N.710(D)
13.0 3U.~
9.6.0
0.0-
-
2.4-.
6.9 01
4.6-
123
0.1
0.0 It
1.4 IF
7.0 "
521
1.0
24.4
11PA11W
19A5601(D)
Lv1#D.~)o
8OR2
20130
C(U.1AD)
fi86R13(D
|{SN.WM
2
eN.31O(A
_J.1300(D)
IOAF.~D)
24J14300(A)
-
1.3-10
166
-
9.7
-
6.4
-
1.1
-
210SO(A)
62
5.3- - 50.40(D)
6.0- -
-
0.0
0.0
SS.28AD)
MB.1120A
3.6
1.6
P2M.1500(A
ze
SY.0
a.1'I
7.65
7.6 r
9F.~60(D)
{t7K
201025(D)
2.21
ilAK9S0(D)
9A.1700(A)
6C.1350(D)
5AD.200~
P2.1700(A)
50.1300(D)
138.5A
3.6
1.5 t
1.0
52
26.7-
2AG 340(A)
1U.U0(D)
0.01
o00.0 w
ir
SN.0)
Me.~
8D.1000(D)
41(20D)
02 J
1.0
1.6
48660(D)
4
1U.5
9AH.510(A)
17AF.A
1.1.725A)
SOR20(A
180
1.0
r
38R02~D)
-
aN.12000D)
-
&4-
9AL1110A)
21
13Y.1400A
21 7
1.07
16.0-
9F.1600)
101Q1060(A
14U.900(D)
4W.16 )
| 21(430(D)
ne51700(A
14ALm~
t
16.1270(A)
40.1250)
7.4-
a7C0O
7.8
20K875)
7.6
#
-
12
6.3-
140.96()
P83.50(D)
714C()
0.0
-
22
19.1
-
6 F.7D)
14Y.~25A
4C.11~OA
16L65A
4K750A)
E.7W
24.3-
-1ST.320O(A)
15.0
-
-101Z117A)
13.0-
-60.360(D)
-1k650(D)
141J.60(
14WF.65
7711320(D)
9.53-
SN
4Y.350A)
-71750(A
6A5225D)
20
0.0
00
17AE600(D)
12.-
100.1560A)
)
88.660
15P2
F2IO
O(D)
)
7
3AG.1330(A)
4J.600(
8.0
1.0
2
.
4,
1SV.71DA
161.1190(D)
50.12
)
.I-V.600(D)
17P.1150
4.2
1001500(
1.
13815001D)
4.2
7.4 -
I:- U0(D)
1.0
*
1
/
1.2*
21.9
-
9.12.5
11.0-
S2
63-
20.575w
17X700~AD)
10AF.112D(D)
14A0.90D)
20R1100(D)
1SO.m(AJD)
6U.145A)
13AM70(D)
1SR300(D)
SNI.1310(D)
-14J.90D)
-F.1450(D)
2M.7M0D)
-1SRSO0(AD)
061 ( D)
PPOA.610(
)
3.1
9a
0.0
Fig. 1. A genetic map for the zebrafish. DNA samples were prepared from
94 haploid progeny of a DAR/AB heterozygous female fish and subjected
to PCR with 10-nucleotide-long primers (9). Amplified fragments were
separated by electrophoresis on agarose gels. Linkage groups were
identified with the program Linker (14), and maps were constructed with
MAPMAKER (14). Scoring error was less than 1% (34), and although most
marker orders will be correct, distances between markers will vary
because of statistical fluctuations and genetic background effects. DNA
1MA .190(A
IS0.12 00(A)
02
1
1.0
|7N.5OM(D)
80.6
ADA. ~)67
11 .7
20B.740
7A.000)
13C.1100(D)
.-900(D)
17Y.1075D)
-01-
IV.675(A)
14U.150()
O .1O
6.4 -U
)
2.8
9-
I)
OR 1260(D)
4W.000
1.2
-12AE1480(D)
10-
21
1.9
__11.0
5.0
. 1.770O
4.2:
20F.71O(A)
-1.6
10.4
214475D)
4
13Y.600(A)
10G.11tA)
-11A.1A)
LOWXI
as-~~~~~~~~~~~~1.
