Introduction

Implantable neural probes sense neural signals within the brain, holding immense potential in electrophysiological research, facilitation of motor restoration, and regulation of neurological disorders1,2,3. Commercial invasive probes, typically constructed with rigid metal or silicon materials, trigger an intracerebral immune response because of the severe mechanical mismatch between these rigid probes and the brain tissue, thus hampering the long-term stability of the signal acquirements4. Notably, the non-compliance of the probe probably causes intraoperative bleeding and postoperative sustained injuries on fragile brain tissue, resulting in leaky blood-brain interface and even trauma due to the relative micromotion between probe and tissue5,6.

Previous studies have developed various strategies to alleviate postoperative foreign body responses. These strategies include reducing probe dimensions, using removable or biodegradable coatings, and reformulating responsive substrates or matrices to improve biocompatibility7,8. Advanced research has demonstrated that micromachining processes can effectively reduce the bending stiffness of microscale metal probes to accommodate brain micromotion5. However, it is imperative to recognize that diminishing the scale of the probes entails a trade-off between increased pliability (the reciprocal of bending stiffness associated with diameter and length) and accuracy in penetration5,9. Although the penetration can be assisted by utilizing shuttle instruments and encapsulating biodegradable materials, the remaining chemo-mechanical incompatibility between the re-exposed rigid probe and tissue leads to device failure caused by glial scar proliferation at the peri-probe region10,11,12,13,14. Responsive hydrogel has been developed to relieve the mismatch between the device and tissue, such as skin, muscle, and nerve fiber15. Integrated responsive hydrogel substrates/matrices have fueled the development of probes with intrinsic adaptability to the physiologic environment16,17, combined with their superior chemical-mechanical similarity to tissue18,19,20, thus alleviating concerns about biocompatibility and surgical complexity. However, the rigid-soft interface between the rigid conductive materials with Young’s modulus exceeding 100 GPa (including silicon, metals, carbon fibers, or nanotubes) and the hydrogel (hundred kPa) is susceptible to rupture due to asynchronous swelling rates21. Though chemically orthogonal waterproofing has been developed for micropatterning fully encapsulated and elastic electrodes, the swelling of stretchable polymers hinders the three-dimensional fabrication due to their porous nature22.

Of particular importance, a clinical feasibility concern arises from the high risk of bleeding during the direct insertion of the probe, considering that the aforementioned probes must maintain rigid throughout penetration. The reported incidence rate of bleeding during the implantation of invasive neural probes is 2.86%, with a corresponding mortality rate of 0.02% due to bleeding5. Bleeding-induced tissue compression and increased intracranial pressure trigger neuronal somatic membrane break that could precipitate ultimate neuronal death23. The ensuing dysfunction in blood supply is one of the causative factors in neurodegenerative diseases and even affects overall normal brain function, leading to stroke-like effects24. Despite advanced precision equipment that could reduce the risk of vascular damage with the aid of optical guidance systems or highly sensitive mechanical sensors, their reliance on sophisticated design, intricate preparation, and requirements for specialization pose challenges for implementation25,26,27. Scientists are beginning to consider the development of probes that could become compliant during implantation. However, insufficient bending stiffness makes such probes highly susceptible to deformation under thrust rather than being implanted into the target brain region as expected. We reasoned that these challenges could optionally be met by the rational design of polymer network alignment.

Here, we present a uniaxial extending neural probe (UENP) for sensing within the brain while avoiding intraoperative bleeding. The probe trunk comprises self-assembled conductive polymers (PEDOT) in a UV-illumination pre-oriented polyethylene glycol diacrylate (PEG-DA) scaffold. The assemblies (positively charged PEDOT chains) pack in a parallel fashion along the negatively charged polystyrene sulfonic acid (PSS) scaffold, generating strong interchain interaction. The subsequent dehydration rises to decrease chain distances between PEDOT and PEG-DA, significantly enhancing the interchain interaction. As a result, the restriction of cross-sectional expansion in the global scale between the microstructural cordlike-backbones is strengthened (the effect of in-plane deformation inhibition), enabling our probes to remain extended along the principal axis of implantation (self-calibrated implant). Meanwhile, the UENP can quickly soften (in tens of seconds), effectively avoiding angiorrhexis during insertion. On the contrary, the rigid probe tends to puncture the blood vessels (Fig. 1). UENP colliding with and slipping over a microvessel and postcapillary venules in vivo was recorded by a two-photon microscope. No leakage of vascular contents after retraction occurs, indicating the intactness of the vessel throughout the implantation process. In the postoperative period, our UENP caused negligible stress in the surrounding brain tissues even when micromotion happened, as confirmed by finite element analysis. This is consistent with the minimal foreign-body reaction revealed by the histopathological analyses. In anesthetized rats, our UENP offered reliable electroencephalogram monitoring with a signal-noise ratio (SNR) comparable to commercial electrodes due to the ideal impedance (~8 kΩ before and ~10 kΩ after implantation at 1 kHz, respectively). We further verified its clinical potential utility by evaluating the imageological compatibility of UENP using a micro-computed tomography scan (micro-CT) and magnetic resonance imaging (MRI). We envision that this strategy of cerebrospinal fluid (CSF)-responsive uniaxial extending will enable autonomous navigation and significantly reduce surgical risks.

Fig. 1: Conceptual illustration of the key features of the UENP.
figure 1

Before insertion, the probe is rigid and straight, allowing facile brain tissue penetration. During the operation, the inserted part of the probe becomes soft once inside biological tissue due to the absorption of cerebrospinal fluid, thus avoiding the rupture of blood vessels during insertion. In contrast, rigid probes would puncture blood vessels directly and cause bleeding. In the long-term application, the UENP exhibits high compliance, low stress, and minimal injury to tissue during brain motion.

