aiding in its installation, M. Levitt for numerous computer graphics used in the analysis of molecularhelpful and critical discussions, Z. Wasserman for dynamics simulations.important contributions to code development andmethodology, and R. Hilmer for the interactive 29 June 1987; accepted 1 September 1987

A Nerve Growth Factor-Induced Gene Encodes aPossible Transcriptional Regulatory Factor

JEFFREY MILBRANDT

Nerve growth factor (NGF) is a trophic agent that promotes the outgrowth of nervefibers from sympathetic and sensory ganglia. The neuronal differentiation stimulatedby this hormone was examined in the NGF-responsive cell line PC12. Differentialhybridization was used to screen a complementary DNA library constructed fromPC12 cells treated with NGF and cycloheximide. One of the complementary DNAclones that was rapidly induced by NGF was found to have a nucleotide sequence thatpredicts a 54-kilodalton protein with homology to transcriptional regulatory proteins.This clone, NGFI-A, contains three tandemly repeated copies ofthe 28- to 30-aminoacid "zinc finger" domain present in Xenopus laevis TFIIIA and other DNA-bindingproteins. It also contains another highly conserved unit of eight amino acids that isrepeated at least 11 times. The NGFI-A gene is expressed at relatively high levels in thebrain, lung, and superior cervical ganglion of the adult rat.

with NGF and CHX for 3 hours. Threedifferent NGF-induced cDNAs were identi-fied from this screen: c-fos, a protein thatbinds DNA (16) and may function as atrans-acting factor (17), NGFI-A, andNGFI-B.NGFI-A was selected for further study.

This cDNA hybridized to a -3.3-kb mRNAthat was induced by NGF (Fig. IA, lane 3)but accumulated to a much higher level inPC12 cells treated with NGF and CHX(lane 4). In PC12 cells grown without NGFthis mRNA could not be detected (lane 1).However, when cells were treated withCHX alone, the level of this transcript wasslightly increased (lane 2).The time course of induction of the

NGFI-A mRNA was determined by isolat-ing mRNA from PC12 cells treated withNGF for various lengths of time. Northernanalysis with the NGFI-A cDNA probeshowed that the basal level of this transcriptwas very low in PC12 cells (Fig. 1B), butwithin 15 minutes after the addition ofNGF (lane 2) the level of NGFI-A mRNA

NERVE GROWTH FACTOR (NGF) isa polypeptide hormone that is re-quired for the development and

survival of sympathetic and neural crest-derived sensory neurons in vivo and in vitro(1). NGF also participates in nerve regenera-tion after injury (2) and plays a role in thecentral nervous system (3). The effects ofNGF on the differentiation of sympatheticneurons are manifested by an acceleratedoutgrowth of nerve fibers and the inductionof several enzymes involved in neurotrans-mitter biosynthesis (4). During the differen-tiation process, specific genes are turned onand off at precise times. This ordered patternof gene expression, resulting in the estab-lishment of the neuronal phenotype, is pre-sumably controlled by DNA-binding pro-teins that act to regulate transcription.The best characterized transcriptional ac-

tivating factor is the TFIIIA protein thatregulates the expression of 5S RNA genesduring development in Xenopus laevis (5).The DNA-binding region of this protein iscomposed of nine tandemly repeated unitsthat contain two invariant pairs of Cys andHis residues (6). Each repeat unit is foldedaround a single zinc ion that is coordinatelybonded to the Cys and His residues to forma DNA-binding domain or "Zn finger."Amino acids capable of interacting with theDNA helix (Lys, His, Asn, Gln, Thr, andArg) are located at the tip of each "Znfinger." This 28- to 30-amino acid DNA-binding domain has now been identified in

Division of Laboratory Medicine, Departments of Pa-thology and Internal Medicine, Washington UniversitySchool of Medicine, St. Louis, MO 63110.

