3D-printed optogenetic neural probe integrated with microfluidic tube for opsin/drug delivery

Abstract

Optogenetics, known for its precision in neural stimulation, is integral to behavioral research, enabling the study of neural circuits involved in decision-making, memory, social interaction, and movement. Traditional methodologies require two separate surgeries: the first to deliver a viral vector containing the opsin gene to the targeted brain region, and the second to implant an opto-probe for light stimulation. This dual-step process increases the risk of tissue damage and misalignment between the injection and implantation sites. In this study, we present a 3D-printed multimodal optogenetic neural probe that combines light delivery and fluid injection into a single device. By integrating a commercially available microfluidic tube with a 3D-printed opto-probe, the device offers rapid and customizable assembly for diverse applications. The probe was implanted in the subthalamic nucleus of mice, enabling viral vector delivery and device implantation in a single procedure. Following viral expression, behavioral experiments demonstrated that optical stimulation increased travel distance and velocity, confirming effective neuronal activation. Immunohistochemistry analysis revealed successful expression of Channelrhodopsin-2 (ChR2(H134R)) through mCherry labeling of neurons, reduced astrocytic (GFAP) and microglial (ED1) activation around the implantation site, and preserved neuronal populations as confirmed by NeuN staining. These results highlight the device’s biocompatibility, minimal inflammatory response, and suitability for long-term neural modulation, with potential applications in research and clinical settings.

Introduction

Optogenetics is a well-established neuromodulation technique that employs light to regulate the membrane dynamics of targeted neurons¹. This method has revolutionized neuroscience research by allowing precise manipulation of neural circuits and behaviors. However, traditional optogenetic approaches face several challenges, including the need for multiple surgeries and limitations in device design flexibility. This optical modulation technique works by introducing light-sensitive proteins called opsins into the neuronal cell membranes typically using viral vectors. When these opsins are exposed to light of particular wavelengths, they become activated, allowing precise control over the neurons’ membrane dynamics and, consequently, their activities^2–5. For example, channelrhodopsin-2 (ChR2) is a prototypical opsin capable of neuronal excitation through light exposure within the 450 to 470 nm range by opening cation channels on the neuronal membrane⁶. Conversely, the natronomonas pharaonis halorhodopsin (NpHR) opsin can inhibit neural activity with light of a wavelength around 590 nm, which activates chloride ion pumps on the membrane⁷. Optical modulation of neural activity can influence brain functions or behaviors of the subject. The applications of optical neural modulation encompass the excitation and inhibition of sensory-motor movements⁸, appetite⁹, and respiration¹⁰. Furthermore, optogenetic stimulation can modulate sensory systems such as auditory^11,12olfactory^13,14 and visual pathways^15,16. A notable example of optogenetic modulation involves the subthalamic nucleus (STN), a critical component of the basal ganglia motor circuit. The STN receives excitatory cortical input and projects to downstream nuclei, including the globus pallidus and substantia nigra, thereby exerting a substantial influence on motor output. Optogenetic stimulation of the STN, particularly using Channelrhodopsin-2 (ChR2), enables temporally precise excitation of STN neurons via blue light, which increases neuronal firing rates and modulates downstream basal ganglia pathways. This modulation has been shown to affect motor behavior significantly, including enhanced locomotor activity characterized by increased travel distance and velocity.

These effects are well-supported in prior studies. Gradinaru et al.¹⁷ demonstrated that optical stimulation of the STN in rodent models of Parkinson’s disease could ameliorate pathological motor behaviors. Furthermore, Kravitz et al.¹⁸ provided evidence that optogenetic control of basal ganglia substructures, including the STN, can bidirectionally regulate movement, highlighting the STN’s pivotal role in motor behavior. Thus, optogenetic activation of the STN represents a powerful approach for probing and modulating neural circuits underlying motor control, with direct relevance to both basic neuroscience and therapeutic strategies.

The procedure for implanting an optogenetic device along with virus delivery involves several critical steps. Initially, a specific viral vector containing the gene for the light-sensitive opsin is injected into the target brain region to enable gene expression in the desired neurons, as shown in Fig. 1a. This injection is typically performed using stereotactic surgery, which ensures precise targeting. Following viral vector injection, a waiting period (1–2 weeks) is necessary to allow for sufficient opsin expression in the neurons. During this time, the surgical site is closed, only to be reopened later for a second surgery. During this surgery, an optogenetic device, such as a miniaturized optical neural probe or an optical fiber coupled to a light source, is implanted into the same brain region¹. The device is then securely fixed to the skull using skull screws and dental cement to ensure stability (Fig. 1a). Once implanted, the device can deliver controlled light pulses to modulate neuronal activity in vivo, enabling the study of specific neural circuits and behaviors¹⁹.

Fig. 1.

(a) The conventional procedure involves stereotactic injection of a viral vector to express light-sensitive opsins in target neurons, followed by surgical implantation of an optical probe to deliver controlled light pulses for neuronal modulation. (b) The combined procedure for optogenetic device implantation integrated with a microfluidic tube for virus delivery. It involves the stereotactic placement of the device in the target brain region, where the microfluidic tube delivers the viral vector to express light-sensitive opsins in specific neurons, followed by the simultaneous or subsequent activation of the optical probe to deliver controlled light pulses for precise neuronal modulation.

However, this two-surgery approach may have more risks in damaging sensitive tissues and provoking more immune responses. This response may lead to activation of microglia and formation of glial barrier by astrocytes, which may reduce the effectiveness of light stimulation by obstructing light penetration. Traditionally, the in vivo application of optogenetics has depended on the use of optical fibers linked to external lasers^19,20. This method has been the predominant technique since the inception of optogenetics. Despite its simplicity and efficacy, the optical fiber approach is constrained by its tethered operation and limited design flexibility.

