We also identified phenotype-associated SNPs causing the top 10% of structural changes and the lowest 10%. The phenotype-associated mutations with the greatest changes in DNA structure oc- curred significantly more often (P 10���2, Fisher���s exact test) in evolutionarily constrained regions of the genome (56% for high structure-change regions versus 29% for low structure-change re- gions) (21). This suggests that noncoding DNA may be under selective constraint, which may prevent changes in DNA structure. Because the severity of structural change might help identify functional SNPs, we constructed a database of changes in the structural profile for all known SNPs in the human genome (21). Finally, we demonstrated that structural changes affect biological function in noncoding evolutionarily constrained regions identified by the Chai algorithm. We chose 12 predicted enhancer-containing regions (24): 5 regions over- lap elements detected only by the Chai algorithm 7 regions overlap a combination of Chai- and binCons-detected elements (table S2). We cloned the 300 bp surrounding each of these genomic regions in a luciferase reporter construct and transfected them into 293T cells. Eight of the 12 constructs displayed luciferase activity that was significantly greater (P ��� 0.05, Wilcoxon rank sum test) than that of random control sequences (Fig. 4). Three of the five constructs overlapping only Chai-detected elements were positive (table S2). Given the plethora of regulatory functions that a genome encodes and the three-dimensional genomic architecture required to orchestrate these events (28), it may not be surprising that there is widespread conservation of local DNA topogra- phy. Perhaps only a subset of local structural configurations can accommodate the functional requirements of a genomic locus [for example, see (29)]. Once the molecular topography of a locus is permissive to a regulatory function, this structure may be maintained within the genome. Our high-resolution topography-based constraint- detection method reveals that structure-informed constraint is widespread in the human genome and that these regions overlap known noncoding func- tional sites. Because different DNA sequences can have similar local structures (20), these regions might escape detection with sequence-based conservation���identification methods. References and Notes 1. International Human Genome Sequencing Consortium, Nature 431, 931 (2004). 2. M. D. Wilson et al., Science 322, 434 (2008). 3. L. Elnitski et al., Genome Res. 13, 64 (2003). 4. M. Kellis, N. Patterson, M. Endrizzi, B. Birren, E. S. Lander, Nature 423, 241 (2003). 5. G. G. Loots et al., Science 288, 136 (2000). 6. L. A. Pennacchio, E. M. Rubin, Nat. Rev. Genet. 2, 100 (2001). 7. W. W. Wasserman, M. Palumbo, W. Thompson, J. W. Fickett, C. E. Lawrence, Nat. Genet. 26, 225 (2000). 8. E. H. Margulies, M. Blanchette, NISC Comparative Sequencing Program, D. Haussler, E. D. Green, Genome Res. 13, 2507 (2003). 9. A. Siepel et al., Genome Res. 15, 1034 (2005). 10. G. M. Cooper et al., Genome Res. 15, 901 (2005). 11. S. Asthana, M. Roytberg, J. Stamatoyannopoulos, S. Sunyaev, PLoS Comp. Biol. 3, e254 (2007). 12. S. Fisher, E. A. Grice, R. M. Vinton, S. L. Bessling, A. S. McCallion, Science 312, 276 (2006). 13. D. M. McGaughey et al., Genome Res. 18, 252 (2008). 14. H. M. Petrykowska, C. M. Vockley, L. Elnitski, Genome Res. 18, 1238 (2008). 15. The ENCODE Project Consortium, Nature 447, 799 (2007). 16. T. Ohyama, Ed., DNA Conformation and Transcription (Landes Bioscience/Eurekah.com, Georgetown, TX, 2005). 17. W. K. Olson, A. A. Gorin, X. J. Lu, L. M. Hock, V. B. Zhurkin, Proc. Natl. Acad. Sci. U.S.A. 95, 11163 (1998). 18. R. E. Dickerson, Methods Enzymol. 211, 67 (1992). 19. M. A. Price, T. D. Tullius, Methods Enzymol. 212, 194 (1992). 20. J. A. Greenbaum, B. Pang, T. D. Tullius, Genome Res. 17, 947 (2007). 21. Materials and methods are available as supporting material on Science Online. 22. E. H. Margulies et al., Genome Res. 17, 760 (2007). 23. Mouse Genome Sequencing Consortium, Nature 420, 520 (2002). 24. N. D. Heintzman et al., Nat. Genet. 39, 311 (2007). 25. N. P. Pavletich, C. O. Pabo, Science 252, 809 (1991). 26. M. L. Bulyk, P. L. F. Johnson, G. M. Church, Nucleic Acids Res. 30, 1255 (2002). 27. B. Giardine et al., Hum. Mutat. 28, 554 (2007). 28. T. Misteli, Bioessays 27, 477 (2005). 29. R. Joshi et al., Cell 131, 530 (2007). 30. We thank E. D. Green and L. C. Brody for feedback on the manuscript, E. Bishop and D. Landsman for discussion, and G. K. McEwen for assistance with experimental steps. Funded by the National Human Genome Research Institute (NHGRI) of the NIH (grant R01 HG003541) to T.D.T. and by the Research Corporation for Science Advancement. E.H.M. was supported by the Intramural Research Program of the NHGRI, NIH. S.C.J.P. was supported by a National Academies Ford Foundation Dissertation Fellowship. Supporting Online Material www.sciencemag.org/cgi/content/full/1169050/DC1 Materials and Methods Figs. S1 to S5 Tables S1 and S2 References 26 November 2008 accepted 23 February 2009 Published online 12 March 2009 10.1126/science.1169050 Include this information when citing this paper. In Vivo Analysis of Dendritic Cell Development and Homeostasis Kang Liu,1* Gabriel D. Victora,1��� Tanja A. Schwickert,1��� Pierre Guermonprez,1 Matthew M. Meredith,1 Kaihui Yao,1 Fei-Fan Chu,1 Gwendalyn J. Randolph,2 Alexander Y. Rudensky,3,4 Michel