several other proteins involved in transcrip-tional regulation, including yeast ADR1-encoded protein (7) and the Kruppel geneofDrosophila (8).One model of neuronal differentiation is

the PC12 cell line (9), whose NGF-mediat-ed transition from replicating adrenal chro-maffin-like cells to sympathetic neuron-likecells is prevented by the addition of RNAsynthesis inhibitors (10). This cell line, de-rived from a rat pheochromocytoma, re-sponds to NGF by extending neurites andby increasing the transcription of severalgenes, including ornithine decarboxylase(11), GAP43 (12), and c-fos (13). The acti-vation of the proto-oncogene c-fos is rapidbut transient in NGF-treated PC12 cells.The c-fos transcripts accumulate to evenhigher levels in cells stimulated in the pres-ence of cycloheximide (CHX), in part be-cause of an increase in messenger RNA(mRNA) half-life (14). Using c-fos as amodel of an early activated gene, I havetaken advantage of this CHX-mediated su-perinduction to identify other early, NGF-induced genes that may regulate neuronaldifferentiation.A complementary DNA (cDNA) library

containing 8 x 10 independent clones wasconstructed in the vector XgtlO (15), byusing mRNA isolated from PC12 cells thathad been treated with NGF (50 ng/ml) andCHX (10 ,ug/ml) for 3 hours. Initially, 8000recombinants (1% of total library) werescreened by differential hybridization withsingle-stranded cDNA probes complemen-tary to mRNA from either unstimulatedPC12 cells (no NGF) or PC12 cells treated

A 1 2 3 4

-28S

-18S

2 3 4 5 6 7 8 9

-3.3 kb

Fig. 1. Identification ofan NGF-regulated cDNAclone. Total cellular RNA was isolated by themethod of Chirgwin (24). The cDNA library wasscreened by differential hybridization with single-stranded cDNA probes complementary to mRNAfrom either naive PC12 cells (no NGF) or PC12cells treated with NGF and CHX for 3 hours (25).Phage from the NGFI-A cDNA clone were isolat-ed on DEAE columns (26). DNA was purifiedand digested with Eco RI. The cDNA insert wasisolated by electrophoresis on low-melting agar-ose and labeled with 32P-labeled deoxyadenosinetriphosphate by oligo labeling (27). RNA electro-phoresis and Northern blot analysis were per-formed as described (13). (A) A Northern blotcontaining RNAs from PC12 cells treated for 3hours with (lane 1) no addition, (lane 2) CHX(10 ,tg/ml), (lane 3) NGF (50 ng/ml), or (lane 4)NGF and CHX, was probed with the NGFI-AcDNA. The locations of the 28S and 18S ribo-somal RNAs are indicated. (B) Time course oftheNGF-mediated induction of NGFI-A mRNA.RNA samples from PC12 cells grown in thepresence ofNGF for 0, 2, 15, 30, 45, 60, and 120minutes, 13 hours, and 6 days (lanes 1 through 9,respectively) were hybridized to 32P-labeledNGFI-A cDNA insert. The 3.3-kb NGFI-AmRNA is indicated.

6 NOVEMBER 1987

B 1

ARTICLES 797

on

Dec

embe

r 14

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

increased. The NGFI-A mRNA level peakedapproximately 30 to 45 minutes after theaddition ofNGF (lanes 3 and 4) and persist-ed for up to 6 days (lane 9). The rate ofinduction of the NGFI-A gene was similarto that of the NGF induction of c-fos (13),but the rate of decay of NGFI-A was muchslower. The NGFI-A gene was also inducedby treating PC12 cells with the phorbol 12-myristic 13-acetate (PMA) and the calciumionophore A23187 as has been demonstrat-ed for the c-fos gene (18).As shown in Fig. 2, the NGFI-A cDNA is

3118 nucleotides in length and contains anopen reading frame from nucleotide 269 tonucleotide 1876. Several in-frame Met co-dons are present, but only the Met at nucle-otide 353 contains the translational initia-tion consensus sequence (CCG/ACCAT-GG) (19). On the basis of initiation at this