28.7-
1M0F.1010(A)
-F.1500(D)
D)
161.1800D)
17P.700(A)
201600D)
\'10Z.6(A
OU.480(A)
O.0J
Le XIII
12KM~
10140()
12.6-
113,e107f
16.0
X
9.0-
LO Xll
1B.710(AD)
2.1
1.56
7.1 -
-
3.3
0.0
P.1750(D)
LO Xl
LOX
192-
J
-2G.325A)
17.5 -
1C1J.1170(D)
'21.1225W
1OGO.7(D)
-
1 3.5
2AD.O25()
1T36D)
2A~0A~)
4t950()
0.0
6141150(D)
130A.0(D)
GAB 63(D)
1B.8SO
13Y.725A
612K30(D)
162
13C.A)
5.3-
_SE.8(AID)
6A.310tA
P143.40(A)
50.510(A)
-
9AMH200D)
12
6.6
-
20.8
-
0.0
11.3
2.8-
14AG330(D
8SR2qAD)
-
1.7
2N.30(A)
-
2.3
17.3
17V.500(D)
AMG.610(A)
Lla^e
21L127A
4.33.S -0
1
00.11()
700
#
11.9
-K1300~4D)
3.7
11 _
3.5
8.4-
7K14)
-71(148(A)
-14Y30)
-
11
4X1000~
5.4
7AE.50(D)
-
La IX
La Vill
3.1 -E
3.1-
6.3
60790(A)
-8 SkOOD)
-
-
P1O(D)
-
^E.1180(A
5.6
4.32
0.0
13C2250(ID)
-
1.1
M4U1350(D)
437
-
0.0
15.9
42
/
¶
0.0
3.7 -
AC.100m
17Y.135)
a
-
129
-
6.9
1.0
1.0r
-17P.875(A)
9.7
13.1
-20Y.~
Z
11.9
3.6
IW.2(D)
MLU0(A)
,B
1AD.107()
141.1060(D)
80
2a1)
CL4J
127-
1.3
80 X
IISAD.97O(D)
15R600~)
ws14QLOt)
-1
80"
1.2
4.S
s1506.
4C.11i0w
23
-7N.1750(AiD)
14AG.1140(D)
-
LO VII
LO Vl
LO V
LG IV
La HI
La 1
LOa
strategy was illustrated by the mapping of
brass, a pigment pattern mutation (25).
Homozygous brass females were mated to
DAR males. From one of the resulting F1
heterozygous females, we collected 38 brass
and 38 wild-type haploid offspring. We
collected DNA from each embryo individually and then made two DNA pools-one
from 20 brass haploids and one from 20
wild-type haploids. We screened this pair of
DNA pools with 48 different decamer primers and compared their patterns of amplified
fragments. Genetic markers unlinked to
Downloaded from www.sciencemag.org on September 20, 2011
somes x 2 telomeres per chromosome X an
average of 5.8/2 cM from end marker to
telomere), and a minimum of 261 cM to
account for at least nine unfilled gaps (5
unlinked markers plus 4 too many linkage
groups multiplied by 29 cM, the maximum
distance accepted in the map), to give a
minimum estimate of the complete map size
of 2720 cM. The length of the map can also
be estimated from considerations of marker
density (20), which yield a value of about
2550 cM. Because the haploid genome of
zebrafish contains about 1.7 x 109 bp (2 1 ),
)
M.00~
11711225(A)
-8F1~
from the same 94 haploid siblings was scored for SSRs as described (8),
except that 4% agarose gels were used to resolve the DNA fragments.
The MHC class 11 P chain locus was mapped in the linkage map cross with
the primer pair Tu385 and Tu360 (15). The eight mutations visible in
haploid embryos were placed on the map by means of the protocol
illustrated in Fig. 2. Because leopard is not distinguishable in haploids
(35), it was mapped in a diploid backcross. Numbers on the left of each
LG indicate genetic distance in centimorgans.
SCIENCE * VOL. 264 * 29 APRIL 1994
`
' "'
E
-11
1
---"` `--- 'I'll
-1.10.
1-,11 "' `
'I
"
'I l'
-1 .".'. '.,. 'l.,. '. . l 1 111-11
` ..
.1
`
-11
I'll
brass should have been present in both
pools, but those closely linked to the brass
locus should have been present predominantly in one pool or the other. The use of
haploids instead of diploids in this method
[called bulked segregant analysis (26)1 eliminates the need to remove heterozygous
individuals from phenotypic pools in order
to identify markers associated in coupling
with either the mutant or the wild-type
allele of the gene under study. The analysis
identified several fragments that were amplified predominantly or exclusively in one
phenotypic pool but not in the other (Fig.
