CORRESPONDENCE

four times higher than in MEFs expressing p53QS-

Val135 (Fig. 1b). This difference is not attributable to differences in half-life (T1/2 of all Trp53QS pro-teins was >8 h, whereas T1/2 of wild-type p53 was <20 min; Fig. 1c,d). Trp53Lsl-QS and wild-type littermate cells do express more Trp53 mRNA (∼1.5-fold more), possibly reflecting strain dif-ferences (Fig. 1a). p53QS may be translated more efficiently than p53QS-Val135, or other factors may contribute to its greater abundance.

We have previously reported that germ-line heterozygosity or homozygosity for Trp53QS-Val135 allows normal survival but causes substantial tumor predisposition equivalent to Trp53 heterozygous and homozygous null mice2. By contrast, hetero-zygosity for Trp53QS causes early embryonic lethality1. What can explain the life-or-death differences caused by the presence of valine or alanine at codon 135, which is at the dimer interface of the p53 DNA-binding domain6,7?

Johnson et al. propose that the early lethality caused by Trp53QS results from hypoxia-induced activation during early embryogenesis. Although p53QS does not show higher apoptotic activity than wild-type p53 under hypoxic conditions1, and p53 normally shows low apoptotic activity during develop-ment8, it is conceivable that the abnormally high levels of this Mdm2-resistant p53 mutant could induce apoptosis. However, Johnson et al. did not examine Trp53QS/– embryos, and thus it remains possible that embryonic lethal-

ity requires the wild-type Trp53 allele. Because we compared p53QS-Val135 and p53QS in homo-zygous cells, we have no data on potential stress-activated effects of p53QS on wild-type p53 in heterozygous cells.

We also reported that p53QS-Val135 binds nor-mally to the chromatin of some p53 target genes but binds weakly to others3. By contrast, p53QS binds very strongly in the presence and absence of stress to all p53 target genes analyzed3. These observations suggest two additional models to explain the lethality of the Trp53QS allele. First, in Trp53QS/+ embryos, p53QS might augment or alter the function of the wild-type protein during early embryonic development or in spe-cific tissues. A dominant effect could arise from increased activity or stability of chromatin-bound wild-type and p53QS heterotetramers, as p53QS does not bind Mdm2 or MdmX. Second, transcriptionally inactive p53QS homotetramers may form and bind to chromatin but may not be cleared from the promoters of the hundreds of estimated p53 target genes9 and intragenic and intergenic loci containing p53 binding sites10. Thus, the early lethality in Trp53QS/+ embryos may be attributable not to a stress-activated func-tion of this mutant allele but rather to the consti-tutive presence of an inactive transcription factor at numerous target loci. A contributing factor could be that the p53QS protein was substantially more abundant in the strain used by Johnson et al. than in the strain used in our studies.

Finally, we note that Johnson et al. did not analyze the tumor suppressor capacity of

p53QS. Thus, we do not know whether the partial function of p53QS in vivo is restricted to the conditions that occur during embry-onic development, or if these alterations to the transactivation domain also influence p53 tumor suppression.

Mengjia Tang & Geoffrey M Wahl

The Salk Institute for Biological Studies, Gene Expression Laboratory, 10010 N. Torrey Pines Road, La Jolla, California 92037, USA.

Monica Nistér

Department of Oncology-Pathology, Karolinska Institutet, Karolinska University Hospital, 171 76 Stockholm, Sweden, and Department of Genetics and Pathology, Rudbeck Laboratory, University of Uppsala, 751 85 Uppsala, Sweden.e-mail: wahl@salk.edu

Note: Supplementary information is available on the Nature Genetics website.

1. Johnson, T.M., Hammond, E.M., Giaccia, A. & Attardi, L.D. Nat. Genet. 37, 145–152 (2005).

2. Jimenez, G.S. et al. Nat. Genet. 26, 37–43 (2000).3. Nister, M. et al. Oncogene 24, 3563–3573 (2005).4. Martinez, J., Georgoff, I. & Levine, A.J. Genes Dev. 5,

151–159 (1991).5. Milner, J. & Medcalf, E.A. J. Mol. Biol. 216, 481–484

(1990).6. Cho, Y., Gorina, S., Jeffrey, P.D. & Pavletich, N.P. Science

265, 346–355 (1994).7. Zhao, K., Chai, X., Johnston, K., Clements, A. &

Marmorstein, R. J. Biol. Chem. 276, 12120–12127 (2001).

8. Gottlieb, E. et al. EMBO J. 16, 1381–1390 (1997).9. Vogelstein, B., Lane, D. & Levine, A.J. Nature 408,

307–310 (2000).10. Cawley, S. et al. Cell 116, 499–509 (2004).

