The global distribution of chromosomes relative to each other after cell division was followed in HeLa cells expressing histone H2B-GFP with the use of local photobleaching (Kanda et al., 1998
). The majority of histone H2B-GFP molecules is incorporated into chromatin and exhibits limited mobility. Therefore, patterns induced by local photobleaching in nuclei of living cells are relatively stable and can be followed in cells over time (Kimura and Cook, 2001
). We followed cells with photobleached patterns through division and observed the return of the patterns in the nuclei of both daughter cells (). Our approach differed from that of previous work with respect to the time the cells were followed through division and with respect to the size of the nuclear area that was marked and followed (Gerlich et al., 2003
; Walter et al., 2003
; Thomson et al., 2004
). We marked global chromosome position by bleaching less than half of the nucleus and then followed the position of chromatin from interphase through mitosis to the next interphase. We observed that the photobleached patterns in chromatin of interphase nuclei were reestablished in the interphase nuclei of both daughter cells. In contrast when almost all H2B-GFP fluorescence is bleached leaving only a small region at the periphery of the nucleus, the reappearance of this pattern in daughter cells after division is not always observed (Walter et al., 2003
). In addition, monitoring small chromosomal loci through binding of GFP-LacI, reveals no evidence for stable relative position through mitosis (Thomson et al., 2004
). Consistent with our observations, marking large regions of the nucleus, patterns are conserved from prophase to metaphase as well as in anaphase and telophase when followed in relatively condensed chromosomes for only a short time period around mitosis (Gerlich et al., 2003
Figure 1. Time-lapse imaging of locally photobleached histone H2B-GFP–expressing cells. Hela cells stably expressing histone H2B-GFP were subjected to a local bleach pulse (area indicated by the dotted line) and imaged in 4D through mitosis. (A) An interphase (more ...)
The reappearance of the photobleached pattern of histone H2B-GFP in daughter nuclei indicated that global chromosomal organization is reestablished after division. There is some uncertainty in this method as newly synthesized histone H2B can exchange and redistribute, albeit to a limited extent, in the nucleus (Kimura and Cook, 2001
). Therefore, we also used a completely independent method to mark the DNA of chromosomal domains by local application of UV light through nitrocellulose filters with 5-μm pores (Mone et al., 2001
; Volker et al., 2001
; Hoogstraten et al., 2002
). For these experiments we used CHO cells in which UV-induced cyclobutyl pyrimidine dimers (CPDs) persist in DNA because they are not effectively removed by nucleotide excision repair (Hwang et al., 1998
). The position of the UV-marked chromosomal domains was visualized in fixed cells by immunofluorescence with CPD-specific antibodies (). Metaphase chromosomes from locally irradiated cells stained for CPDs confirmed that this method marked chromosomal domains () in a manner analogous to irradiation of cells with a laser-UV-microbeam (Cremer et al., 1982
). In addition, this technique reveals nuclear organization by marking domains on separate chromosomes that were in close proximity at the time of irradiation (for example, see ). These cells were also stained for PCNA, a protein involved in both replicative and repair DNA synthesis. Simultaneous monitoring of CPDs and PCNA by immunofluorescence revealed that CPDs and PCNA could colocalize (). On top of the normal PCNA staining pattern that reveals the specific stage of the cell cycle (Leonhardt et al., 2000
), we observed accumulation of PCNA at the same position as the CPDs. Therefore, PCNA could be used as marker for the position of CPD-containing chromosomal domains.
Figure 2. Visualization of chromosomes locally marked by UV-light in interphase nuclei of PCNA-GFP–expressing cells. (A) Local accumulation of CPDs in interphase cells. Cells were covered with a UV-blocking membrane containing 5-μm-wide pores and (more ...)
