Epigenetics is the study of heritable modifications to gene expression that are not caused by alterations to the underlying genetic code. Classically, three broad types of epigenetic regulation have been described: DNA methylation, regulatory RNAs and histone modifications (Jaenisch and Bird, 2003; Allis and Jenuwein, 2016; Statello et al., 2021). Unlike genetic mutation, where changes in DNA sequence are notoriously difficult to reverse, epigenetic information can be viewed as intrinsically reversible, and a plethora of enzymes and inhibitors have been described that can alter chromatin states and thereby gene expression. This capacity for reversibly altering gene expression has sparked new interest in using “epidrugs”, particularly where enhancing or suppresing gene expression is of therapeutic benefit (Heerboth et al., 2014; Salarinia et al., 2016; Wolff et al., 2017; Miranda Furtado et al., 2019).

Reporter genes provide a mechanism to examine gene regulation and the effect of these epidrugs in a preclinical setting. The first reporter genes to be described utilised enzymes from Escherichia coli (E. coli) to visualise colorimetric changes: ß-glucuronidase (GUS) was used in plant systems (Jefferson et al., 1987) and ß-galactosidase (lacZ) in mammals (An et al., 1982) and eukaryotic cells (Casadaban et al., 1983). ß-galactosidase has been extensively used to examine gene expression in tissue sections, for optical projection tomography (OPT) (Sharpe et al., 2002) and fluorescence activated cell sorting (FACS) (Nolan et al., 1988), offering sensitive cellular resolution (Serganova and Blasberg, 2019). However, as such assays compromise cell viability, fluorescent and bioluminescent systems have been developed to enable longitudinal studies. Free from the requirement of an enzymatic substrate, fluorescent proteins are ideal for obtaining cellular and sub-cellular resolution in living cells but are limited by low sensitivity (Serganova and Blasberg, 2019). As an example, endogenous autofluorescence in a typical mammalian cell is equivalent to 1 µM concentration of green fluorescent protein (GFP), and thereby detection, particularly of low level signals, is limited (Niswender et al., 1995). In contrast, bioluminescent reporters offer high sensitivity, combined with a capacity to dynamically monitor signal within living cells and tissues, rendering them ideal as tools for epigenetic research (Mezzanotte et al., 2017; Dimond et al., 2020). In this mini review we describe how bioluminescent imaging can be harnessed for epigenetic research and preclinical drug discovery. Bioluminescent reporter genes inserted into endogenous loci are subject to the same genomic and epigenomic control as the gene of interest and can be used to track gene expression longitudinally and across generations.

Bioluminescence is a naturally occurring chemiluminescent phenomenon, where a luciferase enzyme produces light as a by-product of oxygenation of its luciferin substrate (Sharifian et al., 2018; Dimond et al., 2020). Hundreds of luciferin/luciferase pairs have been identified throughout the natural world, used to deter predators, attract prey and as a form of communication, as previously discussed in more detail (Kaskova et al., 2016; Syed and Anderson, 2021; Baljinnyam et al., 2023). The majority of bioluminescent organisms are found in marine life (Sharifian et al., 2018), where coelenterazine is used as a substrate, with oxygen (O₂) as a cofactor, to produce light in the blue spectrum (Mezzanotte et al., 2017; Serganova and Blasberg, 2019). The most common examples are Renilla luciferase (RLuc) found in sea pansy (Ward and Cormier, 1979; Loening et al., 2007) and Gaussia luciferase (GLuc) from the marine copepod Gaussia princepsin (Tannous et al., 2005). In addition to O₂, terrestrial bioluminescent organisms require adenosine triphosphate (ATP) and magnesium (Mg²⁺) as cofactors (Mezzanotte et al., 2017; Serganova and Blasberg, 2019). Firefly luciferase (FLuc) from the North American firefly (Gould and Subramani, 1988), and click beetle luciferases (CBGLuc/CBRLuc) from Pyrophorus plagiophthalamus, are commonly used, with D-luciferin as their substrate (Mezzanotte et al., 2017; Serganova and Blasberg, 2019; Dimond et al., 2020). In certain bacteria and fungi, the genetic pathways encoding both the luciferase enzyme (encoded by luxAB genes in bacteria or nnLuz in tropical fungi) and the metabolic enzymes required to naturally generate the luciferin substrate have been fully characterised (Kotlobay et al., 2018; Gregor et al., 2019; Dimond et al., 2020).

