Benefits of a Skull‐Interfaced Flexible and Implantable Multilight Emitting Diode Array for Photobiomodulation in Ischemic Stroke

Abstract Photobiomodulation (PBM) has received attention due to its potential for improving tissue function and enhancing regeneration in stroke. A lightweight, compact, and simple system of miniaturized electronic devices consisting of packaged light‐emitting diodes (LEDs) that incorporates a flexible substrate for in vivo brain PBM in a mouse model is developed. Using this device platform, the preventive and therapeutic effects of PBM affixed to the exposed skull of mice in the photothrombosis and middle cerebral artery occlusion stroke model are evaluated. Among the wavelength range of 630, 850, and 940 nm LED array, the PBM with 630‐nm LED array is proved to be the most effective for reducing the infarction volume and neurological impairment after ischemic stroke. Moreover, the PBM with 630 nm LED array remarkably improves the capability of spatial learning and memory in the chronic poststroke phase, attenuates AIM2 inflammasome activation and inflammasome‐mediated pyroptosis, and modulates microglial polarization in the hippocampus and cortex 7 days following ischemic stroke. Thus, PBM may prevent tissue and functional damage in acute ischemic injury, thereby attenuating the development of cognitive impairment after stroke.


Introduction
Interest in tissue-electronic interfaces has increased recently due to a variety of potential applications including light-based Here, we demonstrate an implantable PBM system based on multi-LED arrays that are stably in contact with the skull over a targeted cortical area, which enabled the study of the long-term therapeutic effects. Taking the key advantages of the existing technology into account, we performed an in vivo mouse experiment to evaluate the efficacy of the LLLT for minimizing acute ischemic brain damage and chronic cognitive impairment after focal cerebral ischemia. The long-term therapeutic effects of PBM on various disease models could be evaluated with suitable device fixation and the precise shining of light on brain lesions due to the advantageous tissue-adhesive optoelectronic devices implemented on the flexible printed circuit board (FPCB) with optimal wavelength. Such capabilities would enable examination of biological mechanisms in model animals, a major goal of modern neurocognitive behavior research. The major contribution of brain PBM therapy may be associated with various biological processes, such as increasing the metabolic capacity of neurons and stimulating anti-inflammatory, anti-apoptotic, and antioxidant responses, as well as with neurogenesis and synaptogenesis. [30,31] Neuroinflammation is one of the crucial pathophysiological features of ischemic stroke and aging, which likely have a detrimental effect on cognitive function after stroke. [32][33][34] In recent studies, we found that absent in melanoma 2 (AIM2) inflammasomemediated inflammation and pyroptosis contributed to chronic poststroke cognitive impairment (PSCI), and we reported beneficial pre-conditioning effects of brain PBM therapy for the reduction of cortical tumor necrosing factor-and interleukin (IL)-1 expression at 24 h postischemia. [29,35] Furthermore, brain PBM in a mouse photothrombotic stroke model reduced IL-1 and IL-18 levels at 72 h post stroke via inhibition of the NLRP3 inflammasome signaling pathway. [26] Therefore, it is expected that brain PBM using an implantable optoelectronic device could attenuate the development of PSCI by controlling the inflammatory response. In addition, we first examined the preventive and therapeutic effects of brain PBM on infarct volume and performed behavioral assessments of neurological and motor functions after ischemic brain injury. Further, we also evaluated whether brain PBM therapy in the acute phase of stroke attenuates the development of PSCI by controlling AIM2 inflammasome activation and the inflammatory response. The aim of this study was to demonstrate the robustness of this miniaturized optoelectronic device implemented via brain implants and its practical utility for PBM therapy in live animals. The experimental design of the implantable multi-LED device, conformally covering a local brain area for a mouse model study. The right image shows a fully implanted multi-LED device except for interconnection leads. c) Schematic illustration of the LED circuit array placed outside the skull, above the brain. d) The relative intensity of three different wavelengths (i.e., 630, 850, and 940 nm), generated from the multi-LED arrays used in the in vivo experiments. e) Digital images of mice skulls for the measurement of the transcranial transmission to determine the extent of LED light penetration across the skull barrier, and the measurement laser setup. f) The transmittance data measured from hydrated/dehydrated skull samples. g) Digital images of the illuminating six-LED cores of the device, and the angledependent spatial distribution of light sources. h) The measured surface power density of multi-LED array with or without PDMS encapsulant. i) The reliability test of PDMS encapsulant for the multi-LED array.