Results

Design, fabrication, and characterization of UENPs

Our strategy to design the probe with CSF-responsive uniaxial extending along the principal axis is schematically illustrated in Fig. 2a. The strategy relies on a global macromolecule alignment, which includes (i) forming a UV-illumination pre-oriented polyethylene glycol diacrylate (PEG-DA)/poly (styrene sulfonate) (PSS) scaffold due to the hydrogen-bonding interaction between PEG-DA and PSS (which seeds the EDOT growth) and (ii) orienting the in-situ polymerized PEDOT chains along the scaffold resulting from the electrostatic effect between the PEDOT and PSS28,29, (iii) decreasing the distance between adjacent polymer chains (interchain spacing) after dehydration. Thus, the enhanced near-range interactions including interchain interaction between PEG-DA and PEDOT30, along with the intrachain electrostatic interaction of PEDOT and PSS, synergistically restrain the cross-sectional expansion of the probe. Simultaneously, the extension along the principal axis upon implantation is free (Fig. 2b). The fabrication details are described in the methods section and shown in Supplementary Fig. 1a. Briefly, slender PEG-DA scaffolds were cured under UV light with a customized mask plate (Supplementary Fig. 1b). EDOT was infiltrated into the scaffold with the assistance of ethanol and then polymerized under the catalytic influence of sulfuric acid and iron p-toluenesulfonate. The sulfur concentration showed a notable increase, providing evidence for the in situ polymerization of PEDOT:PSS (Supplementary Table 1).

Fig. 2: Design and operating principles of UENPs.
figure 2

a The fabrication schema of UENP. The EDOT is polymerized in the PEG-DA scaffold and cured under UV. The PEDOT chains are firmly coiled on the scaffold due to the electrostatic interaction between them and the PSS chains. After dehydration, the interchain interaction is increased between PEDOT and PEG-DA. b The cross-sectional in-plane deformation inhibition effect during rehydration by CSF, while the relaxation of the chains along the principal axis is free. c SEM images of hydrated UENP show a porous surface with aligned fibrillar bridges. The direction of illumination is vertical. The scale bar is 250 μm in the left image and 90 μm in the right image. d SEM images show the dehydrated double network hydrogel utilized in-situ polymerization of PEDOT has regular cords (left), whereas the hydrogels directly doped with PEDOT:PSS present a randomly dispersed particle structure (right). Scale bar, 50 μm. e The statistics proved that the cords are aligned along the direction of illumination, while the particles are not. f The Raman spectra of the probe before dehydration, after dehydration, rehydration, and the final UENP which is treated with DMSO. g The vertical elongation of the UENP is much greater than its lateral elongation following DMSO treatment (n = 6). h Anisotropic elongation of the UENP was recorded by a high-speed camera (in a phantom brain, 0.6% agarose gel). Scale bar, 500 μm.

To investigate whether UV light induces the orientational scaffold, we altered the direction of illumination and observed a corresponding change in the orientational structure within the PEG-DA/PSS hydrogel (Supplementary Fig. 2a, b). In contrast, the orientational structure vanished when we substituted light with heating as the trigger for polymerization (Supplementary Fig. 2c). This phenomenon can be attributed to a diffusive mass flow resulting from a concentration gradient induced by the UV intensity gradient, which serves as the primary driving force for molecules within our aligned hydrogel scaffold31. This gradient, absent in thermo-curing hydrogels, serves as the primary force propelling molecules within our aligned hydrogel scaffold. As expected, insufficient mass flow through diffusion caused by low illumination intensities leads to incomplete and disordered polymerization (Supplementary Fig. 2d). The porous structure with aligned fibrillar bridges within hydrated UENP remains after in situ polymerized PEDOT:PSS (Fig. 2c), while the internal structure of the dehydrated probe displays a cordlike morphology of its backbones (Fig. 2d). Statistical analysis of the distribution of backbone angles indicates that the global alignment of polymer chains remains oriented in the direction of UV-illumination after dehydration. By contrast, the microstructure of the scaffold formed by directly doping PEDOT:PSS in PEG-DA exhibits unordered blocks, confirmed by the angle analysis of the block edges (Fig. 2e). The reason for this phenomenon lies in the fact that an abundance of PEG-DA hinders the dissociation of PSSH into PSS anions and protons. A less dissociated PSS chain had much weaker Coulombic interactions with PEDOT. Therefore, the isolating PSS shell around PEDOT grains is reduced, forming the water-insoluble physically crosslinked PEDOT network even before UV crosslinking30. Furthermore, the hindrance of light propagation by PEDOT:PSS results in the insufficient length of the probe32.

The conformational characteristics of PEDOT in the dehydrated and rehydrated state of UENP were further investigated by Raman spectrum, aiming to reveal the underlying mechanism behind the inhibition of lateral expansion in UENPs (Fig. 2f). The band in 1439 cm−1 is assigned to the Cα = Cβ symmetrical stretching vibration of thiophene rings in PEDOT and the band from 1496 cm−1 to 1571 cm−1 is assigned to the Cα = Cβ asymmetrical stretching33. After dehydration of UENPs, the reduced distance between PEG-DA and PEDOT enhances their interaction, leading to phase separation between PEDOT and PSS. This is evident in the Raman spectrum by a notable decrease in the peak intensity of Cα = Cβ asymmetrical stretching and a broadening of the symmetrical stretching peak30. It is noteworthy that the phase separation between PEDOT and PSS remains after rehydration, demonstrated by a relative decrease in peak intensities of Cα-Cα’ inter-ring and Cβ-Cβ stretching34. When water enters the network, the enhanced interchain interactions (between PEG-DA and PEDOT) and electrostatic forces (between PEDOT and PSS) compete with the lateral expansion of UENPs caused by water entering the hydrophilic scaffold. Additionally, dimethyl sulfoxide (DMSO) treatment can remove excess PSS35. A significant decrease in the intensity of PSS characteristic peaks, along with a redshift and narrowing of the Cα = Cβ symmetrical stretching peak in the Raman spectrum was observed. These changes suggest a further transformation of PEDOT chains into rigid quinoid structures36,37. It was confirmed by the increased vertical and decreased lateral elongation of UENPs after DMSO treatment (Fig. 2g).

