Download Library subtraction of in vitro cDNA libraries to identify differentially

Nucleic Acids Research, Vol. 18, No. 9 2789
Library subtraction of in vitro cDNA libraries to identify
differentially expressed genes in scrapie infection
John R.Duguid 123 and Mary C.Dinauer45
1
Geriatric Research, Education and Clinical Center, Edith N.Rogers Memorial Veterans Hospital,
Bedford, MA 01730, 2 Departments of Biochemistry and 3Neurology, Boston University School of
Medicine, Boston, MA 02118, 4Division of Hematology-Oncology, Children's Hospital, Boston, MA
02115 and 5 Dana-Farber Cancer Institute, Boston, MA 02115, USA
Received September 25, 1989, Revised and Accepted December 4, 1989
EMBL accession nos. X17002, M29894,
M29895
ABSTRACT
We have developed a system where double-stranded
cDNA can be amplified using a synthetic
ollgonucleotide primer and the polymerase chain
reaction, generating cDNA libraries In vitro. Using a
library subtraction strategy (1), scrapie and control
brain In vitro cDNA libraries were used to identify
sequences whose expression is modulated In scrapie
infection. One of these sequences represents /3-2
microglobulin, while the other two have not been
previously described. The use of in vitro libraries offers
increased speed and efficiency of construction, and
their subtraction is more efficient and powerful,
compared with the previous system (1).
INTRODUCTION
Scrapie is a transmissible neurodegenerative disease of sheep and
goats, and has become the experimental prototype of the
spongiform encephalopathies, manifest in humans as kuru and
Creutzfeld-Jakob disease (2,3,4). While the nature of the scrapie
agent is controversial (2,3,4), one possibility is that the agent
lacks a nucleic acid genome, as suggested by Prusiner in the pnon
hypothesis (5).
Identifying the changes in brain gene expression that occur in
scrapie might contribute to the understanding of the pathogenesis
of this condition. A library subtraction strategy has recently been
used to isolate several genes whose transcription products are
increased in scrapie infected hamster brain (1,6). This strategy
was based on the generation of single-stranded cDNA libraries
from scrapie and control brains, and the subtractive hybridization
of these cDNA libraries to generate a subtracted library enriched
for scrapie-specific sequences. In this study we ligated a duplex
oligonucleotide, referred to as an oligovector, to double-stranded
cDNA and used the polymerase chain reaction to amplify the
sequences, generating cDNA libraries in vitro. These libraries
are well suited to the library subtraction strategy described
previously (1), and were used to identify three additional genes
whose expression is significantly increased in scrapie infected
brain. These may reflect the gene expression of a cell type with
increased prevalence in the scrapie infected brain (e.g. fibrous
astrocytes) or a change in gene expression in cells present in the
normal brain caused by the disease process.
MATERIALS AND METHODS
Library construction—The hamster brain tissue, RNA preparation
and cDNA synthesis have been previously described (1). Two
oligonucleotides, dCTCTTGCTTGAATTCGGACTA and
dTAGTCCGAATTCAAGCAAGAGCACA were synthesized
using an Applied Biosystems apparatus. After phosphorylation
using T4 polynucleotide kinase (New England Biolabs, Beverly,
MA), the two oligonucleotides (1 mg/ml) were hybridized in the
kinase buffer at 45° for 10 min (7). This yielded the duplex
oligovector, which was blunt on one end, had a four nucleotide
3' overhang on the other end, and had an EcoRl restriction site
7 nucleotides from the blunt end. The ds-cDNA from 1.5 /*g
mRNA was phenol extracted and precipitated with ethanol, then
mixed with 5 /ig of the oligovector and incubated with 5 units
of T4 ligase (New England Biolabs) as described (7) in a 100
/tl reaction for 8 hrs at 20°. The reaction product was fractionated
to remove excess oligovector by centrifugation through a 5 -20%
potassium acetate gradient containing 1 /tg/ml ethidium bromide
(7) for 4 hrs at 60,000 rpm using a Sorval TST 60.4 rotor. The
cDNA was removed from the bottom of the tube using a
hypodermic needle, leaving the oligovectors, visualized with UV
light, behind. The cDNA was precipitated with ethanol. Onehalf of the product was amplified with Taq polymerase through
20 cycles in a 100 /tl reaction containing 1 fig of the 21-mer
ollgonucleotide specified above, using a Perkin-Elmer/Cetus
(Norwalk, CT) thermal cycler as described by the manufacturer,
using the following parameters: denaturation—30 sec, 94°;
annealing-30 sec, 50°; elongation-120 sec, 72°. cDNA libraries
from both control and scrapie-infected brain were constructed
and amplified in this manner and represent the starting material
used in the steps below.
