Unraveling the complex role of intestinal microbiota in drug permeability: Insights from N-acetylcysteine

The Biopharmaceutics Classification System (BCS) serves as a foundational framework for drug development and, streamlines generic drugs approval by categorizing them based on solubility and intestinal permeability [1]. However, this system overlooks the influence of the intestinal microbiota on drug permeability. Recent research has revealed that intestinal bacteria can directly metabolize pharmaceutical compounds or indirectly influence drug absorption by altering the intestinal barrier function, enzyme activity, and transport systems. These microbial-mediated modifications can significantly affect the pharmacokinetic profile of a drug, potentially enhancing or reducing its bio-availability [[2], [3], [4], [5]]. Hence, there is a pressing need to refine the traditional BCS framework and integrate the intestinal microbiota as a critical variable in the assessment of drug permeability after oral administration. This paradigm shift is especially critical for drugs with narrow therapeutic windows or those that exhibit high inter-individual variability in clinical response. The growing field of pharmaco-microbiomics promises to deliver more personalized medication regimens by accounting individual unique microbial signatures in drug permeability assessments.

In a recent study published in Nature Communications, Zhang et al. revolutionized this paradigm by demonstrating that the intestinal microbiome, particularly Bacteroides species, can profoundly affect the permeability of BCS Class I or Class III drugs [6] (Fig. 1). To capture this interaction, the authors employed a single-pass intestinal perfusion (SPIP) platform to compare the in situ permeability of drugs with varying permeability classifications between germ-free (GF) and specific-pathogen-free (SPF) rats. In contrast to conventional permeability models such as Caco-2, which exclude microbial activity, the SPIP platform inherently incorporates microbial components. This provides a more physiologically relevant basis for predicting drug absorption and helps to reconcile discrepancies between in vitro predictions and in vivo outcomes [7,8]. They screened the permeability of 32 oral drugs, among which N-acetylcysteine (NAC) was identified as a pivotal model drug due to its inconclusive BCS classification and significant side effects linked to its pharmacokinetics [9]. A key finding was the significantly increased permeability of NAC in GF rats than that in SPF controls. This difference was attributed to the active bacterial metabolism by Bacteroides, which converts NAC-derived cysteine into cysteine-3-ketosphinganine, thereby sequestering the drug component and impeding its absorption.

Fig. 1.

Intestinal microbiota-mediated drug permeability variation.

Further mechanistic exploration proposed a “cysteinate sphingolipid metabolism pathway”, in which bacterial serine palmitoyltransferase (SPT) directly catalyzes covalent conjugation of NAC-derived cysteine with palmitoyl-CoA. This pathway diverts the drug component into bacterial sphingolipid biosynthesis, distinct from the canonical de novo pathway that uses. Of note, the SPT-deficient mutants of B. thetaiotaomicron and B. fragilis not only abolished the synthesis of cysteine-linked sphingolipids but also completely reversed the suppression of NAC permeability compared to wild-type strains, confirming the role of SPT in hindering NAC permeability.

To further verify the aforementioned microbial-dependent characteristics, the researchers integrated clinical data from a prospective cohort of 319 participants, and extracted 19 initial features including demographic, physiological, and microbiota-related variables. Through feature selection and hyperparameter optimization via fivefold cross-validation, a 15-feature eXtreme Gradient Boosting (XGB) model was developed to predict the fraction of the NAC dose entering the systemic circulation. This model achieved a mean absolute error of 2.453 and an area under the receiver operating characteristic curve of 0.867, confirming that cysteine palmitoylation by the intestinal microbiota significantly contributes to variability in NAC permeability and its ambiguous biopharmaceutical classification.

The development of the XGB machine learning model that incorporates intestinal microbial features represents a significant advancement in the prediction of drug permeability [6]. Evaluation using the net reclassification improvement (NRI) index confirmed that the XGB model provided a statistically significant refinement over the widely used ACAT model (NRI = 0.551, P < 0.001). Positive, although not statistically significant, trends toward improvement were also observed when compared with the GITA (NRI = 0.075, P = 0.464) and ADAM (NRI = 0.168, P = 0.135) models. The inclusion of microbial characteristics in the model underscores the importance of considering the gut microbiota in drug development and personalized medicine. These findings have potential implications for identifying drug classifications with variable oral bioavailability, despite high aqueous solubility [10]. By understanding the microbial factors that influence drug permeability, strategies can be devised to identify BCS classes and optimize drug delivery.

In summary, Zhang et al. [6] highlighted the intricate relationship between intestinal microbiota and drug permeability, with significant implications for the bioequivalence of orally delivered drugs. By unraveling the underlying mechanisms and developing predictive models, this study underscores the need for more sophisticated models that incorporate a wider range of factors affecting drug absorption, including diet, drug interactions, and physiological conditions. Future research may integrate microbial ecology into biopharmaceutical classification to guide the exemption of bioequivalence evaluations.

CRediT authorship contribution statement

Chutong Tian: Writing – original draft. Zhonggui He: Writing – review & editing, Conceptualization.

Conflicts of interest

The authors report no conflicts of interest.