2A). These fragments became candidates
for genetic markers linked to brass.
To determine whether any of the candidate markers were indeed closely linked to
brass, we scored the 38 brass and 38 wildtype haploid siblings individually for these
markers. These experiments confirmed that
two of the 48 original primers (13AI and
16AI) amplified DNA fragments that identify genetic markers linked to brass (Fig.
2B). The 16A 1 190 DNA fragment amplified from all of the 10 wild-type haploids
shown but from only 1 of the 10 brass
haploids; thus, there was 1 recombinant
among the 20 haploids shown. In all, there
were 6 recombinants between brass and
16A. 1 190 out of 76 haploids, indicating a
32- -2D.810(A)
5.46.4 -
13.3
-3U.100A)
1M7X.6A)
-
2.AF.-
D)
0.5-
127
-
3.4
-
17.4
-
38.510(A/D)
-AF.0)
7.80.0
9E.100A)
0.0w
0.0w
6.7.
IV.300()
* lIA 1300A)
wp
D1AF.WA
.07
14.7
14.6
-
-
S7 -_
32 &S - -
-
-40.120(AD)
6.61200()
15.4
-
1.1 -
4.5
.2 - 1.6
11.6
140.460(A)
150.375(D
-12t4900AD)
-20K1150D)
1.9
-
170.1100(D)
-
.710(D)
13.6
5.1
-
11.5
-
_
1.1
SM775(A)
SW.4~()
4.3
85
71(6KED)
1.1
5.5 -5.5
1.2
270(D)
7A.550(D)
3.7110
81
-
60.1000D)
h.
201675(A)
MC.1000(A)
38050(A)
L 2A420(A)
L 5.4~()
'11'
7J.71O(D)
32 -
1.1'
4.8
4.37
ms Jr
2068.0D)
o15sA0E.I)
14.1 7 _
2AFF.7~)
12.1 315-
6AC.~7~)
14P.10)
138.10)
84
13.170(D)
- 4C.725(A)
14U.1300A)
16AL000(D)
407.70)
11.7
-
20AF.6 )
4Y.775(D)
42
0.0
208.410(A)
t¶
0.0
0.0
34-
4K~OO(D)
~~~~~3.7
10I49A(
88F112(ND)
4G.120A)
MC.650)
-2056140(D)
1
1230A)
MEX 60(D)
14P.1475(A)
-
17X1000(A
IS-V.60(D)
8H44()
L1Z.io10A)
- 20V.7SO(D)
17AA.0A)
V.325A
1
eS37(A)
-
3.3-h
1-A.1275(A)
4.4- --IOMF.1470(D)
4A1040(A)
2.5-
12
0.0
337
)
B.12OA
s 14CL70)
-\ 20H.625(AD)
M700(D)
SI.A)
6F.1430~D)
3.6D)
10A163(D
LA XI
22
19-
2N.5A
6L14M00)
L
L. XXII
L0 XXI
3AF.50(A)
17Y.1
:
-OD
_ 5SK~)
4A1220(D)
-
7.0
4.6
3.1
, AC.1200(A)
SNIAMD)
5.6-
LGO x
LGXX
10.7-
1.0 _
3.5
_
4.8
4.3
16.4
_ SD.1000(D)
-
-7J.70O(D)
16.6
0.0
2AD.11M6(A)
-12L600(A)
2AFA460(A)
2N.6000)
324
SW.15~0(A)
-AA620(D)
7.6
LO XIX
LA XVIII
&7-
25.5
eG.1300(D)
34
1.1
LOX1
La XVl
LG XV
XIX) (7), silent heartbl"9 (LG XXVII), and
cyclopsbl6 (LG XII) (5). Linkage analysis
revealed that the cyclopsbl6 mutation is
associated with a chromosomal rearrangement that obscures the precise location of
cyclops with respect to the other markers
shown. The map showed that the homologs
of two genes that are linked 15 cM apart on
chromosome 17 in mice [MHC class II gene
H-2Ab (15) and Brachyuryj (23) are unlinked in zebrafish [MHC class II locus
(DAB) (15) and no tail (7)1 (Fig. 1).