These limitations impede swift adaptation to the varied requirements of complex neuroscience experiments, where critical factors include the number of light sources, the location, depth, direction of light and fluid delivery, and device implantation. Recent advancements in materials science and microfabrication techniques have sought to address these challenges by developing optical neural probes that can be customized in size, dimensions, and functions for minimally invasive and versatile operations²⁰. This microfabrication-based approach enables the utilization of various materials, such as silicon^21–23, SU-8²⁴, polyimide (PI)^25,26 and polyethylene terephthalate (PET)^27,28 to create highly customizable designs. Additionally, it facilitates the integration of optogenetic probes with micron-scale light sources and wireless control units, paving the way for untethered, chronic optogenetics in awake, behaving animals^25–32. Nevertheless, these methods require expensive, bulky equipment and specialized cleanroom facilities, resulting in high manufacturing costs and limited accessibility for laboratories lacking the necessary equipment or expertise. Furthermore, design modifications are time-consuming and challenging, necessitating the creation of new photomasks and/or the development of new fabrication processes. This difficulty in customization impedes the rapid adjustment and optimization of neural devices to meet specific application requirements.

To address these limitations, we propose a novel approach using 3D printing technology to create an optogenetic neural probe that combines light and fluid delivery in a single device. This innovative design allows for minimally invasive implantation and precise, localized delivery of both light and opsins or drugs. Figure 2a shows the overall design of the proposed device. Our device consists of a 3D-printed probe integrated with a micro-sized light-emitting diode ($\mu$LED) and a microfluidic channel. This microfluidic channel and injection method enables minimally invasive delivery of fluid into specific brain areas in comparison to standard injection techniques as shown in Fig. 1b. The use of 3D printing in the fabrication of optogenetic stimulation devices simplifies the manufacturing process, enabling rapid customization without the need for complex microfabrication techniques. This approach not only streamlines the fabrication process but also makes it more accessible to a wider range of laboratories. Following the completion of the device assembly and manufacturing procedures, the electrical, optical, and thermal properties of the device are characterized in this paper. Figure 2b shows the fabricated device with its dimensions emitting blue light generated by 1 mA current.

Fig. 2.

(a) Overall system view of the proposed optrode device integrated with microfluidic channel for opsin/drug delivery. (b) Images of the fabricated Microfluidic-tube Integrated Optrode (MIO) device. The device emits blue light when driven with a current of 1 mA. The main panel shows the illuminated optrode tip. The right insets show the device width ($\sim$  270 µm) and thickness ($\sim$ 150 µm), respectively. A comparison image with a US one-cent coin (inset, upper left) is included to indicate the overall device scale.

In this study, we present the design, fabrication, and characterization of our 3D-printed optogenetic neural probe integrating a microfluidic tube. We demonstrate its functionality through in vivo experiments in mice, targeting the Subthalamic nucleus (STN) region of the brain. Furthermore, we assess the device’s biocompatibility and efficacy using immunohistochemistry analysis, examining markers for opsin expression, inflammatory response, and neuronal preservation. By combining viral delivery and device implantation in a single surgical procedure, our approach aims to minimize tissue damage and immune response while providing a versatile tool for optogenetic studies. Results from these analyses showed reduced inflammatory response, suggesting the device’s implantation and operation were well-tolerated by brain tissue. The preservation of neuronal cells and successful opsin expression further underscored the biocompatibility of the device and its potential for safe, long-term applications in neural modulation. This work represents a step forward in making optogenetic techniques more accessible and adaptable for complex neuroscience experiments.

Results

Electrical and optical characterization

The device’s electrical characterization results are depicted in Fig. 3a, showing the current–voltage (I–V) curve of the proposed device. For the characterization, five samples were utilized. The $\mu$LED current was measured over the voltage applied across it by utilizing a precision source measurement unit (B2901A, Keysight, USA). The characterization current of the proposed device was measured as approximately 12 mA at an input voltage of 3.0 V. To stimulate neurons with light, light with sufficient intensity must be delivered to the target area. In order to stimulate neurons with ChR2, a minimum of 1 mW/mm² of optical output power is necessary at a wavelength of approximately 470 nm³³. Consequently, the optical properties were assessed under various current conditions utilizing an integrating sphere (S140C, Thorlabs, USA) and a power measuring console (PM400, Thorlabs, USA). The measured light intensities of five samples across a current range of 1 to 15 mA are depicted in Fig. 3b. Evidently, the mean light intensity can readily surpass 1 mW/mm², a threshold that is sufficient to induce neural activities. The apparatus’s light spectrum measured by a spectrometer (CCS100, Thorlabs, USA) is depicted in Fig. 3c, in conjunction with the light-sensitive spectrum of the ChR2³. The device’s light output reaches its maximum at 465 nm, a wavelength that falls well within the response spectrum of the ChR2.

Fig. 3.

(a) Current–voltage (I–V) curves of the $\mu$LED (n = 5). (b) Measured light intensity of the devices over the current input (n = 5). (c) Measured relative light intensity of the $\mu$LED used in this work (blue curve) and the ChR2 response spectrum (black curve)³.

Thermal validation

Tissue damage may result from the heat produced by the device during stimulation. In order to mitigate potential harm to brain tissues, it is advisable to keep the temperature variation to a minimum of 2 °C ($\mathrm{\Delta }$T < 2 $_{}^{\circ }$C)³⁴. Maintaining a low temperature variation is critical, as excessive heat can cause irreversible tissue damage, including necrosis and inflammation. The temperature measurement setup is illustrated in Fig. 4a. To quantify the heat emitted by the device, a type-T thermocouple copper-constantan sensor (MT-29/3HT Needle Microprobe, Physitemp Instruments INC) was connected to a microprobe thermometer (BAT-12, Physitemp Instruments INC). These sensors provide high sensitivity and a quick response to temperature changes, ensuring accurate monitoring during the experiment. In order to simulate the conditions of actual implantation, dental cement was utilized to encase the device prior to its injection into a 0.6$\%$ agar gel, which closely resembles the composition of brain tissues (see Fig.4a)³⁵. This setup allows for controlled experiments while mimicking the thermal properties of brain tissue. We recorded the temperature of the apparatus at various pulse widths and stimulation rates throughout the procedure.

Fig. 4.

(a) Test setup of thermal validation. (b) Measured temperature change ($\mathrm{\Delta }T$) for various stimulation rates and pulse widths at an input current of 5 mA. The dotted line at $\mathrm{\Delta }T=2_{}^{\circ }\text{C}$ indicates the safety threshold beyond which thermal damage to neural tissue may occur.

