Dynamic proteomics in individual human cells uncovers widespread cell-cycle dependence of nuclear proteins Alex Sigal 1,6 , Ron Milo 1,2,6 , Ariel Cohen 1 , Naama Geva-Zatorsky 1 , Yael Klein 1 , Inbal Alaluf 1 , Naamah Swerdlin 1 , Natalie Perzov 1 , Tamar Danon 1 , Yuvalal Liron 1 , Tal Raveh 3 , Anne E Carpenter 4 , Galit Lahav 5 & Uri Alon 1,2 We examined cell cycle–dependent changes in the proteome of human cells by systematically measuring protein dynamics in individual living cells. We used time-lapse microscopy to measure the dynamics of a random subset of 20 nuclear proteins, each tagged with yellow fluorescent protein (YFP) at its endogenous chromosomal location. We synchronized the cells in silico by aligning protein dynamics in each cell between consecutive divisions. We observed widespread (40%) cell-cycle dependence of nuclear protein levels and detected previously unknown cell cycle–dependent localization changes. This approach to dynamic proteomics can aid in discovery and accurate quantification of the extensive regulation of protein concentration and localization in individual living cells. A long-term goal of biology is a quantitative understanding of gene and protein networks of human cells and their responses to stimuli. This requires the ability to accurately measure protein levels 1–4 and localizations 5–12 . An ideal proteomic measurement system would (i) work at the level of individual cells, as experiments that average over cell populations can overlook events happening in only a subset of cells, all-or-none effects and variability between cells; (ii) follow the same cell over extended periods of time to uncover phenomena such as oscillations and temporal programs; (iii) make minimal perturbations to the state of the cell and (iv) measure both protein levels and localization. One cellular process that produces widespread changes in the proteome is the cell cycle. Systematic studies of cell cycle–regulated genes using DNA arrays showed that about 10% of genes were cell cycle dependent in yeast and Arabidopsis thaliana 13–16 , compared with 2–3% of genes in human cells (whether primary fibroblast 17 or HeLa carcinoma 18 ). These differences may reflect fundamental differences between plant and yeast cells on one hand and human cells on the other, or increased experimental noise arising from problems in synchronization of human cells relative to yeast and A. thaliana. We examined the cell-cycle dependence of nuclear proteins (which constitute 90% of known cell cycle–dependent proteins 18 ) as a proof of principle for a new dynamic proteomics approach in individual living cells. Our approach was based on automated time- lapse microscopy and image analysis of a library of cells, each with a different fluorescently tagged protein expressed from its endogen- ous chromosomal location. We were able to use a non-perturbing method of synchronization, monitoring tagged protein levels in non-synchronized individual cells from one cell division to the next and retroactively aligning protein dynamics of all cells between consecutive division events. We discovered that in the random subset of nuclear proteins we examined, 40% showed cell cycle– dependent dynamics. RESULTS Construction of a YFP CD-tagged reporter clone library To measure protein level and localization in individual living cells, we developed a system for dynamic proteomics. We generated a library of cell clones in which each clone contained a different fluorescently tagged protein expressed from its endogenous chro- mosomal location. We fluorescently labeled endogenous proteins by the central dogma (CD) tagging approach 19,20 . For tagging, we used a retrovirus to insert the YFP coding region flanked by splice signals into the genome of H1299 lung carcinoma cells in a nondirected manner (Fig. 1a). When integrated in the proper frame and orientation into an intron of an expressed gene, the YFP was spliced into the mRNA as a new exon. This resulted in a full- length fluorescent protein fusion expressed from its endogenous chromosomal locus. The CD-tagging method allows construction of tagged mammalian cell libraries, resembling the library of endogenous GFP protein fusions constructed in yeast by homo- logous recombination 11 . Unlike gene trap methods 21–24 , this tag- ging approach results in a full-length protein. CD-tagging preserves wild-type localization and function in a large fraction of the tagged proteins 19,25,26 and avoids overexpression concerns associated with fluorescent protein fusions expressed from exogenous promoters. After tagging with the virus, we selected cells with tagged proteins by flow cytometric sorting of YFP-positive cells into multiwell plates; each tagged cell was expanded into a clone (Fig. 1b,c). In p u o r G g n i h s i l b u P e r u t a N 6 0 0 2 © e r u t a n / m o c . e r u t a n . w w w / / : p t t h s d o h t e m RECEIVED 22 MARCH; ACCEPTED 23 MAY; PUBLISHED ONLINE 21 JUNE 2006; DOI:10.1038/NMETH892 1 Department of Molecular Cell Biology and 2 Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot 76100, Israel. 3 Department of Developmental Biology, Stanford University, Palo Alto, California 94305-5439, USA. 4 Whitehead Institute for Biomedical Research, Cambridge, Massachusetts 02142, USA. 5 Department of Systems Biology, Harvard Medical School, Boston, Massachusetts 02115, USA. 6 These authors contributed equally to this work. Correspondence should be addressed to U.A. (uri.alon@weizmann.ac.il). NATURE METHODS | VOL.3 NO.7 | JULY 2006 | 525 ARTICLES