Our results indicate that zebrafish embryonic lethal mutations segregating in various genetic backgrounds can be integrated
quickly into the linkage map by simply
mating a mutant stock to either the DAR or
AB lines (or both) and performing haploidbased bulked segregant analysis as described
here. A screen with RAPD primers of a
number of zebrafish stocks, including the
Singapore, Hong Kong, and German lines,
showed that most lines had many bands in
common with line AB (9), which suggests
that most of the loci described here will
segregate in typical crosses. The mapping of
most of the published SSRs (Fig. 1) further
demonstrates that this type of molecular
marker is also easily integrated into the
current linkage map.
Haploid-based bulked segregant analysis
genetic distance of about 8 cM. Because
RAPD marker 1 6A. 1 190 had already been
localized to LG XIII (Fig. 1), brass must also
be localized to LG XIII. Another DNA
fragment amplified by the same primer
(fragment 16AI.900) (Fig. 2B) was unlinked to brass (39 recombinants out of 76
haploids) and maps to LG XX (Fig. 1).
Primer 13A1 amplified marker 13AI.800,
which had 0 recombinants with brass out of
76 haploids. Because 13AI.800 did not
segregate in the original linkage map cross
(Fig. 1), however, it was not informative
for LG assignment. To verify the location
of brass and to make a more complete local
map (Fig. 2C), we scored the 76 haploid
siblings segregating for brass for other nearby markers in LG XIII. All LG XIII markers
that we tested and that were segregating in
the cross were found to be linked to brass.
The combination of the original linkage
map and the local map gave a composite
map (Fig. 2C).
With the use of this general strategy, we
mapped a total of nine loci identified by
mutation. These include the visible mutations leopard (LG I) (27), brassb2 (LG XIII)
(25), sparseb5 (LG XX) (25), and albinoN
(LG XXI) (25), as well as the embryonic
lethal mutations floating headn` (LG XIII)
(28), throblessb212 (LG VIII), no tailbI60 (LG
LA XXV
LO XXV
138.16
IA.135
-8 SRISAD)
)
14.1 -
50.050(
1SV.1000(A)
SSRO1I(D)
MS.4MA)
1.6)
14U
10.5
hoW
X.0(D)
3.0
9.8
LG XXVII
3-
3.6
S
4.7
7
MC.
A)
1J0(A
-
450()
as - -6F.775A
13y.D)
AMH.120D
-13A115s(A)
4.2
162 -
Lo XXiI
-
-2AD.O0)
82
-13B.100(D)
14.0
21.5-
1.3 _~
1.3
0.0
13.1
QO-
1N.~)
8.52s(D)
15~11D)
SY.350A
1AD.1150A)
7NM5(D)
1-7
LA )XX
17G.W)
87
O44D)
-88R1I~D)
10 mpr unh
4K460(D)
7J.1000()
A4K760
aA.~)
Fig 1. (Continued).
SCIENCE
*
VOL. 264
*
29 APRIL 1994
701
Downloaded from www.sciencemag.org on September 20, 2011
.I'.l".".
. .1. . . . . '.,. 111
A 11Al 12A1
L
13A1 14A1 15AI 16Ai1.
L.
L.
L-
L.
initiation of chromosome walks (,29).
The genetic map (Fig. 1) perrmits the
analysis of chromosome rearranLgements,
which can expedite cloning experiments.
For example, Talbot (29) has shiown that
the cyclopsb213 mutation is a tranLslocation
between LG II and LG XII. Meapping of
noncomplementing mutations thaIt produce
related but distinct phenotypes can also
help test hypotheses regarding the ir genetic
basis. In addition, analysis of episttatic gene
interactions requires distinguishirig double
and triple mutants from single mlatant embryos. Ambiguity can be resolve d by the
extraction of DNA from pieces off the c>.t
dal fin of breeding adults or frorn a parts of
experimental embryos and then b)y assay for
RAPD markers closely linked andA flanking
the mutation.
Such genomic analysis of fish nnay prove
useful in investigations of mamnialian genomes. Genetic linkage relations in ma
mals have been locally conserved over distances averaging up to 8 cM dturing 100
million years of mammalian dlivergence
(23). Linkage analysis of fish and amphibian genomes has lagged behind, b)ut studies
have revealed linkage relations that have
survived 400 million years of vertebrate
evolution (32), which suggests that the
primitive vertebrate gene arrangeiment may
have been largely preserved durinj gfish evolution. Brenner and co-workers (33) have
patsofIau
17A1 18A1 19A1 20A
L.
