Background

The ability to monitor molecular events in living tissues noninvasively is a long-standing goal of biomedical research [1]. Understanding how diseases such as cancer, neurodegeneration, and inflammation develop at the molecular level requires tools that can report on biological events such as enzyme activation, protein–protein interactions, signaling, and drug action, within intact, optically opaque experimental models. Magnetic resonance imaging (MRI) is uniquely suited to this challenge: it generates high-resolution (from millimeters in humans to ~100 μm in rodents), three-dimensional images of soft tissue without utilizing ionizing radiation and has no depth constraint, making it one of the few modalities compatible with longitudinal molecular imaging in whole animals in both research and clinical settings.

Consequently, the development of genetically encoded reporters that produce detectable MRI signals has been an active area of research for over two decades [2–6]. Genetically encoded reporters for MRI can be broadly categorized by their contrast mechanism: metal-based reporters that alter T1 or T2/T2* relaxation times by accumulating paramagnetic ions, and water-exchange reporters that increase the apparent diffusion coefficient (ADC) of cellular water detectable by diffusion-weighted MRI. The latter is the foundation of the modular aquaporin-based protease-activatable probe for enhanced reporting (MAPPER) platform described in this protocol. Early approaches focused on metalloprotein overexpression by engineering cells to accumulate iron via ferritin or transferrin receptor overexpression, producing local perturbations in T₂ or T₂* relaxation rates that could be detected using standard T₂-weighted imaging protocols similar to blood oxygenation level–dependent functional MRI (BOLD-fMRI) [7–9]. While these reporters have been used successfully in cell-tracking studies, they carry notable limitations. They are constitutively active, signal magnitude depends on expression level rather than on a specific molecular event, and, critically, require exogenous iron supplementation to generate sufficient MRI contrast. Together, these properties make them poorly suited to biosensing applications that require a switchable, activity-dependent readout. A second class of reporters exploits paramagnetic metal-binding proteins, including engineered variants of the dopamine- and serotonin-sensing MRI reporters developed by directed evolution [2], and calcium-responsive reporters based on calprotectin-driven chelation of manganese [10]. These provide genuine biosensing functionality but still require exogenous metal cofactors, limiting their applicability in vivo and raising translational concerns. A third class exploits chemical exchange saturation transfer (CEST), in which highly charged polypeptides, such as polylysine, exchange protons with bulk water at a distinct frequency offset, generating contrast without paramagnetic metals [11]. While promising, CEST reporters suffer from relatively low sensitivity compared to metal-based approaches. More recently, gas vesicle–based acoustic reporters derived from aquatic microorganisms have been developed and shown to function in mammalian cells, offering a distinct and genetically encodable contrast mechanism [12]. However, the large genetic payload required for gas vesicle biosynthesis, combined with the lower depth penetration of ultrasound compared to MRI, constrains their widespread practical deployment.

Aquaporin-based reporters represent a conceptually different and practically attractive strategy. Aquaporins are compact, single-gene transmembrane water channels that accelerate transcellular water exchange, producing a measurable increase in the ADC detectable by diffusion-weighted MRI (DW-MRI), a sequence available on virtually every clinical and preclinical MRI scanner with no exogenous contrast agent required. Human aquaporin-1 (hAqp1) was established as a genetically encoded MRI reporter by our lab in 2016 [5], and subsequent work demonstrated its safety and utility across diverse mammalian cell types [13] and in vivo. Notably, hAqp1 has been used as a gene reporter to track viral delivery across the blood–brain barrier (BBB), trace neural connectivity in the rodent brain, and monitor tumor progression in mice. Despite these advances, a key limitation remained: aquaporin-based reporters lacked a general, modular framework for coupling the reporter to arbitrary molecular inputs. Each new biosensor application required independent engineering of the aquaporin protein itself, with no guarantee that the sensing domain would modulate channel activity as intended.

The MAPPER platform described in this protocol addresses this limitation by decoupling the reporter (hAqp1) from the sensing module [14]. Rather than engineering the aquaporin directly, MAPPER places hAqp1 under the post-translational control of synthetic viral protease circuits, a strategy that exploits the well-characterized programmability of protease-based signaling architectures developed in the synthetic biology field [15,16]. Two complementary mechanisms are provided: DD-MAPPER, in which hAqp1 stability is controlled by a protease-cleavable destabilizing domain (DD) [17–19], and ER-MAPPER, in which hAqp1 membrane trafficking is regulated by a protease-cleavable endoplasmic reticulum retention signal (ER) [20]. Because the protease recognition sequence and not the aquaporin itself is the point of input integration, a new molecular target can be detected simply by rewiring the upstream protease circuit: substituting a different protease, splitting the protease across two interacting proteins to sense protein–protein interactions, or placing protease expression under ligand-dependent post-transcriptional control. This modularity sharply reduces the engineering burden compared with prior reporter strategies and, critically, has been validated across diverse mammalian cell types and multiple target classes without requiring optimization of the core reporter for each application.

One potential limitation of MAPPER-based sensors is the latency associated with protease-based regulation; signal changes reflect cumulative protease activity over several minutes to hours, rather than capturing instantaneous molecular events. This mechanism is similar to that of integral sensors in optical imaging and offers several practical benefits, such as a higher signal-to-noise ratio and applicability in awake, ambulatory animals. Another limitation of MAPPER is the larger genetic footprint of these sensors because of their circuit-based (rather than conventional single protein–based) architecture. Although most MAPPER constructs remain within the packaging limits of lentiviral vectors (~9.2 kb), delivering MAPPERs via adeno-associated viral vectors (AAVs) will likely require splitting the sensor across multiple AAVs for co-transduction. Within this context, it is worth noting that a distinct advantage of MAPPERs is their entirely post-transcriptional operational mechanism, which potentially allows for delivery as mRNA-encoded devices in lipid nanoparticles. This approach circumvents the packaging limitations inherent to viral gene delivery vectors and may be particularly beneficial for the in vivo delivery of gene-based diagnostics, which do not require sustained or permanent genetic modification of cells and tissues.

In summary, the primary advantage of MAPPER over existing MRI reporter technologies is its programmability, offering a sensor engineering platform vis-à-vis a single bespoke biosensor. Beyond the applications demonstrated in the original publication, we anticipate that MAPPER will be useful across a broad biomedical space, from capturing various stages of signal transduction in physiological and disease states to longitudinally tracking the functional outcomes of gene- and cell-based medicines.