Since the cross-sectional expansion of the UENP is restricted owing to the strong interchain interactions between the adjacent, side-by-side aligned backbones, our UENPs pose a special self-implantation behavior. We verified this character by inserting UENPs into the brain simulated by 0.6% agarose gel process and recorded the anisotropic elongation by a high-speed camera (Fig. 2h, and Supplementary Movie 1). It is worth noting that slow insertion velocity is required after penetrating the brain surface to improve acute recording qualities38. The average velocity of UENP upon self-implantation is about 11.67 μm s−1 which is much lower than the velocity widely used in standard implantation protocol (200 μm s−1), further confirming the reduced surgical risk of our UENPs.

Real-time characterization of the effects on the vasculature during implantation

The visualization of the internal structure within the superficial cerebral cortex in the living rat by a two-photon microscope enables us to monitor the condition of blood vessels along the track of probe implantation. Benefiting from the high resolution and layer-scanning of two-photon microscopy, we can quickly identify vessels that are in the focus plane of the probe. To distinguish blood vessels from probes, we extensively labeled the rat vascular endothelium with red fluorescence and labeled UENP with green. Given the narrow maneuverable space under the microscope, we inserted the probe obliquely into the cerebral cortex below the cover glass slide (Fig. 3a). Tracing the UENP’s tip revealed the collision and friction between the microvasculature and the UENP (Supplementary Movie 2). The probe was gently inserted into the brain tissue and unsurprisingly collided with a microvessel (approximately 90 μm) which smoothly shifted aside from the probe. The microvessel then was slightly dragged by the probe and swiftly stopped (Fig. 3b i). The vessel synchronously repositioned as the probe was drawn back to its initial position (Fig. 3bii, c). Significantly, the displacement of the microvessel is sufferable and safe during insertion and retraction of UENP, rather than being forcefully pushed or dragged away by the implant till the vascular rupture occurs39. We attribute this to the decreased friction between the vessel and UENP, primarily influenced by the minimized additional pressure exerted on the vessel due to the inhibited lateral expansion of UENP. For the isotropic extended probes, the compression of the probe on the brain tissue increases due to lateral expansion. According to the formula of sliding friction, f = μN, the friction (f) between the probe and brain tissue increases with the increased pressure (N), where μ represents the dynamic friction factor. Consequently, the isotropic probe poses a greater risk of vascular rupture compared to our UENPs.

Fig. 3: Spatial and temporal orchestration of vasculature during UENP implantation.
figure 3

a A photograph of a mouse implanted with the UENP before two-photon imaging. Scale bar: 1 cm. b The movement of a microvessel during the insertion (i) and retraction (ii) of UENP. The initial site before insertion (red dash line) and final site after retraction (blue dash line) in c demonstrate the minimal displacement of this microvessel. Scale bar: 100 μm. d The displacement magnitude evolution from distal to proximal postcapillary venule as UENP insertion and retraction. The results of specific values are shown in e. f The successive footage of UENP’s self-calibrated insertion into a phantom brain which contains the isolated microvessel. The frame rate of the images in the red box is 1 s. Scale bar, 100 mm. The vertical and lateral length changes in UENPs from a batch are shown in g, n = 6.

In addition, we noted several postcapillary venules (diameters less than 20 μm) with vascular wall structures similar to those of capillaries. The displacement of the clearest postcapillary venule in view (identified in Supplementary Fig. 3a, b, upper) was investigated by particle image velocimetry (PIV) analysis. The results revealed that the proximal vessel (relative to the microvessel) slowly moved along the insertion direction under the pushing of UENP (Fig. 3d, upper). As insertion, the friction between the postcapillary venule and UENP was gradually transmitted down to the distal vessel. As the probe was retracted, the vessels gradually repositioned as expected (Fig. 3d lower). The statistical analysis according to the thermograms showed that the maximum displacement of the proximal vessel was 8 μm, while the distal one was 5 μm (Fig. 3e). The less displacement of the distal is due to the unaffected intercellular tight junctions between the distal neurovascular unit40, preventing the overall displacement of the vasculature. Even though these postcapillary venules are extremely fragile, there was no one ruptured. In addition, the local deformation and movement directions of this postcapillary venule were similar to that of the proximally connected microvessel (Supplementary Fig. 3a, b, lower). The contrastive observation demonstrates that silica probes are highly susceptible to puncturing blood vessels during implantation, causing bleeding (Supplementary Movie 3). By the way, we were unable to utilize two-photon microscopy for real-time recording of the precise moment when the silica probe encountered the blood vessel in vivo, because of frequent bleeding occurrences during probe implantation, which necessitated immediate termination of the experiment. Supplementary Movie 4 provides a comparable microscopic perspective of the UENP electrode implantation procedure, illustrating the absence of bleeding resulting from our UENP insertion.

The implementation of temporary hardening strategies significantly facilitates the implantation process9. However, the degradation of the stiffener may extend for several weeks (Supplementary Table 2), surpassing the duration of probe insertion. Consequently, concerns regarding intraoperative bleeding remain. Moreover, the accumulation of degradation products in the local brain tissue results in neurotoxicity5. We constructed a temporarily hardened probe by combining soft fiber with PEG 4000 as the control group. PEG is a commonly used stiffener in neural probe applications. The resulting probe diameter was approximately 230 μm, consistent with the size of our UENPs (Supplementary Fig. 4a). Upon retracting the temporarily hardened probe, a notable outflow of blood occurred, as illustrated in Supplementary Fig. 4b, indicating a significant risk of blood vessel rupture during the implantation of the temporary hardened probe.

