nature biotechnology • VOLUME 21 • MAY 2003 • www.nature.com/naturebiotechnology

CORRESPONDENCE

492

A genome-wide view of antisense

To the editor:The ability to study gene expression on agenomic scale is allowing a deeper under-standing of specificity of antisense inhibi-tion. Although concerns about nonse-quence specific antisense effects have beenraised1, few genome-wide studies have sys-tematically investigated these effects. Wedraw readers’ attention to one examplefrom our laboratory where antisense treat-ment has clearly been shown to alterexpression of genes that appear to have norelationship to the intended antisense-tar-geted gene.

Antisense inhibition of gene expressionrelies on the simple rules of Watson-Crickbase pairing. A synthetic small single-

stranded oligonucleotide (generally a13–25-mer) that is complementary to aspecific gene is introduced into cells, bindswith mRNA, and inhibits translation.Hybridization of the antisense oligonu-cleotide with the target mRNA can physi-cally block the translation machinery oractivate RNase H cleavage at theRNA–DNA duplex site2–4. Targeting geneexpression at the RNA level allows proteinproduction to be turned off, even if RNA isabundant. When the protein product oftranslation is important for cell growthand/or viability, antisense inhibition ofgene expression can produce a lethal phe-notype. Because a particular 15- to 17-mersequence has been estimated to occur onlyonce in the entire human genome1, theo-retically, antisense inhibition of geneexpression should be exquisitely specific.

High-density cDNA microarrays enableparallel analysis of the expression of thou-sands of genes in a single hybridization forcomplex biological systems5. We have pre-viously examined genomic effects of anti-sense inhibition of protein kinase A RIαexpression in tumor cells using cDNAmicroarrays6. Using in vivo tumor modelsof PC3M human prostate carcinomagrown in nude mice, the specificity of anti-sense effects on gene expression signatureswas critically assessed using three oligonu-cleotides that differed in sequence orchemical modification: an immunostimu-latory phosphorothioate oligonucleotidedirected against human RIα, a second-gen-eration immunoinhibitory RNA-DNAmixed-backbone oligonucleotide, and anon-immunostimulatory phosphoroth-

ioate oligonucleotide targetedto mouse RIα (this oligonu-cleotide cross-hybridizes withhuman RIα).

Antisense treatment wasfound to affect one cluster, orsignature, of genes involved inproliferation and anotherinvolved in differentiation(Fig. 1)6. These expression sig-natures were quiescent andunaltered in the livers of anti-sense-treated animals, indicat-ing that distinct cAMP signal-ing pathways regulate growthfor normal and cancer cells.

A careful analysis of themicroarray data reveals thatthe antisense modulates manygenes that appear to have a ten-uous or no relationship withthe targeted gene (RIα)6. A fewexamples include remarkableupregulation of genes encod-ing Cdc42, RAP1A, and

cytoskeleton regulatory proteins, andmarked downregulation of genes codingfor MAP kinase 5, collagen type 4, catalase,and M-phase inducer phosphatase 2.

While recent clinical success with GentaPharmaceuticals’ (Berkeley Heights, NJ)Genasense against BCL-2 demonstrates thepromise of antisense as an adjunct to moreconventional chemotherapeutics, our resultsdemonstrate that downregulating a specificprotein will likely have unforeseen conse-quences in multiple cellular signaling path-ways, at least with phosphorothioateoligonucleotides. It is therefore crucial toexamine the antisense effect at the genomiclevel rather than at the level of a single targetgene. Our results indicate that microarraystudies can facilitate the study of oligonu-cleotide pharmacokinetics, sequence speci-ficity, non-sequence-specific effects, andtoxicity. Further adoption of this technologywill facilitate development of nucleic acidmedicines with higher target specificity andminimized side effects.

Yoon S. Cho-Chung,Cellular Biochemistry Section,

Basic Research Laboratory,National Cancer Institute,

Bethesda, MD and Kevin G. Becker,

DNA Array Unit,National Institute on Aging,

Baltimore, MD (chochung@helix.nih. gov).

1. Stein, C. & Cheng, Y.-A. in Principles and Practiceof Oncology. DeVita, V., Rosenberg, S., Hellman,S. (eds.) 3059–3074 (Lippincott, New York; 1997).

2. Agrawal, S. Trends Biotechnol. 14, 376–387(1996).

3. Bennett, C.F. Biochem. Pharmacol. 55, 9–19(1998).

4. Crooke, S. (ed.). Antisense Research andApplication. (Springer, New York; 1998).

5. Schena, M., Shalon, D., Davis, R.W. & Brown, P.O.Science 270, 467–470 (1995).

6. Cho, Y.S. et al. Proc. Natl .Acad. Sci. USA 98,9819–9823 (2001).

Figure 1. Molecular portrait of the prostate carcinoma revertedphenotype (reproduced from ref. 6). cDNA microarray analysisof the RIα antisense-induced expression profile shows up- anddownregulation of coordinately expressed gene clusters thatproduce the molecular portrait of a reverted tumor-cellphenotype. These studies also reveal that antisensemodulates many genes that appear to have little relationshipwith the targeted gene (RIα).

Delivering zinc fingers

To the editor:In his News & Views on advances in the useof zinc-fingers as DNA binding modules inartificial transcription factors in the Marchissue (Nat. Biotechnol. 21, 242–243, 2003),Ansari mentions “the loftier goal of usingartificial transcription factors as therapeu-tic agents.” In this context, he states severalchallenges, including the need to evade theimmune system, delivery, and the ability toregulate their function based upon intra-and extra-cellular signals.

For the delivery issue, transgenes encod-ing chimeric transcription factors can betransferred via retroviral gene therapy,although this method suffers from the

Rafael Rangel-Aldao,Polar Technology Center,

Empresas Polar,Los Cortijos de Lourdes,

Caracas, Venezuela(rafael.rangel@empresas-polar.com)

1. Kitano, H. Science 295, 1662–1664 (2002).2. Lee, T.I. et al. Science 298, 799–804 (2002).3. Milo, R. et al. Science 298, 824–827 (2002).4. Barabási, A.-L. & Albert, R. Science 286. 509–512

(1999).5. Luscombe, N.M. et al. Genome Biol. 3, 8,

Research 0040.1 (2002).6. Koonin, E.V. et al. Nature 420, 218–223 (2002).7. Jeong, H. et al. Nature 407, 651–654 (2000).8. Buchman, T.G. Nature 420, 246–251 (2002).9. Tausssig, R. et al. Nature 420, 703–706 (2002).

10. Hasty, J. et al. Nature 420, 224–230 (2002).11. Rosenwald, A. et al. N. Engl. J. Med. 346,

1937–1947 (2002).12. Bemporad, M. ASOVAC National Database,

1996–2002.

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reb

iote

chn

olo

gy