Gene, 121 (1992) 383-386 0 1992 Elsevier Science Publishers

GENE

B.V. All rights reserved.

383

0378-1119/92/$05.00

06754

A rat brain-derived neurotrophic factor-encoding gene generates multiple transcripts through alternative use of Sexons and polyadenylation sites (Splicing;

intron;

transcription;

Osamu Ohara, Yoshinari Shionogi Research Laboratories, Received

by Y. Sakaki:

sequencing;

Gahara,

nerve growth factor; polymerase

Hiroshi Teraoka

Shionogi & Co.. Ltd. 5-12-4 Sag&

13 May 1992; Revised/Accepted:

and Tadahisa

Fukushima-ku,

chain reaction)

Kitamura

Osaka 553. Japan

4 July/l 1 July 1992; Received

at publishers:

27 July 1992

SUMMARY

As a first step toward clarification of the transcriptional controls of the gene encoding brain-derived neurotrophic factor (BDNF), we cloned and sequenced a rat genomic DNA fragment carrying this gene. RNA blotting analysis using a probe derived from the 3’-flanking region of BDNF revealed that alternative use of 3’-polyadenylation sites generates at least two BDNF transcripts that differ in the size of the 3’-noncoding region. Furthermore, sequence analysis of the 5’-end of the BDNF cDNA revealed the presence of at least six different types of transcripts which were probably derived through alternative use of the multiple 5’-exons. Therefore, a single BDNF gene could produce multiple types of transcripts with different noncoding sequences through alternative use of both 5’-exons and 3’-transcription termination sites.

INTRODUCTION

The brain-derived neurotrophic factor (BDNF) is one of the best characterized mammalian neurotrophic factors in the nerve growth factor family. It is now evident that the function and production pattern of BDNF are distinguishable from those of other members of the nerve growth factor family. The next important issue to be addressed is the molecular regulatory mechanism of BDNF expression in the brain. This requires characterization of the BDNF transcript present in the brain as basic information, since not enough is known about it mainly because of its scar-

Correspondence to: Dr. 0. Ohara, nogi & Co., Ltd., Tel. (81-6)458-5861;

5-12-4

Shionogi

Sagisu,

Research

Fukushima-ku,

Laboratories, Osaka

Shio-

553, Japan.

Fax (81-6)458-0987.

Abbreviations: BDNF, brain-derived neurotrophic factor; BDNF, gene (DNA) encoding BDNF; bp, base pair(s); kb, kilobase or 1000 bp; nt, nucleotide(s);

PCR, polymerase

chain reaction.

city. In this study, we utilized a one-sided PCR strategy to determine the 5’-end structure of the transcript and were able to solve the problems related to the sensitivity. Here we describe the results of analysis of BDNF transcripts in the rat brain, which indicate that multiple BDNF transcripts with different noncoding sequences were generated through alternative use of 5’-exons and polyadenylation sites from a single BDNF gene.

EXPERIMENTAL

AND DISCUSSION

(a) Cloning and sequencing of the genomic BDNF fragment In preliminary genomic DNA blotting analyses using a mouse BDNF probe (prepared by PCR), we found that BamHI digestion of rat genomic DNA generated a 4-kb DNA fragment containing BDNF. Thus, a sub-genomic plasmid library was constructed with the BamHI-digested genomic DNA fragment (about 4 kb; size-selected on an agarose gel) and screened by colony hybridization using the mouse BDNF probe. This 4-kb BDNF genomic fragment

384 thus cloned was entirely sequenced by the dideoxy chaintermination method (Fig. 1A). Although the same fragment had already been cloned by Maisonpierre et al. (1991), they did not report on the nt sequence of the flanking regions of the BDNF gene. Using the sequence information of these noncoding regions, we tried to characterize the structure of the BDNF transcript in detail as described below.