Fig. 2. Nucleotide and deducedamino acid sequence of the NGFI-A cDNA. The nucleotide sequencewas determined by the method ofSanger et al. (28). The NGFI-AcDNA was cloned in both orienta-tions into the Eco RI site of theBluescript plasmid (Stratagene),and deletions were constructedwith Exonuclease III and mungbean nuclease as described (29).The numbers on the left refer to thenucleotide sequence (upper) andthe amino acid sequence (lower).The region corresponding to theZn-finger domains is boxed. Theregions containing the eight amino

acid repeats are underlined. Thelong stretch of Ser and Gly residuesis underscored by a heavy bar. Apotential phosphorylation site iscircled. Potential sites for N-linkedglycosylation are dotted.

ATF II IA

NGFI -ANGFI-ANGF I -AKruppelADR I

Consensus

Repeat 1Repeat 2Repeat 3

site, the NGFI-A protein is 508 amino acidslong with an unmodified molecular size of53,934 kD. Consensus polyadenylation sitesare located at nucleotide 1920, 3132, and3143. Multiple copies of the sequenceACAAAA are also present; this sequence hasbeen postulated to provide a protein-bind-ing domain in either the gene or the corre-sponding mRNA of the Kruppel gene (8).A search of the NBRF protein database

(20) revealed that the NGFI-A protein ishomologous to the X. Iaevis transcriptionalfactor, TFIIIA. As shown in Fig. 2, theregion of NGFI-A that is homologous toTFIIIA extends from nucleotide 1277 tonucleotide 1534. This region contains three28- to 30-amino acid repeats. In theTFIIIA molecule, the area of homologycontains nine copies of a similar 28- to 30-amino acid repeat. In Fig. 3A, the three

CC6AACTT6666^6AC6CC6CC6CCGCATTCGCCGCC6CC6CCU6C TTCCGCC6.CC6CAAATC6GCCCC TGCCCUAGCC TCCGCGGCAGCCC TGCGTCCACCACG66CC6C66CCACC6CCA6CCT66G66CCCACCTACACTCCCC6CAGT6TGCCCCT6CACCCCCATGTAACCCGGCAACATCCGGCGAGTGTGCCCTCAGTAGCTTCGGCCCC666CT6C6CACUCACCCAACATUCAGTCTCUCAGTCGCACGTCCG666AT6GGCACGGCCAAG6CG6A6ATGCAA T TGATGTCTCCGC TGCAGA TCTCTGACCC6.TTC6GC TCCTTTCCTCACTCACCACUCATGGACAATACCCCAAACTGGArAGAATGATGC TGCTGAGCAACG66GC TCCCCAGT TCC TCGGTGC

N D N Y P K L E E( M L L S N G A P E F L G A

TGCC6GAWCCCCAGAGGGCA6CGGCGGCAAT _ GT6T6GG GGGGGGGCACAACA6C GCGGCAGCACGCTT TA G T P E G S G G 11 11 S S S S S S S S S G G G G 6 G G S k S G S S A F

CAATCCTCAAGGGGAGCCGAG cCGAACACCTACGAGCACCTGACACAGAr.TCCTTTTCTGACATCGCTCTGAATAACGAGAAr.GCGCTGGTGGAGACAAGTTAN P E G E P S EE P Y E H L T T E S F S AA L K N E K A L V E,T S Y

TCCCAGCCSACATACCCGGTTGCCTCCCATCCCGTATACTGGCCGCTTCTCCCTGGACCTGCACCCACAGTG6CACACTTTGTGGCCTGAACCCCTTTTCAGP S Q T T R L P P T Y(G G R F S L E P A P N S G N T L W P E P L F S

CCTAGTCATGGCCTTGTGAGCATGACCAcCCTCCAACCTCTTCTCCTuCACCTTCTCuCATGCTTCTCGTCTTCCTCTGCCTCCCAGAGCCCACCCCTL V S 6 L V S M T N P P T S S S S A P S P A A S S S S S A S S S P P L

