A neural probe for concurrent real-time measurement of multiple neurochemicals with electrophysiology in multiple brain regions in vivo

Significance Real-time monitoring of multiple neurochemicals has been challenging due to the crosstalk between sensors. The proposed system enables multiple measurements through the monolithically integrated microfluidic chip on a neural probe. Also, the proposed real-time bimodal neural probe allows the concurrent observation of electrophysiological signals and different neurochemical concentrations from multiple brain regions of a live animal in real time. In this manuscript, we successfully measured bimodal neural activities in vivoin the medial prefrontal cortex and the medial dorsal thalamus of an intact mouse brain. We expect that the system will contribute to not only investigating the role of neurochemicals in neural circuits related to brain function but also developing the drug for a variety of neurochemical-related brain disorders.


S1. Fabrication of the RTBM MEMS neural probe
Our RTBM MEMS neural probe was fabricated with a microfabrication process similar to the previously reported probe(1). We selected an SOI wafer of 40 μm top silicon and formed trenches for integrating microfluidic channels through deep reactive-ion etching (DRIE) processes on the top silicon layer. The trenches consisted of a center cavity (25-μm high and 30-μm wide) for the glass layer anchoring and two sets of five cavities (25-μm high and 6-μm wide) for microfluidic channels located on both sides based on the center cavity. Then, the microchannels were formed through a glass reflow process.
We patterned the gold (Au) signal line (20-nm-thick titanium (Ti) and 300-nm-thick Au) on the silicon substrate with microfluidic channels to electrically connect the electrical and chemical signals. Afterward, 400-nm-thick SiO2 was deposited on the surface of the wafer using plasma-enhanced chemical vapor deposition (PECVD). The reactive-ion etching (RIE) process was performed to remove the SiO2 layer at the position where the pads for wire-bonding and the platinum (Pt) electrodes (i.e., recording and biosensor electrodes) would be formed. The recording and biosensor electrodes were deposited with 20-nm-thick Ti and 150-nm-thick Pt using a sputter and defined by a lift-off process. Finally, to define the shape of the probe, the top and bottom of the sample were etched through the DRIE process and a neural probe was released.
To fabricate the PMDS-based interface chip, we first fabricated a master mold via the SU-8 patterning process, the previously reported photolithography method (2). We patterned a 12-μm thick SU-8 layer (SU-8 3010, Kayaku Advanced Materials Inc., USA) as the first layer with multiple drug delivery channels and channel resistance patterns in the extraction channel. Afterward, a 60-μm thick SU-8 layer was secondly patterned in the extraction channel except for the resistance pattern. Finally, the wafer with patterns of microfluidic channels was cut according to the size of the PDMS interface chip with a laser cutter.

SI-3
We packaged the RTBM neural probe to provide the fluid and electrical interface with external instruments for performing multiple functionalities (i.e., real-time bimodal monitoring and chemical stimulation). First, we attached the fabricated neural probe to a PCB and conducted wire-bonding between the pads of the probe body and of the PCB to provide electrical connections. Then, we applied thermal epoxy to part of the bonded wires and pads and cured the device in an oven at 80 °C for 30 minutes.
Afterward, we completed enzyme layer patterning on the working electrode array described above. Additionally, we fabricated the interface PDMS chip that the mixed PDMS with the base to curing agent ratio of 10:1 was poured into a duralumin master mold covered with a piece of silicon with SU-8 microfluidic channels pattern and cured thoroughly at 80 °C for 2 hours. Then, we punched the ports where the tubes were to be connected with a custom punch (inner diameter (ID) of 0.3 mm and an outer diameter (OD) of 0.7 mm). Finally, we applied O2 plasma bonding (100 W, 50 mTorr for 40 seconds) between the fabricated PDMS interface chip and the body of probe with an enzyme-coated biosensor array.

S3. Pt black electroplating
We prepared an electroplating solution by adding 30 mg⋅ml -1 of chloroplatinic acid hydrate, 25 μM HCl, and 0.25 mg⋅ml -1 in DI water according to the previously reported protocol(1). Then, we immersed the recording electrodes (working electrode) of the neural probe and an Ag/AgCl wire (reference electrode) in the electroplating solution.
Pt black was electroplated by applying -0.2 V to the recording electrode for 35 seconds to reference the Ag/AgCl wire using the chronoamperometry method of the twoelectrode system (PalmSens3, PalmSens, Netherlands).

