A three-dimensional self-opening intraneural peripheral interface (SELINE)

The SELINE consists of a loop structure made of a main shaft with two wings per each side. The shaft is 360 μm wide; each wing has a total length of 400 μm and a width of 150 μm. The device has a sensitive area in which active sites (0.0037 mm²) are located. On both sides of the device there are one active site on the main shaft and two active sites on each wing (figure 1(e)). The length of the sensitive area has been designed to fit the diameter of the rat sciatic nerve, however the design can be modified to match different size nerves. Wing shape and electrode geometry are in compliance with the optimized design presented in our previous study (Cutrone et al 2011).

Zoom In Zoom Out Reset image size

The materials for the SELINE were chosen in order to satisfy biocompatibility, mechanical integrity and electrical efficacy. The insulation of the electrode is made of polyimide, a biocompatible, flexible and stable polymer (Rubehn and Stieglitz 2010); active sites and electrical tracks are made of gold with titanium as adhesion layer. The microfabrication of the device follows standard lithographic techniques used for thin-film electrodes (Boretius et al 2010).

A silicon wafer (Si-Mat, silicon materials) was used as substrate for the development of the electrode layer by layer. Polyimide PI2611 (HDMicroSystem) was spin-coated at 3000 rpm for 30 s (thickness = 6 μm, figure 1(a)). Samples were hard baked in an oven with nitrogen flux at 350 °C for 1 h (Carbolite). Primer, LOL and S1813 (Microposit) were spin-coated on the wafer respectively at 3000 rpm for 30 s, 1000 rpm for 20 s, 4500 rpm for 30 s; samples were exposed to UV (Karl Suss) using a glass photomask (Frontrange Photomask) with an exposure dose of 180 mJ cm⁻² (i-line). Samples were developed for 30 s using MF-319, rinsed with DI water and dried. Oxygen plasma (Gambetti) was used to mechanically functionalize the surface and increase the roughness of the layers so enhancing consequent adhesion with the metals (100 W, 15 sccm of O₂, 300 mTorr). Titanium and gold were sputtered (Sistec–Leybold) on the samples with a thickness of 20 nm (90 W—1 min) and 250 nm (70 W—3 min) respectively at a vacuum of 4 × 10⁻⁶ Torr. Lift-off was made by plunging the wafer in the photoresist remover with ultrasound for 20 min (figure 1(b)). Samples were washed with DI water and dried. Samples were spin-coated with the second layer of polyimide 6 μm thick (3000 rpm for 20 s) and hard baked at 350 °C for 1 h. Aluminum was thermally evaporated on samples (1 × 10⁻⁶ Torr; 1.3 A) at a thickness of 200 nm (TecnoService). S1813 (Microposit) was spinned on samples (4500 rpm for 30 s) that were subsequently exposed to UV (exposure dose = 180 mJ cm⁻², i-line). Development was done for 30 s using MF-319 and samples were rinsed and dried. Wet etching of aluminum was done with a solution of HNO₃:H₃PO₄:HAc:H₂O = 4:1:1:4 for 15 min at room temperature. Samples were dry etched with oxygen plasma with a flux of oxygen of 40 sccm at 170 W for 30 min. The aluminum mask was etched away (figure 1(c)). Electrodes were peeled off from the wafer (figure 1(d)).

This design of SELINE provides an anchorage system to the nerve tissue during implantation. A memorization strategy is implemented to switch wings from the normal position (passive configuration) to the open position (active configuration). A mechanical plastic deformation of the electrode was applied to promote wing opening at different amplitudes. Electrodes were attached vertically to a 10 N load cell, while the base was constrained with a press. A panel was located behind the electrode to allow the wings opening on one side only. Electrodes were plastically deformed by applying stress over polyimide yield point (0.6 N mm⁻²) with a loading velocity of 2 mm min⁻¹. Different angle amplitudes were obtained by regulating the applied stress to the electrode (figures 2(b)–(c)). The setup used for the memorization of wing profile is shown in figure 2(e).

Zoom In Zoom Out Reset image size

Electrodes were rinsed in ethanol and dried. A printed circuit board (PCB) was designed (CadLine) in order to miniaturize the bulk of the whole electrode (length = 8.5 mm; width = 5.7 mm; thickness = 0.2 mm). Insulated wires (Road Runner) were soldered to the PCB. The final part of the electrode with connection pads was attached to the PCB for each side of the electrode with cyanoacrylate glue. Electrical connections between electrode pads and PCB contacts were made by using silver conductive glue (Ablestik-Henkel); followed by thermal treatment in furnace to stabilize and dry silver solderings (2 h at 150 °C). Two-component biocompatible silicone (Silbione-Bluestar Silicones) was applied on the PCB and solderings to avoid contact of the solder material with biological tissue. Polymerization of the silicone for its final stabilization was achieved by thermal treatment at 100 °C for 30 min.