h
'
C
h
DAR X AB
Composite
DAR X brass
-16A1.1190 -16Al.1190 -16Al.1190
1
2 3 4
13AJ.800
5
~13A1.800
brass
B 16AI
-
5Y.450
-
1V.800
brass
, 5Y.450
Il
- 1V.800
- 1V.800
-
1 2 3 4 5 6 7 8 9 10 1112 131415 16 171819 202122
5 map uinits
Fig. 2. Mapping the brass locus. We collected haploid embryos from a heterozygous brass/DAR
female, scored their pigment patterns, and extracted their DNA individually. (A) DNA saimples from
20 phenotypically brass (b) haploids were pooled, as was DNA from 20 of their wild-tyl,pe siblings
(+). DNA samples from the two pools were amplified with 48 different RAPD primers, 110 of which
are shown (lanes 1 to 4, 6 to 13, and 15 to 22; lanes 5 and 14 are 1 00-bp size standards; ;the bright,
third band from the bottom of the gel is 600 bp). Most fragments appeared in both pools, but
fragments 1 3A1.800 and 1 6A1.1 190 and a few others seemed to be present only in one phenotypic
pool or the other, as expected for DNA fragments from closely linked genetic markers ('26). (B) To
determine if marker 16A1. 1190 was closely linked to brass, we scored the DNA of 38 birass (brs+ )
and 38 wild-type haploid siblings (brs-) from a brass/DAR mother for the ability to .amplify the
1 6A1.1 190 fragment (upper arrowhead); the results from 20 of these haploids are shown here (lanes
2 to 21). The phenotype of the embryo in lane 20 reflects a recombination event boietween the
16A1. 1190 marker and the brass locus. Marker 16A1.900 (lower arrowhead) assorts indlependently
of brass. Similar experiments were conducted with the other candidates identified in the bulked
segregant analysis experiments. (C) To integrate brass on the linkage map, we scorecJ in haploid
offspring of the brass/DAR mother other markers previously shown (Fig. 1) to be linked to 16A1. 1190.
Marker 1V.800 showed 6 recombinants with brass out of 60 haploids. Marker 5Y4150 did not
segregate in the cross. The local map (DAR x brass) constructed from these data was in corporated
into the larger linkage map with markers segregating in both crosses.
702
suggested that the ordering of genes in a
fish may prove useful in the analysis of the
human genome. Because one can systematically collect embryonic lethal mutations in zebrafish (2) [some of which are
phenotypically similar to mutations in the
homologous genes in mammals (7)] and
because the methods reported here will
facilitate map-based moleular isolation of
these genes, studies of zebrafish embryos
are likely to further our understanding of
developmental genetic mechanisms conserved among all vertebrates, including
humans.
SCIENCE * VOL. 264 * 29 APRIL 1994
REFERENCES AND NOTES
1. A.
Meyer, C. H. Biermann, G. Orti, Proc. R. Soc.
London Ser. B 252, 231 (1993).
2. C. Kimmel, Trends Genet. 5, 283 (1989); J.
Rossant and N. Hopkins, Genes Dev. 6,1 (1992);
M. C. Mullins and C. Nusslein-Volhard, Curr. Opin.
Gen. Dev. 3, 648 (1993).
3. G. Streisinger et al., Nature 291, 293 (1981); C.
Walker and G. Streisinger, Genetics 103, 125
(1983); D. J. Grunwald and G. Streisinger, Genet.
Res. 59, 93 (1992); ibid., p. 103.
4. G. Stuart, J. Vielkind, J. McMurray, M. Westerfield,
Development 109, 577 (1990).
5. K. Hatta et a)., Nature 350, 339 (1991).
6. C. B. Kimmel etal., ibid. 337, 358 (1989); R. K. Ho
and D. A. Kane, ibid. 348, 728 (1990); A. L.
Felsenfeld et al., Development 108, 443 (1990).
7. M. E. Halpern etal., Cell75, 99 (1993); S. Schulte-
Merker, F. van Eeden, M. E. Halpern, C. B. Kimmel, C. Nusslein-Volhard, Development, in press;
D. Sepich and M. Westerfield, Soc. Neurosci.
Abstr. 19, 1294 (1993).
8. D. Goff et al., Genomics 14, 200 (1992).
9. S. L. Johnson, C. N. Midson, E. W. Ballinger, J. H.
Postlethwait, ibid. 19, 152 (1994).
10. J. G. K. Williams, A. R. Kubelik, K. J. Livak, J. A.
Rafalski, S. V. Tingey, Nucleic Acids Res. 18,
6531 (1990); J. Welsh and M. McClelland, ibid., p.