To confirm the minor vascular injury caused by our UENP compared to a rigid probe, we replicated the UENP implantation process in vitro using agarose gel wrapped around isolated microvessels. Saline containing an oily blue injection compound was selected to perfuse the vascular system of live mice to simulate the physiological filling pressure of blood vessels (Supplementary Fig. 5). To avoid the premature extension of UENP, we quickly inserted it to where it just touched the blood vessel. As the images shown in Fig. 3f (red box), the UENP automatically uniaxial extends and passes alongside the blood vessel. In contrast, a silicon probe of the same size pushes the vessel downward significantly (Supplementary Fig. 6 and Movie 5). The recording and statistics of vertical and lateral length variations for the same batch of probes prove their similar self-calibrated implantation capabilities (Fig. 3g).

Mechanical analysis of UENPs

Because of the low adhesion between cells and PEG hydrogel with biological inertness, the interaction force between the probe-tissue interface is predominantly by the bending and torsion of probes after implantation13,41,42. The negligible probability of torsion allows us to consider bending stiffness as the critical mechanical parameter in our study. The bending stiffness (Ktheory) of cylinder-like geometry is determined by the modulus (E) and size (length and diameter) according to the formula of theoretical value as

$${K}_{{theory}}={({3{\rm{\pi }}{Ed}}^{4})({64\,{{L}}}^{3})}^{-1}$$
(1)

The modulus of UENP obtained by compression measurements is higher than that of a pristine PEG-DA scaffold due to the introduction of rigid PEDOT (Fig. 4a). After overnight hydration, the modulus decreases about 10 times (from 1.72 MPa to 135.26 kPa). In the initial hydration, the modulus drops to 400 kPa in 60 s, and then the reduction rate decreases as displayed in Fig. 4b. Finite element analysis (FEA) was utilized to quantify the stiffness (KFEA), which was predicted using the classic cantilever beam model, where a force (F = 10–6 N) is applied upon the free end of the probe. The relationship between K and F is given by

$$K={{Fu}}^{-1}$$
(2)

where u is the displacement of the free ends of the probes (u is shown in Supplementary Fig. 7, red boxes). Compared with the stainless steel (31635.56 N m−1) and silica (8045.05 N m−1) probes which are widely used in brain-machine interfaces, the fully hydrated UENPs exhibit an extremely low bending stiffness (0.02 N m−1) in identical scales. The simulation results agree well with the theoretical results (Fig. 4c). It is noteworthy that dehydrated UENPs exhibit a high degree of bending stiffness (0.28 N m−1), which aids in their smooth penetration of the brain surface. During the initial implantation, researchers commonly employ a method of selecting vascular-free sites for probe insertion under assistance to avoid puncturing blood vessels43. In this study, we followed this approach. However, the microscopic navigation only prevents the probe from puncturing surface vessels. The potential for puncturing internal vessels remains a risk that is evaded when using our UENPs for implantation.

Fig. 4: The mechanical characterizations of UENPs.
figure 4

a The compression stress-strain curves of UENPs and the pristine probe that absences with the PEDOT:PSS, at hydrated and dehydrated states. b Modulus changes as the probe is immersed in PBS solution in 90 s. c The bending stiffness for stainless steel, silica, and the dehydrated and hydrated UENP. d A schematic illustration of the FEA model depicting the probes fixed to the skull during brain micromotion. The stress fields to surrounding brain tissue are produced by (e i) stainless steel probe, (e ii) silica probe, and (e iii) UENP. The maximum stresses are demonstrated in f. Insertion of the fully hydrated (g) and dehydrated (h) UENP into the phantom brain at a speed of 1 mm s−1. Scale bar, 1 mm.

Brain micromotion caused by physical behaviors and daily activities inevitably generates repetitive injuries to peripheral tissues each time the contact interface shifts between the implanted probe and the brain, especially in the devices tethered in skulls44. Thus, we used an FEA model to investigate the stress fields in tissue caused by the probes (tethered with skulls)45, where the brain movement has an amplitude of 100 µm in the horizontal direction (Fig. 4d). The results shown in Fig. 4e, suggest that our UENP generates significantly lower stress fields in the surrounding tissue compared to the stainless steel and silica probes. The stress field of UENP is relatively concentrated in its fixed end, while that of the stainless steel and silica is more concentrated at their tips where the sensing and modulation of neural activity are carried out. The increased stress can trigger astrocyte proliferation and wrapping around the probe, inducing chronic signal loss. The specific value is shown in Fig. 4f. It is expected that the reduced stress field of UENPs will improve their long-term reliability. In terms of the common practical concern about compliant probes, the bulking tips of these probes produce depressions onto the brain surface rather than puncture and deterministic implantation into the target regions (Fig. 4g). Our dehydrated UENP with higher bending stiffness is able to penetrate the phantom brain at a high speed of 1 mm s−1 and keep perpendicular insertion into the brain as shown in Fig. 4h.

To extend the potential of probes for applications in multi-site neural activity detection, we employed the stretchable silicon-based elastomer as the substrate to construct the UENPs array. This array possesses stable positioning to the brain surface based on the anchoring effect rather than chemical bonding, facilitating the maintenance of high-quality signal recording and reducing the potential risk of tissue damage46. To investigate the anchoring effect of the UENPs, we carried out a pull-out test in a phantom brain, and compared the stability of planar ECoG and our UENPs array (Supplementary Fig. 8a). The initial process of load rises as pulling is called the debonding stage, then an inflection point (Pmax) of load occurs. In this stage, the contact area of the UENPs array and brain model is invariable (Supplementary Fig. 8b i). Then the motor of the UENPs array changes into a slippage regime, and the pullout load decreases with the displacement increase because the contact area of the array-brain interface keeps reducing (Supplementary Fig. 8b ii). Until the probe is extracted from the brain, the pullout load equals the friction along the shear direction. Different from the UENPs arrays, planar ECoGs are readily pulled away with a much smaller load (Pmax2). The changes in pullout loads of the UENPs are similar to the fiber pullout curve47 (Supplementary Fig. 8c), allowing us to evaluate the stability of UENPs by the Pmax. We evaluate the shear strength of the planar ECoG and UENPs array as the equation

$${\rm{Shear\; Strength}}=\frac{{P}_{\max }}{{WL}}$$
(3)

where L is the length and W is the width of the array.