S4. Characterization of the microfluidic system
We made customized fluidic adapters by plugging pre-cut 23-gauge needles into the Tygon Microbore tubes (ID: 0.05 cm, OD: 0.15 cm). Then, we connected the front end of the customized fluidic adapters to the PDMS interface chip port and the back end of the customized fluidic adapters to the external pressure controllers. We used mass flow controllers (ITV0090-3BL and ITV0051-3BL, SMC Corporation, Japan) to precisely regulate the pressure. Flow controllers require both positive and negative SI-4 pressure to operate. The positive pressure was applied to the nitrogen gas tank and the negative pressure was applied to the vacuum pump.
We immersed the shank of the neural probe in 0.1 M PBS and measured the flow rate by precisely controlling the pressure to notice the positive and negative pressure at which the sampling flow rate was 100 nl⋅min -1 . Thereafter, an experiment to confirm the time when the drug and the aCSF solution were switched was performed in the same manner as previously reported(1). Briefly, two delivery ports were filled with redand blue-dye-infused water, and then the pressure was controlled so that the red dye was injected into PBS for 5 minutes. Afterward, the time at which the blue dye was observed at the probe tip was measured by switching the pressure applied to two ports.

S5. Brain slice preparation and ex vivo experimental procedure
Mouse was anesthetized with isoflurane and decapitated. The brain was then immediately removed and placed in ice-cold slice cutting solution (0-4 °C) containing the following aCSF: 250 mM sucrose, 26 mM NaHCO3, 11 mM KCl, 1.2 mM NaH2PO4, 7 mM MgCl2, and 0.5 mM CaCl2. Coronal slices of the hippocampus, 300 μm thick, were obtained using a Vibratome (Leica VT1200S; Leica, Germany). For recording purposes, brain slices were transferred to the following aCSF: 126 mM NaCl, 3.5 mM KCl, 1.2 mM NaH2PO4, 1.3 mM MgCl2, 2 mM CaCl2, 25 mM NaHCO3, and 11 mM Dglucose, bubbled with a gas mixture of 5% CO2 / 95% O2 to maintain a pH of 7.4. The slices were incubated at 28 °C for 30 minutes to recover. All recordings were performed within 6-8 hours from recovery.
To measure the change in neurochemicals induced by the electrical stimuli, current pulses (6-ms long, 50-Hz two-paired monophasic pulses with amplitudes ranging 300 μA) were applied to 1 to 6 recording electrodes combined to the anode and to 7 to 12 recording electrodes combined to the cathode in combination with a stimulus isolator (model DS3, Digimeter Ltd., UK) and a waveform generator (33500B series, Agilent Technologies, USA).

S6. In vivo experimental procedure
The experimental procedures on animals that were approved by the Korea Institute of Science and Technology (KIST), Seoul, Korea, were performed following the ethical standards outlined in the Animal Care and Use Guidelines of KIST. We used adult male mice (C57BL/6, 10-12 weeks of age) for all in vivo experiments, and the in vivo SI-5 experimental procedure was similar to that previously reported(1). Briefly, we mounted a mouse anesthetized with urethane (400 mg ⋅kg -1 , intraperitoneal injection) in a stereotaxic frame (David Kopf Instruments, USA), incised the scalp, and made small circular holes in the skull by drilling to reach the target location according to Paxinos and Franklin(3). Then, an RTBM neural probe was fixed to the stereotaxic frame so that the probe shank could vertically lower and reach the brain target regions. A preamplifier (Intan Technologies, USA, the RHD2000 system acquired at 20 kS⋅s -1 per channel, band-pass filtered at 0.3 kHz-6 kHz for the detection of action potentials) and customized connectors linked three electrodes were connected to the omnetics and biosensing connectors on an RTBM neural probe, and delivery tubes and extraction tubes were connected to the interface chip. Before implanting the probe shank into the brain tissue, we injected the aCSF into the extraction tubes at 100 ml⋅min -1 and simultaneously applied a working potential to the biosensors for 30 minutes to stabilize the biosensors and remove bubbles in the extraction channels. We calculated the concentration of neurochemicals by comparing the stabilized currents with the current measured on in vivo experiment. Next, the probe slowly lowered at a speed of 1 mm⋅ min -1 until the probe tips reached the target regions. Subsequently, the sampling flow rates were matched, and the stabilization of electrical signals and biosensors was performed for an additional 30 minutes. Then the in vivo experiment was performed to investigate the functional connectivity and the correlation of electrical and chemical activities. The high KCl solution (100 nl, 100 mM KCl in aCSF) delivered to modulate neural activity was injected over 1 minute at 30-minute intervals and repeated 3 times in total.

