Thransport of macromolecules from the cerebrospinal fluid to the CNS parenchyma

Objectives: The goals of our studies were to (i) investigate the poorly understood physiological processes occurring during the transport of intrathecally (IT) administered macromolecules and particles to the CNS, and (ii) evaluate the feasibility of these processes for the delivery of biopharmaceuticals to the CNS.

Methods: Recombinant human enzymes, virions, soluble polymers and nanoparticles were labeled with ¹²⁴I or ⁸⁹Zr and administered IT to rats and cynomolgus monkeys. Dynamic PET imaging data (0-30 min post-injection) and multiple whole-body images over at least 48 hours were acquired. Images were analyzed to determine the rates and patterns of the label spread within the CSF and then into CNS from the injection site. The translocation of model macromolecules labeled with fluorophores from the CSF into CNS along perivascular conduits was studied by fluorescence microscopy.

Results: The general patterns of solute transport in the CSF of rats and monkeys were similar. The initial solute distribution in the CSF greatly depended on the volume injected. In monkeys, lumbar administration at >0.5 ml/kg resulted in the immediate delivery of up to 90% of the injected volume to the cerebro-cervical CSF cisternae. The subsequent solute spread depended on the initial location and was slow in the spinal CSF (millimeters per hour) but fast in the cerebral CSF (complete equilibration within 30 min). No evidence of directional solute flows anywhere in the CSF was found. PET imaging showed solute penetration into both white and gray matter regions of the brain within 2-5 hours (depending on the probe) after the injection, with subsequent biphasic clearance. Fluorescence microscopy further demonstrated the penetration of the intrathecally administered material along numerous perivascular channels originating from both external and internal (fissures) boundaries throughout the CNS (brain, cerebellum, spinal cord) with indications of probe exits to the parenchyma. The rate and depth of the entrance exclude diffusion-driven mechanisms. The data suggest a strong size dependence of solute clearance from the CSF, likely through at least three different mechanisms, including two types of pore-mediated transport and diffusion. No evidence of significant lymphatic drainage from the CSF was found in monkeys. However, in rats, some lymphatic drainage was detected in the deep anterior cervical area (nodal uptake of ca. 3% of the injected dose).

Conclusion: The overall mechanistic landscape of the cerebrospinal solute transport significantly differs from the paradigm suggesting that CSF bulk flows prevail outside the CNS, while interstitial flows prevail within. The major factor defining the initial solute distribution in the CSF is the hydrostatic compliance of the compartment, whereas hydrodynamic “remixing” of the compartment plays the major role in the subsequent fast (head) and slow (spine) solute spread in the CSF. Solute translocation along the perivascular conduits likely relies on hydrodynamic remixing as well. The ubiquity of such conduits within the CNS deems the intrathecal route potentially feasible for the delivery of biopharmaceuticals to a variety of CNS targets. The mechanisms of active macromolecule transport in the CSF and into the CNS are generic and do not depend on the type of the macromolecule or particle. Research Support: This study was supported by NIH grants R21NS090049 and R01NS092838.