The result shows that the load required to pull the UENPs arrays out of the brain tissue is about 4.5 times higher than that of planar ECoGs (Supplementary Fig. 8d), indicating the anchoring effect enhances the stability of the interface between UENPs and the brain. The force changes in UENPs while being pulled out are analyzed. Once the pullout load is applied, the UENPs start to suffer internal stress (F) from the surrounding tissue and tend to bend. After UENPs bending, a composite force including part of the internal stress (F cos θ), the frictions from tissue to probe (Fh sin θ) and to the substrate (Fs) resists to the pullout load. In the slippage regime, the internal stress vanishes. Hereto, Fs plus Fh equals P (Supplementary Fig. 8b). Furthermore, our UENPs maintained their structural integrity throughout these three stages, even though the UENPs were bent 90° during the slippage stage. In addition, even on a moving brain, our UENPs arrays keep stably attaching to slippery surfaces. We applied the arrays and planar ECoGs onto a shaking brain model with artificial CSF on the surface (Supplementary Fig. 8e). After rotation for 1 hour at 60 rpm, like the rhythm of the heartbeat, our arrays were as motionless as a statue while the ECoGs rotated and moved away from the initial position (Supplementary Fig. 8f). Enhanced connective stability between the UENPs array and the brain is anticipated to improve the reliability of the recorded signals.

Electrocorticography monitoring in vivo and the electrical properties of UENPs

A UENPs array was subsequently applied to cover multiple brain regions for acquiring neuroelectric potentials (Fig. 5a i). To confirm the recorded potential changes are from the neuronal activities rather than the electrical noise, we set a platinum (Pt) wire electrode as a standard reference electrode. Figure 4a ii shows the clear potential signals recorded by the UENPs array with a sufficiently high signal-noise ratio (SNR) (Fig. 5b). Spectral analysis of the obtained signal exhibits typical anesthetic characteristic peaks (Supplementary Fig. 9). The electrical properties of the UENPs were investigated as well. The hydrated UENPs exhibit the lowest impedance at 1 kHz, 8.05 ± 3.17 kΩ, which is 13 times lower than the dehydrated UENPs (102.39 ± 65.58 kΩ) (Fig. 5c, d). The reduction in impedance is facilitated by immersing the UENPs in ion-rich artificial cerebrospinal fluid (ACSF), thereby imparting them with mixed ionic-electronic conduction capabilities. The conductivity (8.47 ± 0.49 S cm−1) of the hydrated UENP was characterized by a four-point probe measurement. The impedance changes of the PEG-DA scaffolds before and after being hydrated also prove the influence of ACSF on conductivity improvement. The phase angle for hydrated UENPS is near 0°, indicating the interface represented less phase delay (Fig. 5e). The decreasing impedance of UENPs as the hydration time prolongs confirms the mixed ionic-electronic conduction phenomenon in UENPs (Fig. 5f).

Fig. 5: Electrocorticography monitoring in vivo and the electrical properties of UENPs.
figure 5

a i The implantation sites of a seven-channel UENPs array with a Pt reference probe. The obtained field potential is shown in (ii). b There is no significant difference in the signal-to-noise ratio between the signal acquired by our UENPs and the platinum (Pt) probes. (n = 7, paired two-sided Student’s t test: p = 0.5232) ce Tips’ impedance (c) and phase angles (e) of the pristine probes and UENPs at the dehydrated and hydrated states. The impedance in 1 kHz is extracted and shown in d. f The impedance of UENPs after hydrating for 0, 3, 7, 9, and 13 min. g The impedance and phase angle of UENPs after implanting 14 days (n = 6). h The impedance of UENPs before and after bending 90° (n = 6).

Rejection response activation causes astrocytes to wrap around the surface of the implant and form a compact glial sheath, resulting in elevated impedance of the probe, which is unfavorable for long-term signal acquisition5. To value the physiological aqueous instability of our UENPs, we compared the impedance before and after implantation for 14 days. The similar impedance and phase angles indicate the long-term biocompatibility of our UENPs (Fig. 5g). In addition, there were no characteristic peaks of PEDOT and PSS found in the UV spectrum of the washing solution after multiple times washing of the UENP, indicating that our UENPs pose no risk of leakage during long-term application (Supplementary Fig. 10). For practical application, the impedance of UENPs remains virtually unchanged even when bent up to 90° (Fig. 4h), which is significant in minimizing equipment damage due to unavoidable bending. The integrity of UENP after 14 days of implantation further demonstrated their recyclability (Supplementary Fig. 11).