The stimulation frequency was varied from 10 to 60 pulses per second for three different pulse widths of 5, 10, and 15 ms at 5 mA of the current through the device. The thermal measurement results of the stimulator are illustrated in Fig. 4b. These results help to understand the thermal behavior of the device under different stimulation conditions. On the basis of these measurement outcomes, the stimulation parameters for the in vivo experiments detailed in the subsequent section were determined.

Assessment of efficacy of optogenetic device in subthalamic nucleus stimulation

We conducted a behavioral study to assess the efficacy of optogenetic manipulation using the proposed device in the subthalamic nucleus (STN) of mice. Two groups were evaluated: an experimental group implanted with the proposed Microfluidic-tube Integrated Optrode (MIO) device and a control group implanted with a lab-made optrode of similar structure³⁶.

Both devices had identical length and width dimensions (7 mm in length and 270 µm in width), but differed in thickness: the MIO device measured 150 µm, while the control device measured 85 µm due to the absence of an integrated microfluidic tube.The study aimed to compare the behavioral responses elicited by optical stimulation across pre-stimulation, stimulation, and post-stimulation phases, providing insights into the effectiveness and biocompatibility of the MIO device. The MIO device, combining an optical probe and microfluidic tube, allowed for simultaneous viral vector delivery and implantation into the STN. In contrast, the control group underwent a two-step process: separate viral vector injection followed by optrode implantation.

After post-surgery recovery, we performed weekly behavioral assessments over seven weeks. Each session consisted of three five-minute phases: pre-stimulation (no stimulation), stimulation (10 pulses/second, 13 ms pulse width)³⁷, and post-stimulation (no stimulation). During the stimulation phase, light pulses were delivered at a frequency of 10 pulses per second, with a pulse width of 13 ms and a current of 5 mA. The 13 ms pulse width was selected as a compromise between effective ChR2 activation and thermal safety, based on in vitro validations. The 5 mA stimulation current was determined from preliminary optical and thermal testing, which demonstrated that this level reliably produced sufficient light intensity (>1 mW/mm²) for ChR2 activation while limiting temperature increases to below 2 °C, ensuring safe operation during in vivo stimulation.

During these sessions, we continuously monitored the mice’s locomotor activity, measuring distance traveled and velocity. As an example, a video of one behavioral analysis session for a single mouse, covering pre-stimulation, stimulation, and post-stimulation phases, is provided in the supplementary information along with its recorded track. Our goal was to compare the MIO device with conventional optrodes by assessing the precision of viral delivery and the overall behavioral impact. This provided insights into the long-term stability and effectiveness of optogenetic manipulation using our integrated device versus conventional techniques.

Figure 5 illustrates a comparative analysis of weekly distance traveled and velocity in mice (mean ± SEM, n = 5) implanted with the conventional optical device or the microfluidic tube integrated optical device (MIO) in the pre-stimulation, stimulation, and post-stimulation phases. For statistical analysis, a non-parametric one-way ANOVA (Kruskal-Wallis test) was used to assess differences across the three phases for both devices. Post hoc comparisons were conducted using Dunn’s test with correction for multiple comparisons. Significant differences are denoted by *$p<0.5$, **$p<0.01$, ***$p<0.001$, ****$p<0.0001$.

Fig. 5.

Comparative Analysis of Locomotor Activity in Mice: Optrode vs. MIO Device. (a) Optrode device: Weekly distance traveled (top) and velocity (middle) presented as Mean ± SEM ($n=5$) for pre-stimulation, stimulation, and post-stimulation phases across seven weeks. Bottom panels show mean total distance (left) and velocity (right) for all phases, averaged across all mice. (b) MIO device: Weekly distance traveled (top) and velocity (middle) as Mean ± SEM ($n=5$) for pre-stimulation, stimulation, and post-stimulation phases. Bottom panels show mean total distance (left) and velocity (right) for all phases across seven weeks.

In mice implanted with the optrode device(Fig. 5a), the weekly distance traveled (top panel) and velocity (middle panel) varied across the pre-stimulation, stimulation, and post-stimulation phases over seven weeks, with marked increases observed during the stimulation phase. Analysis of mean ± SEM for all animals over all the weeks for distance traveled (bottom left panel) showed a significant increase during the stimulation phase compared to pre-stimulation (p < 0.01)phase, while no significant difference was detected between stimulation and post-stimulation phases. The mean velocity (bottom right panel) did not show statistically significant differences across the three phases.

For mice implanted with the MIO device (Fig. 5b), the weekly distance traveled (top panel) and velocity (middle panel) demonstrated distinct changes across the pre-stimulation, stimulation, and post-stimulation phases over seven weeks, with notable increases observed during the stimulation phase. Mean ± SEM for all animals over all the weeks for distance traveled (bottom left panel) revealed significant differences, with stimulation showing higher values compared to pre-stimulation (p < 0.001). However, the velocity (bottom right panel) did not show statistically significant differences between phases.

Overall, Both devices demonstrated enhanced locomotor activity during optogenetic stimulation, confirming effective neural modulation. However, the MIO device exhibited more consistent performance over seven weeks, supporting its superior design for long-term behavioral studies. These findings highlight the MIO device’s potential as an improved tool for simultaneous neural stimulation and fluid delivery in optogenetic research.

Evaluation of expression of ChR2

The expression of Channelrhodopsin-2 (ChR2) in the mouse brain was evaluated using the viral construct pAAV-CaMKIIa-hChR2(H134R)-mCherry (AAV9). To confirm successful transduction and expression of ChR2, the associated mCherry fluorescence was examined in the brain sections. mCherry, which is co-expressed with ChR2, served as a reliable fluorescent marker for identifying neurons that had incorporated the viral vector. The expression of mCherry was assessed in brain sections from both the MIO device and the control device groups. The brain tissue was imaged using fluorescence microscopy, with an excitation wavelength of approximately 587 nm, specific for mCherry. This allowed for the visualization of mCherry expression (see Fig. 6a,b), which directly corresponded to ChR2 expression in the targeted neuronal populations. Both the MIO device and the control device showed effective transduction, with clear mCherry expression in neurons, confirming the successful delivery and integration of ChR2 via both methods. The newly developed MIO device efficiently combined viral delivery and device implantation into a single procedure, potentially minimizing tissue damage and the associated immune response.