S7. Analysis of data in vivo
The analysis of electrical signals starts with sorting electrical neural spikes using previously reported custom MATLAB(4) (Mathworks, USA). We sorted the neural spikes by setting the spike latency to 1 millisecond and negative and positive threshold voltages. We calculated the firing rates by dividing the number of sorted spikes by time and displayed them as heat maps. In the sorted clusters, autocorrelation was calculated through the signals recorded for 60 seconds (2400-2460 seconds) during the second KCl modulation. The significance of the difference in firing rates was evaluated using the Student's paired t-test (Prism, GraphPad Software Inc., USA). The

SI-6
firing rate values used for the significance of difference were the average firing rates measured at 5 minutes before and after the KCl injection and during the KCl injection each for 1 minute.
In the chemical signal analysis, the concentration of neurochemicals was expressed by substituting the currents with the concentration as functions of the currents measured in the in vitro characteristics. The quantitative value of concentration was obtained by subtracting the background current obtained from the measured saturated current before implanting the probe shank into the brain tissue.
The significance of the difference in neurochemicals was also evaluated using the Student's paired t-test. The significance of the difference in concentrations of glutamate, lactate, and choline was calculated as the average of concentrations before and after KCl injection at 5 minutes and during KCl injection at the maximum neurochemical concentration each for 10 seconds. In glucose, the significance of the difference was calculated by the average of the concentrations at 5 minutes before KCl injection, at the minimum concentration during KCl injection, and at the maximum concentration after injection each for 10 seconds. We explored the correlation between neural signals in the mPFC and MD regions by analyzing their cross-correlation.
Initially, we determined the firing rate of all recorded neurons at intervals of 30 seconds and calculated the change in their firing rates. We obtained the change in firing rate by subtracting the firing rate of the current event from that of the previous event and dividing the result by the sum of the two firing rates. Then, we multiplied the change in the firing rate of neurons in the MD by the change in the firing rate of neurons in the mPFC to obtain the cross-correlation value. If the neural activity in the MD region responds to the neural activity in the mPFC region, the cross-correlation value will be close to 1. On the contrary, if the neural activity in the MD region does not respond to the neural activity in the mPFC region, the cross-correlation value will be close to 0.

S8. Viral tracing experimental procedure on the mPFC-MD neural circuit
We performed anterograde tracing to confirm the mPFC-MD neural circuit. We conducted the procedure on adult C57BL/6J male mice (8 weeks old) under isoflurane anesthesia (5% induction, 1% maintenance), which were fixed on a stereotactic apparatus (Ultra Precise Mouse Stereotaxic Instruments; Stoelting Co., USA). We made an incision in the scalp and drilled a craniotomy for injecting the virus into the mPFC using a 30-gauge microinjection cannula (P1 Technologies, USA) and an

SI-7
UltraMicroPump III (World Precision Instruments, USA) at a rate of 0.06 µl per minute.
After injecting approximately 0.3 µl virus into the mPFC (coordinates: AP +1.70 mm, ML -0.35 mm, DV -1.85 mm from the bregma), we closed the incision with 9-mm autoclips (205016, MikRon Precision, Inc., USA), and administered antibiotics and analgesics to the mice. We allowed the mice to recover from anesthesia and then kept them in their home cages for 2 weeks to allow the AAV to take effect. To identify GFP neurons or terminals, we anesthetized the mice with a mixture of alfaxalone (40 mg⋅ per -1 ⋅kg -1 ) : xylazine (10 mg⋅per -1 ⋅kg -1 ) and perfused them with 0.9% saline and then 4% paraformaldehyde 2 weeks after virus injection. After postfixation, we cut the brains into 40-µm-thick coronal sections with a cryotome (CM300, Leica Camera, Germany); the sectioned brain regions encompassed the mPFC or MD. We washed the sections with PBS three times for 5 minutes at room temperature and then incubated them with Hoechst 33342 (H3570, Thermo Fisher Scientific, USA), a blue fluorescent stain that binds double-stranded DNA, at room temperature for 10 minutes.
Finally, we immersed the stained sections in mounting solution for 30 minutes at 37 °C and viewed and photographed them using a TCS SP8 dichroic/CS microscope (Leica). presented as mean ± s.d. (n = 3 for all, Two-tailed t-test). In C1, **P = 0.0032, t = 6.302