7213; S. V. Tingey, J. A. Rafalski, J. G. K. Williams,
in Applications of RAPD Technology to Plant
Breeding (Crop Science Society of America, Minneapolis, 1992), pp. 3-8.
11. D. Neale and R. Sederoff, Probe 1, 1 (1991); R.
Reiter etal., Proc. Nat). Acad. Sci. U.S.A. 89,1477
(1992); S. M. Al-Janabi, R. J. Honeycutt, M. McClelland, B. Sobral, Genetics 134,1249 (1993).
12. Primers and their sequences can be obtained
from Operon Technologies (Alameda, CA). Marker names use the inverse of Operon's primer label
(to avoid confusion with zebrafish allele designations) followed by the apparent size of the amplified fragment [for example, genetic marker
20Y670 (A) results in a 670-bp DNA fragment
amplified from the AB haplotype by Operon primer Y20; a (D) suffix indicates marker origin in the
DAR stock].
13. Haploid embryos were produced by in vitro fertilization. Eggs obtained by gently pressing the
abdomens of anesthetized animals were mixed
with sperm similarly obtained and treated with
ultraviolet light. Such sperm activate embryonic
development but do not make a genetic contribu-
tion to the offspring. Haploids developed until
hatching at 3 days, when their DNA was collected. Several hundred PCR assays can be per-
formed on the DNA from a single haploid embryo.
14. Linker (a program written by M. Hutchinson) compares all possible marker pairs and computes the
frequency of recombinants between pairs, whether they are in a coupling or repulsion phase.
not prethat 40
were
DNA
fragments
Linkerin ignored
chi
to 60%,
(range,
50% of the
samples
sent
square test; P = 0.02). Linker analyzed the data
set pairwise and showed significant linkages (chi
square test; 0.001 > P > 0). The data set was
filtered to include only marker pairs with less than
a 26% frequency of recombinants, which converts
Downloaded from www.sciencemag.org on September 20, 2011
with RAPD markers allows the rapid identification of DNA polymorphisms linked to
either the mutant or wild-type allele of any
gene in zebrafish. In mapping the nine
mutations mentioned above, about 1% of
the primers screened (18 out of 1778) identified a RAPD marker located within about
5 cM of any given mutant locus; thus, only
about 1000 decamer primers should be
needed on average to find a marker about
0.5 cM from any mutation. Some of the
mutant genes studied were very close to
their nearest RAPD marker-for example,
brass failed to recombine with its nearest
marker among 76 haploids and floating head
failed to recombine with its nearest marker
among 1332 haploids (29). A marker 0.5
cM away from a mutation should be on
average only about 300 kb from the mutant
locus, a distance that corresponds to about
four P1 clones and is shorter than a single,
large-insert yeast artificial chromosome
(30). Once closely linked markers are identified, the PCR products can be readily
cloned and sequenced. We have already
cloned more than a third of the codominant
markers plus a number of dominant markers
closely linked to various mutants. Some of
the cloned dominant markers segregate as
codominants in other crosses or can be
converted to codominant markers by restriction enzyme digests (31). The cloned
markers can also be used as probes in the
16.
17.
18.
were 42 such doubles
error or may be bona fide recombination events.
Markers with a high error rate often tend to map
spuriously to the ends of LGs because that loca-
tion causes errors to appear as single crossovers
rather than as double crossovers. In our data set,
terminal intervals were on average a bit longer
(8.7 ± 7.6 cM SD) than the average interval on the
map (5.8 ± 5.9 cM), but the difference was not
statistically significant. A total of 48 markers
(about 12% of the total) on 10 LGs were retested
for LG assignment in a number of crosses in
different genetic backgrounds. The location of all
markers tested was confirmed, indicating the robustness of LG assignments.
35. S. L. Johnson, unpublished results.
36. This paper is dedicated to C. Walker for pioneer-
(1968).
(1992).
3 January 1994; accepted 29 March 1994
Pietro De Togni,* Joseph Goellner, Nancy H. Ruddle,
Philip R. Streeter, Andrea Fick, Sanjeev Mariathasan,
Stacy C. Smith, Rebecca Carlson, Laurie P. Shornick,
Jena Strauss-Schoenberger, John H. Russell,
Robert Karr, David D. Chaplint
e-2Nd/D.
23. N. Copeland et al., Science 262, 57 (1993).
24. C. Wicking and B. Williamson, Trends Genet. 7,
288 (1991).