Fig. 6.

Representative horizontal section showing mCherry expression near the device tract in the subthalamic nucleus (STN) region. (a) mCherry expression in the mouse brain using the method where the viral vector was delivered via a Hamilton syringe, followed by a separate surgery to implant the optrode after a waiting period of 1–2 weeks. (b) mCherry expression in the mouse brain following viral delivery using the proposed MIO device. This method allowed for simultaneous viral injection and optrode implantation, reducing the procedure to a single surgery. Both methods successfully demonstrated mCherry expression, indicating effective ChR2 transduction.

Assessing tissue response: optrode device versus microfluidic-tube integrated optrode (MIO

• Neuronal density: To evaluate the neuronal impact of implanting conventional Optrode Devices versus Microfluidic-tube Integrated Optrodes (MIO), we analyzed neuronal density in different regions of interest (ROIs) surrounding the implantation sites. Specifically, we examined three ROIs: ROI-1 (50–100 µm from the implantation site), ROI-2 (100–150 µm), and ROI-3 (150–200 µm).

  The results shown in Fig. 7 demonstrate that neuronal density was significantly higher with the MIO device compared to the Optrode Device in all three ROIs. In ROI-1, the MIO device exhibited a significantly higher density (P < 0.05). Similarly, ROI-2 showed a higher neuronal density with the MIO device (P < 0.01). In ROI-3, the MIO device also demonstrated a statistically significant difference in neuronal density (P < 0.05). These findings suggest that the MIO device causes less disruption to neuronal populations in the vicinity of the implant compared to the Optrode Device, even at greater distances from the implantation site.

• Astrocytic response: Astrocytic response was evaluated using GFAP staining to detect reactive astrocytes in the regions surrounding the implant sites. Our findings indicated that the astrocytic response was significantly lower in regions near the Microfluidic-tube Integrated Optrodes (MIO) compared to the conventional Optrode Devices, specifically in ROI-1 (P < 0.05). In ROI-2 and ROI-3, no significant differences were observed between the two devices. These results suggest that MIO devices may induce less astrocytic activation in the immediate vicinity of the implant, potentially reducing glial scarring and promoting better tissue integration.

• Microglial activation : Microglial activation was evaluated using ED1 staining to assess the inflammatory response in regions surrounding the implant sites. Quantification was performed across three regions of interest (ROIs): ROI-1 (50–100 µm), ROI-2 (100–150 µm), and ROI-3 (150–200 µm) from the implantation site. The results, as shown in Fig. 9, indicate no significant differences in ED1 density between the Microfluidic-tube Integrated Optrode (MIO) device and the Optrode Device in any of the ROIs. These findings suggest that both devices elicit a comparable microglial response, indicating similar levels of inflammation at the implantation site.

Fig. 7.

Representative horizontal brain section showing NeuN staining near the device tract in the subthalamic nucleus (STN) region. (a) Representative image of NeuN staining in brain tissue implanted with the Optrode Device. (b) Representative image of NeuN staining in brain tissue implanted with the Microfluidic-tube Integrated Optrode (MIO) device. Both images are shown at 10$\times$ magnification with a scale bar of 200 µm. (c–e) Quantification of NeuN density in ROI-1 (50–100 µm), ROI-2 (100–150 µm), and ROI-3 (150–200 µm) from the implantation site for the Optrode Device and the MIO device, respectively. Neuronal density is expressed as the fraction of the stained area (above threshold) to the total area of the ROI. Statistical comparisons between the Optrode Device and the MIO device were conducted using T-tests (n = 5 per group), with significant differences indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001).

Fig. 9.

Representative horizontal brain section showing ED1 staining near the device tract in the subthalamic nucleus (STN) region. (a) Representative image of ED1 staining in brain tissue implanted with the conventional Optrode Device. (b) Representative image of ED1 staining in brain tissue implanted with the Microfluidic-tube Integrated Optrode (MIO) device. Both images are shown at 10$\times$ magnification with a scale bar of 200 µm. (c–e) Quantification of ED1 density in ROI-1 (50–100 µm), ROI-2 (100–150 µm), and ROI-3 (150–200 µm) from the implantation site for the Optrode Device and the MIO device, respectively. ED1 density is expressed as the fraction of the stained area (above threshold) to the total area of the ROI. Statistical comparisons between the Optrode Device and the MIO device were conducted using T-tests (n = 5 per group), with significant differences indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001).

Discussion

The comprehensive validation of the 3D-printed optogenetic neural probe integrated with a microfluidic tube demonstrates its suitability for in vivo applications, specifically for stimulating neuronal cells in the brain. The electrical, optical, and thermal properties of the device have been rigorously tested to ensure its functionality and safety. The electrical tests confirm that the device is capable of consistently lighting up the LED, which is crucial for delivering precise light pulses to the target neurons (Fig. 3). The stability of the electrical connections and the robustness of the device under operational conditions indicate a reliable performance during implantation.

Optical validation shows that the blue light emitted by the LEDs falls within the ChR2 opsin activation spectrum, which is critical for effective neuronal stimulation. The spectral alignment ensures that the light can efficiently activate the ChR2 opsins, leading to desired neural responses (Fig. 3c). This optical compatibility underscores the device’s potential for precise control over neuronal activity through optogenetic methods.

Thermal analysis reveals that the device operates within a safe temperature range that does not cause tissue damage (Fig. 4). Maintaining the temperature within physiological limits is essential to prevent thermal injury to the brain tissues during prolonged stimulation sessions. The device’s ability to manage heat dissipation effectively supports its long-term use in experimental and therapeutic settings.

Following these validations, the device was successfully implanted into the subthalamic nucleus (STN), a key structure within the basal ganglia involved in motor control. Behavioral studies conducted at the post-implantation provide compelling evidence of the MIO device’s efficacy in modulating neural activity. Upon light stimulation, mice implanted with the MIO device exhibited a significant increase in locomotor activity, as measured by distance traveled and velocity (Fig. 5). This response indicates successful activation of the target neuronal pathways in the subthalamic nucleus and underscores the device’s functionality in driving optogenetic stimulation.