GAGC TGT6CCGTGCCGTCCACGACAGCAGTCCCATTTACTUCATGUCCACCTTTCCTACTCCUACAC TACATTTTTCCTGAGCCCCAAAGCCAGGCCTTS C A V P S N D S S P Y S A A,P T F P T P N T,D F P E P Q S,Q A F

TCCTGGCTCTGCGGCACAGCCTTGCAGTACCCGCCTCCTGCCTACCCTGCCACCAAGGGTGGTTTCCAGGTTCCCATGATCCCTGACTATCTGTTTCCACAACAP G S A G T A L Q Y P P P A Y P A T K G G F Q V P M P D Y L F P Q Q

1051 ACAGGGAGACCTGAGCCTGGGCACCCCAGACCAGAAGCCCTTCCAG6GTCTGGAGAACCGTACCCAGCAGCCTTCGCTCACTCCACTATCCACTATCAAAGCCTT234 Q 6 D L S L G T P E Q K P F Q G L E A R T Q Q P S L T P L S T K A F

1156 CGCCAC TCAGTCGGGC TCCCAGGACTTAAAGGC TC TTAATAACACCTACCAGTCCCAAC TCATCAAACCCA6CCGCATGCGCAAG TACCCCAACCGGCCCAGCAA269 A T 0 S G S E D L K A L NI T Y E S Q L K P S R N R K Y P AR P S K

12 61 GACACCCCCCCATGAACGCCCGTATGCTTGCCC TGTTGAGTCCTGCGATCGCCGCT TTTC TCGCTCGGAT6AGC TTACACGCCACATCCGCATCCATACAGGCCA304 T P P H E AR P Y A C P V E S C D R R F S R S D E L T R H R H T G 5|

1366 GAAGCCCTTCCAGTGTCGEATCTGCATGCGTAATTTCAGTCGTAGTGACCACCTTACCACCCACATCCGCACCCACAGGCGAGAAGCCTTTTGCCTGTGACAT339 AK P F Q C R C A R N F S R S E H L T T H R T H T G E|IK P F A C D

471 TTGTGGGAGAA6T T TGCCAGGAGTGATG6AACGCA6AGAGCATACCAAAATCCAC TTAAGACAGAAG6ACAAGAAAGCAGACAAAAGT GTCGTGGCC TC C TCAGC374 C G R K F A R S D E R K R H T K H L R O|K D K K A E K S V V A S S A

1576 TGCC TC TTCCCTCTCT TCC TACCCATCCCCAGTGGC TACC TCC TACCCUTCCCCCGCCACCACCTCATT TCCATCCCCAGTGCCCACC TC TTACTCC TC TCCGGG409 A S S LI S S Y P S P V A, T S Y P S P A T, T S F P S P V P, T S Y S S P G

'681 C TC CTCTACCTACCCGTC TCC TGCACACAGTGGCTTCCCATCGCCC TCGGTGGCCACCACC TATGCC TCCGTCCCACC TGC T TTCCC TGCCCAGG TCAGCACCT T444 5 5 T Y P S P A H, S G F P S P S V A, T T Y A S V P P, A F P A Q V, S T F

1786 CCAGTC TGCAGGGGTCAGCAAC TCC T TCAGCACCTCAACGGGTCT T TCAGACATGACAGCAACCTT TTCTCC TAGGACAAT TGAAAT TTGC TAAA6G6AATGAAA479 5 S A G VA S N S F S T S T G L S DE T A T F S P R T E C ENE

GGGTAGGAGYT6ATGTTTC TG TTAGAGCACGATGAGGAAGAGGCTGAGCTGAGCTTTGGTTCTCCAGAATGTAAGAAGAAMATTTAACAAAAATC TGAAC TCTCAAAAGTC TATT TT T TTA,ACTGAMATGTAGAT TTATCCATGT TCGGGAGTTC6"GCAT6GCGGTT*ACC TAC TGAGTAGGCGGTGACT T TTG TAT GC TATGAACATGAAGT TCAT TAT T TTGTGGT *TTTAT T TTAC TTCGTACTTGTGT T TGCTTAAkACAAAGTGAC T TGTTT6GCTTATAAACACAT TGAATGCGCT TTA