(A)

5’

mllpqjmllH

I I EcchiIE&N

hIHI

I I Pst1 PstIpatIIPstI

Psti

3’ SphI

EM-II

(b) RNA blotting analysis of BDNF mRNA using the 3’flanking region probes RNA blotting analysis reveals two poly(A)’ BDNF transcripts of different size (1.6 and 4.2 kb) in the mammalian brain, although BDNF is encoded by a single-copy gene (Maisonpierre et al., 1991; Jones et al., 1990; Hofer

(6)

et al., 1990). In order to clarify why a single-copy BDNF gene generates such messages in two size classes, we performed RNA blotting analysis using two different probes (designated probes A and B) derived from the isolated rat genomic BDNF fragment. Probe A corresponds to the reported BDNF cDNA (Maisonpierre et al., 1991), whereas probe B is derived from the 3’-flanking region of the obtained genomic BDNF clone which is located just downstream from probe A (Fig. IA). The RNA blotting analysis revealed that probe B visualized only the longer BDNF message while probe A could become hybridized to both messages, indicating that the longer BDNF transcript carried a longer 3’-noncoding region than that of the shorter one (Fig. 1B). Thus, it was concluded that a single BDNF gene directs the synthesis of at least two transcripts that differ in the size of the 3’-noncoding region. (c) Characterization of the 5’-end of BDNF cDNA Previous investigators have so far reported one or two 5’-end sequences of the BDNF cDNA from each mammal; pig, mouse, rat, and two human types (Leibrock et al., 1989; Hofer et al., 1990; Maisonpierre et al., 1991; Shintani et al., 1992). However, no extensive studies have been done to characterize the 5’-end structure of BDNF mRNA mainly due to the scarcity of the message. In order to characterize the 5’-end of BDNF mRNA present in the brain, we took advantage of the power of the one-sided PCR strategy (Ohara et al., 1989; Ishizaki et al., 1989). We could successfully amplify the 5’-end BDNF cDNA fragments using this method; some DNAs of about 300 bp at the first round of the one-sided PCR were found to be the amplified BDNF cDNA fragments by DNA blotting analysis with a 5’-end BDNF-specific probe. Thus, the DNAs in this size range after the second-round PCR were cloned into an EcoRV site of pBluescript KS( - ) vector by the dideoxythymidine-tailing method (Holton and Graham, 1991), and then 20 independent clones of the 5’-end BDNF cDNA were subjected to DNA sequencing. The results showed that there were six types of 5’-end cDNA clones

c

c

L

Fig. 1. RNA blotting Restriction together

analysis

of total RNA isolated

sites for AalII, EcoRI, BarnHI,

with the location

sonpierre

kb

1.6

kb

B

A

PROBE

4.2

from rat brain. (A)

PstI and SphI are illustrated

of the BDNF cDNA region reported

tion with AatII + SphI and SphI + BamHI, respectively. nt sequence under

by Mai-

et al. (1991). Probes A and B were derived by restriction was deposited

accession

with the DDBJ/GenBank/EMBL

No. D10938.

(Panel B) Total brain

databases

RNA (20 pg/lane)

was run on a 1 y0 agarose gel containing 2.2 M formaldehyde, onto a nylon membrane, and then crosslinked by ultraviolet Hybridization The probes

was carried

(Church

transferred irradiation.

and Gilbert,

were labeled with [ a-j*P JdCTP using a PRIME-IT

labeling kit (Stratagene, was washed

out as described

La Jolla, CA). After hybridization,

several times with 15 mM NaCl/l.S

sodium dodecylsulfate

at 65°C

diges-

The determined

1984). random

the membrane

mM Na,scitrate/O.l

and then subjected

y0

to autoradiography.

which carried different nt sequences, as shown in Fig. 2. The generation of multiple 5’-end sequences could be well explained by alternative usage of multiple 5’-exons, since the nt sequences varied just upstream from a putative splice site (Maisonpierre et al., 1991). The obtained sequences of transcripts l-3 were highly homologous or identical to the ones published for mouse/pig, rat, and human BDNF

