Apparent activity in high-throughput screening: origins of compound-dependent assay interference

Abstract

Expansive compound collections made up of structurally heterogeneous chemicals, the activities of which are largely undefined, present challenging problems for high-throughput screening (HTS). Foremost is differentiating whether the activity for a given compound in an assay is directed against the targeted biology, or is the result of surreptitious compound activity involving the assay detection system. Such compound interference can be especially difficult to identify if it is reproducible and concentration-dependent - characteristics generally attributed to compounds with genuine activity. While reactive chemical groups on compounds were once thought to be the primary source of compound interference in assays used in HTS, recent work suggests that other factors, such as compound aggregation, may play a more significant role in many assay formats. Considerable progress has been made to profile representative compound libraries in an effort to identify chemical classes susceptible to producing compound interference, such as compounds commonly found to inhibit the reporter enzyme firefly luciferase. Such work has also led to the development of practices that have the potential to significantly reduce compound interference, for example, through the addition of non-ionic detergent to assay buffer to reduce aggregation-based inhibition.

Figure 1. Assay interference by compounds can be reproducible and demonstrate concentration dependence, producing false positives in a high-throughput screen (HTS)

(A) False-positives in the primary assay used for HTS are generally eliminated upon testing in an orthogonal assay (see Box 1) conducted on all compounds with greater than average activity. (i and ii) Each dot in the scatter plot represents a compound, plotted based on its percent activity in an assay, in this case, an assay to identify inhibitors. In the example shown, a 3 standard deviation (s.d.) threshold value (dotted line) was used to qualify a compound as active. (i) Results from a primary assay to identify compounds with inhibitor activity, with the percentage of active compounds at 5% of the total screened (found below 3s.d. line). (ii) Following an orthogonal assay, 95% of the active compounds are found to be false positives due to assay interference, leaving only a few genuine actives (red dots; 0.25% of compounds screened). (B) Fluorescent compounds (an example of which is shown) behave much like genuine actives, demonstrating concentration dependence (blue dots in scatter plot) that is reproducible, as shown here for results of a fluorescence-based HTS run in the blue spectral range. Sporadic fluorescence interference is also seen (grey dots), but is not reproducible and does not show concentration dependence (adapted from Simeonov et al., 2008 [14]). (C) Compound interference that directly interacts with an enzymatic reporter, such as firefly luciferase, is shown here. This type of compound interference also demonstrates compound concentration dependence (orange concentration response curve – CRC), and in this specific case, the compound is a competitive inhibitor of firefly luciferase (green CRCs).

Figure 2. Different types of assay interference: compound fluorescence, inhibition of the reporter enzyme firefly luciferase (FLuc), aggregation-based inhibition, and redox cycling compounds (RCC)

(A) Self-organizing map (SOM) depicting the relationship between chemical structure and fluorescence for compounds in a representative library whose fluorescence profile overlaps with AlexaFluor 350 or resorufin [14]. Each hexagon represents a cluster of compounds that are structurally related. Clusters shaded red/orange indicate an enrichment of compounds in the structural class that fluoresce (compared to the average), whereas clusters shaded blue indicate a structural class with fewer than average fluorescent compounds. As can be seen, significantly more compound classes fluoresce in the blue-spectral range, compared to those that fluorescence in the orange region. Additionally, compound classes that fluoresce tend to be related, as demonstrated by an enrichment in fluorescent compounds in adjacent clusters [14]. (B) Some compounds that inhibit the FLuc enzyme (red CRC for biochemical assay) appear to be active in cell-based FLuc reporter gene assays (green and black CRCs) due to the interaction of the compound with the FLuc enzyme. Apparent activation in cell-based assays due inhibitor-based FLuc stabilization can generate different types of CRCs, with two example CRCs shown here. If the inhibitory compound is competed off the enzyme by excess FLuc substrates found in the detection mix, or by washing of the cells prior to addition of the detection mix, a CRC similar to the one shown by the black-dotted line can be generated. If inhibition is not relieved at high compound concentrations, then a CRC similar to the one indicated by the green hatched-line is commonly seen. (C) Addition of Triton X-100 to the assay buffer significantly relieves inhibition of a compound that forms aggregates. Aggregation-based inhibition shows concentration dependence and the tendency for a compound to form aggregates is assay-specific. (D) Enzyme assay behavior of an example redox cycling compound (RCC). Data shown are from caspase-1 assays performed in the presence of 10 mM DTT (PubChem AID: 896) or 10 mM Cysteine (PubChem AID: 929). Changing the reduction potential of the buffer from a strong (DTT) to a weak (Cys) redox potential alleviates the apparent inhibition by the compound shown (PubChem CID: 647501), a RCC that generates peroxide upon exposure to DTT (PubChem AIDs: 672, 878, 787, 1234, 2035).