K P YTV ]X X D G C D R F T K X X L K REX X X H TIG E

RP YIA CIP V E SIC D RIR FSRS DEL; T RIH RIEH T|G QK P F Q C R I --CIM R N F S IRsDHT TIH I R| TH T G EK P F A C D I -C G R K F A R S D E R K RIH T K,IH LR QK PF D C K I C S R S F G Y H V L Q. N HE T G E

FV C E V- T R FA HLKR Y S NE

B Residues Sequence

RL 91-98 T Y P 5 Q T TR2 180-187 P T F P T P N TR3 188-195 D I F P E P Q S

R4 413-420 5 Y P 5 P V AR5 421-428 T 5 Y P 5 P A TR6 429- 436 T 5 F P S P V PR7 437-444 T S Y S 5 P G SR8 445-452 S T Y P S P A HR9 453-461 S G F P 5 P V ARAL 462- 468 T T Y A V P PRAl 476-483 S T F Q S A G V

Consensus sequence r[][5Iy[] P P X X

Fig. 3. (A) Alignment of the sequence of amino acids of the "Zn fingerAY repeats in the NGFI-A proteinwith themselves and with the DNA-binding regions of several eukaryotic transcriptional regulatoryproteins. (B) Alignment of the highly conserved eight-amino acid repeat of the NGFI-A gene. Aconsensus sequence is derived for positions at which at least 8 of the 11 repeats contain identical or

conservatively substituted residues.

798

NGFI-A repeat sequences are aligned withthe TFIIIA repeat and similar sequencesfrom other DNA-binding proteins that reg-ulate transcription. The high degree of simi-larity is apparent from this comparison. Themost highly conserved residues are boxed;these include the residues determined to bemost important in the nine TFIIIA repeats.The NGFI-A sequence contains several

other unique features. It contains a stretchof predominantly basic residues which ex-tends from Arg 380 to Lys 401. Immediate-ly following this basic region (at nucleotide1556 and extending to the COOH-termi-nus) is a hydrophilic region that is highlyenriched (59 of 107 residues) with Pro, Ser,and Thr. Within this COOH-terminal por-tion of the molecule another tandemly re-peated unit can be distinguished. This isillustrated in Fig. 4, where the predictedamino acid sequence is compared to itselfand displayed in a dot matrix format (21).The central, continuous diagonal resultsfrom the perfect match of NGFI-A proteinwith itself; the short diagonals offset fromthe central one represent repeat units. Oneset of three repeat units is located betweenresidues 309 and 394 and represents the"Zn finger" domains discussed above. An-other site of repeating units can be seenbetween residues 413 and 487. This repeathas a periodicity of eight amino acids and isreiterated eight times. Three additionalcopies of this repeat are located in the NH2-terminal region of the molecule (Fig. 2).These repeats are aligned with each other,along with a consensus sequence, in Fig. 3B.The NH2-terminal portion of the NGFI-

A molecule is also composed of a highpercentage of Pro, Ser, and Thr (45%) fromnucleotide 455 to nucleotide 1045. A po-tential phosphorylation site (Thr/Ser-Gly-Arg) is present at nucleotide 668. Severalpotential N-linked, and many potential 0-linked, glycosylation sites are also present. Acurious feature of this region is the stretch ofnine Ser residues directly followed by sevenGly residues (Fig. 2). Interestingly, the samecodon, AGC, is used for each ofthe nine Serresidues in this stretch. This triplet repeat,called opa, has been previously recognized inthe Notch gene, a gene that is involved inDrosophila neurogenesis (22). The poly(Gly)residues are encoded by a repeat clustercalled pen, which consists of clusters ofGGN (where N is any nucleotide) (23). Thejuxtaposition of these two triplet repeatsmay indicate that the critical functional fea-ture of this region is the underlying nucleo-tide sequence, rather than the amino acidsequence. The repeating AGC or GGN mo-tifs could represent protein binding sites inthe gene or in the RNA transcript.To examine the expression ofthe NGFI-A