25. G. Streisinger, F. Singer, C. Walker, D. Knauber,
N. Dower, Genetics 112, 311 (1986). The brass
locus was formerly called golden-2.
26. R. Michelmore et al., Proc. Nati. Acad. Sci. U.S.A.
88, 9828 (1991); J. G. K. Williams et al., Nucleic
Acids Res. 21, 2697 (1993).
27. F. Kirschbaum, Wilhelm Roux's Arch. Dev. Biol.
177, 129 (1975).
28. W. Trevarrow, M. E. Halpern, T. Jowett, Soc.
Neurosci. Abstr. 19, 217 (1993).
29. W. S. Talbot, unpublished results.
30. J. C. Pierce, B. Sauer, N. Sternberg, Proc. Nati.
Acad. Sci. U.S.A. 89, 2056 (1992); F. L. Chartier,
Nat. Genet. 1, 132 (1992).
31. R. V. Kesseli, I. Paran, R. W. Michelmore, in
Applications of RAPD Technology to Plant Breeding (Crop Science Society of America, Minneapolis, 1992), pp. 31-36; J. A. Rafalski and S. B.
Tingey, Trends Genet. 9, 275 (1993); I. Paran and
R. W. Michelmore, Theor. Appl. Genet. 85, 985
(1993).
32. J. D. Graf, Genetics 123, 389 (1989); D. Morizot,
in Isozymes: Structure, Function, and Use in Biology and Medicine, Z.-l. Ogita and C. L. Markert,
Eds. (Liss-Wiley, New York, 1990), pp. 207-234;
D. Morizot, S. Slaugenhaupt, K. Kallman, A.
Chakravarti, Genetics 127, 399 (1991).
33. S. Brenner et al., Nature 366, 265 (1993).
34. To detect potential errors, two people scored gels
independently. Markers with discrepancies greater than 5% were excluded from analysis. In other
cases, discrepancies were resolved by reexamination of the original photograph or by reamplification. Duplicate reactions of some primers involving nearly 1000 data points indicated scoring
error (because of incorrect amplification) of about
1% or less. After linkage groups were ordered
with MAPMAKER, double recombinants in short
(<10 cM) intervals [an indication of error in the
data set; W. Dietrich et al., Genetics 131, 423
(1992)] were identified and rechecked against
original gel pictures or by new reactions. There
ing and continuing work in zebrafish genetics. We
acknowledge support from the following: NIH
grant 1RO1A126734, Medical Research Fund of
Oregon, and American Heart Association Oregon
Affiliate (J.H.P.); NIH grant HD07470 (S.L.J.); NIH
grant NS23915 and Research Career Development Award grant NS01476 (J.S.E.); and the
University of Oregon Zebrafish Program Project
grant 1 PO1 HD22486 (J. Weston). W.S.T. is a
fellow of the Jane Coffin Childs Memorial Fund for
Medical Research. We thank R. Sederoff and R.
Lande for discussions; M. Halpern, E. Selker, G.
Sprague, and M. Westerfield for helpful comments on the manuscript; and W. Potts for the
MHC primers. Linker and the mapping stocks are
available (AB from C.B.K. and a partially inbred
Darjeeling stock from S.L.J.).
Abnormal Development of Peripheral Lymphoid
Organs in Mice Deficient in Lymphotoxin
19. H. Jacob etal., Cell 67, 213 (1991). For Nmarkers
in a genome of size D, the proportion of the
genome within d cM of at least one marker is 1 20. S. H. Hulbert et al., Genetics 120, 947 (1987). The
fraction of marker pairs less than x cM apart is
equal to 2x divided by the genome size. For x =
20 cM, the genome size is estimated to be 2570
cM; for x = 15 cM, it is 2510 cM.
21. R. Hinegardner and D. E. Rosen, Am. Nat. 166,
621 (1972); M. D. Bennett and J. B. Smith, Philos.
Trans. R. Soc. London Ser. B 274, 227 (1976).