The investigation of biocompatibility and immunological rejection

To evaluate the immune rejection caused by UENPs, we implant them in the M1 cerebral cortex of rats for 14 days and 2 months. We select three representative biomarkers to investigate the foreign body response by immunohistochemical analysis of the immune cells on coronal sections. Astrocytes and microglia are the main immune cells of the brain’s response to implanted needle-like materials48. The upregulation of glial fibrillary acidic protein (GFAP) is one of the characteristics of the injury-induced transformation of astrocytes to a “reactive” phenotype49. Meanwhile, ionized calcium-binding adapter molecule 1 (Iba1) is one of the biomarkers of microglia. We also stained a cluster of differentiation 68 (CD68) to identify activated macrophages. Two common commercial brain probes (silica and stainless steel) with similar dimensions were utilized as control groups. Three types of inserts were fastened to the skull in the same way as dental cement. As shown in Fig. 6a, unlike the obviously compact glial sheath around the silica and stainless steel probes, our UENPs show extremely less macrophage activation, astrocyte reactivity, and microglial infiltration, indicating lower levels of acute tissue reaction caused by our UENPs. The less glial sheath around UENPs explains the similar impedance and phase angles of UENPs before and after implantation (Fig. 5g). Noteworthily, the accumulations of inflammatory cells around the damaged blood vessels (green arrow in Fig. 6b) are observed in the brain tissue implanted silica and stainless steel probes while barely appearing in the vicinity of the UENPs. Meanwhile, the missing area (kill zone) of brain tissue caused by stainless steel is much bigger than the area of the probe, which confirmed the FEA results of the stress fields caused by the high bending stiffness of stainless steel under brain micromotion. In terms of chronic tissue reaction, we also stained and qualified these three biomarkers in the vicinity of the probes after two months of implantation. Long-term tissue response is mainly determined by chronic tissue-device interactions (e.g. micromotion), so we extended the implantation period to 2 months. Our UENPs elicit substantially reduced chronic foreign body response as compared to both silica and steel probes (Fig. 6c and Supplementary Fig. 12). In addition, the staining of living/dead cells after cocultured PC12 cells with UENPs for 1, 3, and 7 days further confirms the lower cytotoxicity of UENPs (Fig. 6d and Supplementary Fig. 13).

Fig. 6: The short-term and long-term biocompatibility of UENPs.
figure 6

a Representative image of fluorescently labeled cells after implanting for 14 days. Red indicates CD68, GFAP, and Iba1 positive cells, whereas blue indicates cell nucleus. Scale bar, 200 µm. Average immunofluorescence area quantifying the presence of CD68, GFAP, and Iba1 in the vicinity of the UENPs, silica, and stainless steel probes for 14 days (b), 2 months (c) following implantation, respectively. Data are shown as mean ± SEM, n = 9 images/3 rats, One-way ANOVA test, *p < 0.05, **p < 0.01, ***p < 0.001. d The cell cytotoxicity of UENPs for 1, 3, and 7 days compared with blank and PEG-DA hydrogel. One-way ANOVA test, ns means none significant, n = 6.

The clinical imaging compatibility of UENPs

Clinical imaging, such as magnetic resonance imaging (MRI) and computerized tomography (CT), is increasingly essential for diagnosis and surgical navigation. However, commonly implanted metallic neural electrodes may render MRI unsafe or affect the restruction of images and significantly limit its diagnostic utility50. Besides, in CT imaging, the metallic artifacts are produced by the beam hardening effect and scattering effect, which severely reduce the quality of subsequent filtered back projection reconstruction, restricting its diagnostic accuracy51,52. This poses a significant clinical challenge as it is precisely these patients with implanted devices who require imaging assessments53. As a result, magnetic field compatibilities are required in implantable bioelectronic devices.

Controlling the content of PEDOT:PSS in electrodes or utilizing the composite of PEDOT:PSS can avoid the reflection of generated RF signal54,55,56. On this basis, we perceive that our UENP will not block the magnetic field and RF signal, avoiding artifacts from implants. To demonstrate the MRI compatibility of our UENP, we imaged a metal probe and a UENP-embedded cuboidal 0.6% agarose gel in a 7T high-field research MRI system. The metal probe produced a vast artifact, blocking out a big area of surroundings (Fig. 7a i). On the contrary, our UENP was adequately infiltrated by CSF during the long-term application so that it was nearly invisible in MR imaging, regardless of the scan sequence (Fig. 7a ii). More importantly, the location of the metal probe moved from vertical to horizontal, which introduced potential security problems for the testee. Consistent with in vitro results, we observed severe metallic artifacts in vivo (up to 18 mm diameter spherical shadow) and deformation of the image (Fig. 7b i), while no artifacts around our UENPs (Fig. 7b ii). During the examination, the respiratory rate of the rat was a serious disorder (Supplementary Fig. 14a).

Fig. 7: MRI and CT compatibility of UENPs.
figure 7

High-field 7T MRI scans for metal probe (i) and UENP (ii) embedded in a phantom brain (a) and an anesthetized rodentine brain (b). The metal probe presents a large shadow under MRI scans while the UENP does not. Micro-CT scans for metal probe (i) and UENP (ii) embedded in a phantom brain (c) and an anesthetized rodentine brain (d). The metal probe presents a typical X-ray scattering artifact under micro-CT scans while the UENP does not.

The aforementioned metallic artifacts in CT imaging fundamentally resulting from the high density and large atomic mass pose challenges to conventional metal electrodes56. The density of pure PEDOT:PSS is about 1.00 g cm-1, which is much lower than common metal materials, such as stainless steel, platinum (Pt), gold (Au), and aluminum (Al). According to this, we consider that our UENP could minimize scattering artifacts in CT. To verify the CT compatibility of our UENP, we first imaged the same sample of the simulated brain in an MRI test by a micro-CT scanner. As expected, we observed typical X-ray scattering artifacts around the metal electrode (Fig. 6c i), but not around the UENP (Fig. 7c ii). Similarly, we observed the silver conductive adhesive in the UENPs array guiding the position of simple UENPs, while UENPs were almost invisible in CT imaging (Supplementary Fig. 14b). Finally, we compared the metal probe and UENP in the rat where the notch of the skull in Fig. 7f was the surgical window. The artifact of a metal probe obscured the surrounding brain tissue to some extent (Fig. 7f i). On the contrary, there was no artifact around our UENP (Fig. 7f ii). Thus, the dental cement used to encapsulate the skull window was visible.