SCIENCE, VOL. 238

'996 TCCTTCTAGTCAGTAGAAGGCCCGTTGGCCACCAGCCCTTTCACTTAGCGTCCCTGCCCTCCCCAGTCCCGGTCCTTTTGACTTCAGCTGCCTGAAACAGCCACG2101 TCCAAGTTCTTCACCTCTATCCAAAGGACTTGATTTGCATGGTATTGGATAAACCATTTCAGCATCATCTCCACCACATGCCTGGCCCTTGCTCCCTTCAGCACT2206 AGAACATCAAGTTGGCTGAAAAAAAAAATGGGTCTGGGCCCTCAGAACCCTGCCCTGTATCTTTGTACAGCATCTGTGCCATGGATTTTGTTTTCCTTGGGGTAT

2416 TTTGTGATGATCCTGCTGTGACATTACGTTTGAAACTTTTTTTTTTTTTTGAAGCAGCAGTCCTAGGTATTAACTGGAGCATGTGTCAGAGTGTTGTTCCGTTAA2521 TTTTGTAAATACTGCTCGACTGTAACTCTCACATGTGACAAAATACGGTTTGTTTGGTTGGGTTTTTTGTTGTTTTTGAAAAAAAAATTTTTTTTTTGCCCGTCC

1106211316

42124

52659

63194

736129

841164

946199

273128362943046

on

Dec

embe

r 14

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Fig. 4. Dot matrix analysis of the amino acidsequence of the NGFI-A protein. The amino acidsequence is represented on both the horizontaland vertical axes. Each 30-residue segment iscompared to all other 30-residue segments. Ahomology score is calculated for each comparisonusing the mutation data scoring matrix (20). Adot is placed at the position given by the midpointofthe two segments when the comparison score isgreater than 20.

Fig. 5. Expression of the NGFI-A gene in theadult rat. Poly(A)+ RNA (7.5 p.g) from theindicated tissues of 6-week-old Sprague-Dawleyrats was electrophoresed and transferred to nitro-cellulose. The RNA blots were hybridized to theNGFI-A cDNA. In the SCG lane only 12 ,ug oftotal RNA was analyzed. The lane labeled PC12contains 250 ng of poly(A)+ RNA from PC12cells treated with NGF and CHX. Tissues: PL,placenta; LU, lung; HT, heart; BR, brain; SCG,superior cervical ganglia; AD, adrenal; LV, liver;and KD, kidney.

gene, RNA transfer analysis was performedon samples from various tissues of the adultrat. The NGFI-A gene is expressed at highlevels in the brain, lung, and superior cervi-cal ganglia, and at a moderate level in theadrenal gland; however, the amount ofNGFI-A mRNA is much greater in PC12cells treated with NGF and CHX than in anyof the tissues tested (Fig. 5). In other tissuesNGFI-A transcripts are detectable but verylow in number. The signal detected in theSCG lane reflects a high level of expressionin these cells, since approximately one-tenthas much mRNA was loaded in this sample(12 ,ug of total RNA versus 7.5 pug ofpolyadenylated [poly(A)+] RNA for theother tissues). The detection of NGFI-Atranscripts in the brain is interesting in lightof recent evidence that NGF-responsiveneurons are present within the central ner-vous system (3). The expression in the lungand the detection of a second, slightly larger,transcript in this organ is not understood.The widespread expression observed for theNGFI-A gene opens the possibility that thisgene may be regulated by factors other thanNGF in some tissues.The homology of the protein encoded by

the NGFI-A gene to other transcriptionalregulatory proteins suggests that it may havea similar function. The recent demonstrationthat c-fos is a DNA-binding protein, coupledwith the discovery of a third NGF-inducedgene, the NGFI-B gene, which is also ho-mologous to transcriptional regulatory pro-teins, supports the hypothesis that these

6 NOVEMBER I987

.0E

q

o)

0

Residue number

{L X I ID <: J Y X-j :) cc C. 0 > 0aa. MD 4 id a.

rapidly induced genes encode proteins thatmay play a role in initiating the differentia-tion program elicited by NGF. It is likelythat genes that are rapidly induced in re-

sponse to other environmental cues or hor-mones may also encode transcriptional regu-latory factors.

