Cancer represents a critical threat to human survival [1]. Current clinical treatments, particularly chemotherapy, targeted therapy and photodynamic therapy, continue to rely on small-molecule drugs [2]. However, their efficacy is often limited by unfavorable physicochemical properties, inadequate pharmacokinetic profiles, and off-target distribution, which collectively complicate formulation design and reduce in vivo delivery efficiency [3], [4]. The emergence of next-generation therapeutics, such as peptide [5], proteins [6] and nucleic acids [7], has introduced additional delivery challenges, as these macromolecules often suffer from poor stability, enzymatic susceptibility, and limited membrane permeability [8]. Nanotechnology-based carriers have attempted to overcome these drawbacks by improving drug solubility and stability, prolonging systemic circulation, and enhancing tumor accumulation. Nonetheless, the in vitro assembled nanostructures often render them susceptible to premature disassembly or functional inactivation under dynamic physiological environments. Therefore, conventional nanoassemblies still face several bottlenecks, including inadequate targeting efficiency, limited intratumoral penetration, potential carrier-related toxicity, and complexities in scalable manufacturing [9], [10].
To address these challenges, in vivo self-assembly has emerged as a transformative strategy, inspired by the self-organizing behavior of natural biomacromolecules [11]. The core concept involves the administration of biocompatible molecular “precursors” that remain inert and stable during systemic circulation, but undergo selective activation within the tumor microenvironment (TME) in response to endogenous cues (e.g., enzymatic activity, pH gradients, redox states, receptor overexpression) or exogenous triggers (e.g., light, temperature, ultrasound, magnetic fields) [12], [13], [14]. The activation induces structural transformations of precursors, such as changes in molecular conformation, hydrophobicity/hydrophilicity, or surface charge [15], [16]. These changes further drive localized self-assembly through various non-covalent intermolecular forces, including electrostatic interactions, hydrophobic interactions, π-π stacking, and hydrogen bonding, or covalent interactions such as coordination bonds, condensation reactions, and click chemistry-mediated cross-linking [17], [18]. This “small-to-large” dynamic process elegantly integrates the strengths of small molecules and nanomaterials: precursors, by virtue of their minute size, achieve deep penetration through the dense tumor stroma, while the subsequently formed assemblies become effectively trapped and retained within the tumor, resulting in high local drug concentration and sustained therapeutic exposure.
Such finely tuned processes underscore the critical importance of rational precursor design for successful in vivo self-assembly. For optimal therapeutic performance, precursors must accumulate selectively at pathological sites, and undergo stimulus-triggered molecular transformations that initiate in situ assembly via well-defined intermolecular interactions. Although existing reviews have extensively summarized in vivo self-assembled nanomaterials and categorized self-assembly strategies according to material types (e.g., peptides, polymers) and specific stimuli (e.g., enzymes, redox), they generally adopt descriptive or classificatory approaches and leave a gap in providing a unified and predictable design framework. This review aims to fill this gap by proposing a transformative perspective centered on a modular design strategy. Besides the necessary active pharmaceutical ingredients, we classify assembly precursors into modular components based on their functional properties, including self-assembly modules, response modules, and targeting modules. We further clarify the unique biological effects achieved through in vivo self-assembly: assembly-induced retention effect, assembly-enhanced binding effect, and assembly-enhanced uptake effect. On this basis, we discuss advances in therapeutic modalities involving tumor chemotherapy, immune regulation, imaging diagnosis, and multi-application therapy (Scheme 1). Overall, this review provides rational guiding principles for the development of a new generation of in vivo self-assembled nanomedicines, shifting the research field from purely descriptive classification toward predictive engineering design.