26. MegAlign program, DNASTAR, Madison, WI.27. L. Redmond, S.-R. Oh, C. Hicks, G. Weinmaster, A.Ghosh, Nature Neurosci. 3, 30 (2000).28. To create the CREST antibody, We used PCR toamplify full-length crest cDNA from P0 mousebrain total RNA and cloned it into pET21 vector forexpression of His-tagged CREST protein in bacteria.After isopropyl-␤-D-thiogalactopyranoside induc-tion, CREST protein was mainly collected in theinclusion body of the bacteria and extracted with 6M urea. We further purified the protein with Ni-NTAagarose fractionation. After SDS–polyacrylamide gelelectrophoresis (PAGE), the Coomassie brilliant blue–stained CREST band was excised and injected intorabbits to produce anti-CREST antiserum.29. To generate the crest knockout mice, crest genom-ic DNA was cloned by screening a 129SVJ mousegenomic phage library (gift of A. Kolodkin) withfull-length mouse crest cDNA as a probe. Aftermapping the restriction enzyme sites, we generat-ed a knockout vector with a Neo gene cassette(Fig. 4). We deleted the poly(A) addition signalfrom the Neo gene cassette for poly(A) trappingmethod of gene targeting. The cassette replaced allof exon 4 and a 5⬘ portion of exon 5. ES cellstransfected with the targeting vector and resistantto G418 and gancyclovir were expanded andscreened by genomic Southern blotting. Correctlytargeted ES cells were injected into C57B6J-de-rived blastocysts and resulted in the generation ofseveral high-percentage chimeras, which producedgermline targeted offspring. Genotyping of micewas performed on tail clip DNA by PCR.30. We thank D. Livingston for the UAS-CAT construct; A.Lanahan for advice on library construction; A. Kolod-kin for rat phage libraries; L. Redmond, A. Datwani, K.Whitford, M.-R. Song, and G. Ince for various proce-dures; J. Nathans, C. Montell, D. Ginty, P. Worley, S.Snyder, D. Linden, D. Murphy, P. Kim, M. Molliver fordiscussions; and M. Greenberg, D. Ginty, A. Kolodkin,and S. Snyder for comments on the manuscript.Supported by grants from NIH (MH60598 andNS39993), the March of Dimes Birth Defects Foun-dation (A.G.), the Klingenstein Foundation (A.G.), anda Merck Scholar Award (A.G.). H.A. was supported bya Uehara Memorial Foundation Research Fellowship.Supporting Online Materialwww.sciencemag.org/cgi/content/full/303/5655/197/DC1Materials and MethodsFigs. S1 to S330 July 2003; accepted 28 October 200314C Activity and Global CarbonCycle Changes over the Past50,000 YearsK. Hughen,1* S. Lehman,3J. Southon,4J. Overpeck,5,6O. Marchal,2C. Herring,1J. Turnbull3A series of14C measurements in Ocean Drilling Program cores from the tropicalCariaco Basin, which have been correlated to the annual-layer counted chronologyfor the Greenland Ice Sheet Project 2 (GISP2) ice core, provides a high-resolutioncalibration of the radiocarbon time scale back to 50,000 years before the present.Independent radiometric dating of events correlated to GISP2 suggests that thecalibration is accurate. Reconstructed14C activities varied substantially during thelast glacial period, including sharp peaks synchronous with the Laschamp and MonoLake geomagnetic field intensity minimal and cosmogenic nuclide peaks in ice coresand marine sediments. Simulations with a geochemical box model suggest thatmuch of the variability can be explained by geomagnetically modulated changesin14C production rate together with plausible changes in deep-ocean ventilationand the global carbon cycle during glaciation.Radiocarbon age may deviate significantlyfrom calendar age as a result of time-varyingprocesses affecting14C production in the atmo-sphere, as well as the distribution of14C amongthe active global carbon reservoirs (1). To ac-count for such changes, radiocarbon age deter-minations must be calibrated against indepen-dent estimates of calendar age, but existingcalibration data sets often lack temporal rangeand/or resolution. The current standard calibra-tion, Intcal98 (2), extends at high resolutionback to just ⬃14,600 calendar years before thepresent (14.6 cal. ka B.P.) on the basis of annualtree rings (3) and varved (annually layered)marine sediments (4). Paired14C and U/Thages on corals (5) provide additional calibrationpoints back to ⬃40 cal. ka B.P., but at muchlower resolution. Varved lake sediments (6),U/Th ages on speleothems (7) and lake sedi-ments (8), and marine sediments correlated toGreenland ice core chronologies (9, 10) havealso been used to constrain calibration and ini-tial14C activity [expressed as ⌬14C(11)] be-yond the range of Intcal98. In many cases, theserecords suggest that extremely large and rapidshifts in ⌬14C have occurred; however, theserecords also show disagreements before ⬃25cal. ka B.P. that are as large as the reconstructedanomalies. Thus, considerable uncertainty re-mains in calibrating the older half of the14Ctime scale. Here, we present a calibration andreconstruction of ⌬14C back to 50 cal. ka B.P.on the basis of the correlation of14C data fromCariaco Basin sediments with the annual-layertime scale of the GISP2 Greenland ice core(12). Similarity between reconstructed ⌬14Cand variations in14C production rate estimatedfrom independent paleomagnetic and geochro-nologic data suggests that the calibration and⌬14C reconstruction are accurate despite thelack of in situ calendric age control.Our14C series (Fig. 1) is constructed from280 accelerator mass spectrometry (AMS)14Cmeasurements on planktonic foraminifera ex-tracted from discrete sediment samples in holes1002D and 1002E from Ocean Drilling Program(ODP) leg 165, site 1002, in the Cariaco Basin(10°42.73⬘N, 65°10.18⬘W, 893-m water depth).The results span a14C age range of 55 to 12 kaB.P. and complement 355 varve-age calibrated14C measurements for the interval from ⬃15 to10 cal. ka B.P. based on our prior studies ofnearby Cariaco Basin sediment piston cores (4,13). AMS14C target preparation and measure-ment was conducted at three different institu-tions: Center for Accelerator Mass Spectrometryat Lawrence Livermore National Laboratory(CAMS-LLNL) (n ⫽ 127), Laboratory forAMS Radiocarbon Preparation and Research atUniversity of Colorado, Boulder, and NationalOcean Sciences AMS at WHOI (NSRL-NOSAMS) (n ⫽ 118), and the Keck CCAMSFacility at University of California, Irvine (n ⫽35) (Fig. 1 and table S1) (14). A constant 420-year marine reservoir age correction (differencebetween14C ages of surface water and atmo-sphere) was applied to all14C ages, in accor-dance with