385

GENOHIC TRANSCRIPT TRANSCRIPT TRANSCRIPT TRANSCRIPT TRANSCRIPT TRANSCRIPT

GENOMIC TRANSCRIPT TRANSCRIPT TRANSCRIPT TRANSCRIPT TRANSCRIPT TRANSCRIPT

Fig. 2. The nt sequence with a primer modified sultant

of the 5’-ends

1 2 3 4 5 6

40 20 30 1 10 -GTCTGAAATTACAAGCAGATGGGCCACATGGTGTCCCCAAG~AGTAAGG -AAGAGTCTAGAACCTTGGGGACCGGTCTTCCCCAGAGCAG -GTGTGATCCGGGCGAGCAGAGTCCATTCAGCACCTTGGACA -CCCAGGGCAGGGCGCAGGGACCAGGAGCGTGACAACAATGTGAC -TGACTTCAAACAAGACACATTACCTTCCAGCATCTGTTGGGGAGACGAGA -GGTCCCCTCATTGAGCTCGCTGAAGTTGGCTTC -GGGAAATGCAAGTGTTTATCT 70 60 60 TCTAACCTGTTCTGTGTCTGTCTCTGCTTCCTTCCCACAG CTGCCTTGATGTTTACTTTGACAAGTAGTGACTGAAAAAG GAGCCAGCGGATTTGTCCGAGGTGGTAGTACTTCATCCAG TCCACTGCCGGGGATCCGAGAGCTTTGTGTGGACCCTGAG TTTTAAGACACTGAGTCTCCAGGACAGCAAAGCCACAATG CTAGCGGTGTAGGCTGGAATAGACTCTTGGCAAGCTCCGG CCAGGATCTAGCCACCGGGGTGGTGTAAGCCGCAAAGAAG

1 2 3 4 5 6

of BDNF cDNAs

[5’-ACCGAAGTATGAAATAACC;

Moloney cDNA

murine leukemia

obtained

virus reverse transcriptase

was tailed with dATP

(5’-AGTAAGGAAAAGGATGGTC;

and terminal the

by the one-sided

the complementary

(SuperScript

deoxynucleotidyl

complementary

sequence

RNaseH

transferase,

sequence

to

nt

90

100 TTCCACCAGG TTCCACCAGG TTCCACCAGG TTCCACCAGG TTCCACCAGG TTCCACCAGG TTCCACCAGG tPCR primer Splicing Site

PCR method. to nt 96-114

Rat brain total RNA (20 pg) was reverse-transcribed

of rat BDNF cDNA

_ reverse transcriptase, and subjected

75-93

of

rat

50

to first-round BDNF

(Maisonpierre

Bethesda

Research

et al., 1991)] and a Laboratories).

PCR with a first-round

cDNA)

and

a

dT-tailed

The re-

PCR

anchored

primer primer

as described (Ohara et al., 1989; Ishizaki et al., 1989). The thermal cycle number of the (5’-TAAGATCTAATACGACTCACTATAGGGAAGC(T),,) first-round PCR was 20 (denaturation, 94°C for 30 s; annealing, 55°C for 30 s; extension, 72°C for 30 s; the last extension, 72°C for 5 min). An ahquot

(10~1)

of the first-round

(5’-ACTCTTCTCACCTGGTGG; TATAGGG) products

mixture

was

directly

the complementary

in a 100~~1 scale under the same thermal

were analyzed,

two independent The 5’-flanking

PCR

although

the cloned

used

sequence

derived from the isolated

round

(150-200

genomic

of PCR

of rat BDNF cDNA)

cycle profile as in the first-round

PCR products

clones are shown in this figure. The further upstream sequence

for the second

to nt 54-71

(30 cycles)

PCR. Twenty

BDNF fragment

a second-round

PCR

primer

primer (5’-TAATACGACTCAC-

independent

bp) were smaller than expected. sequences,

with

and an anchored

plasmids

carrying

Only the nt sequences

which may include (an) artifact(s)