Discussion

In this study, we develop a uniaxial extending neural probe (UENP) with compliance. The UENP exhibits a CSF-responsive self-implantation along the insertion direction with a simultaneous reduction in bending stiffness. The anisotropic elongation of UENP originates from cross-sectional in-plane deformation inhibition resulting from the globally aligned polymer chains and the subsequently enhanced interchain interactions. As a result, probe buckling before and during implantation is effectively prevented. More importantly, intraoperative bleeding associated with vessel puncture by probes is avoided, significantly decreasing the surgical complexity and the requirement for surgical navigation. The mechanical similarity of UENP and brain tissue substantially reduces the stress and strain in the surrounding neural tissues during the micro-motion of the brain concerning the skull. Thus, both the acute/chronic immune responses and foreign body responses caused by the probe are minimized. Furthermore, the MRI and micro-CT compatibilities are beneficial for patients implanted with BCI to receive a comprehensive postoperative condition assessment.

To expand the detectable space, we sought to integrate UENP to construct a multi-channel array for sensing the potential signals from various regions, including the primary somatosensory cortex, motor cortex, retrosplenial dysgranular cortex (RSD), and visual cortex. Importantly, the implantation of the UENPs array does not result in any bleeding. Furthermore, the presence of a minimal scaled wire composed of metal nanoparticles does not compromise the image compatibility of the UENP array. However, for practical clinical applications, concerns regarding mass manufacturing remain. The UV-cured molding process necessitates individual probe separation from a liquid pool which results in isolated unassembled components, hindering the automated alignment during assembly. In future endeavors, we aim to establish a liquid-phase compatible integrated manufacturing platform that combines the strengths of our current strategy with state-of-the-art methods of automated arraying driven by acoustic radiation force. This development in scalability is advantageous for its application in regulating multi-organ functions through the central nervous system.

Methods

Fabrication of UENPs

PEG-DA (~700 Da, Macklin Inc.) aqueous solution (50 vol. %) was vortex mixed, and then PSS (MW ~ 75000, 18 wt. % in H2O, Sigma Aldrich) was added in 60 vol. % of the PEG-DA. The concentration of hydrosoluble photoinitiator (2-hydroxy-4’-(2-hydroxyethoxy)-2-methylpropiophenone, Macklin Inc.) is 1 wt. % (relative to PEG-DA). The well-mixed pre-polymerization solution was added to a small pool and covered with a plasma-treated masking plate on top. A vertical light source was illuminated at a distance of 2.5 cm from the mask plate with a curing time of 15 s and a power of 340 mW cm−2. After being washed with water, the produced scaffolds were immersed in the mixed solution with EDOT monomer (90 vol. % in ethanol) for 3 h. Then, the scaffolds were transferred to an ethanol solution containing iron (III) p-toluenesulfonate hexahydrate (Aladdin Inc) and sulfuric acid and reacted at room temperature for 48 h. The probes all turned dark blue and were treated in 50 vol. % dimethyl sulfoxide (DMSO, Aladdin Inc.) aqueous solution for 12 h. Finally, they were annealed and dried.

Material characterization and measurements

The surface morphological features of dehydrated UENP and hydrogels directly doped with PEDOT:PSS were obtained using field-emission scanning electron microscopy (FE-SEM, JSM-7900F, JEOL). Angle analysis was performed after edge extraction using ImageJ. The mechanical property tests of the UENPs were performed on a universal testing machine (Criterion Electromechanical Test System, C42.503, MTS) with a compression rate of 10 mm min−1. Young’s modulus was calculated according to the slope of the stress-strain curves (within 0–10% of strain values). Impedance measurements were conducted over a 1 Hz to 100 kHz frequency range using a commercially available LCR meter (IM3533–01, HIOKI). Conductivities were characterized by four-point probe measurement (HP-504, 4Probes Tech Ltd). The washing solution was measured by ultraviolet spectrophotometer (UH5300, HITACHI).

Characterization of attachment stability UENPs array on the brain surface

A bulk 0.6% agarose gel was used to simulate the brain environment in vitro. UENPs were inserted into the brain model, and a high-speed video camera was used to record the UENP self-implantation process. UENPs array, which utilized the stretchable silicon-based elastomer as the substrate, was constructed to acquire multi-site neuroelectric potentials. The conductive wire was constructed by evaporated Au on the substrate, and UENPs were connected with the Au point by commercial silver paste. To investigate the anchoring stability of UENP in the brain, a pull-out test was performed on planar ECoG and our UENPs array to analyze their force changes during the pullout process. To further assess the adhesion characteristics of the implanted UENPs, both UENPs and controls were implanted into the brain model and placed on a shaking bed. The speed of the shaking bed was set to 60 rpm to simulate the heart rate under normal physiological conditions. The displacements of UENPs and controls were observed after one hour.

Biocompatibility tests

PC12 cells (CRL-1721, ECACC)were chosen to coculture with the UENP for 1, 3, and 7 days. The cell viability was assessed using living and dead cell staining (Calcein-AM/PI Double Stain Kit, Beijing Solarbio Science & Technology Co., Ltd following established protocol. Stained cells were visualized under a confocal fluorescence microscope (Eclipse 80i, Nikon, Japan). Cell counts were analyzed using ImageJ software.

UENP implantation

Male Sprague-Dawley rats (SD rats, 6–8 weeks of age) were anesthetized with 5% isoflurane. The rats were secured onto the rat brain stereotaxic apparatus and kept anesthetized with 1.5–2% isoflurane. Cut the skin and tissue overlying the skull to expose the surface of the skull. Subsequently, created windows in the skull with a cranial drill and carefully removed the dura and pia. The UENP was guided into the target site using a micro-manipulator arm attached to the stereotaxic frame. Finally, the UENP was fixed to the skull using dental cement. After the surgery, the rats were returned to their cage and placed on a heating pad at 37 °C until they fully recovered from anesthesia. All animal procedures were approved by the Nanjing Medical University Committee on Animal Care and carried out in accordance with the Guide for the Care and Use of Laboratory Animals.

Fabrication and implantation of the temporally stiffened probe

Both ends of a cotton thread were held with tweezers and then immersed in molten PEG4000 solution. After 1 min, the thread was slowly lifted and allowed to cool for hardening. Subsequently, it was sterilized and implanted into the rat brain using the UENP method.