To secure the flexible multi-LED array on the skull, a thin layer of biomedical glue (≈1.3 μm) was used to prevent unwanted detachment in repeated attempts by the mouse to remove it with its paws upon sensing a foreign body. [36] In our PBM system, three different wavelengths were carefully selected and generated from the LED circuit, including 630, 850, and 940 nm, to evaluate the potential efficacy of the therapeutic approach (Figure 1d). [37] As the improvement effects of the PBM may be dependent on the quantity of LED light that crosses the skull barrier, transcranial transmission was investigated to determine the extent of LED light penetration across the skull barrier using the laser setup and photodetector diagrammatically represented in Figure 1e. [25] In this experiment, skulls were dissected, and comparisons were performed between hydrated and dehydrated skulls (thickness ≈ 0.26-0.27 mm) to evaluate light penetration in an ambient condition with a power density of ≈17 mW cm −2 (Table S1, Supporting  Information). At all wavelengths, the hydrated state allowed for higher penetration of laser radiation due to the existence of water. A statistically significant decrease in penetration (≈41%) was detected for the wavelength of 630 nm, as shown in Figure 1f. When the multi-LED array was tested with conformal contact with the skull, the apparent penetration profiles differed between the hydrated and dehydrated states, as shown in Figure 1g. Extensive diffusion and scattering patterns were observed in-andaround the centers of the illuminated six-LED cores.
As the light passes through a heterogeneous and opaque environment such as living tissues, the trajectory of the incident light is dynamically deviated by the barrier component. Thus, only a partial fraction of light waves propagating inside biological tissues can reach the desired target area, while most are scattered and randomly diffused. [38] In this regard, the hydrated skull serves to diffuse the light, attenuating it, but effectively covering the light (i.e., light diffuser) over a target area. Different diffused areas are shown in the photograph (top panel in Figure 1g). Moreover, the spatial directivity of LED radiation intensity can be estimated by applying a multi-LED directivity pattern in wide-angle responses defined by the average from adjacent individual LEDs, as schematically represented in the bottom panel in Figure 1g. [39] The surface power density was measured by a power meter (PM100D; Thorlabs, Newton, NJ, USA) with a skull-mounted multi-LED array, as featured in Figure 1h. In a view of the real application, this measurement can be a practical guide for the light-dose per treatment area. The gradual attenuation of light penetration was naturally expected because of the hydrated skull barrier and an encapsulant (i.e., transparent elastomer, polydimethylsiloxane, PDMS: film thickness = 700 μm) wrapping the electric circuit. [40] Thus, the density of power delivered to the cortical surface would be decreased with respect to the series of compartmentalized barriers prior to producing a biological effect. Additional barriers such as the encapsulant (i.e., PDMS) and the skull affected the overall density of the performance of the multi-LED array in the far-field range ( Figure 1h). The originally manufactured multi-LED array exhibited a surface power density of ≈22 mW cm −2 when conformally contacted with the power meter, and appeared separately as ≈17 mW cm −2 in the encapsulated condition ( Figure S1, Supporting Information). Although the PDMS packaging process decreased overall LED efficiency by ≈22.7%, encapsulation was inevitable for implantation. On the other hand, the hydrated skull barrier distance ranged ≈0.26-0.27 mm and resulted in a power density of ≈10 mW cm −2 , which has been reported as a beneficial range of power density in thin-skull-based PBM animal studies (i.e., ≈10-25 mW cm −2 ), leading to an optimized trial design in our experimental approach. In addition, to test the degree of the chemical and water-proof properties of the multi-LED array in the biological environment (i.e., hydrated condition), the device was placed in phosphate-buffered saline (PBS) buffer solution (pH = 7.4) at 37°C for 7 days as demonstrated in Figure 1i. [3] The absorbance was continuously measured using ultraviolet-visible spectroscopy to detect possible dissociated molecules from the encapsulated LED device affected by the changes in the pH of the PBS buffer solution. PDMS encapsulation exhibited superior chemical stability; no changes were found, as shown in the graph. Clearly, the inset photograph of the luminous multi-LED array soaking in the PBS buffer solution is presented in the inset after 7 days, delivering the same power density of ≈10 mW cm −2 , without degradation in its performance.  