Luciferase reporters have been prevalent since the 1980s, when Fluc was first used to study E.coli (Nussbaum and Cohen, 1988) and circadian rhythm in plants (Millar et al., 1992). Upon exposure to the substrate luciferin, cells or organisms expressing luciferase can be imaged with a charge-coupled device camera to produce a quantitative and localised signal (Dimond et al., 2020). As a reporter, luciferase benefits from a high signal to noise ratio, a wide dynamic range and rapidly quantifiable measurements (Roda et al., 2003; Mezzanotte et al., 2017; Dimond et al., 2020). It is estimated to be 10–1,000 times more sensitive than fluorescence reporters (Allard, 2008). This can be attributed to the fact that bioluminescence is the byproduct of a chemical reaction; oxidation of luciferin is catalysed by luciferase to form an excited-state species (oxyluciferin) that relaxes to its ground state by giving off a photon of light [outlined in more detail in Kaskova et al. (2016)]. Conversely, fluorescent proteins require excitation from an external source to produce light. External excitation generates significant background (autofluorescence) and struggles to penetrate tissue below the surface. Bioluminescence imaging (BLI), however, requires no excitation, resulting in minimal background autoluminescence and permitting a more accurate measurement of change. Indeed, imaging of fluorescent and bioluminescent reporters with similar emission wavelengths proved that bioluminescence is 500 times more sensitive than fluorescence in vivo (Yan et al., 2013). Furthermore, the substrate, luciferin, is non-toxic, has a relatively short half-life and there are no adverse issues associated with phototoxicity or bleaching (Serganova and Blasberg, 2019). This makes BLI ideal for longitudinal and transgenerational studies where repeated sampling is required (Figure 1A, left hand panels).

Luciferase assays have been extensively used in vitro, both in cell-free systems where the target molecule is purified and activity of a drug tested directly, and in cell-based systems where the activity of a drug is evaluated in a genetically-modified cell (Conway et al., 2000; Goetz et al., 2000; Roda et al., 2003; Campana et al., 2016). In vivo, bioluminescence systems have traditionally been used for cell tracking where the growth of a luminescent tumour, or progression of infection, is measured in a living organism to provide spatial, longitudinal and quantitative information (Edinger et al., 2003; Cao et al., 2008; Rabinovich et al., 2008; Eun Kim et al., 2013; Lee et al., 2013; Yang et al., 2013; Kim et al., 2015). A unique advantage of bioluminescence for in vivo imaging is the lack of autoluminescence produced from mammalian cells, providing extremely low background (Mezzanotte et al., 2017; Dimond et al., 2020), whilst imaging at depth is aided by the lack of requirement for excitation by an external source. Nonetheless, as with any optical method, detection is limited by signal attenuation from deep and dark tissues; the scatter of light by cells, lipids and other components increases with depth, and substances such as haemoglobin absorb blue-green light (Choy et al., 2003; Zhao et al., 2005). These issues are being addressed through development of novel luciferase/luciferin variants designed for deeper and more sensitive signal detection (Zambito et al., 2021). For example, mutant luciferases have been designed which shift the emission spectrum of FLuc and CBRLuc into the far-red or near-infrared wavelengths, reducing attenuation of signal by haemoglobin and expanding the possibility of dual-coloured BLI (Mezzanotte et al., 2017; Zambito et al., 2021). Unmixing algorithms can be used to distinguish signal from spectrally distinct luciferases (Mezzanotte et al., 2011; Mezzanotte et al., 2013), as demonstrated by Mezzanotte and others using a red shifted firefly luciferase (PpyRE8) and a green click beetle luciferase (CBG99Luc) (Mezzanotte et al., 2011), allowing multiple reporters to be utilised in the same model.

Bioluminescence can be used to visualise gene activity, by inserting a luciferase reporter gene downstream of the regulatory elements of a gene of interest (GOI) (Figure 1A, right-hand panels). Advances in gene editing now enables reporter genes to be routinely inserted directly into the endogenous allele (Dimond et al., 2020). Insertion of the reporter into the endogenous allele provides the full genomic and epigenomic context of the GOI, including DNA and histone modifications, and accessibility of regulatory regions. This is particularly useful for studying epigenetic regulation, including epidrug interventions, since the reporter resides within the same chromatin environment as the GOI, and post-transcriptional regulation via the 3′ or 5’ untranslated regions (UTRs) can be retained. Changes to the chromatin regulatory environment can be transient or stable and heritable and since BLI allows individual samples to be imaged multiple times, the durability of epigenetic changes can be assessed across cell and organismal generations (Figure 1A). To interrogate mechanisms of GOI regulation, endogenous bioluminescent reporters could be combined with mice deficient in epigenetic regulators; for example, bioluminescence imaging of an endogenous reporter crossed with a conditional knockout of Dnmt1 (Golshani et al., 2005), Dicer (Cobb et al., 2005) or Ezh2 (Su et al., 2003) could be used to assess the impact of altered miRNA, H3K27me3 or DNA methylation on GOI expression.