Library subtraction—The strategy used in the preparation of subtracted libraries is presented in Figure 1. 60 ng of the normal
hamster brain library DNA was amplified in 600 fil reaction mix
as described above, yielding approximately 30 /ig of DNA. The
DNA was digested with 400 units of EcoRl (New England
2790 Nucleic Acids Research, Vol. 18, No. 9
Biolabs) for 2 hrs to cleave the oligovector and reduce the
amplification potential of the normal library DNA After phenol
extraction and ethanol precipitation, the DNA was dissolved in
50 /tl 5 mM HEPES, 1 mM EDTA, pH 7.5, and boiled for 5
min. The DNA was cooled and mixed with 50 /tl photo-probe
biotin (Vector Labs, Burlingame, CA), irradiated with a sunlamp,
and processed as described (1). 15 /tg of the biotinylated normal
library DNA was mixed with 2.5 /tg of scrapie-infected brain
library DNA, denatured and hybridized, after which the mixture
was subjected to a stringency incubation and applied to a biocytin
affinity resin (1). Alternatively, biotinylated molecules were
removed with streptavidin and phenol extraction (8), with similar
efficiency (95%). The resulting subtracted DNA was ethanol
precipitated and hybridized with another 15 /tl biotinylated normal
library DNA and the cycle repeated. The subtracted DNA was
amplified through 15 cycles as described above, in a 100 /xl
reaction. 2.5 /tg of the subtracted and amplified scrapie library
DNA was subjected to two cycles of hybridization and
substraction as described above, and then amplified. The process
was repeated for a total of 6 hybridization/subtraction cycles,
with amplification of the product after each two of these cycles.
In a similar manner, the normal library was subtracted with itself,
to generate a subtracted library enriched for low abundance
sequences, which would serve as a source of nucleic acid to
generate probe for the minus arm in plus/minus screening.
Library screening—250 ng of the subtracted scrapie library
DNA was amplified through 3 cycles as described, in a 100 /tl
reaction. This step was included to assure that the final product
was fully double-stranded, with blunt ends. The product was
cleaved with EcoRl (20 units in a 100 /tl reaction, for 2 hrs)
and subjected to phenol extraction and ethanol precipitation. The
product was ligated with 1 /tg BstXl-EcoRl adapter (the duplex
was composed of dCTGGCGGG and dAATTCCCGCCAGCACA) in a 100 y\ reaction, as described above The
product was purified by potassium acetate gradient centnfugation
to remove excess adapters, and ethanol precipitated. One-tenth
of the tailed cDNA was ligated with 100 ng BstXl -cleaved pH3M
(9) in a 100 /tl reaction, and used to transform competent FG-2
cells (1), yielding approximately 20,000 colonies. The colonies
(2,000/10 cm plate) were transferred to GeneScreen Plus (New
England Nuclear, Boston, MA) (10). Probes were generated from
the subtracted libraries (100 ng heat denatured DNA in a 100
/tl reaction) using MMLV reverse transcriptase (BRL,
Gaithersburg, MD) and the 21-mer component of the oligovector
as a primer (1). The colony replicas were hybridized with the
normal probe, to which was added probes representing the five
recombinants previously isolated (1,6), washed, and
autoradiographed for 24 hrs. The filters were then hybridized
with the scrapie probe. The resulting autoradiographs were
aligned and differentially expressed recombinants were picked
and colony purified. These recombinants were characterized by
DNA sequencing and their transcripts were identified by RNA
blot analysis.