In reply:Our laboratory recently generated conditional Trp53 knock-in mice in which codons 25 and 26 were mutated to glutamine and serine (QS) to address the role of transactivation in p53 function1. Our Trp53QS mice have a different phenotype from those described by Wahl and colleagues2, which also carry a mutation in the DNA binding domain at codon 135 (Trp53QS-

Val135)3. Our analysis of Trp53QS mice demonstrates that this mutant retains select p53 activities. In contrast, Wahl and colleagues show that the Trp53QS-Val135 allele behaves indistinguishably from a Trp53-null allele.

Analyses of p53QS by our laboratory and others have unequivocally indicated that p53QS retains partial transactivation function, as well as activity in apoptosis and growth suppression1,4–8. In cells derived from our knock-in mice, we showed that although p53QS is compromised for transactivation of many target genes, it retains near–wild-type activity on certain targets such as Bax (ref. 1) and Apaf-1 (data not shown). Furthermore, we showed that p53QS induces

substantial apoptosis in response to select stresses, including serum starvation and hypoxia. Consistent with our findings, Tang et al. observed Bax transactivation using our Trp53QS MEFs, but not with their Trp53QS-

Val135 MEFs. They did not, however, investigate the apoptotic response to serum deprivation or hypoxia, conditions in which we observed p53QS apoptotic activity.

A vast body of literature has shown that the A135V mutation severely compromises p53 function. The seminal studies establishing p53 as a tumor suppressor showed that while co-introduction of p53 with Ras and E1A into rat embryonic fibroblasts suppressed transformation, expression of p53Val135 not only failed to suppress transformation but actually enhanced transformation, indicating dominant-negative activity toward endogenous p53 (refs. 9,10). Further evidence for the inactivity of p53Val135 has accumulated in numerous subsequent studies. For example, Trp53-null mice expressing multiple copies of a Trp53Val135 transgene develop tumors at the same rate as Trp53-null mice lacking the

transgene, indicating that p53Val135 is unable to suppress tumorigenesis11. Moreover, the experiments by Tang et al. showing that p53QS-expressing MEFs display Bax transactivation, whereas those expressing p53QS-Val135 do not, further demonstrate that p53QS-Val135 is inactive. Given that the A135V mutation severely compromises p53 function, it is impossible to conclude that transactivation is essential for p53 function from analysis of Trp53QS-Val135 mice.

An additional consequence of the QS mutations is that p53QS no longer binds the Mdm2 ubiquitin ligase and therefore shows enhanced stability relative to wild-type p53 (ref. 12). Because unrestrained p53 in the context of Mdm2 deficiency induces early embryonic lethality13, we hypothesized that, should p53QS retain biological activity, its expression might similarly trigger lethality; hence, we created our mice using a conditional strategy. Indeed, expression of p53QS causes embryonic lethality1, whereas homozygous Trp53QS-Val135 mice are viable2. The most straightforward interpretation of

396 VOLUME 38 | NUMBER 4 | APRIL 2006 | NATURE GENETICS

©20

06 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

ureg

enet

ics

CORRESPONDENCE

NATURE GENETICS | VOLUME 38 | NUMBER 4 | APRIL 2006 397

Enrichment of regulatory motifs upstream of predicted DAF-16 targets

To the Editor:In their recent study, Oh et al.1 mistak-enly reported that only 11% of the 514 genes whose expression was most strongly and consistently influenced by DAF-16 activity in our previously published microarray analysis2 contain the sites GTAAAc/tAA and/or CTTATCA in the 5-kb region upstream of their promoters. As we described2, we identified these two

sequences as potential regulatory targets because they were overrepresented in the 1-kb upstream promoter regions of the top ∼50 of these 514 genes (prioritized solely according to the magnitude and consis-tency of their expression profile changes). We never examined whether these two sequences were also present in the remain-ing ∼460 genes from our original list of 514 genes. We have done this now using public

databases (Caenorhabditis elegans EnsEMBL Build 33; http://rsat.ulb.ac.be/rsat/), and we find that 88% (rather than 11%) of these 514 genes contain one or both sites in their 5-kb upstream regulatory regions. In fact, 57 of the 58 genes (98%) that we chose for further study (based solely on the strength and consistency of their expression pro-files) contain these elements in their 5-kb upstream regulatory regions. Because only

these findings is that retention of select p53 functions, such as induction of apoptosis in response to hypoxic stress, in the absence of regulation by Mdm2, accounts for p53QS-induced embryonic lethality, an idea supported by both our MEF experiments and the literature. Importantly, our MEF experiments have shown that p53QS does not display activity unless subjected to stress1, indicating that p53QS retains crucial regulation in the absence of Mdm2 binding and that it does not act in a nonspecific manner.

Tang et al. advance two alternative hypotheses for the observed p53QS-induced embryonic lethality. First, they propose that activation of wild-type p53 by the p53QS mutant induces lethality. Although formally possible, this is highly improbable, as expression of p53QS along with wild-type p53 in MEFs does not enhance target gene expression, either in the absence or presence of DNA damage, relative to Trp53+/− MEFs

(Fig. 1). This question is also readily addressed through breeding experiments to determine if Trp53QS/– embryos display lethality, studies that are currently in progress.