REFERENCES AND NOTES

1. B. Yankner and E. Shooter,Annu. Rev. Biochem. 51,845 (1982).

2. S. Korsching, R. Heumann, A. Davies, H. Thoenen,Soc. Ncurosci. Abstr. 12, 1096 (1986); C. E. Bandt-low, R. Heumann, M. E. Schwab, H. Thoenen,EMBOJ. 6, 891 (1987); M. Taniuchi, H. B. Clark,E. M. Johnson, Jr., Proc. Natl. Acad. Sci. USA. 83,4094 (1986).

3. A. Szutowicz, W. Frazier, R. Bradshaw, J. Biol.Chem. 251, 1516 (1976); T. H. Large ct al., Science234, 352 (1986); S. Korsching, G. Auberger, R.Heumann, J. Scott, H. Thoenen, EMBOJ. 4, 1389(1985); S. Whittemore ct al., Proc. Natl. Acad. Sci.U.SA. 83, 817 (1986); M. Taniuchi, J. B. Schweit-zer, E. M. Johnson, Jr., ibid., p. 1950.

4. H. Thoenen, Y.-A. Barde, D. Edgar, in Neurosecrc-tion and Brain Pcptides, vol. 28 of Advances inBiochemical PAchopharacolqgy, J. B. Martin, S.Reichlin, K. L. Bick, Eds. (Raven, New York,1981), pp. 263-272.

5. D. Engelke, S.-Y. Ng, B. S. Shastry, R. G. Roeder,Cel 19, 717 (1980).

6. J. Miller, A. D. McLachlan, A. Klug, EABO J. 41609 (1985).

7. T. A. Hartshorne, H. Blumberg, E. T. Young,Nature (London) 320, 283 (1986).

8. U. B. Rosenberg et al., ibid. 319, 336 (1986).9. L. Greene and A. Tischler, in Advances in Celular

Neurobiology, S. Federoff and L. Hertz, Eds. (Aca-dermic Press, New York, 1982), vol. 3, pp. 373-414.

10. D. E. Burstein and L. A. Greene, Proc. Nail. Acad.Sci. USA. 75, 6059 (1978).

11. S. C. Feinstein, S. L. Dana, L. McConlogue, E. M.Shooter, P. Coffino, ibid. 82, 5761 (1985).

12. L. R. Karns, S.-C. Ng, J. A. Freeman, M. C.Fishman, Science 236, 597 (1987).

13. T. Curran andJ. I. Morgan, ibid. 229, 1265 (1985);M. Greenberg, L. Greene, E. Ziff, J. Biol. Chem.260, 14101 (1985); W. Kruijer, D. Schubert, I. M.Verma, Proc. Natl. Acad. Sci. U.SA. 82, 7330(1985); J. D. Milbrandt, ibid. 83, 4789 (1986).