PCR-amplified

confirmed

by more than

of PCR, are available

upon request.

is also shown as GENOMIC.

cDNAs, respectively (Leibrock et al., 1989; Hofer et al., 1990; Maisonpierre et al., 1991; Shintani et al., 1992). Therefore, the reported differences in the 5’-noncoding sequence of BDNF cDNAs isolated from various mammals seemed to be only apparent and each of the 5’-exon sequences is thought to be highly conserved during evolution as in the case of the 3’-noncoding region of this gene. In contrast, the 5’-end sequences of transcripts 4-6 have not been reported yet. Transcripts 4-6 were not spurious since the presence of these transcripts in the brain was confirmed by conventional PCR with an upstream primer specific for each of the 5’-end sequences and the downstream common primer in the BDNF-encoding region (data not shown). Furthermore, conventional PCR revealed the presence of the transcript which carried the two 5’-exons corresponding to transcripts 5 and 6 in tandem, indicating that alternative splicing played some role in generating these multiple transcripts.

multiple types of transcripts in the brain. The selection of the 5’-exons as well as polyadenylation sites may imply some biological importance related to the function of BDNF in the central nervous system.

(d) Conclusions We demonstrated that a single BDNF gene generated multiple transcripts by alternative use of 5’-exons and polyadenylation sites in the rat brain. Although the BDNF transcripts differ only in the noncoding region, the results indicate that we should study the regulatory mechanism of BDNF expression taking into consideration the presence of

Jones, R.K. and Reichardt,

REFERENCES Church, G.M. and Gilbert, W.: Genomic Sci. USA 81 (1984) 1991-1995. Hofer, M., Pagliusi,

S.R., Hohn,

gional distribution Holton,

A., Leibrock,

of brain-derived

adult mouse brain.

sequencing.

Proc. Natl. Acad.

J. and Barde, Y.-A.: Re-

neurotrophic

factor mRNA in the

EMBO J. 9 (1990) 2459-2464.

T.A. and Graham,

M.W.: A simple and efficient method

rect cloning of PCR products

using ddT-tailed

vectors.

for di-

Nucleic Acids

Res. 19 (1991) 1156. Ishizaki, J., Ohara, Yoshida,

O., Nakamura,

N., Teraoka,

E., Tamaki,

M., Ono, T., Kanda,

H., Tojo, H. and Okamoto,

A.,

M.: cDNA cloning

and sequence determination of rat membrane-associated phospholipase A,. Biochem. Biophys. Res. Commun. 162 (1989) 10301036. L.F.: Molecular

cloning of a human gene that

is a member of the nerve growth factor family. Proc. Natl. Acad. USA 87 (1990) 8060-8064. Leibrock,

J., Lottspeich,

kowski,

P., Thoenen,

expression

F., Hohn, A., Hofer, M., Hengerer, H. and Barde,

of brain-derived

Y.-A.: Molecular

neurotrophic

factor.

Nature

Sci.

B., Masiacloning

and

341 (1989)

149-152. Maisonpierre,

P.C., LeBeau, M.M., Espinosa

III, R., Ip, N.Y., Belluscio,

386 L., de la Monte,

S.M., Squinto,

G.D.:

and

Human

neurotrophin-3: cahzations. Ohara,

S., Furth,

brain-derived

gene structures,

Genomics

O., Dorit,

rat

M.E. and Yancopoulos, neurotrophic

distributions,

factor

and chromosomal

and lo-

10 (1991) 558-568.

R.L. and Gilbert,

W.: One-sided

reaction:

the amplification

Shintani,

A., Ono, Y., Kaisho,

the 5’-flanking polymerase

chain

of cDNA.

Proc. Natl. Acad. Sci. USA 86

(1989) 5673-5677. Y. and Igarashi.

region of the human

tor gene. Biochem.

Biophys.

K.: Characterization

brain-derived

Res. Commun.

neurotrophic

182 (1992) 325-332.

of fac-