Nussenzweig1,4* Dendritic cells (DCs) in lymphoid tissue arise from precursors that also produce monocytes and plasmacytoid DCs (pDCs). Where DC and monocyte lineage commitment occurs and the nature of the DC precursor that migrates from the bone marrow to peripheral lymphoid organs are unknown. We show that DC development progresses from the macrophage and DC precursor to common DC precursors that give rise to pDCs and classical spleen DCs (cDCs), but not monocytes, and finally to committed precursors of cDCs (pre-cDCs). Pre-cDCs enter lymph nodes through and migrate along high endothelial venules and later disperse and integrate into the DC network. Further cDC development involves cell division, which is controlled in part by regulatory T cells and fms-like tyrosine kinase receptor-3. Dpresent endritic cells (DCs) are immune cells that are specialized to capture, process, and antigens to T lymphocytes in or- der to induce immunity or tolerance (1). Where commitment to DC development takes place, at what stage the monocyte lineage diverges from DCs, and the precise nature of the migrating DC precursor that moves from the bone marrow to Fig. 4. Luciferase-based reporter activity of 12 re- gions containing Chai- detected elements (21). These regions overlapped predicted enhancer regions. Plotted for each element is the luciferase activity rela- tive to the median activity from 100 random control constructs (y axis see fig. S5). Error bars represent 1 SD from the mean of four experimental replicates, and asterisks denote P ��� 0.05 (Wilcoxon rank sum test). Control 1 5 7 10 11 12 Luciferase activity * * * * * * * * 0.1 1 10 100 Reporter construct 2 3 4 6 8 9 17 APRIL 2009 VOL 324 SCIENCE www.sciencemag.org 392 REPORTS

the peripheral lymphoid organs are not known. These questions have been difficult to resolve in part because DC subsets are functionally and phenotypically diverse (2). For example, classi- cal spleen DCs (cDCs) include two major func- tionally distinct subsets that are distinguished by the expression of a variety of C-type lectins and CD8 (2���4). Spleen and other tissues also contain plasmacytoid DCs (pDCs) that primar- ily initiate immune responses to nucleic acids (5, 6). Lymphoid tissue cDCs, pDCs, and monocytes share a common progenitor called the macrophage and DC precursor (MDP) that is identified by its surface phenotype (Lin��� cKithi CD115+ CX3CR1+ Flt3+) (7, 8), whereas a distinct progenitor called the common DC precursor (CDP) (Lin��� cKitlo CD115+ Flt3+) is restricted to producing cDCs and pDCs (9, 10). Although monocytes can de- velop many of the phenotypic features of DCs under inflammatory conditions (11���13), the cDC, pDC, and monocyte lineages separate by the time they reach tissues, and neither monocytes nor pDCs develop into cDCs under steady-state conditions (8, 14). Unlike monocytes and pDCs, cDCs in lymphoid tissues are thought to emerge from the bone marrow as immature cells that must further differentiate and divide in lymphoid organs (15, 16). Consistent with this idea, pre-cDCs that are re- stricted to the cDC lineage were isolated from the spleen and bone-marrow cultures containing fms- like tyrosine kinase receptor-3���ligand (Flt3-L) (17, 18). However, the relationship between MDP, CDP, and pre-cDC in vivo and the question of where the monocyte, pDC, and cDC lineages split have not been addressed (10, 14, 18). We searched for MDPs and CDPs in the blood and spleen by means of flow cytometry but could only detect them in the bone marrow (Fig. 1A and fig. S1). Although pre-cDCs can be identified in the spleen by combining density centrifugation and flow cytometry (18), we speculated that these cells could be identified directly by expression of Flt3 and the chemokine receptor CX3CR1, which are expressed on other DC progenitors and also on mature cDCs (7, 10, 19). Indeed, we found a small but distinct population of lineage-negative CD11c+ 1 Laboratory of Molecular Immunology, Rockefeller Uni- versity, New York, NY 10065, USA. 2Department of Gene and Cell Medicine, Mount Sinai School of Medicine, New York, NY 10029, USA. 3Department of Immunology and Howard Hughes Medical Institute (HHMI), University of Washington, Seattle, WA 98195, USA. 4HHMI, Rockefeller University, New York, NY 10021, USA. *To whom correspondence should be addressed. E-mail: liuk@rockefeller.edu (L.K.) nussen@rockefeller.edu (M.N.) ���These authors contributed equally to this work. Fig. 1. Isolation of pre-cDCs. (A) Presence of MDPs and CDPs in the bone marrow (BM), blood, or spleen (SP). NF, not found. (B) Identification of pre-cDCs in bone marrow, LNs, spleen, and blood. The bar graph shows the percentage of pre-cDC in each organ. 1 �� 106 cells were acquired per sample. Lin-indicates cells that do not express CD3, CD19, Ter119, NK 1.1, or B220 antigens. (C) Donor-derived spleen cells (CD45.2+) were analyzed for cDC (CD11c+ MHC class II+), pDC (CD11cint B220+), and monocytes (CD11b+ CD11clo/���) 7 days after intravenous injection of 2 �� 106 bone marrow cells or 1 �� 105 pre-cDCs. (D) Analysis of donor-derived splenocytes after transfer with the indicated number of pre-cDCs from bone marrow, spleen, and blood. (E) Chimerism in parabiotic mice. In (B) and (C), the left side shows representative dot plots. The bar graphs summarize 2 to 4 independent experiments with 3 to 4 mice each. Error bars represent the mean T SEM. www.sciencemag.org SCIENCE VOL 324 17 APRIL 2009 393 REPORTS