Biodegradable polyphosphoester micelles act as both background-free ³¹P magnetic resonance imaging agents and drug nanocarriers

The study addresses the need for noninvasive, quantitative tracking of polymers used in biomedical applications, such as drug delivery and tissue regeneration. While 1H MRI is widely used, heteronuclear “hot spot” MRI provides high specificity and quantitative imaging but typically relies on 19F agents, often per- and polyfluoroalkyl substances (PFAS), which are amphiphobic, challenging to stabilize or functionalize in physiological conditions, can persist in organs, and are environmentally persistent pollutants. The authors propose an alternative based on biocompatible and biodegradable polyphosphoesters (PPEs) that can be directly imaged via the 100% abundant NMR-active 31P nucleus in the polymer backbone. Key challenges for 31P MRI include endogenous phosphate background, lower MR sensitivity than 1H, unfavorable relaxation times, and J coupling. The hypothesis is that molecular design of PPEs, self-assembly into nanostructures to increase local 31P concentration, and tailoring microstructure and dynamics (e.g., using polymers above their glass transition temperature) can overcome these limitations to achieve background-free, sensitive 31P MRI while enabling drug delivery.

Prior work in heteronuclear MRI has focused on 19F MRI for imaging diseases, cell therapies, nanomedicines, and biomaterials. However, 19F agents often use PFAS/PFCs, which are difficult to stabilize and functionalize, can accumulate in organs, and pose environmental concerns due to chemical persistence. Efforts to improve 19F MRI include partly fluorinated polymers and inorganic nanoparticles. 31P MR spectroscopy has been used for pH sensing and detecting metabolites (phosphocreatine, ATP) and some non-biodegradable polymers, but requires long acquisition times and suffers from endogenous background and sensitivity limits. Polyphosphoesters have been explored as stealth, biocompatible, and degradable materials and PEG alternatives. A recent pioneering 31P MRI probe involved a phosphorus-containing polymeric zwitterion for biosensing. The present work builds on these developments by designing PPEs with P–C bonds to shift 31P resonances away from endogenous phosphates and by engineering microstructure to enhance relaxation times and signal-to-noise.

• Polymer design and synthesis: Amphiphilic PPE-based copolymers were synthesized, including block copolymers with 31P-containing hydrophobic cores (Ph-PPn) or hydrophilic shells (Et-PPn) and gradient copolymers (PPnGRAD) via single-step anionic ring-opening copolymerization of phenyl- and ethyl-functional monomers. Kinetic studies and Monte Carlo simulations established gradient microstructures and composition drift.
• Self-assembly and characterization: Copolymers self-assembled into micelles in water, characterized by DLS (radii ~9–10 nm, low PDI < 0.1), low critical micelle concentration determined by pyrene fluorescence (CMC ~14 mg L−1), and AFM for morphology. SEC (DMF/LiBr) with polystyrene calibration was used for molar mass analysis. DSC measured Tg under nitrogen from −80 to 50 °C.
• NMR spectroscopy: 1H/31P NMR (400 MHz class spectrometers) measured chemical shifts and relaxation times. T1 via inversion recovery; T2 via CPMG with at least 10 echo times; interscan delay d1=5×T1; samples in H2O/D2O mixtures (10% D2O for lock). Data processed with Mestrenova and Origin.
• MRI sequences and phantoms: 31P spin-echo imaging used for proof-of-concept on aqueous dispersions of micelles at 9.4 T, FOV 20×20 or 30×30 mm2, matrix 64×64, acquisition time TAcq 17 min. Multi-chemical selective imaging (mCSSI; spin-echo-based) implemented to simultaneously image multiple 31P resonances (Ph- and Et- units) and sum images to boost SNR while avoiding chemical shift artefacts. 1H MRI provided anatomical reference.
• In vitro and ex vivo demonstrations: Physalis berry injection demonstrated tissue imaging feasibility with 1H/31P overlay using mCSSI.
• In vivo model: Manduca sexta caterpillars used as alternative animal model (hemolymph volume ~1–2 mL). PPnGRAD micelles (15 mg in 100 µL; dose aligned with typical 19F agent doses) were injected into the dorsal vessel (circulation) or directly into the gut. 1H/31P MRI performed immediately and at 24 h to assess distribution and persistence. Feces collected at 24 h were analyzed by 31P NMR to detect degradation products.
• Drug formulation and cell assays: PROTAC ARV-825 was encapsulated by nanoprecipitation into PPnGRAD micelles to assess theranostic potential. Stability assessed by DLS over 2 months. Cell viability in HeLa cells compared for free PROTAC, PROTAC-loaded micelles, and unloaded micelles (n=3, mean±SD). Flow cytometry (Annexin V/PI) evaluated apoptosis at 1 µM ARV-825.