Second, Tang et al. posit that inappropriate occupancy of p53 response elements by transcriptionally inactive p53QS could interfere with p53 target gene expression, leading to embryonic lethality. This phenomenon would be expected to phenocopy a Trp53-null phenotype, or possibly a deficiency in the p53 family members p63 and p73. However, as genetic deficiency of Trp53, p63 or p73 does not result in early embryonic lethality14, this represents an unlikely cause of death in our embryos.

Instead, the lethality we observe is consistent with the finding that the p53QS mutant retains biological activity and is not normally regulated by Mdm2, phenocopying the early embryonic lethality seen in the presence of wild-type p53 in an Mdm2−/− background13 (Supplementary Fig. 1 online).

Our phenotypic differences are not likely to be due to nonspecific, gain-of-function activities of the p53QS protein in our knock-in mice, as suggested by Tang et al., but rather to the complete loss of p53 function resulting from the presence of the additional mutation in the Trp53QS-Val135 mice generated by Wahl and colleagues. The continued analysis of Trp53QS mice will prove highly informative for defining the role of p53 transactivation for tumor suppression in a variety of physiologically relevant contexts.

Thomas M Johnson

Stanford University School of Medicine, Department of Radiation Oncology, Stanford, California 94305, USA.

Laura D Attardi

Stanford University School of Medicine, Departments of Radiation Oncology and Genetics, Stanford, California 94305, USA. e-mail: attardi@stanford.edu

Note: Supplementary information is available on the Nature Genetics website.

1. Johnson, T.M., Hammond, E.M., Giaccia, A. & Attardi, L.D. Nat. Genet. 37, 145–152 (2005).

2. Jimenez, G.S. et al. Nat. Genet. 26, 37–43 (2000).3. Nister, M. et al. Oncogene 24, 3563–3573 (2005).4. Haupt, Y., Rowan, S., Shaulian, E., Vousden, K.H. &

Oren, M. Genes Dev. 9, 2170–2183 (1995).5. Zhu, J., Zhou, W., Jiang, J. & Chen, X. J. Biol. Chem. 273,

13030–13036 (1998).6. Baptiste, N., Friedlander, P., Chen, X. & Prives, C.

Oncogene 21, 9–21 (2002).7. Venot, C., Maratrat, M., Sierra, V., Conseiller, E. &

Debussche, L. Oncogene 18, 2405–2410 (1999).8. Sabbatini, P., Lin, J., Levine, A.J. & White, E. Genes Dev.

9, 2184–2192 (1995).9. Finlay, C.A., Hinds, P.W. & Levine, A.J. Cell 57, 1083–

1093 (1989).10. Levine, A.J., Finlay, C.A. & Hinds, P.W. Cell 116, S67–

S69 (2004).11. Harvey, M. et al. Nat. Genet. 9, 305–311 (1995).12. Lin, J., Chen, J., Elenbaas, B. & Levine, A.J. Genes Dev.

8, 1235–1246 (1994).13. Montes de Oca Luna, R., Wagner, D.S. & Lozano, G.

Nature 378, 203–206 (1995).14. Yang, A., Kaghad, M., Caput, D. & McKeon, F. Trends

Genet. 18, 90–95 (2002).

Perp

Cdkn1a

Mdm2

Gapd

Trp53

LSL-

QS-w

t/LSL-

QS-w

t

Trp53

LSL-

LW/L

SL-

LW

Trp53LSL-QS-wt/+ Trp53

LSL-

QS-w

t/LSL-

QS-w

t

Cre

Empty

Dox +++

++++

+++

+–

– –

– –

–

––––

p53 status: +/– +/– QS/+ QS/+ QS/

QS

+/+ –/–

Figure 1 p53QS does not hyperactivate wild-type p53. Trp53LSL-QS/+ MEFs were infected with adenoviruses either expressing Cre (adeno-Cre) or not expressing Cre (adeno-empty) and were either left untreated or were treated with 0.2 µg/ml doxorubicin for 8 h. Trp53LSL-QS/+ MEFs infected with adeno-Cre, but not treated with DNA damage, do not show upregulation of the p53 target genes Perp, Cdkn1a or Mdm2, relative to Trp53LSL-QS/+ MEFs infected with adeno-empty, consistent with the hypothesis that p53QS is not hyperactivating the wild-type p53 protein. In addition, Trp53LSL-

QS/+ MEFs infected with either adeno-empty or adeno-Cre show similar target gene upregulation in response to DNA damage. Trp53LSL-QS/LSL-QS MEFs and Trp53LSL-LW/LSL-LW MEFs infected with adeno-Cre and Trp53LSL-QS/LSL-QS MEFs infected with adeno-empty serve as controls. Gapd serves as a loading control. On average, more than 95% of cells infected with adeno-Cre express p53QS protein, as detectable by immunofluorescence staining (data not shown).

©20

06 N

atur

e P

ublis

hing

Gro

up

http

://w

ww

.nat

ure.

com

/nat

ureg

enet

ics