22. L. D. Brooks and R. W. Marks, Genetics 114, 525
(1986); S. D. Tanksley et al., ibid. 132, 1141
remaining out of about
39,000 data points [94 haploids x (401 RAPDs +
13 SSRs + 3 cloned genes)]. These may reflect
Mice rendered deficient in lymphotoxin (LT) by gene targeting in embryonic stem cells have
morphologically detectable lymph nodes or Peyer's patches, although development of
the thymus appears normal. Within the white pulp of the spleen, there is failure of normal
segregation of B and T cells. Spleen and peripheral blood contain CD4+CD8- and
CD4-CD81 T cells in a normal ratio, and both T cell subsets have an apparently normal
lytic function. Lymphocytes positive for immunoglobulin M are present in increased numbers in both the spleen and peripheral blood. These data suggest an essential role for LT
in the normal development of peripheral lymphoid organs.
no
Lymphotoxin (LT, also designated TNF,) is a soluble product of activated lymphocytes that was first defined by its cytotoxic
activity against fibroblasts (1-3). LT is now
recognized to be produced by activated
CD4' T helper cell type 1 (THi) lymphocytes, CD8+ lymphocytes, and certain B
lymphoblastoid and monocytoid cell lines
(4-6). The gene encoding murine LT is
located 1100 base pairs (bp) upstream of the
evolutionarily related gene encoding tumor
P. De Togni, S. C. Smith, L. P. Shornick, Department of
Internal Medicine, Washington University School of
Medicine, St. Louis, MO 63110, USA.
J. Goellner, P. R. Streeter, J. Strauss-Schoenberger,
R. Karr, Department of Immunology, Monsanto Company, St. Louis, MO 63198, USA.
N. H. Ruddle, Department of Epidemiology and Public
Health, Yale University School of Medicine, New Haven, CT 06510, USA.
A. Fick, S. Mariathasan, R. Carlson, D. D. Chaplin,
Howard Hughes Medical Institute, Washington University School of Medicine, St. Louis, MO 63110, USA.
J. H. Russell, Department of Molecular Biology and
Pharmacology, Washington University School of Medicine, St. Louis, MO 63110, USA.
*Present address: Department of Pathology, University of Arkansas for Medical Sciences and John McClellan Veterans' Hospital, Little Rock, AR 72205, USA.
tTo whom correspondence should be addressed.
SCIENCE
*
VOL. 264 * 29 APRIL 1994
necrosis factor-a (TNF-a) within the major histocompatibility complex (MHC) (7,
8). The LT and TNF-a proteins are structurally related and show similar activities in
vitro and when given to experimental animals (9). In solution, LT is a homotrimer
with a conformation similar to that of
TNF-a. A membrane-associated form of LT
has been described, consisting of a heterotrimeric complex containing two LT monomers together with a 33-kD transmembrane
protein designated LT-P (10). The gene
encoding LT-I is located immediately centromeric to the gene encoding TNF-a. The
biological effects of LT and TNF-a are
mediated by two receptors, designated p55
and p75 (11).
In vitro, LT and TNF-a can modulate
many immune and inflammatory functions.
As implied by their names, both cytokines
are cytotoxic for a variety of transformed
and normal cell types (12-14). In order to
be killed, target cells must express LTTNF-a receptors, with the p55 receptor
appearing to mediate the cytotoxic response
(15). LT and TNF-a can augment the
proliferation of activated thymocytes (16)
703
Downloaded from www.sciencemag.org on September 20, 2011
15.
to about 28.8 cM with the use of the Kosambi
mapping function [D. D. Kosambi, Ann. Eugen.
12, 172 (1944)]. MAPMAKER [E. S. Lander et al.,
Genomics 1, 185 (1987); modified by S. Tingey for
the Macintosh] allows use of either the Haldane
[J. B. S. Haldane, J. Genet. 8, 299 (1919)] or the
Kosambi mapping function. Because the Kosambi function assumes a greater amount of
positive interference [which occurs in zebrafish
(3)], it was selected. For analysis in MAPMAKER,
markers in repulsion were inverted. MAPMAKER
provided scores of the logarithm of the likelihood
ratio for linkage (lod scores; all linkages in this
map were supported by a lod score of >3) and
converted the frequency of recombinants into
map distance in centimorgans.
H. Ono et al., Proc. Nati. Acad. Sci. U.S.A. 89,
1 1886 (1992); L. M. Silver et al., Mamm. Genome
3, S241 (1992).
C. Thisse, B. Thisse, T. Schilling, J. H. Postlethwait,
Development 1 19, 1203 (1993); M. Hammerschmidt
and C. Nusslein-Volhardt, ibid., p. 1107.
M.-A. Akimenko, S. L. Johnson, M. Westerfield, M.
Ekker, unpublished results.
A. Endo and T. H. Ingalls, J. Hered. 59, 382