A wide range of endogenous bioluminescent reporters have been generated so far (Table 1) and fall into two main classes. In the first, luciferase replaces the GOI at one or both alleles and is expressed under the control of the endogenous promoter. This approach was used to visualise p16 ^INKa expression during aging and tumorigenesis (Burd et al., 2013) and p21 following TP53 mutation (Tinkum et al., 2011). The main limitation of this approach is that the resulting cells or animals lack at least one copy of the endogenous gene, whilst internal regulatory elements may also be lost. A second ‘non-disruptive’ approach aims to couple the expression of the luciferase gene with the GOI, while preserving its function. This can be achieved using either an internal ribosomal entry site (IRES), as demonstrated for Vegfr3 (Olmeda et al., 2017) and Utrn (Vuorinen et al., 2021), or using a 2A self-cleaving peptide sequence (T2A site) (Van de Pette et al., 2017; Son et al., 2021; Pham et al., 2022; Van de Pette et al., 2022; Gleneadie et al., 2023). During translation, T2A cleavage results in expression of endogenous protein and a separate luciferase reporter. This is particularly beneficial for multi-reporter models where a GOI is coupled to more than one reporter gene. Dual reporter models have been developed for Cidea (Son et al., 2021), Utrn (Gleneadie et al., 2023), Cdkn1c (Van de Pette et al., 2017), Dlk1 (Van de Pette et al., 2022), Fgf21 (Pham et al., 2022) and Dpd (Smith et al., 2022) (Table 1) where the reporters were separated from each other and from the endogenous gene by T2A sites. These models combine the high sensitivity and real-time dynamics of bioluminescence for in vitro screening and in vivo imaging with the cellular resolution of lacZ or fluorescent reporters (Day et al., 1998; Contag et al., 2000). The ever-expanding diversity of luciferin/luciferase systems provides the opportunity to visualise expression of more than one gene of interest in the same animal using luciferase proteins which either require different substrates for activation (Jones et al., 2017) or produce spectrally distinct luminescence signal (Mezzanotte et al., 2011; Gleneadie et al., 2023; McMorrow et al., 2023) (exemplified in Figure 1A, right-hand panel, red and green).

BLI offers a powerful tool for preclinical research. Reporters can be used to visualise progression of a tumour (O’Farrell et al., 2013) or pathogen (Chang et al., 2014; Vande Velde et al., 2014; Daniel et al., 2015), or can represent expression of a disease-relevant gene, including epigenetically-silenced imprinted (Van de Pette et al., 2017; Van de Pette et al., 2022; Dimond et al., 2023) or X-linked genes (Tinsley et al., 2011; Moorwood and Khurana, 2013; Sripathy et al., 2017; Lee et al., 2020; Grimm and Lee, 2022; Gleneadie et al., 2023). A key benefit of using bioluminescent reporters for drug screening is that one genetic model can be used for each stage of preclinical research, from in vitro drug screening to in vivo testing of candidate molecules (as illustrated in Figure 1B).

Bioluminescence offers an exciting tool for in vivo imaging and drug development, with the repertoire of bioluminescent tools continuously expanding. However, a number of limitations should be considered when designing such experiments. The oxygenation of luciferin varies in response to the environment. The temperature, pH, and, importantly, the oxygen content of a cell, or extracellular space, determines the rate of luciferin turnover. This is a particular problem in cancer research, where conditions within the tumour can differ greatly from the norm (Alsawaftah et al., 2021). Due to the absorbance and scatter of light by biological materials, bioluminescent signal can also be attenuated at depth in vivo, particularly for wavelengths below 600 nm. Using directed evolution of firefly luciferase, a novel variant, termed AkaLuc was developed and paired with a D-luciferin analogue, Akalumine-HCL, to produce signal 100–1,000X higher than conventional D-luc/Fluc imaging, with an emission peak at 650 nm (Iwano et al., 2018). Excitingly, AkaBLI allows visualisation of small numbers of cells in free moving animals, in deep tissue in mice and in a female marmoset (Iwano et al., 2018). Similarly, as D-luciferin has poor permeability in the brain (Kim et al., 2022), the substrate for NanoLuc was optimised to produce cephalofurimazine (CFz), which, when paired with the NanoLuc-CyOFP fusion protein, Antares, emits 20X more luminescence than conventional D-luc/Fluc and, at high doses, outperforms AkaBLI threefold (Su et al., 2023). The development of new luciferin/luciferase pairs, such as these, improves sensitivity in vivo, and expands the options for luciferase multiplexing (Mezzanotte et al., 2017; Zambito et al., 2021). Using multiple spectrally distinct luciferases, different target genes can be visualised simultaneously, paving the way for research into gene expression networks and drug targets.