Compared to the conventional Optrode Device, the MIO device consistently elicited greater motor activity during stimulation, with significant difference observed in distance traveled ($p<0.001$) and ($p<0.01$) across multiple weeks of the behavioral assesment. Analysis of movement over the seven-week period revealed a trend of increasing movement during the stimulation phase for both devices, with the MIO device showing consistently higher movement. Although the pre- and post-stimulation phases fluctuated and returned to baseline, the stimulation phase consistently exhibited greater movement over all the time from week 1 to week 7, suggesting sustained neuronal activation and engagement of neural circuits. This progressive increase in movement supports the idea that the MIO device facilitates effective and long-term optogenetic modulation.

The observed trend of increasing movement across the weeks, while not always reaching statistical significance in the pre- and post-stimulation phases, highlights the potential for the MIO device to support long-term optogenetic studies with minimal disruption to neuronal and glial environments. The significant increase in movement during the stimulation phase, as shown in Fig. 5, further confirms the sustained efficacy of the MIO device. This suggests that the MIO device not only effectively stimulates neural activity but also enhances behavioral responses, likely due to its more efficient integration and reduced tissue disruption.

The integration of a microfluidic tube for opsin delivery further enhances the device’s versatility, allowing for precise, localized delivery of opsins and other therapeutic agents directly to neural tissues. These features improve the specificity and efficacy of optogenetic interventions, facilitating more effective and stable neuronal activation during stimulation. Furthermore, the validated electrical, optical, and thermal properties of the MIO device ensure safe and effective operation, minimizing adverse effects on surrounding brain tissues. The observed behavioral outcomes in mice highlight its potential for practical applications in both research and clinical settings, representing a significant advancement in neural interface technologies.

Immunohistochemistry (IHC) representative images (Fig. 7) show histological images of brain sections stained for NeuN, used to assess neuronal density following implantation. These images compare the effects of stab injuries generated using the Optrode Device and the MIO device. NeuN staining revealed a higher neuronal density in regions near the MIO device compared to the Optrode Device, particularly in ROI-1 (50–100 µm) and ROI-2 (100–150 µm). This suggests reduced neuronal disruption in the immediate vicinity of the MIO implant. Panels (c–e) present bar graphs comparing the neuronal density between the two devices in three regions of interest (ROIs). The data show that the MIO device preserved significantly more neurons in ROI-1,  ROI-2, and ROI-3 ($P<0.05$ , $P<0.01$, and P<0.05 respectively), indicating reduced neuronal disruption near the implant site. These findings underscore the advantages of the MIO device’s single-surgery approach, which minimizes tissue damage and enhances neuronal preservation at the implant site. This capability could improve the device’s long-term stability and effectiveness in neural modulation

Although the Optrode Device has a smaller cross-sectional footprint compared to the MIO device–owing to the absence of an integrated microfluidic tube–the greater neuronal disruption observed near the control device can be attributed to the double-surgery protocol required for its use. In the control group, two distinct insertions were necessary: the first for viral vector delivery and the second for optrode implantation one to two weeks later. This sequential procedure likely compounded mechanical trauma, exacerbated local inflammation, and disrupted neuronal repair processes between surgeries. Additionally, repeated microglial and astrocytic activation cycles and overlapping inflammatory zones may have further intensified tissue damage in the control group.

In contrast, the MIO device’s integrated design allowed for simultaneous viral injection and optrode implantation through a single insertion track. This single-surgery approach reduced cumulative mechanical insult and minimized repeated immune activation. Moreover, immediate device placement following viral delivery helped avoid the secondary inflammation associated with delayed implantation. These procedural advantages likely outweigh the potential drawbacks of the MIO device’s larger size, resulting in the improved neuronal preservation observed in NeuN-stained sections.

Figure 8 shows representative histological images of brain sections stained for GFAP to assess astrocytic activation following implantation. The images compare the effects of stab injuries generated using the Optrode Device and the Microfluidic-tube Integrated Optrode (MIO) device. These images highlight the distribution and density of GFAP-positive astrocytes in the vicinity of the implantation site. Quantification of GFAP staining (Fig. 8c–e) reveals a significantly lower astrocytic response near the MIO device, particularly in ROI-1 (50–100 µm), suggesting that the MIO device induces less astrocytic activation in the immediate vicinity of the implant. The reduced astrocytic response might be due to the MIO device causing less mechanical disruption or irritation, which lowers the activation threshold for astrocytes. In contrast, the Optrode Device elicited a stronger astrocytic response, likely due to greater tissue disruption during implantation. Interestingly, astrocytes near the MIO device may remain less activated due to the single-surgery approach, which minimizes tissue damage and promotes better tissue integration.

Fig. 8.

Representative horizontal brain section showing GFAP staining near the device tract in the subthalamic nucleus (STN) region. (a) Representative image of GFAP staining in brain tissue implanted with the conventional Optrode Device. (b) Representative image of GFAP staining in brain tissue implanted with the Microfluidic-tube Integrated Optrode (MIO) device. Both images are shown at 10$\times$ magnification with a scale bar of 200 µm. (c–e) Quantification of GFAP density in ROI-1 (50–100 µm), ROI-2 (100–150 µm), and ROI-3 (150–200 µm) from the implantation site for the Optrode Device and the MIO device, respectively. GFAP density is expressed as the fraction of the stained area (above threshold) to the total area of the ROI. Statistical comparisons between the Optrode Device and the MIO device were conducted using T-tests (n = 5 per group), with significant differences indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001).

Figure 9 shows representative histological images of brain sections stained for ED1 to assess microglial and macrophage activation following the implantation. The images compare the effects of stab injuries generated using the Optrode Device and the MIO device. These images highlight the distribution and density of ED1-positive microglia and macrophages in the vicinity of the implantation site. Quantification of ED1 staining (Fig. 9c–e) shows that both devices induced similar levels of microglial activation across the ROIs. This result suggests that microglia respond more uniformly to the presence of a foreign object, leading to similar activation levels regardless of the optrode type. The immune response is likely more generalized, focusing on the body’s reaction to the implant rather than the specific design of the device.