RESULTS
We constructed scrapie-infected and normal brain cDNA libraries
in vitro by bracketing each double-stranded cDNA sequence with
an oligovector, a duplex oligonucleotide that is blunt on one end
and is staggered on the other end. The oligovector ligates to each
end of a given recombmant through its blunt end, so that both
strands of a given recombinant could be amplified using a single
scnptotorary
DNA
hytrkfiz.»«i
control ftwvy
DNA
tmpKy tnrictud
(subtracted In vttm
Ifcrwy) with
pen
l0fts into ptesnld
vsclor to Q H i n t i
subtracted in »4K>
•rcy
Figure 1. The strategy for the preparation of subtracted libraries The steps in
the strategy are numbered The hybridization and subtraction steps were repeated
in this study, pnor to PCR amplification, following the sequence 1 - 2 - 3 - 2 - 4
The subtracted, amplified DNA was then subtracted again, e g 5 - 2 - 3 - 2 - 4
This process was repeated once more prior to ligation of the subtracted sequences
into plasmid xH3M, for transfection and colony hybridization
oligonucleotide pnmer (one of the strands of the oligovector) in
the polymerase chain reaction. An in vitro subtracted library,
enriched for scrapie specific sequences, was generated by
repetitive hybridization and subtraction of the scrapie library using
biotinylated control library DNA (see Figure 1). The ohgovectors
bracketing each of the recombinant sequences in the subtracted
library were cleaved with EcoRl, and then ligated with a
EcoRl-BstXl adapter (see Methods). The in vitro subtracted
library sequences were then ligated into the BstXl site of the
plasmid vector TTH3M (9) for colony hybridization.
The probes used for plus/minus screening were generated from
the scrapie and control libraries themselves, after removal of
abundant sequences by subtractive hybridization. These probes
were used directly for colony hybridization, whereas probes
generated from inserts derived from in vivo libraries (6) were
contaminated by heterodisperse vector sequences which had to
be removed prior to colony hybridization to avoid false positive
signals.
Seven recombinants were isolated which hybridized
significantly more strongly with the scrapie library probe than
with the control library probe. Tworecombinantswere sequenced
and found to represent glial fibnllary acidic protein and
transferrin, genes which had been previously identified (1,6).
Although these previously identified recombinants were included
in the control probe for colony hybridization, the two new isolates
represented regions of these genes not contained in the original
isolates, which explained their detection here.
Three recombinants were found to share sequence similarity
Nucleic Acids Research, Vol. 18, No. 9 2791
Murine p*-2 Microglobulin
P
V
M
251
M
S
I
L
T
I II I I I I I I I I II II I II II II I I I II II I I I I I I I I I I I I I I I II I II II II II
N
D
V
L
Hamster Recombinant
Murine P-2 Microglobulin
K H D S
351
M
L
L
E
F
T
II II
R
E
F
E
lOfl
I I I II I I I I I I I I I I
V
T
I I I I I I II
CCCACTGCGACTGATAAJ
A
K
A
II I I I I I I I I I I I I
I I II II II
II I I I I IIIIIIII I
I I I I II II I I I I I II I I II I I I I ! II I II II
GCAGAGTTTCACACATAACTCTGAAGGAGCCCAAGGTGGTCACCTGGGAACGAGACATGTAATCAAGCGTCATG. . . . .GAGATGCTTCATTTGGATCAG
S H I T L K
V
T
E
*
Hamster Recombinant
Figure 2. The nucleic acid sequence of the isolated recombinant. The sequence of munne 0-2 nucroglobulin (22) is presented (upper sequence), aligned with the
recombinant (accession number X17002) The translated arruno acids are listed at points of discrepancy between the two nucleic aad sequences The coding sequence
for P-2 microglobulin extends to residue 409 and the termination codon is indicated by the asterisk
with /3-2 microglobulin. The partial sequence of one of these
recombinants, 478 base pairs in length, is shown in Figure 2,
aligned with the sequence for murine /S-2 microglobulin. These
sequences share a 77 % identity in the coding region; the translated
amino acid sequences share a 67% identity, with one-third of
the discrepancies representing conservative substitutions. RNA
blot analysis identified transcripts with three different lengths,
all of which were considerably more abundant in the scrapie
samples than in the controls (see Figure 3, panel A).