Another limitation of BLI is poor resolution. As mentioned previously, although a single bioluminescent cell can be visualised in vivo (Iwano et al., 2018), the exact position of these cells is difficult to determine with current BLI approaches. These limitations have led to the development of multimodal reporter models, where luciferase is expressed in conjunction with different reporters. In this context, photoacoustic tomography (PAT), magnetic resonance imaging (MRI) and position emission tomography (PET) offer high spatial resolution but lower sensitivity than BLI (Dimond et al., 2020) and fluorescent and colorimetric reporters, which are not well suited to in vivo imaging, provide cellular resolution ex vivo. Therefore, multimodal reporter models have been developed using combinations of these reporters to complement the benefits of luciferase (Levin et al., 2014; Van de Pette et al., 2017; Liu et al., 2019; Son et al., 2021; Pham et al., 2022; Gleneadie et al., 2023). When directly coupled with fluorescent probes, luciferase signal can be enhanced; bioluminescence resonance energy transfer (BRET) systems involve a luciferase, such as Rluc, fused to a near-infrared fluorescent protein (iRFP). Light produced from the Rluc excites iRFP resulting in a near-infrared signal, optimised for deep-tissue imaging (Schaub et al., 2015).

Currently, BLI in mammalian cells requires administration of luciferin. This can be a limiting factor in experimental design, particularly in vivo where animal welfare regulations limit the number of injections per animal and the permeability of luciferin differs between tissues (Syed and Anderson, 2021). Therefore, the development of autonomously bioluminescent animal models is an appealing concept. In plants, expression of the entire fungal bioluminescence system resulted in self-sustained luminescence which was visible to the naked eye (Khakhar et al., 2020; Mitiouchkina et al., 2020). Likewise, in mammalian cells in vitro, transfection of all components of the bacterial lux (Xu et al., 2014) or fungal nnLuz (Mitiouchkina et al., 2020) systems led to autoluminescence. However, toxicity has been associated with some lux pathway intermediates and the luciferase emits light in the blue spectrum, making it unsuitable for in vivo research. While nnLuz is reported to lose activity at temperatures above 30°C, detectable bioluminescence was emitted from mice injected with nnLuz expressing cells (Kotlobay et al., 2018). Therefore, the fungal bioluminescence system could be adapted to generate autonomously bioluminescent mice, expanding the imaging toolkit for epigenetic research and genetic modelling. Removing the need for substrate administration, coupled with recent advances improving luminescence signal (Iwano et al., 2018), one could envision continuous monitoring of gene expression in free moving animals throughout lifespan or in response to stimuli.

In previous sections we have outlined how endogenous bioluminescent reporters can be used in epigenetic research and inhibitor screening. Such systems report gene expression changes, inferring corresponding alterations to the epigenetic landscape. Bioluminescence is also being harnessed to look at mechanisms of epidrug activity by developing novel split luciferase systems, which directly measure changes in epigenetic modifications (Badran et al., 2011; Shekhawat and Ghosh, 2011; Sekar et al., 2015). For example, to monitor methylation of histone H3 tails at lysine 9 and 27 (H3K9, H3K27), fusion proteins were developed where H3K9 or H3K27 were connected to the methyl-lysine binding domains of Suv31H1 or Pc2 by a flexible linker and inserted between the N- and C- terminals of RLuc (Sekar et al., 2015). Luminescence signal can be detected when H3K9/H3K27 become methylated and bind to the Suv31H1/Pc2 domains, bringing the two-halves of the split RLuc together (Sekar et al., 2015). Global DNA methylation levels have been similarly measured using split luciferases (Badran et al., 2011) and other bioluminescent reporter systems (Wang et al., 2021; Taka et al., 2022). Interestingly, the split protein concept has been further adapted to determine site specific DNA methylation changes by fluorescence; a zinc finger attached to one-half of a GFP molecule targets a specific DNA sequence and a methyl binding domain (MBD) attached to the other half targets an adjacent CpG site (Stains et al., 2006). Fluorescence is therefore only measurable when the site is methylated. It is conceivable to envision an adaption to this technique using a split luciferase, to improve sensitivity for drug screening and in vivo imaging. Moreover, using spectrally distinct luciferases, endogenous bioluminescent reporters could be combined with the tools described above to simultaneously measure target gene expression and alterations to the epigenetic landscape.

Altering gene expression without mutating the underlying DNA sequence is an increasingly appealing therapeutic option for many genetic diseases, paving the way for extensive research into the design and use of epigenetic therapies. As discussed above, luciferase reporter genes provide a unique opportunity to examine the effects of these drugs. Visualising the reactivation of imprinted, X-inactivated or developmentally regulated genes with bioluminescence allows epigenetic changes to be accurately quantified at the molecular, cellular and organismal level, and across generations. Future developments, such as engineering new autonomous bioluminescent models that self-report activity throughout life, and generating multiple, spectrally-distinct reporters to image gene expression during ontogeny and in disease, promise an even brighter future for epigenetic research.