The electrode was folded at its center point and a loop thread attached to a stainless steel needle (STC-6, Ethicon) was placed between the two branches. The two PCBs of the electrode were then glued together. Wires for electrical connections were inserted in a flexible silicone tube and sealed with biocompatible silicone (figure 2(a)).

Ohmic resistance of gold tracks for each active site was measured with a probe tester (Cascade PM5).

Electrochemical impedance spectroscopy was used to measure impedance (Gamry). A three-electrode setup made of a platinum counter electrode, an Ag/AgCl reference electrode and the working electrode was used in 0.9% saline solution. Impedance spectra were characterized between 100 Hz and 100 kHz applying a sinus wave of 5 mV.

Tensile test (Instron) was performed on four samples in order to analyze the mechanical behavior of the electrode. The electrode was folded in its working configuration. The loop of the electrode was attached to a steel hook linked to a load cell of 10 N, whereas the electrode pads were inserted in a ZIF connector and then in a press linked to the Instron. The electrode was stressed until breakage with a loading velocity of 4 mm min⁻¹.

A mechanical study was performed in order to evaluate the efficacy of the three-dimensional anchoring system provided by the lateral wings.

Two rat sciatic nerves were harvested, and stored in PBS at −20 °C for one week. Three types of electrodes, with the same geometry but differing in the presence of open, flat, or no wings, were tested:

• SELINE (with memorization): electrodes with open wings in active configuration (seven trials).
• SELINE without memorization: electrodes with flat wings in passive configuration (four trials).
• TIME: electrodes without wings (four trials).

In the experimental setup (figure 3(A)), the nerve was fixed between two presses: on one side, a load cell of 10 N was attached to measure the pre-stress of the nerve. The electrode was constrained to the load cell of the Instron and vertically inserted into the nerve, in compliance with the SELINE way of insertion (see section 1). In order to simulate slippage of the electrode inside the nerve, the electrode was extracted in the same direction but opposite sense of insertion with a loading velocity of 50 mm min⁻¹. Force and displacement were recorded for each trial. Mean values of maximal extraction forces for the three types of electrodes have been calculated and compared.

Zoom In Zoom Out Reset image size

The working principle of the SELINE implantation is divided into three phases (Cutrone et al 2011 and 2013): (i) piercing the nerve with a needle that serves as a guide; (ii) transversal insertion of the electrode across the nerve, and (iii) slight pulling back of the electrode to allow the anchorage of the wings.

In vivo experiments were made on adult Sprague-Dawley rats. All procedures were performed under ketamine and xylacine anesthesia (90/10 mg kg⁻¹ i.p.), following protocols approved by the Ethics Committee of the Universitat Autònoma de Barcelona in accordance with the European Communities Council Directive 2010/63/EU.

Two types of electrodes were used for the comparison: SELINE and TIME. For both types, the mid point of the structure is bent in a V-shape before implantation, and disinfected with 70% ethanol, before implant. The sciatic nerve was surgically exposed at the midthigh and carefully freed from adherences to surrounding tissues. The electrode is inserted with the help of a straight needle attached to a 10–0 loop thread (STC-6, Ethicon); the thread is passed between the two arms of the electrode and pulls the arrow shaped center of the electrode strip (Badia et al 2011a). Electrodes were obliquely inserted, forming a 45° angle with the longitudinal axis of the nerve, across the peroneal and tibial fascicles of the sciatic nerve, proximal to the trifucation of the nerve at the knee (figures 4(A)–(B)). The insertion was monitored under a dissection microscope to ensure the correct placement of the electrode and the adequate positioning of the active sites inside the nerve.

Zoom In Zoom Out Reset image size

To compare the performance of SELINE (S) and TIME (T) for selective neural stimulation, two groups of rats were used. In both groups (S, n = 8; T, n = 4) the intraneural segment of the electrode was inserted as indicated above across the sciatic nerve. After insertion, the electrode active sites were located in a given position inside the nerve fascicles, named position 0 (P0). P0 was defined as the position where the three gold strips on the loop of the electrode are located all outside the nerve (figure 4(C)).