Two additional differentially expressed recombinants were
isolated and sequenced (accession numbers M29894 and
M29895). These sequences did not contain an open reading frame
and were not identified in either the GenBank (Release 60.0) or
EMBL (Release 19.0) databases. RNA blot analysis showed that
the transcripts of both of these sequences were significantly
modulated in scrapie infected brain, though they are detectable
in the control RNAs, indicating that they represent host genes
(see Figure 3, panels B and C). The intensity of the signals
indicated that the 0-2 microglobulin mRNAs and the 1.5 kb
transcript identified in the panel B had abundances similar to the
previously identified genes (1,6). The transcript identified in panel
C had a significantly lower abundance. The identity of these last
two recombinants is under further study.
1 2 3 4 5 6 7
DISCUSSION
The polymerase chain reaction has brought considerable power
to the isolation of specific nucleic acid sequences (11,12,13).
Belyavsky et al. have extended the scope of this strategy by using
it to generate cDNA libraries from very small amounts of material
(14). Kinzler and Vogelstern have used a similar strategy to purify
specific genomic DNA sequences recognized by DNA binding
proteins (15). We have applied the polymerase chain reaction
strategy to subtractive cloning, and have found that it is
particularly well suited for use in a library subtraction strategy.
This results from the ease with which in vitro libraries can be
constructed, manipulated and amplified. Furthermore, a broader
representation of sequences can be expected because the
inefficient step of bacterial transformation is avoided. The
sequences of interest are enriched by exhaustive subtraction of
the in vitro libraries. Subsequent amplification of the enriched
sequences prior to transformation permits the representation of
even very rare sequences in the final in vivo library.
A further advantage of the oligovector based system is the ease
Figure 3. Identification of the transcripts of the isolated cDNA recombinants
RNA blot analysis using the recombinant inserts as probe were performed using
5 (ig total brain RNA from four scrapie infected animals (lanes 1 —4) and 3 agematched controls (lanes 5—7) The 0-2 microglobulin probe (panel A) identified
transcripts of 1.5 kb, 0 85 kb and 0 6 kb The lengths of the major transcripts
identified by the other two recombinants were 1.5 kb for accession number M29S9S
(panel B) and 2 6 kb for accession number M29894 (panel C) Molecular weight
standards were obtained from BRL (not shown). Several randomly chosen
recombinants yielded signals of equal intensity in the scrapie and control RNA
preparations, as a control for RNA loading and integrity (not shown)
2792 Nucleic Acids Research, Vol. 18, No. 9
with which probe can be generated from a library. The
oligonucleotide used for polymerase chain reaction amplification
was used to prime the synthesis of high specific activity probe
from the in vitro library using reverse transcriptase. In this study
we generated probes from scrapie and control libraries that had
been subtracted and enriched for low abundance sequences; these
probes had a reduced sequence complexity and would be more
sensitive in the detection of low abundance recombinants. Indeed,
one of the modulated sequences we identified appears to be
considerably less abundant than those previously identified using
a less powerful detection system (1).
There are limitations to this strategy. Not all sequences will
be amplified with an equal efficiency and some sequences will
be underrepresented in the in vitro library (16), although this is
also a problem with both phage and plasmid based libraries. Our
experience is similar to that of others (14) who find that short
sequences are preferentially represented in the amplified product.
