The (very) early years

Long after water-filled glass bowls were used to read small letters (Singer 1914), a simple single lens microscope started the microscopic revolution (Bacon 1267) (see Table 1 and Fig. 1). Spurred throughout the ages by accidental inventions (van der Aa 1851), leaps by Galileo (Galilei 1610), and Hooke (1665), it was not until Carl Zeiss started to mass-produce microscopes in 1847 that DNA observation started to take off. Simultaneously, Mendel studied 29,000 pea plants (1866) and Haeckel postulated the containment of hereditary traits in the nucleus (1866) (Dahm 2008; Haeckel 1866), while Miescher put the microscope to good use and purified the nuclei and observed DNA (Miescher 1871). Köhler’s game-changing illumination technique (Köhler 1893) helped to perfect Zeiss UV-microscope together with Siedentopf in 1908. In 1927, shortly after Levene described the nucleic acid structure (1919), Koltsov postulated the semiconservative replication idea (Soyfer 2001).

Table 1 Chronological list of landmarks in microscopy and DNA replication
Full size table
Fig. 1
figure 1

Graphical overview of microscopy developments and their impact on DNA replication studies

Full size image

The race down to the DNA structure and duplication

Phase contrast microscopy (Zernike 1955) and DNA X-ray diffraction images (Astbury 1947) Franklin, 1952, “Photo 52”) lead to fantastic images, new discoveries, and the description of the double helix DNA structure (Watson and Crick 1953). Meselson and Stahl ingeniously demonstrated the semiconservative mode of DNA replication (Meselson and Stahl 1958). The theoretical description of a confocal microscope and the first practical application of a Nipkow disk in microscopy (Egger and Petráň 1967; Petráň et al. 1968) were clear landmarks of the microscopy revolution.

Radioactive labeling and autoradiography allowed Cairns to observe DNA unwinding and the replication fork (Cairns 1963), and Huberman and Riggs observed similar replication structures in mammalian chromosomes (Huberman and Riggs 1966) and Okazaki described the lagging strand synthesis and “its” fragments (Okazaki et al. 1968; Okazaki and Okazaki 1969; Sugimoto et al. 1969; Sugimoto et al. 1968).

Always look on the bright side

Along came Aequorea victoria green fluorescent protein (Shimomura et al. 1962) and brought light into darkness. Where audioradiography once ruled (Huberman and Riggs 1966; Huberman and Riggs 1968; Taylor et al. 1957), immunofluorescence labeling of fixed cells with monoclonal antibodies to modified nucleotides incorporated into newly synthesized DNA took the stage (e.g., Aten et al. 1992; Cardoso et al. 1993; Jackson and Pombo 1998; Jaunin et al. 1998; Ma et al. 1998; Mazzotti et al. 1990; Nakamura et al. 1986; Raska et al. 1989; Raska et al. 1991) only to be outshined by live cell microscopy of fluorescent fusion proteins (Cardoso et al. 1997; Leonhardt et al. 2000). Cell cycle duration (Van Dilla et al. 1969), fork speed, replication rate, and replicon sizes (Kriegstein and Hogness 1974; Taylor 1977; Taylor and Hozier 1976; Wilson and Wilson 1975; Yurov 1977; Yurov 1978; Yurov 1979; Yurov and Liapunova 1977) were all unearthed from the dark.

In parallel, the first affordable home computers made digital image analysis possible through the help of Wayne S. Rasband who developed the milestone in image analysis ImageJ (then, NIH Image) in 1987 (Schneider et al. 2012).

Extensive microscopic analysis in fixed cells followed and provided a spatiotemporal description of replication sites (replication foci; see Fig. 2) in cells throughout S-phase (Nakamura et al. 1986) along with the three main distinctive early, mid, and late S-phase replication foci patterns (Jackson and Pombo 1998; Mills et al. 1989; Nakayasu and Berezney 1989). Alongside, replication origins (Burhans et al. 1990; Burhans et al. 1991) were also reported.

Fig. 2
figure 2

Organization of DNA replication from the genome to the individual replisome/replicon. A fluorescently labeled human HeLa Kyoto cell with a typical late S-phase replication pattern is presented in the top left corner (scale bar = 5 μm). Magnified super-resolution replication foci, with white circles representing individual replication sites displayed in the middle of the top row. A scheme of clustered DNA loops with active replication sites (white) is shown on the right. Starting point of DNA replication, the replication origin (ori), and the region replicated from a single origin displayed in the bottom row. Each replicon is replicated by two replication machineries (magenta), composed of various replication proteins, magnified in the bottom left corner. Adapted from (Chagin et al. 2016; Chagin et al. 2010)

Full size image

The quest to replicate the genome

DNA loops (see Fig. 2) and their “functional” attachments to active transcription units were shown as chromatin organizers during mitosis (Jackson et al. 1992), and replication factories were proposed as clusters of DNA replication sites organized by the nucleoskeleton (Hozák et al. 1993). Molecular combing, refined DNA fiber analysis, and sensitive detection of DNA (Bensimon et al. 1994) opened the door to whole genome stretching and high-resolution studies (Michalet et al. 1997). It allowed analysis of single DNA molecules undergoing replication (see Fig. 2) in a much greater resolution (Herrick and Bensimon 1999) than ever before. Stable replicon clusters were also described as a basis for effective activation and propagation of genome replication during S-phase (Jackson and Pombo 1998) and regulation of replication timing (Dimitrova and Gilbert 1999).

Studies on DNA replication proteins (see Fig. 2) using live-cell fluorescence microscopy produced time lapse movies of replication factories and elucidated basic principles of their dynamic assembly-disassembly behavior (Leonhardt et al. 2000). Different regulatory levels were shown to be necessary to initiate and regulate DNA replication. Not only the chromatin structure, nuclear, and chromosomal locations but also origin recognition complex (ORC) and a whole bunch of other factors were found to define start sites of replication (DePamphilis 2003; Gerbi and Bielinsky 2002; Gerbi et al. 2002; Sasaki and Gilbert 2007).