• Background-free 31P MRI feasibility: Polyphosphonate (PPn) micelles yielded distinct 31P chemical shifts due to P–C bonds, separated by >10 ppm from endogenous phosphates, enabling selective imaging.
• Relaxation enhancements: Moving from solid-like hydrophobic nanoparticles (T2=0.017 s) to amphiphilic micelles improved T2. Block-copolymer micelles showed: PPnCORE (Ph-PPn in core) T1≈1.0 s, T2≈0.049 s; PPnSHELL (Et-PPn in shell) T1≈2.3 s, T2≈0.46 s. Gradient PPnGRAD micelles further increased T2: Et-PPn block T2≈0.9 s; Ph-PPn core T2≈0.13 s.
• Increased 31P content and SNR: Gradient micelles incorporated both 31P-containing blocks, achieving ~20 wt% 31P per micelle and, using mCSSI to sum Ph- and Et- resonances, delivered ~3× higher SNR than block-copolymer micelles at equal polymer dose (TAcq 17 min, 9.4 T). Signal intensity scaled linearly with concentration.
• Comparative sensitivity: Under specified conditions, a perfluoro-15-crown-5 ether emulsion (3.8 M 19F) showed ~5× higher SNR than PPnGRAD (1.5 M 31P), though systems were not directly comparable.
• Tissue demonstration: Injection into physalis berries enabled artefact-free localization with 31P mCSSI and co-registered 1H MRI.
• In vivo imaging in Manduca sexta: After dorsal vessel injection, PPnGRAD micelles distributed homogeneously in hemolymph and remained detectable at 24 h, suggesting prolonged circulation (stealth behavior).
• Biodegradation in vivo: After gut injection, feces collected at 24 h showed sharp 31P NMR resonances at ~16 and ~31 ppm corresponding to low-molar-mass degradation products, alongside broader parent signals (~20 and ~38 ppm), consistent with backbiting hydrolysis and faster degradation of outer Et-PPn.
• Drug delivery function: Nanoprecipitation loaded PROTAC ARV-825 into PPnGRAD micelles forming clear, stable dispersions (radius ~10 nm; stable ≥2 months by DLS). Without polymer, PROTAC precipitated into polydisperse aggregates. PROTAC-loaded micelles reduced HeLa cell viability comparably to free PROTAC and induced early apoptosis at 1 µM, while unloaded micelles served as benign controls.
• Self-assembly metrics: PPnGRAD micelles exhibited low PDI (<0.1) and low CMC (~14 mg L−1), akin to nonionic surfactants.

Engineering PPEs with P–C bonds shifts 31P resonances away from endogenous phosphate signals, enabling background-free hotspot MRI. Self-assembly into micelles increases local 31P concentration, and operating above Tg enhances chain mobility, extending T2 and improving SNR. Gradient microstructures broaden interfacial regions within micelles, further increasing mobility and T2 for both hydrophilic and hydrophobic domains relative to block architectures. Implementing mCSSI allows simultaneous, artefact-free imaging of multiple 31P resonances and summation to maximize SNR. Demonstrations in plant tissue and in vivo in Manduca sexta validate imaging specificity, persistence in circulation consistent with PPE stealth properties, and biodegradability, addressing concerns associated with PFAS-based 19F agents. The ability to encapsulate poorly soluble therapeutics (e.g., PROTACs) underscores the theranostic potential of PPnGRAD micelles, combining drug delivery with quantitative, background-free 31P MRI tracking.

This work establishes biodegradable polyphosphoester micelles as dual-function, background-free 31P MRI agents and drug nanocarriers. By synthesizing amphiphilic gradient polyphosphonates, the authors achieved improved 31P relaxation times, higher effective 31P content per micelle, and superior SNR via mCSSI compared to block copolymers. In vivo imaging in Manduca sexta confirmed circulation and localization, while 31P NMR of feces verified biodegradation. The same platform encapsulated a hydrophobic PROTAC and retained cytotoxic efficacy, demonstrating theranostic capability. Future work should optimize sensitivity and acquisition speed, quantify pharmacokinetics and biodistribution in mammalian models, refine degradation profiles and safety, and expand to other therapeutic cargos and imaging ‘colors’ via chemical shift engineering.

• Sensitivity: 31P MRI showed lower SNR than a high-content 19F standard (~5× under the tested conditions), necessitating optimization for clinical translation.
• Model systems: In vivo validation was performed in Manduca sexta; biodistribution, pharmacokinetics, and safety in mammalian models remain to be fully characterized.
• Acquisition time: Imaging used relatively long acquisition times (≈17 min) to achieve sufficient SNR.
• Degradation timeline: Gut transit (~1.5 h) in Manduca was insufficient for complete degradation; comprehensive in vivo degradation kinetics and metabolite fate require further study.
• Comparative conditions: SNR comparisons to 19F systems are not directly comparable due to differing physicochemical properties and concentrations, limiting quantitative benchmarking.