Overall, the MIO device demonstrates improved neuronal preservation and reduced astrocytic activation, which are essential for maintaining stable and functional neural interfaces in long-term optogenetic studies. The absence of significant differences in microglial response suggests that additional strategies may be needed to further mitigate neuroinflammatory processes at the implant site. Furthermore, the practical benefits of the MIO device’s single-surgery approach–such as simplified surgical procedures, reduced risk, and shorter recovery time–highlight its translational potential.

While our results demonstrate the feasibility and advantages of the 3D-printed optogenetic neural probe with integrated microfluidic delivery, several limitations should be acknowledged. First, the in vivo validation was conducted exclusively in the subthalamic nucleus (STN) of mice. Thus, the broader applicability of the device to other brain regions or animal models remains to be established. Second, the sample size in our behavioral and histological experiments was limited, potentially affecting statistical power and generalizability.

Although the integrated design enabled single-step viral delivery and optrode implantation, we did not quantify the spatial extent or efficiency of viral transduction. Variability in opsin expression across animals could influence both behavioral and histological outcomes. Additionally, our assessment of tissue response and biocompatibility was limited to a relatively short post-implantation period. Long-term studies are necessary to evaluate chronic device stability, gliosis formation, and sustained immune response.

From a design perspective, while the physical dimensions of the MIO device were optimized for functionality, they are larger than those of conventional optrodes, which may restrict use in smaller or more delicate brain regions. Thermal validation was performed in agarose gel as a brain tissue analog. However, in vivo thermal dynamics–modulated by blood flow and tissue heterogeneity–may differ, necessitating future thermal monitoring in live tissue to ensure safe stimulation protocols.

Technically, although 3D printing enables rapid prototyping and design flexibility, variability in device quality between fabrication batches remains a concern. Further work is required to establish reproducible and standardized manufacturing protocols. Moreover, the integration of a microfluidic channel, while beneficial for single-surgery procedures, introduces potential risks such as fluid leakage or channel blockage. The current design also lacks real-time monitoring capabilities for injection dynamics or tissue response.

Addressing these limitations in future works will be essential to optimize device performance, validate long-term safety, and expand the utility of the MIO system across a wider range of experimental and clinical neuroscience applications.

Future studies should investigate the long-term stability and tissue compatibility of the MIO device by extending the observation period and incorporating in vivo imaging to monitor chronic responses. Additionally, optimization of the device’s material properties to further reduce immune activation, along with testing across various brain regions and behavioral paradigms, will help establish the full scope of its applicability. The MIO device represents a meaningful advancement in optogenetic interface technology by integrating optical and fluidic functions into a single platform that enhances precision, biocompatibility, and usability for both research and clinical applications.

Methods

Device design and fabrication process

The fabrication and assembly process of the device is depicted in Fig. 10. This process is divided into three stages (Fig. 10 i,ii, and iii). Initially, the resin (IP-Q, Nanoscribe GmbH, Germany) is prepared on a silicon substrate (i-(a)), and the optrode substrate is 3D printed using two-photon polymerization technology (Photonic Professional GT, Nanoscribe GmbH, Germany) (i-(b)). The printing duration varies based on the device length, ranging from about 20 to 40 minutes.

Fig. 10.

Fabrication and assembly process of the device, divided into three stages: (i) 3D printing process (a–d), (ii) device fabrication and assembly of conductive components and microfluidics (e–k), and (iii) assembly and packaging process (l–s).

For a 7 mm device, it takes approximately 30 minutes. After printing, the structure is immersed in a developer (Propylene glycol monomethyl ether acetate, Sigma-Aldrich, US) for 15 minutes, followed by a one-minute dip in isopropyl alcohol (IPA).

The structure is then UV cured (Form Cure, Formlabs, US) for 5 minutes to enhance stiffness (i-(c),(d)). Next, conductive wires (Bare platinum wire, A-M Systems, US) with a 25-µm diameter are positioned in the substrate’s concave groove (ii-(f)). This wire insertion process takes about 10 minutes using ultra-fine tweezers. Solder balls (TS391LT10, Chipquik, Canada) are then placed in the 3D-printed concave spaces on the pad areas to connect the wires and the LED pads (ii-(g)). The LED emitting blue light (C460TR2227, Cree, US) is positioned on the substrate, and a hot-air gun melts the solder balls to establish electrical connections (ii-(h)). The LEDs measure 220$\times$270 µm$_{}^2$.

A superglue (Gorilla Super Glue, The Gorilla Company, US) secures the wires and the microfluidic tube (150 µm) placed in the groove (ii-(i–k)). Prior to placing the microfluidic tube onto the device, a gold wire was inserted fully through the microfluidic tube (Bare gold wire, A-M Systems, US), extending from the proximal to the distal end. The microfluidic tube, with the gold wire inside, was then positioned onto the device. The optrode device is then removed from the silicon wafer.

To connect the fabricated optrode to a connector, the wires must be attached to electrical connector pins within a customized housing. Figure 10(iii) illustrates the housing and optrode assembly process. Optrode is placed onto the housing as shown in iii-(l). Two connector pins are inserted into the housing body, and a pin holder secures them (iii-(m)). The wires from the optrode are bent and placed on the connector pins (iii-(m)). All housing components are 3D printed (S240, Boston Micro Fabrication, US). Solder balls are used to solder the wires to the pins, and a hot-air gun melts the solder balls (iii-(o)). A 3D-printed cover part encloses the pin area and is secured using an adhesive (Loctite 401, Henkel Adhesive Technologies, Germany)(iii-(p)). The entire device is then closed with a 3D printed cover (iii-(q)). Finally, the entire device is conformally coated with parylene C of thickness 8 µm (iii-(r)) using a chemical vapor deposition system (PDS2010 Labcoater 2, Specialty Coating Systems Inc, US). This parylene C coating ensures the device is biocompatible, making it safe for acute implantation testing. After the coating process was completed, the gold wire was carefully withdrawn from the distal end of the microfluidic tube (iii-(s)), thereby re-establishing a clear and functional microfluidic channel for subsequent viral vector injection.