Factories full of dominos

In addition to the “factory model” (Hozák et al. 1993), more dynamic models ensued (Sadoni et al. 2004; Sporbert et al. 2002) whereby replication at one site induces domino-like activation of neighboring origins, without the need to postulate pre-determined clusters of replicons. The combination with an earlier model postulating that origins of replication would be licensed only during mitosis and this license to replicate would be revoked after one round of replication (Blow and Dutta 2005; Blow and Laskey 1988) elegantly demonstrated how DNA is completely duplicated once, and only once, during each cell cycle. Despite Cvetic wishing for “eukaryotic origins of DNA replication to please be more specific” (Cvetic and Walter 2005), DNA replication origins in higher eukaryotes have been at best elusive. Nonetheless, as a whole, DNA replication is a very robust mechanism and stalled forks can be reactivated or reactivate neighboring origins to close all gaps and provide a perfect copy of billions of nucleotides at every cell division (Langston and O’Donnell 2006; Patel et al. 2006).

The ever elusive origin

The hunt for the elusive consensus motif of DNA replication origins continued with genome-wide high throughput mapping of potential origins and next-generation sequencing methods (Besnard et al. 2012; Cadoret et al. 2008; Cayrou et al. 2012; Dellino et al. 2013; Karnani et al. 2010; Lucas et al. 2007; Martin et al. 2011; Mesner et al. 2013; Mesner et al. 2011; Mukhopadhyay et al. 2014; Picard et al. 2014; Valenzuela et al. 2011) but stalled without a conclusive definition of the mammalian origin of replication. Correlations with specialized DNA structures (e.g., G-quadruplexes) and many others have been suggested but there seems not to be a simple solution and potentially there is no need to have one.

Studies into the epigenomic landscape, epigenetic control of DNA replication, and higher order chromatin organization (Casas-Delucchi and Cardoso 2011; Casas-Delucchi et al. 2012) have provided a link of epigenetic modifications (in particular, histone acetylation level) and temporal control of DNA replication origin firing.

Altogether, even Hyrien’s “Peaks cloaked in the mist,” all out approach was not able to identify possible origins by similarities in thousands of microarrays and/or next-generation sequencing data, suggesting origins form at unspecific DNA sites, but are suppressed by ongoing transcription (Hyrien 2015), which is highly correlated with histone acetylation.

To go where no one has gone before: beyond the Abbe limit

Meanwhile, the microscopy arms race to and beyond the diffraction limit calculated by Abbe continued with the Structured Illumination Microscopy (SIM) (Bailey et al. 1993), the 3D–SIM (Gustafsson et al. 2008) and the stimulated emission depletion (STED) (Schmidt et al. 2008).

The first attempts to label dating back to 1986 (Nakamura et al. 1986) and quantify replication sites in cells yielded numbers on the low hundreds (see Fig. 3). A decade later with the advent of digital imaging and computational image analysis tools, these numbers grew to around one thousand (Berezney et al. 1996; Fox et al. 1991; Jackson and Pombo 1998; Ma et al. 1998), where they remained for several years (see Fig. 3). Such numbers of replication sites were compatible with a concept of clusters of replicons activated together and, thus, visualized together.

Fig. 3
figure 3

Graphical overview of replication foci numbers in correlation with microscopy developments

Full size image

The dramatic increase in spatial resolution made possible with the new super-resolution microscopy techniques (fluorescence nanoscopy) enabled the visualization, for the first time, of smaller replication structures (Baddeley et al. 2010; Cseresnyes et al. 2009; Ligasová et al. 2009). It was now possible to resolve structures well below the Abbe limit, down to 30 nm and smaller. Nanoscopy (Gustafsson et al. 2008; Hell 2003; Hell et al. 1994) is in full swing and let us go where no one has gone before: beyond the Abbe limit. This, on the other hand, created another level of demand upon image analysis tools.

“Myths” confirmed!

The stage was now set to try and unveil the units of genome replication, i.e., the replicons and their associated machinery, the replisome, in cells.

From the earlier studies using light nanoscopy techniques (Baddeley et al. 2010; Cseresnyes et al. 2009) as well as electron microscopy (Koberna et al. 2005), suitable computational image analysis protocols were developed (Chagin et al. 2015). These combined efforts led to a further increase in the numbers of replication sites measured in cells (see Fig. 3), which was now finally compatible and fitting with the predicted numbers of replicons needed to duplicate the genome in human cells (Chagin et al. 2016; Löb et al. 2016).

The microscopic information age had arrived. Previous efforts by Shaw et al. (2010), together with measurements throughout the years culminating on the visualization and quantification of individual replicons in cells in 4D, all supported by 3D–SIM imaging (Chagin et al. 2016) were all combined in a minimalistic but comprehensive 4D replicon simulation model (Löb et al. 2016) displaying previously published replication polarity gradients, replication timing profiles, N/U domains, topologically associating domains, and timing transition regions (Audit et al. 2013; Baker et al. 2012; Chen et al. 2010; Hyrien et al. 2013; Pope et al. 2014).

Journey into the future

Future work should aim to bridge the ever-increasing genome-wide population data, with single molecule and single-cell microscopic data. Novel ways to combine and relate these very different types of information should be developed to get the highest spatial together with the highest temporal resolution without compromising the data on variability between single cells.

Importantly, the available models should be put to work to predict and test genome replication in different cell types and species and under different stress conditions. This would unleash the value of the existing models and lead us into the in silico DNA replication era.