Further, the fidelity of replication using the polymerase chain
reaction has been shown to be less than the fidelity of replication
in vivo (13). However, these potential problems can be
circumvented; a recombinant isolated using in vitro libraries can
be used to identify its full-length cognate from a conventional
library.
The utility of the system was demonstrated by the identification
of three genes whose transcripts were significantly increased in
scrapie infected brain, but which were not detected in previous
studies (1,6). One of the sequences we identified, /3-2
microglobulin, is associated with the major histocompatibility
complex Class I antigen, and its concentration is increased in
the cerebrospinal fluid in many malignant and infectious processes
(17,18,19). In particular, /3-2 microglobulin expression is
increased in conventional viral encephalitides, accompanying the
inflammatory response in these conditions (20,21). It is of interest
that scrapie infection causes increased 0-2 microglobulin
expression. Although a pathological hallmark of this condition
is the absence of signs of inflammation or immune response (2),
the finding of increased /3-2 microglobulin expression suggests
that there is an active response to scrapie infection that is not
apparent histopathologically.
ACKNOWLEDGEMENTS
We would like to thank Brian Seed and Stuart Orkin for helpful
comments, and Maryanne Lemaire for technical assistance. This
work was supported by grants from the Veterans Administration
and the Whitaker Health Sciences Fund.
REFERENCES
1. DuguidJ.R , Rohwer.R G and Seed.B. (1988) Proc. Nail Acad. Sa. USA
85, 5738-5742
2. Gajdusek,D C (1977) Science 197, 943-960
3. Carp,RI., Merz.P A , Kascsak.RJ., Merz.G S. and Wisniewski.H.M
(1985)7 Gen. Virol. 66, 1357-1368
4. Pnisiner.S B. (1987) Annu. Rev Med. 38, 381-398
5. Prusincr.S B. (1982) Science 216, 136-144.
6. DuguxU R , Bohmort.C, Lhi.N and TourteUotte, W.W. (1989) Proc NatL
Acad. Sci. USA 86, 7260-7264.
7. Maniatis.T., Fritsch.E.F. and SambrookJ. (1982) Molecular Cloning- A
Laboratory Manual (Cold Spring Harbor Laboratory, Cold Spring Harbor,
NY).
8. Sive.H.L and Si John,T. (1988) Nucl. Acids Res. 16, 10937
9. Aniffo.A. and Seed.B (1987) Proc Nml Acad. Sa. USA 84, 5937-5941.
10 Granstein.M and Hogness.D (1975) Proc Nail Acad. Sa USA 72,
3961 -3965.
11
12
13
14
15
16
17.
18
19.
20
21
22
Saiki.R K , Scharf.S , Faloona.F el al (1985) Science 230, 1350-1354
Scharf.S , Horn.G.T and Eriich.H A (1986) Science 233, 1076-1078
Saiki.R K , Gelfand.D.H , Stoffel.S a al. (1988) Science 239, 487-491
Belyavsky,A., Vinogradova.T. and Rajewsky,K (1989) Nucl Acids Res.
17, 2919-2932
Kmzler.KW and Vogelstein.B (1989) Nucl Acids Res 17,3645-3654
McConlogue.L., Brown.M.A D and Innis.M.A (1988) Nucl Aads Res.
16, 9869
Mavhgrt.G M , Stuclcey.S E , Cabamllas, F F et al (1980) New Engl J
Med 303, 718-722.
Kodi.T R , Lichtenfeld.K M. and Wiemik.P H. (1983) Cancer 52, 101-104
StarmansJ J.P , Vos.J and van der Helm.H.J (1977) J. Neurol Sci 33
45-49
Sobel.R A , Collins.A B , Colvin.R B and Bhan.A K (1986) Am. J. Pathol
125, 332-338.
Brew.B J , Bhalla.R B , Fleisher.M et al (1989) Neurology 39, 830-834.
Damel.F , MoreUo.D , Le Bad.O , Chambon,P , Cayre.Y and Kounlsky.P.
(1983) EMBO J. 2, 1061-1065