Figure S2 in the Supporting Information). Main LED components on the flexible substrate were assembled by surfacemount technology (SMT), which provided easy access to realize our concept without the limit in designing an electrically and mechanically reliable PBM module. [41,42] The first basic embodiment adopts through-hole via based electric interconnections in the shape of a vertically penetrating configuration, built in a thin and flexible substrate of 25-μm-thick PI. The relatively wide viahole geometry of the FPCB provides simultaneous electrical input channels through sequential first and second metallization to an array of the LED components that can only be contacted on the top electrodes to avoid a complex interconnection design, as shown in Figure 2a. This is the main scheme in the architecture of the integrated LED circuit, which can enable miniaturization of the PBM module by separating the crowded passive electrical paths to the opposite side in a single layer circuit board. [43] Compared with the designs of other optical units assembled in a planar configuration, the slightly changed form factor could reduce the actual size of the PBM module; this allows for some marginal bending, precisely fitting it on the target size of the mouse brain. At the initial stage of the process, with the double-sided thin metal films (1-μm-thick Cu) provided on the PI board, conventional photolithography was used to separately define the top contact pads for LEDs and the lower electrical interconnects by a precise mask-alignment. Next, individual micropackaged LEDs (Wurth Electronics, Niedernhall, Germany) were manually placed on the contact pads joined with the solder paste (lead-free solder alloy) as described in Figure 2b. Subsequently, a reflow process was performed in an oven at eutectic temperature (≈250°C) to induce the self-alignment by complemental wetting and to create permanent solder joints. After the SMT process of the LED components, a thin overlay (12 μm, PI) conformed and bonded to the top and bottom plane in a multilayer format to protect the exposed surface of the electrodes on the circuit. Moreover, this lay-up process could provide some additional physical capabilities, such as electrical insulation, bending resistance, and heat spreading, including the length tailoring in the exposed electrical ends to the connector ( Figure 2c). Next, a water-proof encapsulant of transparent PDMS elastomer (modulus of ≈100 kPa) was applied on the upper side of the surface-mount LEDs to allow for a chemically stable operation in a biological environment by preventing possible dissociation from the solder joints. [42] This final step also has the extra function of filling the bumps and flattening the uneven corrugated LED surface. When contacting the LEDside on the mouse skull, the surface flatness can be one of the key elements for conformity and favorable tackiness. Since previous studies of the constituent materials (PI and PDMS) have revealed no evidence of toxicity, chemical decoupling from the provided support-base would represent chronic stability to study PBM in the implanted area using the prepared optical stimulation array. In a well-established manufacturing process with clear functionality, the lab-scale mass production of mini-PBM modules is suitable for a variety of mouse model studies and other related research in an implantable device format, as demonstrated in Figure 2d (≈100 pieces of modules). The mechanical durability of the multi-LED circuit was simply tested by bending and folding in the physically harsh conditions, associated with the flexible format that well-matched with the soft brain tissue (the right panel in Figure 2d). While the encapsulated LED illuminates a field of light centered on 6-LED sources (Figure 2e), the maximum surface temperature rises slowly over a period of 10 min in the range of ≈34-35.8°C during surgery ( Figure S3, Supporting Information). This excellent thermal stability of the packaged multi-LED array can be attributed to the effective intrinsically embedded heat sinks (i.e., interlayer and overlay of PI, double stacked Cu electrodes, Figure S4a, Supporting Information) and the heat dissipation through the top-coated PDMS layer (Figure 2f, side view of the detailed structure of the monolithic LED component, a total thickness of ≈737 μm). The typical diode characteristic of the LED was measured under the repetitive bending cycle conditions to confirm its reliability in a flexible-shaped configuration, as shown in Figure 2g. Current in-put and output voltages were evaluated under a bending and slight stretching state as presented in Figure 2h, where the connected leads (length = 300 mm) with a connector (Molex, Lisle, IL, USA) were firmly plugged, as shown in Figure 2e and Figure S4b-d (Supporting Information). Moreover, the extreme case was also tested revealing the load resistance as show in Figure 2i. As presented, owing to the mechanical compliance and advanced form factor, our multi-LED array exhibited a significant reduction of the burden of processing with the viable combination of the soft materials and rigid electronic components. [41] Figure 3. Experimental schematic diagram of the preventive and therapeutic effect of various wavelengths of brain PBM using multi-LED array on the infarct volume and neurological and motor function after ischemic stroke. a) A representative photothrombotic stroke mouse model and the position of the multi-LED array. b-e) Experimental procedures for brain PBM treatment and functional outcome tests after ischemic stroke. Mice receive three different wavelengths of 630, 850, or 940 nm (≈17 mW cm −2 , 20 min), twice a day for 3 days, prior to or after the induction of ischemic insult. c,f) Representative photographs of coronal brain sections stained with TTC. d-g) Quantification of direct infarct volume and neurological score and wiregrip results are evaluated to assess recovery of neurologic deficit and vestibular motor function after ischemic brain injury. All data are represented as mean ± SEM. N = 7-11 each. Statistical significance was determined by one-way ANOVA with Tukey post hoc test. *p < 0.05, **p < 0.01, and ***p < 0.001 versus control group (Con).