The stimulation protocol was performed using the STIM'3D stimulator (Andreu et al 2009). Monophasic rectangular pulses at 0.5 Hz, with a width of 10 μs and an intensity from 20 to 300 μA were delivered through each one of the electrode sites of the SELINE and TIME against a small stainless steel needle electrode placed near the nerve. Electromyographic (EMG) signals were simultaneously recorded from tibialis anterior (TA), gastrocnemius medialis (GM) and plantar interossei (PL) muscles, which are innervated by different subfascicles of the sciatic nerve (Badia et al 2010). The compound muscle action potentials (CMAPs) were recorded by means of small needle electrodes placed in each muscle, connected to P511 ac amplifiers (Grass), amplified by x100 or x1000, and filtered (5 Hz, 3 kHz). Digital sampling of the signals was made with a PowerLab recording system (PowerLab16SP) at 20 kHz, and fed into Chart software. The amplitude of the CMAP was measured from baseline to peak, and normalized to the maximum CMAP amplitude obtained for each muscle in the experiment. The same protocol was repeated for all the rats and electrodes. Recruitment curves for each active site were plotted.

Measurements of the CMAP amplitude are used to quantify the activation by a specific stimulating condition to a single muscle i, among the set of muscles j (PL, GM, TA). For each active site, a selectivity index (SI) was calculated as the ratio between the normalized CMAP amplitude for that muscle, CMAPni, and the sum of the normalized CMAP amplitudes elicited in the three muscles, CMAPnj (Veraart et al 1993), as (1):

$$\begin{eqnarray}&&{\rm S}{{{\rm I}}_{i,j}}=\frac{{\rm CMAP}ni}{\sum \limits_{j}^{m} {\rm CMAP}nj}.\end{eqnarray} \tag{ 1 }$$

This index ranges from 0 (no activation of the target muscle) to 1 (activation of only the target muscle). Maximum selectivity obtained for each active site for each muscle was calculated for both electrode designs (SELINE and TIME), and the difference between maximum selectivities plotted. Mean values of maximum selectivity, corresponding stimulating current and CMAP amplitudes, as well as threshold current of activation, were compared for SELINE and TIME.

To assess the anchoring effect of SELINE versus TIME design and thus the stability of the implant, both types of electrodes were tested in two different positions: P0 and P1. The imposition of this transversal displacement to the implanted electrode aimed to simulate eventual slippages inside the nerve when external motion occurs. The reproducibility of the two positions was ensured by considering the gold strips marked on the loop of the electrode after implantation. In the P0 position the three gold strips were located just outside the nerve. The P1 position was obtained by slightly pulling the electrode in the opposite sense of the implant, so that only two gold strips were outside the nerve (figure 4(C)). This displacement is equal to 200 μm since the retracted distance was controlled by the 100 μm separated gold printed strips on the electrode ribbon. At P1 position the wings were expected to be more open in the tissue. Once the electrode was repositioned, stimulation through each active site was performed again.

The in vivo stability was evaluated by considering two pivotal aspects of fascicular stimulation: change of threshold current required for activation of each muscle, and change of selectivity for each active site and for each muscle. Two stability indices were introduced: threshold stability index, ST_th and selectivity stability index, ST_sel.

The threshold stability index ST_th considers the change of threshold current from P0 to P1 for each active site and each muscle, and it is described as (2):

$$\begin{eqnarray}&&{\rm S}{{{\rm T}}_{{\rm th}}}=1-\frac{\left| {\rm Cur}{{{\rm r}}_{{\rm tP}{\bf 0}}}-{\rm Cur}{{{\rm r}}_{{\rm tP}{\bf 1}}} \right|}{{\rm Cur}{{{\rm r}}_{{\rm max} }}},\end{eqnarray} \tag{ 2 }$$

where Curr_tP0 and Curr_tP1 are threshold currents at position P0 and P1 respectively for each active site and each muscle, Curr_max is the maximum stimulating current used in the protocol (300 s the maximum stimulating current us stability over imposed movement) and 1 (high stability). In the case a given muscle is activated in one position and not activated in the other position, the stability index was 0.

The selectivity stability index ST_sel considers the change of maximum selectivity from P0 to P1 for each active site and each muscle and it is described as (3):

$$\begin{eqnarray}&&{\rm S}{{{\rm T}}_{{\rm sel}}}=1-\left| {\rm S}{{{\rm I}}_{{\rm max} {\rm P}{\bf 0}}}-{\rm S}{{{\rm I}}_{{\rm max} {\rm P}{\bf 1}}} \right|,\end{eqnarray} \tag{ 3 }$$

where SI_maxP0 is the maximum selectivity for each active site and each muscle at position P0, and SI_maxP1 is the maximum selectivity for each active site and each muscle at position P1. This value varies between 0 (low stability for selectivity) and 1 (high stability for selectivity). Kolmogorv–Smirnov test was used to statistically evaluate the differences in indices between the two types of electrodes (SELINE and TIME).

