Mechanistic Profiling of Inhalation Sensory Irritants with a Computational Model for Transient Receptor Potential Vanilloid Subfamily Type 1

Abstract

Introduction:

We are using big data mining to develop computational models that predict potential interaction with important biological pathways. Transient receptor potential vanilloid subfamily type 1 (TRPV1) is one of several nociceptors that contribute to sensory irritation. Because sensory irritation is frequently used as a critical effect in setting occupational exposure limits (OELs), we developed a model that predicts interaction with TRPV1 and used it to mechanistically profile two inhalation databases (DBs).

Methods:

We built a random forest machine learning model to predict whether a novel compound will or will not interact with TRPV1 by fingerprinting a large DB curated primarily from public in vitro data. Our model has high sensitivity (90.2%), specificity (99.2%), and balanced accuracy (94.8%). We mechanistically profiled (1) a rodent RD₅₀ DB (concentrations causing a 50% decrease in respiratory rate; N = 190) and (2) a subset of the American Conference of Governmental Industrial Hygienists DB with OELs that were primarily based on sensory irritation (N = 109).

Results:

For both DBs, a high percentage of compounds were identified for potential interaction with TRPV1. Further screening of DBs with a profiler for facile chemical reactivity gave similar results, with many compounds flagged for both mechanisms. The more potent compounds in either DB were often chemically reactive, suggesting potential involvement of a related nociceptor known to serve as a sentinel for electrophiles—transient receptor potential ankyrin subfamily type 1.

Conclusion:

Our findings emphasize the need for an integrated testing approach using tiered in silico and in vitro screening for these nociceptors to derive OELs without using animals.