Fabrication of thermos-curing hydrogel

We employed 1 wt.% ammonium persulphate (APS), a widely utilized thermal initiator, along with 0.1 wt. % N,N,N’,N’-tetramethylethylenediamine (TEMED) as an initiation accelerator. The polymerization was conducted at a temperature of 120 °C for a duration of 1 h.

Miniature two-photon imaging

Animals were anesthetized intraperitoneally with 1–1.5% isoflurane at a rate of 1 L min−1. The animal’s head was fixed in a stereotaxic apparatus and the eyes were kept wet with ointment. Scalp and soft tissue were removed to expose the skull. A craniotomy was performed above the somatosensory cortex using a skull drill. The bone flap, dura, and pia mater were carefully removed. A coverslip (diameter: 4 mm, thickness: 100 um) was placed on the exposed brain and fixed to the skull with dental cement. A metal connector was adhered to the skull by additional dental cement. Imaging was performed using a miniature two-photon microscope (Transcend Vivoscope Biotech Co., Ltd, China) equipped with a large field-of-view (FHIRM-TPM V2.0, Field of view: 190 × 190 μm2; Resolution: ~850 nm; working distance: 1000 μm) water immersion objective. During measurements, the animals were head-fixed. To visualize the vasculature, Texas red Dextran (5% w/v, 70,000 kD MW, Life Technologies catalog number D-1830) was injected intravenously in the tail vein (30 mg/kg). The UENPs were labeled with a green fluorescent dye (ene) (0.5 mg ml−1, Engineering for Life catalog number EFL-DYE-UF-ENE-G). The probe was inserted obliquely into the cortex at a small angle to the brain surface. Texas red Dextran were excited at 1030 nm, the probes were excited at 920 nm. Imaging data were acquired using imaging software (GINKGO-MTPM, Transcend Vivoscope Biotech) at a frame rate of 10 Hz (512 × 512 pixels) with a femtosecond fiber laser (~35 mW at the objective; TVS-FL-01, Transcend Vivoscope Biotech).

Neuroelectric potentials monitoring in vivo

The neuroelectric potentials were acquired by implanting a seven-channel UENPs array with a Pt reference probe into the brain. The raw signals were band-passed (0.1–500 Hz) and amplified 1000 times by Differential AC Amplifier (3500, A-M System) and recorded by PowerLab (PL3516, AD Instruments) at a sampling rate of 10 kHz.

Immunohistochemistry

UENPs, silica probes, and stainless-steel probes are implanted in the M1 cerebral cortex of rats. These electrodes are removed after 14 days (n = 3) and 60 days (n = 3) of implantation. SD rats are deeply anesthetized with 5% isoflurane and then transcardially perfused with saline buffer followed by a paraformaldehyde fixative solution. After that, the rats are decapitated, and the brains are carefully removed. The brains are incubated in 4% paraformaldehyde overnight for fixation and then dehydrated in 30% sucrose (wt/vol) in PBS overnight. Subsequently, the tissues are sliced using a paraffin microtome for immunostaining. The brain slices are incubated with anti-CD68 (GB113109, Servicebio Inc.), anti-GFAP (GB11096, Servicebio Inc.), and anti-Iba1 (gb113503, Servicebio Inc.) imaging at 4 °C overnight. Then, slices were incubated with the secondary antibody (Cy3 conjugated Goat Anti-Rabbit IgG (H + L), GB21303, Servicebio Inc.) at room temperature. After staining the cell nucleus with DAPI, all images are acquired by a confocal microscope. All imaging data are processed using the ImageJ software. All the data are presented as mean ± S.D.

The clinical imaging compatibility

MRI and micro-CT imaging were performed on metal UENPs or 0.6% agarose gel implanted UENP. MRI scanning was conducted using a 7.0-T MR scanner (Bruker BioSpin MRI, Germany). Before MRI scanning, the rats were anesthetized with a 5% isoflurane. The rat’s head was fixed in a coil with the body axis aligned along the center line. T1, T2, and T2* weighted coronal and axial scans were performed with a slice thickness of 1 mm. The imaging parameters were set as follows: TR (repetition time) = 3000 ms, TE (echo time) = 11 ms, matrix size = 256 × 256, and field of view (FOV) = 39.66 × 33 cm2. Micro-CT images of the implantation sites were obtained using a SkyScan 1176 scanner. The voltage for image acquisition was 70 kV, with a current of 329 μA and an exposure time of 100 ms. The images were reconstructed using vendor software.

Finite Element Analysis of bending stiffness and brain micromotions

For both analysis of bending stiffness and brain micromotions, the Finite Element Analysis (FEA) simulations were implemented using the ABAQUS 2016. Each probe used different mechanical properties (Supplementary Table 3) and the same dimension (0.25 mm diameter and 1.5 mm length) to represent probes with different materials. The probe and brain tissue were implemented with C3D8 and C3D8RH, respectively. In the simulations, one extremity of the probe was anchored, while the opposite end was left free to undergo deflection. By applying a small point load F at the free end of the probe, the bending stiffness K was computed by K = Fu-1, where u is the deflection of the probe. A lateral load of 100 μm displacement was applied to the bottom surface of the brain tissue in the simulation. The interaction between the probe and the brain tissue was simulated as surface-to-surface contact (friction coefficient = 0.3).

Statistical analysis

IBM SPSS Statistics 25 software was used to assess the statistical significance of all comparison studies in this work. The normality of data distribution was tested via the Kolmogorov–Smirnov test. The abnormally distributed data were subjected to the Kruskal–Wallis one-way ANOVA test. The normally distributed data were subjected to one-way ANOVA followed by the Scheffe test (for data with equal variances) and Tamhane’s T2 test (for data with unequal variances), employing a significance threshold of *p < 0.05, **p < 0.01, ***p < 0.001.