Fabrication Methods and Chronic In Vivo Validation of Mechanically Adaptive Microfluidic Intracortical Devices

Abstract

Intracortical neural probes are both a powerful tool in basic neuroscience studies of brain function and a critical component of brain computer interfaces (BCIs) designed to restore function to paralyzed patients. Intracortical neural probes can be used both to detect neural activity at single unit resolution and to stimulate small populations of neurons with high resolution. Unfortunately, intracortical neural probes tend to fail at chronic timepoints in large part due to the neuroinflammatory response that follows implantation and persistent dwelling in the cortex. Many promising approaches are under development to circumvent the inflammatory response, including the development of less inflammatory materials/device designs and the delivery of antioxidant or anti-inflammatory therapies. Here, we report on our recent efforts to integrate the neuroprotective effects of both a dynamically softening polymer substrate designed to minimize tissue strain and localized drug delivery at the intracortical neural probe/tissue interface through the incorporation of microfluidic channels within the probe. The fabrication process and device design were both optimized with respect to the resulting device mechanical properties, stability, and microfluidic functionality. The optimized devices were successfully able to deliver an antioxidant solution throughout a six-week in vivo rat study. Histological data indicated that a multi-outlet design was most effective at reducing markers of inflammation. The ability to reduce inflammation through a combined approach of drug delivery and soft materials as a platform technology allows future studies to explore additional therapeutics to further enhance intracortical neural probes performance and longevity for clinical applications.

Keywords: drug delivery; mechanically adaptive; microfabrication; microfluidic; neural interface; polymer.

Figure 1

Steps for film fabrication using three methods: mold-only, emboss-only, and mold-emboss. (A) Solution of PVAc and tCNC in DMF is poured into a negative PDMS mold, and the solvent is allowed to evaporate in an oven to create channel layers for the mold-only and mold-emboss methods. (B) Same solution as in A is poured over a flat PDMS slab and the solvent evaporates in an oven to form a flat film using the emboss-only method. (C) The mold-only method removes the PDMS mold to form channel layer films. (D) The mold-emboss method takes the film and mold prior to separation and embosses between two sheets of Teflon shown here as white rectangles. (E) The emboss-only method removes the film from the flat PDMS slab and embosses the PDMS mold over the film between Teflon sheets. In both the mold-emboss and emboss-only methods, the PDMS molds are removed after embossing. (F) Schematic representation of a 100 mm-diameter film, with rectangular chips. (G) Each chip consists of 16 channels, corresponding to 16 probes (8 channels shown in the schematic). (H) The channel layer film with channels has the width defined as the wall-to-wall distance of the channel. The depth is defined as the top of the film surface to the bottom of the channel trough distance.

Figure 2

(A) Cover layers and channel layers were aligned and thermally bonded. (B) Afterward, individual microfluidic probes were micromachined and removed from the bonded chips.

Figure 3

Single and branched probe designs with single and double outlets. The top row shows the overall design of a probe with 100 µm and 50 µm wide channels in either a single-outlet design or a double-outlet branched design. The middle row shows a close-up of the tips of the probes and their channels. The bottom row shows a cross-section of the probes and the microfluidic channels running through the shanks.

Figure 4

Schematic of setup for functionality validation. (A) A load cell readout is connected to a computer and records the force. (B) A load cell attached to the syringe pump detects force at the syringe plunger. (C) A syringe pump with a Luer-lok syringe filled with air. (D) A completed device connected to the syringe pump and submerged in a beaker of DI water. (E) A microscope camera to record process and detect bubbles at the probe tip.

Figure 5

Diagram showing placement of the probe and osmotic pump in the rats. The probe is placed 2–3 mm anterior and lateral to the bregma. The osmotic pump is placed in a subcutaneous pocket in the mid-scapular region of the rat’s back. (Created with https://www.biorender.com/, accessed on 11 February 2023).

Figure 6

Stylus profilometry and optical imaging results for depth (A) and width (B) measurements of 100 µm-wide microfluidic channels in films fabricated using the mold-only method, embossed-only method, and mold-embossed method. The PDMS mold used was also measured using optical imaging. All methods were significantly different from both the PDMS mold and each other, distinguished with asterisks (*), in width and depth. p < 0.001.

Figure 7

(A) Image of a mold-embossed probe with a 100 µm-wide (S100-me) good channel before packaging. (B) Image of a mold-embossed probe with a 50 µm-wide (S50-me) good channel pre-packaging. (C) Image of a mold-embossed probe with 50 µm-wide channels and two outlets (D50-me) with a good channel pre-packaging. Scale bar = 500 µm. (D) Image of a representative completed device with connector and tubing attached. Scale bar = 3 mm. (E) Cross-section image of a probe with a 50 µm-wide channel. Scale bar = 100 µm.

Figure 8

Histological response to implanted probes using 50 µm–wide channels with a single outlet (N = 3), double outlet (N = 4), and control with no resveratrol delivery (N = 4). (A) Macrophage/monocyte response, (B) blood–brain barrier permeability, (C) oxidative stress response, (D) glial scarring response, and (E) neuronal nuclei density were evaluated to determine which of the probe designs had the best tissue response. p < 0.05. Significance is denoted by the asterisk (*) in the plots.