Animals and surgery

• Animals and ethical approval: All experimental procedures were approved by the New York University Abu Dhabi Animal Care and Use Committee (IACUC protocol number 22-0002), conducted in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. All methods are reported in accordance with ARRIVE guidelines (https://arriveguidelines.org) for the reporting of animal experiments. The study utilized 8–10 weeks old C57Bl/6J male mice (Jackson Laboratories, ME, USA). Animals were housed in standard cages with 12-hour light/dark cycle and ad libitum access to food and water.

• Surgical procedures: Mice were deeply anesthetized using intraperitoneal administration of ketamine: xylazine (100 :10 mg/kg body weight) to ensure loss of consciousness. The animal was carefully placed on the stereotactic device (David Kopf Instruments) and its head was fixed on the frame for precise measurements during the procedure.

Viral injection and optrode implantation

• Methods for the conventional optrode device: Viral Injection: A midline incision was made to expose the skull and visualization of the bregma axis at the brain surface. To target the STN region, a single burr hole was made using the predetermined coordinates: Anteroposterior (AP): − 2.18, Mediolateral (ML): ±1.25, and Dorsoventral (DV): 4.65 (Allen Mouse Brain Atlas). A total of 400nl of viral (pAAV-CaMKIIa-hChR2(H134R)- mCherry_Addgene2697) was delivered at the STN using a microneedle at a rate of 100 nl per minute. To prevent leakage and injury to the animal, the burr hole was sealed closed using dental cement (Eschem dental resin, Spident Co. LTD., South Korea) and a skull screw (PlasticsOne, Roanoke, USA).

  Optrode Implantation: After viral expression was confirmed, a second surgery was performed by reopening the previously drilled Burr hole. A traditional optrode was implanted at the same coordinates as used for viral injection. The optrode was secured in place using dental cement (Eschem dental resin, Spident Co. LTD., South Korea) and a skull screw (PlasticsOne, Roanoke, USA) (Fig. 11a).

• Methods for microfluidic-tube integrated optrode (MIO) device: Single Surgery for Viral Injection and Optrode Implantation: A midline incision was made to expose the skull, and a Burr hole was drilled at coordinates AP: − 2.18, L: ± 1.25, and DV: 4.65. Following the surgical implantation (Fig. 11a), the viral vector (pAAV-CaMKIIa-hChR2(H134R)- mCherry_Addgene2697) was pre-loaded into a syringe and connected to the external end of the implanted microfluidic tube under sterile conditions to avoid contamination (Fig. 11b). The syringe’s tip diameter was designed to fit precisely into the inner diameter of the MIO device’s tube. To ensure a secure connection and prevent leakage during infusion, medical grade superglue was applied around the junction after inserting the syringe into the microfluidic tube. The viral solution was then slowly injected through the microfluidic channel in increments of 100 nl, with a controlled flow rate achieved by manually applying gentle and consistent pressure on the syringe plunger, allowing preise deivery to the target brain region via the device’s fluidic outlet, which had already been positioned during implantation. After each injection, a waiting period of 1 min was observed to allow the solution to diffuse into the target region, and this process was repeated until a total volume of 400 nl was delivered into the subthalamic nucleus (STN) region. Once the injection was completed, the syringe was carefully removed, and the MIO device was secured in place with dental cement (Eschem dental resin, Spident Co. LTD., South Korea) and a skull screw (PlasticsOne, Roanoke, USA), completing the single-surgery procedure without the need for a second surgery.Post-surgery, mice were closely monitored during recovery for any signs of distress or infection. All the animals were allowed to recover in their home cages.

• Perfusion and tissue preparation: On the 8th week post-surgery, mice were deeply anesthetized with 7% chloralhydrate solution (300 µL, intraperitoneally). Lack of response to pain was established using the toe-pinch method, and transcardial perfusion was carried on with cold 20–25 mL phosphate-buffered saline (PBS) to drain maximum blood from the animal. This was followed by immediate circulation of 4% ice-cold paraformaldehyde (PFA, 20mL, pH 7.4) to fix the tissues. Brain tissue was carefully harvested and transferred to a cryoprotectant solution of 30% sucrose for 22–30 hours at 4 °C.

  Following perfusion and fixation, horizontal brain tissue sections were obtained (50 µm thick, n = 16/brain tissue) using a cryostat (Leica CM1950, Leica Biosystems, USA), mounted onto Super Frost Plus glass slides (Menzel-Gläser, Germany), and stored at − 20 °C for imaging and subsequent analyses. All immunohistochemical analyses and imaging were performed on horizontal sections.

  After cryoprotection, the brains were stored in 30$\%$ sucrose in phosphate-buffered saline (PBS) at 4$_{}^{\circ }$C. The brains were kept in the solution until they sank, indicating complete saturation.Horizontal brain tissue sections were obtained (50 µm thick, n =16/brain tissue) using a cryostat (Leica CM1950, Leica Biosystems, USA), mounted onto Super Frost Plus glass slides (Menzel-Gläser, Germany), and stored at − 20 °C for imaging and subsequent analyses.

• Immunohistochemistry: Immunohistological staining was performed on sectioned brain tissue for the assessment of neuronal and immunological response. Frozen sections were thawed to near room temperature (RT) and transferred to a 24-well plate with each well filled with PBS. To wash off the cryomedia and rehydrate the sections, tissues were washed with PBS on a shaker for 10 minutes. At the end of each wash (3 washes), PBS was pipetted out and refilled in each well carefully. To prevent nonspecific binding, the sections were incubated in a solution of 5$\%$ donkey serum (Abcam), and 0.25$\%$ Triton X-100 (Sigma, T8787) in PBS. This was followed by overnight incubation in primary antibody: i) mouse anti-CD68/ED1 (a marker for activated microglia, 1:250, Cat. Nr. MCA341R, Biorad, USA), ii) rabbit anti-glial fibrillary acidic protein (GFAP, an astrocytic cytoskeletal protein; 1:5000, Cat. Nr. Z0334, Dako, Denmark), and iii) rabbit anti-neuronal nuclei (NeuN, a nuclear marker of neuronal activation; 1:500, Cat. Nr. ab104225, Abcam, USA).