This information allowed us to follow CPD-marked chromosomal domains by PCNA-GFP accumulation in living cells. We performed time-lapse imaging of locally irradiated cells to follow the fate of CPD-marked chromosomal domains through cell division. shows an example of a cell, marked in early/mid-S phase with a local concentration of PCNA at the site of UV irradiation, that subsequently progressed through cell division (see also Supplementary Online Movie 1). Although the local PCNA accumulation at the site of CPDs disappeared in late S phase, where the prominent accumulation of PCNA is in replication foci, it was reestablished after division when the cells were in the next S phase. Importantly, this occurred in both daughter cells. We conclude that the locally induced CPDs that were in different, but neighboring, chromosomal domains return to the same position relative to each other in both daughter cells.
Figure 3. Transmission of chromosomes containing locally induced CPDs through mitosis into daughter nuclei. (A) Time-lapse imaging of PCNA-GFP in locally irradiated cells. Local UV irradiation was applied to a PCNA-GFP–expressing cell that was at the transition (more ...)
Similarly, when visualizing CPDs in fixed interphase cells, neighboring cells were observed with similar staining patterns (). The proximity of these cells suggests that they are daughters because they are too close to have resulted from irradiation through separate pores. To determine whether this was the case, we developed a protocol to make in situ metaphase spreads. If neighboring cells were daughters, then neighboring metaphase spreads should display an identical distribution of CPDs on their chromosomes. Indeed, using this protocol we detected neighboring metaphases with identical CPD distributions (). Furthermore, the frequency of neighboring metaphases containing identical CPD patterns should increase when the time between irradiation and preparation of the metaphase spread is increased because more cells will have had the opportunity to divide. Indeed, the percentage of CPD-positive metaphases having a neighbor with an identical CPD pattern increased from 8% at 2 h to 28% at 16 h after irradiation. Thus by marking the DNA in neighboring chromosomal domains in an arbitrary region of the nucleus using UV irradiation, we show by another independent and higher resolution method that the position of chromosomal domains can reappear in a similar relative organization in daughter nuclei.
Although, the relative position of chromosomal domains marked by DNA damage is reestablished in daughter cells after mitosis, the local DNA damage itself could influence nuclear location. In addition, the PCNA-GFP signal was not informative as a chromosomal domain marker at all points in the cell cycle. To follow chromosomal domain positions through the cell cycle, as well as use a method other than DNA damage for marking nuclear location, we performed the following live cell imaging experiment. Cells expressing PCNA-GFP were injected with fluorescently labeled nucleotide precursors (Zink et al., 1998
). This resulted in nuclei in which both PCNA and DNA can be monitored by distinct fluorescent signals in living cells. The cell cycle stage was identified unambiguously for individual cells from the PCNA pattern. The fluorescent DNA was specifically photobleached, while leaving the PCNA-GFP signal intact, to arbitrarily mark chromosomal neighborhoods based on their position at the time of bleaching. The marked chromosomal positions were followed through mitosis by time-lapse imaging of cells ( and Supplementary Online Movie 2). In all cases where cells could be followed through mitosis the general pattern of bleached and unbleached DNA in the daughters resembled that in their mother (). summarizes the cell cycle parameters for 21 individual cells. The time needed to progress through the cell cycle varied greatly for individual cells, emphasizing the necessity for a single cell marker of cell cycle stage such as PCNA. Classifying cell cycle stages by population average for the cell division cycle would have resulted in many misclassifications.
Cell cycle stage at the beginning and end of imaging period and duration of imaging
The experiment described above yielded interesting additional information on relative chromosomal domain positioning. As observed in the histone H2B-GFP photobleaching and the local UV irradiation, daughter cells recapitulated chromosomal domain position patterns marked in the mother cell. However, by following the DNA directly, we observed that there was a portion of the cell cycle, early in G1, where the pattern of bleached and unbleached chromosomal domain positions was lost. Analysis of confocal stacks confirmed that the loss of pattern was not simply due to rotation of the nuclei. To analyze the apparent loss of relative chromosomal domain position more quantitatively, we used the PCNA-GFP signal as an indicator of the total nuclear area and compared that to the portion of the nuclear area covered by Alexa546 fluorescently labeled DNA, representing the localization of the marked chromatin (). After photobleaching the Alexa546 signal covered half of the nuclear area (, time 0 min), at mitosis when the cells rounded up both PCNA-GFP and DNA-Alexa546 signals were lost from view. In early G1 the cells adhered to the coverslip and coherent images of their nuclei could again be obtained. This quantification reassuringly resulted in detection of half of the original PCNA-GFP and DNA-Alexa546 fluorescence in the daughter cells (; compare total fluorescent area at time 0 and 180 min).