For the study of biocompatibility and safety of chronic in vivo implants, six devices comprised of the polyimide and metal contacts of each SELINE with memorization, SELINE without memorization and TIME, were obliquely implanted, as described in 2.3.2, in the sciatic nerve of adult rats. The sensitive portion of the device was inserted within the tibial and peroneal fascicles, leaving the ribbon part of the device outside and parallel to the nerve (figure 4(A)). Then, the wound was sutured by planes, so that no parts of the electrode were subcutaneous or external to the animal. The PCB connector and wires were not implanted in order to directly assess the effect of the presence of the wings on the nerve, avoiding tethering forces on the electrode. After insertion, the SELINEs were retracted to allow the wings anchoring to the tissue (figure 4(B)). All the animals were followed for 90 d to obtain evidence of possible functional and histological damage induced on the nerve by the implanted electrode. At 10, 30, 60 and 90 d post-implantation, electrophysiological and functional properties of the implanted nerve were evaluated by means of nerve conduction, mechanical algesimetry and walking track tests. The intact contralateral sciatic nerve was used as control.

For nerve conduction tests, rats were anesthetized and the sciatic nerve was stimulated with single electrical pulses (200 μs duration at supramaximal intensity) delivered by monopolar needles placed at the sciatic notch. The CMAP of TA, GM and PL muscles were recorded with monopolar needles inserted into the muscle and recorded with an EMG machine. The latency and amplitude of the CMAPs were measured for each implanted and control nerve (Badia et al 2011a).

The walking track test was carried out to assess locomotion function. The plantar surface of the rat hindpaws was painted with ink and the rat was left to walk along a corridor to a dark cage at one end. Footprints corresponding to the implanted and intact paws were identified, and measurements made to calculate the sciatic functional index (SFI) (De Medinaceli et al 1982) to quantify changes in walking pattern.

The nociceptive responses to mechanical stimuli were evaluated by means of an electronic Von Frey algesimeter (Bioseb). Rats were placed on a wire net platform in plastic chambers. Mechanical nociceptive threshold was measured as the force (in grams) at which rats withdrew the paw in response to the stimulus. The mean of three tests was calculated and expressed as the percentage between the operated and the intact hindlimb of each animal (Casals-Dìaz et al 2009).

Three months after implantation, the sciatic nerves were harvested to evaluate morphological changes in the nerve. Rats were deeply anesthetized and perfused with 4% paraformaldehyde in PBS. The sciatic nerve segment including the electrode implant was harvested and postfixed in the same perfusion solution for 4 h. Then samples were washed and embedded in paraffin. Transverse sections (10 μm thick) were cut, mounted in silane-coated slides, and viewed under light microscopy to assess the electrode–nerve interaction. A nerve segment from 2 mm distal to the electrode implant was postfixed in 3% glutaraldehyde-3% paraformaldehyde in buffer solution, and processed for embedding in epon. Semithin 0.5 μm thick sections were stained with toluidine blue and examined. Measurement of cross-section area of the distal segment of the sciatic nerve, and counts of the number of myelinated nerve fibers were carried out using Image J software (Badia et al 2011a) to assess potential damage leading to axonal degeneration.

Data is presented as mean ± SD or SEM. Results were statistically analyzed by using GraphPad Prism software. One and two-way ANOVA followed by Bonferroni's post hoc test for comparison between groups were used when appropriate. Statistical significance was considered when P values were <0.05.

Electrodes were peeled off from the silicon wafer without breaks around the wings by exploiting the low adhesion forces between polyimide and silicon (figure 2(b)). Plastic deformation of the wings allows the memorization of a curved profile in the active configuration. Figure 2(a) shows the final device, whereas panels (b) and (c) show the successful memorization of wing profiles. For in vivo tests, wings have been memorized to form an angle of 30° with the horizontal plane represented by the main shaft of the electrode.

Results of electrical and electrochemical characterizations are shown in figure 3(E). Mean Ohmic resistance of the different active sites was between 94 and 105 Ω. These results are in compliance with theoretical evaluation as illustrated in (Cutrone et al 2013). Typical impedance values at 1 kHz were around 90 kΩ.

Regarding mechanical characterization, yield and fracture points were plotted: the maximum force that caused breakage of the electrode (mean ± SD) averaged 2.14 ± 0.23 N (figure 3(B)). Since the value of electrode insertion forces inside nerves is usually in the order of 10 mN (Jensen et al 2007), it is demonstrated that SELINE tolerates well the insertion procedure, avoiding breakage of the wings and of the shaft.

