The mechanisms of DNA replication initiation and associated protein–protein interactions (PPI) have potential as targets for the development of safe and effective antibiotics (Table 1). Though this has been repeatedly recognised over the past 10 years [9,14], limited progress has been made towards clinical trials for hits targeting bacterial initiator, helicase, and primase proteins [30]. Largely, this is due to the formal characterisation of these initiation and primosome proteins historically being limited to the model bacteria, E. coli and Bacillus subtilis, and some ESKAPE pathogens (Enterococcus faecium, S. aureus, K. pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterococcus spp). Indeed, several essential DNA replication proteins and PPIs have been found to differ significantly between the E. coli model and other bacteria, limiting broad-spectrum application [3,132]. In addition, there are other compounding issues associated with screening assays targeting replication processes involving multiprotein complexes, e.g., difficulty to decipher the specific mechanism of action of a hit, and failure to demonstrate in vivo activity due to various target access barriers (Table 2).

While most in silico and in vitro screening methods fail to identify compounds that demonstrate in vivo activity, cell-based assays cannot identify a direct mechanism of action or ensure selectivity. Table 2 presents a selection of promising methods available to characterise the functions and interactions of the proteins that have been or can be used to screen libraries of compounds.

Many activity assays used in past drug screening initiatives for replisome proteins were performed in low throughput, such as the [α-³²P]ATP-based filter binding assay for E. coli DnaA (and primosome) developed by Mizushima et al. [41] and the thermally denaturing HPLC primer synthesis assay for E. coli DnaG developed by Griep et al. [113]. In some cases, low-throughput activity assays could be adapted to a high-throughput format, as seen with the helicase activity assays used by Lin and Huang [40] where both the ATPase and 5′-3′ DNA helicase activity assays have been adapted to 96-well plate formats for helicases from E. coli [138], Bacillus anthracis and S. aureus [115,140]. The filter binding assay developed for the E. coli primosome [41] is another example that could be adapted to a 96-well plate format using a cell harvester and filter plates. More recent activity assays have been developed with higher throughput formats, including the coupled colorimetric primase–pyrophosphatase assay developed by Biswas et al. for M. tuberculosis DnaG [39,119] and the GFP reporter-based minichromosome assay developed by Klitgaard et al. for E. coli DnaA [134].

Most DNA replication initiation inhibitors have yet to demonstrate activity in in vivo trials [40,119,140,141,143]. For example, inhibitors of Bacillus stearothermophilus helicase–primase interaction [143] identified with a reverse yeast three-hybrid assay were unable to demonstrate antibacterial activity [30]. In another example, in vitro inhibitors of S. aureus and B. anthracis helicase identified with a fluorescence resonance energy transfer (FRET)-based assay demonstrated low activity in vivo, with insufficient inhibition to obtain MIC values [140]. Furthermore, specific inhibitors of B. anthracis primase identified with the coupled colorimetric primase–pyrophosphatase assay also failed to show in vivo activity, due to an inability to penetrate the bacterial envelope [119]. More recent studies applying molecular docking approaches [121,136] and/or fragment-based screens [117,122] have identified inhibitors with their activity yet to demonstrate in vivo. Multi-target approaches increase the potential for identification of lead compounds [128]. Dallmann et al. [145] used high-throughput parallel multiplicative target screening approach building on a PicoGreen fluorescence-based assay for the screening of E. coli and B. subtilis replisome. However, as these include multiple proteins, identifying the mechanism of action can be challenging.

One other major issue with these assays is the possibility of non-specific DNA interaction leading to assay failure. For example, Biswas et al. [119] identified doxorubicin as an inhibitor of B. anthracis primase with bacteriostatic effects in vivo, yet were unable to determine its direct mechanism of action. Doxorubicin is a DNA intercalator which could explain its activity [153]. Another example of this can be seen in the use of the assay developed by Fossum et al. [133] for identifying inhibitors of DnaA in E. coli, instead detecting inhibitors of DNA gyrase [137]. Furthermore, deferoxamine identified using a cell-based assay with an E. coli ATP-DnaA-‘locked’ strain [135] failed to restrict the growth of wild type cells and was found to act via iron-chelating activity. Unfortunately, these potential complications arise in many current biochemical activity assays for replisomal proteins when DNA is essential to their operation, leading to the inability to distinguish whether inhibition is due to targeting of the enzyme or interactions with the DNA itself.

In the last two decades, large scale drug screening campaigns have shifted to using methods that can detect the physical interaction of a compound with its protein target [147,149,154]. Fragment-based approaches involving high-throughput biophysical screening techniques have become a common feature in drug discovery. High-throughput surface plasmon resonance (SPR) is now sensitive enough to be used as a primary screen [155]. Of note, the interactions of helicases and primases have been characterised by SPR validating the utility of this technique for compound screening with these proteins [78,101]. In addition, SPR can be used for subsequent validation and kinetic characterisation of the interactions. However, stability of the target can be problematic and the cost of screening is high; both of which are important aspects to consider [147,148,155]. Specialised MS techniques to screen small molecules [156,157,158] include affinity selection-mass spectrometry (AS-MS) and pulsed ultrafiltration-mass spectrometry (PUF-MS) [159,160]. They are suitable for high-throughput screening of large compound libraries or natural product extracts on protein targets [161,162]. The helicase has been examined by native MS [163,164] and as such, should be amenable to AS-MS. Differential scanning fluorimetry (DSF) [150,165], also known as Thermofluor, is probably the most-commonly used technique as a first step in the process of large compound library screening due to its technical simplicity and low cost. A derivative of this technique, DSF of GFP-tagged proteins (DSF-GTP), has been validated with several E. coli and Burkholderia pseudomallei GFP-tagged replisomal proteins such as Tus, DnaA, DnaB and DnaG [151,152,166,167]. DSF-GTP has several advantages over classic DSF in that it can be used with protein mixtures and extracts to evaluate target access [152], providing a powerful platform for future large scale drug screening campaigns.