For 5 of 21 cells imaged through G1 there were time points in early G1 when the DNA-Alexa546 pattern was lost. DNA-Alexa546 fluorescence was no longer limited to half of the nucleus but often coincided almost completely with the PCNA signal representing the complete nuclear area (, 90- and 100-min image and corresponding time points in ). In the remaining 16 cells this early stage of G1 was missed because of the 10-min intervals between image acquisitions. This mixing of PCNA and DNA-Alexa546 fluorescence was observed in all five examples where at least one daughter cell was followed throughout the very early portion of G1. In addition, the apparent loss of organization of the DNA-Alexa546 pattern is unlikely to be due to changes in nuclear size and shape in early G1. At the time points where the DNA-Alexa546 fluorescence was no longer limited to half of the nucleus, the cells were clearly no longer in mitosis. They were reattached to the coverslip and their nuclei had again flattened, as evident by detection of PCNA fluorescence in an area about half as large as in the mother cell (see , 110 min and recovery of PCNA fluorescent area by 100–120 min in ). The measurements presented in were taken from a 1.5-μm-thick confocal slice closest to the coverslip. To confirm that the loss of the DNA-Alexa546 pattern occurs in the whole nucleus and was not due to a change in orientation presenting a different portion of the nucleus in the confocal slice analyzed, we performed the same analysis on the second and third confocal slices of one set of mother and two daughter cells. Here again in this segment of the nucleus, the pattern of DNA-Alexa546 was lost in both daughters in early G1 (relative fluorescence profile the same as in ). We estimate that the confocal slices analyzed represented more than 70% of the nuclear volume in which there was no apparent pattern to the DNA-Alexa546 fluorescence, suggesting that relative chromosomal domain positions were lost at these time points. However, the DNA-Alexa546 fluorescent pattern, localized to half of the nucleus, was reestablished at later times (). This qualitative and quantitative analysis indicates that there was a period early in G1 when the chromosomes were uncondensed but not yet organized into the same or similar neighborhoods as in the mother nucleus.
We have shown by three independent DNA marking methods that the relative organization of chromosomal domain position is similar between mother and daughter cells. In addition, there was a period early in G1 where this organization was apparently lost. This loss of organization was not due to a specific orientation of the bleached boundary relative to the spindle axis as the cells analyzed included examples of all relative orientations, parallel, perpendicular, and diagonal. Chromosome decondensation in G1 could account for the loss of relative chromosomal domain organization. If this event is not highly coordinated among chromosomes and occurs asynchronously, organization maintained at metaphase would be disturbed. The purpose of maintenance of relative chromosomal domain position and the mechanism for its reestablishment after cell division remain intriguing. The simplest reason to organize chromosomes into domains in the nucleus is to prevent them from becoming entangled. Establishing chromosomal domains into neighborhoods may not be a deterministic process but could result from the physical properties of chromosomes, the time of their separation, their rate of decondensation, and perhaps their differential interaction with components of the reforming nuclear envelope. We followed the general pattern of chromosomal domain organization through one cell division. Though the daughter nuclei resembled their mothers, the similarity was not absolute. Thus, the organization of chromosomal neighborhoods changed somewhat through cell divisions. If different chromosomal neighborhood organization is functionally important, this variation would allow additional epigenetic mechanisms to affect cell differentiation during development.