14. K. Kelly, B. Cochran, C. Stiles, P. Leder, CeU 35,603 (1983).

15. T. V. Huynh, R. A. Young, R. W. Davis, in DNACloning, D. M. Glover, Ed. (Practical ApproachSeries, IRL Press, Oxford, England, 1985), vol. 1,pp. 49-78. The cDNA library was constructed asfollows: poly(A)+ RNA was selected by two pas-sages over oligo(dT)-cellulose. Single-strandedcDNA was synthesized from 5 ,ug ofpoly(A)+ RNAwith 40 units of reverse transcriptase (Life Sciences)in the presence of 100 mM Tris, pH 8.3 at 250C,140 mM potassium acetate, 10 mM magnesiumacetate, oligo(dT) (80 iLg/ml), 800 pM deoxynu-deoside 5'-triphosphates by incubation at 37'C for1 hour. The second strand was synthesized accord-ing to the method ofH. Okayama and P. Berg [Mol.Cell. Biol. 2, 161 (1982)]. The double-strandedcDNAs were treated with Eco RI methylase, treatedwith T4 polymerase, ligated to Eco RI linkers,digested with Eco RI, and size-selected by chroma-tography on BioGel A5M. The size-selected cDNAswere ligated to Eco RI-digested AgtlO and pack-aged in vitro (Stratagene). The resulting phage wereused to infect Eschbrichia coli C600 A hfl.

16. L. C. Sambucetti and T. Curran, Scienac 234, 1417(1986).

17. C. Setoyama, R. Frunzio, G. Liau, M. Mudryj, B. deCrombrugghe, Proc. Natl. Acad. Sci. U.SA. 83,3213 (1986).

18. J. Milbrandt, unpublished data.19. M. Kozak, CeU 44, 283 (1986).20. Protein SequenceDataBase oftheProtein Identifcation

Resource (National Biomedical Research Founda-tion, no. 10.0, Washington, DC, 1986).

21. M. 0. Dayhoff, W. C. Barker, L. T. Hunt, MethodsEnzymol. 91, 524 (1983).

22. K. A. Wharton, B. Yedvobnick, V. G. Finnerty, S.Artavanis-Tsakonas, CeU 40, 55 (1985).

23. S. R. Haynes et al., Proc. Natl. Acad. Sci. U.SA. 84,1819 (1987).

24. J. Chirgwin, A. Przybyla, R. MacDonald, W. Rut-ter, Biochemistr 18, 5294 (1979).

25. The cDNA library was screened by differential hy-bridization as follows: the phage were plated at lowdensity (1600 plaque-forming units per 88 mm2),and four replica filters were prepared from eachplate. Probes were prepared by synthesizing single-stranded cDNA from 2 Fg of poly(A)+ RNA fromeither naive PC12 cells (no NGF) or PC12 cellstreated with NGF and CHX. Synthesis conditionsfor cDNA were similar to above with the exceptionthat 100 ILCi of 32P-labeled deoxycytidine 5'-tri-phosphate were included. The RNA template wasremoved by incubation at 68°C for 1 hour with 50mrM NaOH, and unincorporated nucleotides wereremoved by centrifugation in a Centricon-30 (Ami-con). Two filters from each plate were hybridizedwith 1 x 106 cpm/ml of either type of single-strand-ed cDNA probe in 5x standard saline citrate, 50%formamide, 5x Denhardt's, single-stranded DNA(100 p.g/ml), 0.1% SDS, polyadenylic acid (1,ug/ml), at 42°C for 48 hours. Filters were washedtwice for 10 minutes in 2x standard saline citrateand 0.1% SDS at 250C, followed by two 60-minutewashes in 0.2x standard saline citrate and 0.1%SDS at 65°C. Plaques that produced a differentialsignal were identified by inspection and picked forsecondary screening.

26. G. Helmns, M. Grahamn, J. Dutchik, M. Olson, DNA(New York) 4, 39 (1985).

27. A. Feinberg and B. Vogelstein, Anal. Biochem. 67,15 (1984).

28. F. Sanger, S. Nicklen, A. R. Coulson, Proc. Natl.Acad. Sci. U.SA. 74, 5463 (1977).

29. S. Henikoff, Gene (Amsterdam) 28, 351 (1984).30. I thank T. Fahmer for technical assistance and P.

Manning, E. Johnson, and M. Thomas for advice.This work was supported by grant RG1779-A fromthe Multiple Sclerosis Foundation and grantNS01018 from the National Institute ofNeurologi-cal and Communicative Diseases and Stroke.

4 June 1987; accepted 28 August 1987

ARTICLES 799

on

Dec

embe

r 14

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from