Typical extraction curves for the three electrode types are represented in figure 3(C). Electrodes with wings (memorized and non-memorized) show two peak forces at the beginning of the curve corresponding to the extraction of the two sets of wings. No remarkable peaks were found in the extraction of electrodes without wings. The comparison between extraction forces confirms the efficacy of the anchoring system. Maximum extraction force (figure 3(D) for SELINE with memorized wings was 10 times higher than for TIME (44 mN versus 4 mN). Statistical differences were found between SELINEs and TIMEs (p = 0.0083), SELINEs with non-memorized wings and TIMEs (p = 0.04), but not between SELINE and SELINE with non-memorized wings (p = 0.287).

Stimulation tests after acute implantation proved that the two types of electrodes (SELINE and TIME) allowed excitation of different axonal subpopulations of the sciatic nerve, thus selectively activating PL, GM and TA muscles through different active sites. At low intensity stimulation, selective activation of one muscle could be detected, due to the close contact of one electrode site with the corresponding muscular nerve fascicle. By increasing the stimulus intensity, the CMAP amplitude increased to reach the maximal value (i.e., activation of the whole muscle); however, the stimulus spread to other surrounding fascicles activating two or all three of the tested muscles. Figure 5(A) shows an example of recruitment curve obtained with a SELINE that produces selective enrollment of the three muscles.

Zoom In Zoom Out Reset image size

For the two types of electrodes, SELINE and TIME, the minimum current of stimulation that elicited a CMAP of at least 5% of the maximal amplitude (threshold current) ranged between 20 and 200 μA with pulses of 10 μs duration. There were no significant differences between threshold mean values of SELINE compared to the TIME (see figure 5(F)).

Differences in selectivity between SELINE and TIME for each active site and muscle are shown in figures 5(B)–(C). For PL muscle, higher selectivity was achieved with TIME (six active sites more selective than SELINE); for TA muscle, similar selectivity was achieved using both electrodes, while for GM muscle, higher selectivity was achieved with SELINE on all active sites. This is confirmed also in figure 5(F) where mean values of maximum selectivity for both electrodes are summarized.

Stability indices (ST_th and ST_Sel) were calculated by comparing results obtained from positions P0 to P1 for SELINE and TIME implants. Both stability indices presented significant differences between the two electrodes for each muscle. Mean values and SEM for stability indices are shown in figures 5(D) and (E) respectively. The threshold stability index ST_th in SELINEs was 4.8 times higher than in TIMEs for PL and 3.8 times higher for GM muscles, whereas for TA muscle it was 2.7 times higher. The selectivity stability index ST_sel for SELINEs was 1.7 times higher than for TIMEs for PL, 1.5 times higher for GM muscles and 1.2 for the TA muscle.

During the three months of implant, there were no significant changes in the amplitude and latency of the CMAPs recorded during nerve conduction tests performed in animals with SELINE and TIME electrodes, and values were comparable to control ones (figures 6(A)–(D)).

Zoom In Zoom Out Reset image size

Walking track measurements did not show variations between the groups with the three tested electrodes from 10 d after the implant until 90 d. The SFI averaged between −20 and 0 during follow-up (figure 6(E)). Slightly low values observed in some rats at 10 and 30 d after implant are attributable to the surgical procedure itself since values returned to normal levels in the three groups at 30 d. The algesimetry test yielded similar values of pain threshold for withdrawal between the implanted and the contralateral sides (figure 6(F)), without evidence of hyperalgesia that might had been induced by nerve compression or injury.

The electrodes were found in place within the sciatic nerve in all studied animals and no abnormalities were observed during dissection. Light microscopy observations of nerve sections show that the electrode strip crossed the sciatic nerve fascicles and wings of memorized SELINEs were seen open inside the nerve (figures 7(A)–(B)); in contrast there was no consistent observation of open wings in non-memorized SELINEs, indicating that the memorization process is important for the chronic stability of the wings positioning. Transverse sections of the nerve segment distal to the insertion site showed that the fascicular organization and microarchitecture of the implanted nerves was similar to control nerves, and there were no signs of axonal degeneration (figures 7(C)–(H)). The number of myelinated fibers counted in the distal level for tibial and peroneal branches of the sciatic nerve were similar in the three groups of rats with SELINE, SELINE without memorization and TIME implanted, without significant differences with respect to intact sciatic nerves (figure 8). Control and TIME values are in accordance to those previously published (Badia et al 2011a), and from these data it can be concluded that the SELINE design did not cause damage to the implanted nerve.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size