  The following day, the sections were washed with PBS to remove unbound antibodies (3 $\times$ 10 min). Next, the sections were incubated for 2 hours (RT) in a solution of DAPI (4′, 6-diamidino-2-phenylindole;counterstain at 1:1000, Invitrogen, USA) and secondary antibodies: i) goat anti-rabbit Alexa 594 (1:500, Cat Nr. A11012, Invitrogen, USA) and ii) goat anti-mouse Alexa 488 (1:500, Cat Nr. A11001 Invitrogen, USA). After a final round of PBS wash (3 $\times$ 10 min), the antibody-tagged sections were mounted on glass slides (VWR, USA) using Vectashield, an antifade mounting media (Vector Laboratories, USA).

• Image acquisition and analysis:

  Mouse brains were sectioned into 50-µm-thick horizontal slices using a cryostat (Leica CM1950, Leica Biosystems, USA), covering a range from 2650 µm to 6150 µm (centered at 4.65 ± 2 mm, − 1.5 mm from the bregma). This range corresponds to 70 sections (3500 µm $÷$ 50 µm), which were systematically divided into 12 groups. For NeuN staining, 12 sections within the range of 2750–5750 µm and 3050–6050 µm were analyzed. For GFAP and ED1 staining, 12 sections within the range of 2800–5800 µm and 3100–6100 µm were analyzed.

  The regions of interest (ROIs) were defined based on their distance from the device implantation site: ROI 1 (50–100 µm), ROI 2 (100–150 µm), ROI 3 (150–200 µm), and ROI 4 (300–350 µm) as the reference ROI. A schematic representation of the implant site and concentric ROIs is shown in Fig. 12. Immunohistochemistry-stained sections were imaged using a widefield microscope (Leica DMI6000, Leica Microsystems, Germany) at 10$\times$ magnification. To ensure consistency, uniform gain, contrast, and exposure settings were applied across all images for each marker (NeuN, GFAP, ED1). ImageJ software was utilized for processing and quantification, with the binary threshold function used to identify positive staining.

  For each animal (n = 5 per group), fluorescence images of the 12 sections were analyzed. The raw cell count in each ROI was divided by the corresponding area of the ROI to calculate the density (count/µm$_{}^2$) for NeuN, GFAP, and ED1. The density values for ROI 1, ROI 2, and ROI 3 were normalized by dividing each value by the density of ROI 4 (reference region):

  $$\text{Normalized Density for ROI-X}=\frac{\text{Density of ROI-X (count/area)}}{\text{Density of ROI 4 (count/area)}}$$

  This normalization process accounted for baseline variations and enabled direct comparisons between ROIs and experimental groups. The normalized density values were averaged across animals within each group to evaluate neuronal density (NeuN staining) and glial responses (GFAP and ED1 staining) surrounding the implantation sites of the Optrode and MIO devices.

• Statistical analysis:

  The distance traveled by mice in the arena was measured across the pre-stimulation, stimulation, and post-stimulation phases over a 7-week period for both the Optrode device and the microfluidic tube integrated optrode (MIO) device. To assess locomotor activity during the behavioral tests, the total distance traveled (in meters) was calculated using video-based tracking data from the entire arena, acquired with TopScan software (CleverSys, Inc., Reston, VA, USA). Velocity was computed by dividing the total distance traveled by the total time the animal was actively moving, excluding any periods when the animal was stationary. For each group (n = 5 animals per group), the distance traveled during each phase was recorded weekly. The change in distance traveled per week was analyzed and compared between groups.

  The mean distance traveled during each phase, with error bars representing the standard error of the mean (SEM), was plotted for each group. These measurements were assessed across the pre, during, and post phases. Similarly, mean velocity was calculated for each phase and plotted with SEM for each group.

  Additionally, the overall average distance traveled and mean velocity for each group (Optrode and MIO) across all three phases (pre, during, and post) were calculated over the 7-week period. These averages were compared between the groups using t-tests, with each bar representing the average value and error bars indicating SEM.

  For NeuN, GFAP, and ED1 staining, the quantification of activated areas in the regions of interest (ROIs) was performed. The ROIs were defined as follows: ROI 1 (50–100 µm), ROI 2 (100–150 µm), and ROI 3 (150–200 µm) surrounding the implantation site. The neuronal density was calculated as the fraction of the stained area (above threshold) relative to the total area of each respective ROI.

  Statistical comparisons between the Optrode device and the MIO device were conducted using t-tests (n = 5 per group) for each marker (NeuN, GFAP, and ED1) within each ROI. Significant differences between the groups are indicated by asterisks (*$p<0.05$, **$p<0.01$, ***$p<0.001$). The results of these t-tests were presented in bar graphs showing the mean density for each ROI, with error bars representing the SEM.

  All analyses were conducted using GraphPad Prism 8.1.2 software (GraphPad Software Inc., USA).

Fig. 11.

(a) Image showing the mouse positioned on the surgery table, with the skull exposed and a skull screw placed for stabilization, preparing it for surgery. (b) Image showing the Microfluidic Tube Integrated Optrode (MIO) device implanted on the skull. The inset within (b) provides a schematic of the MIO device, illustrating the MIO device inserted into the brain and secured with dental cement. The connection of the syringe to the microfluidic tube, enabling the injection of the viral vector, is marked with a dashed circular outline in (b). This connection facilitates the virus injection into the brain, ensuring precise delivery of the viral vector into the targeted brain region.

Fig. 12.

Schematic of the implant site highlighting concentric circular regions of interest (ROIs) used for histological quantification: Implant site (grey), 50–100 µm (light blue), 100–150 µm (green), 150–200 µm (yellow), and 300–350 µm (red). Representative dorsal view of a mouse brain indicating horizontal sectioning planes from dorsal to ventral for systematic histological examination, adapted from the Mouse Brain Atlas: C57BL/6J Horizontal³⁸.

Author contributions

R.S. and S.H. conceptualized the proposed device. R.S. fabricated the device and conducted the property measurements with assistance from K.O. R.S. and M.M. wrote the manuscript. M.M., R.S., and M.S. conducted the in-vivo experiments and IHC. R.S. analyzed the results and performed statistical evaluations. D.C. provided guidance on the in-vivo experiments, and S.H. supervised the overall project and provided revisions to the manuscript. All authors reviewed and approved the final manuscript.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare that there are no conflicts of interest in this article.