Brain PBM Assessment of Acute Brain Damage Following an Ischemic Stroke
The fabricated multi-LED array was applied to a stroke mouse model for brain PBM therapy. We first examined the preventive effect of brain PBM using a multi-LED array on the tissue and the functional outcomes following focal cerebral ischemia (Figure 3). In this study, we adopted photothrombotic occlusion of cerebral microvessels in mice, which is based on a photochemical reaction triggered by systemic administration of Rose Bengal and focal illumination of the skull. Illumination leads to local activation of Rose Bengal, which results in free radical formation, disturbance of endothelial function, and local thrombotic occlusion of small cortical vessels (Figure 3a). The illuminated area occupied the sensorimotor cortex which indicated that the lesion would largely destroy the forelimb cortex and hence produce a deficit in the use of the forepaws. Mice underwent PBM therapy at different wavelengths of 630, 850, or 940 nm (≈17 mW cm −2 , 20 min, twice a day for 3 days) prior to the induction of ischemia as well as 1 h prior to the procedure (Figure 3b). Thus, the outcome of the acute injury was evaluated on day 1 after focal cerebral ischemia. Notably, as shown in Figure 3c,d, triphenyl tetrazolium chloride (TTC) staining revealed that PBM therapy employing an LED array with a 630 nm wavelength significantly reduced direct infarct volume, compared to the control cases, which was measured to be 30.6 ± 5.0 and 62.4 ± 5.0 mm 3 for PBM at wavelengths of 630 nm and control, respectively (p < 0.05 (p = 0.046)), whereas other PBM therapy with light sources of 850 and 940 nm had no significant effect. In addition, the mice treated with 630 nm PBM therapy showed significant improvement in neurologic and motor functions, suggesting that the reduced infarct volumes translated into better functional outcomes in the PBM-treated group at wavelengths of 630 nm (Figure 3c,d).
Subsequently, we evaluated whether postischemic treatment of brain PBM with a multi-LED array exerts a therapeutic effect on acute ischemic brain injury. The mice underwent PBM Figure 4. Effect of 630 nm brain PBM using multi-LED array on brain cells following ischemic brain injury. Mice were pretreated with 630 nm PBM (≈17 mW cm −2 , 20 min) twice a day for 3 days prior to the induction of ischemic insult. a,b) Representative photographs and quantification graphs of NeuN (red, neuronal marker) and CD31 (green, endothelial cell marker). c,d) Representative photographs and quantification graphs of GFAP (red, astrocyte marker) and Iba-1 (green, microglia marker) at the peri-infarct regions. DAPI is labeled with blue fluorescence. Scale bar = 50 μm. All data are represented as mean ± SEM. N = 4 each. Statistical significance was determined by unpaired, two-tailed Student's t-tests. *p < 0.05 and ***p < 0.001 versus control group (Con). Abbreviations: LED, light-emitting diode; PBM, photobiomodulation; GFAP, astrocyte marker; DAPI, 4′,6-Diamidino-2-phenylindole; NeuN, neuronal marker; CD31, endothelial cell marker; Iba-1, microglia marker; SEM, standard error of the mean. therapy (20 min, twice a day for 3 days), commencing at 4 h after the ischemic insult (Figure 3e). The PBM therapy at the wavelengths of 630 and 940 nm critically improved neurologic and motor function, compared to the control cases when measured 72 h after ischemic brain injury. However, the infarct volume was not reduced by the brain PBM, as presented in Figure 3f,g. The above combined results indicate that the ipsilateral cortical application of PBM, specifically with a wavelength of 630 nm, had not only a preventive effect on the tissue and the functional outcomes but also a therapeutic effect on the functional recovery after ischemic brain damage.

Brain PBM Assessment for Glial Cell Activation After Ischemic Brain Injury
To evaluate the effectiveness of the PBM therapy, the brain cells in the ischemic cortex were examined by the measurement of the expression levels of the immunostained markers, such as NeuN (neuronal cell), CD31 (endothelial cell), GFAP (astrocyte), and Iba-1 (microglia), at the peri-infarct regions using fluorescence microscopy as presented in Figure 4. The results indicate that neurons were not affected by the brain PBM at a wavelength of 630 nm, whereas the numbers of CD31 + cells were significantly increased in the PBM-treated cases, compared to the con-trol groups (Figure 4a,b), which implies that brain PBM treated with a 630 nm wavelength helped to preserve blood vessels in the ischemic area. Moreover, numerous astrocytes and microglia in the ischemic cortex were significantly decreased in the PBMtreated groups (Figure 4c,d). To evaluate the level of effectiveness of PBM treatment, the same experiment was performed on the normal brain without any stroke. Interestingly, neurons and astrocytes were not affected by the brain PBM, but the numbers of CD31 + and Iba-1 + cells were slightly affected by the brain PBM at the wavelength of 630 nm, as presented in Figure S5 in the Supporting Information. These results suggest that brain PBM decisively attenuates glial cell activation in the ischemic brain, whereas it increases blood vessels and decreases microglia activation even in the normal brain.

Brain PBM Assessment of PSCI Using a Multi-LED Array
Using a multi-LED array, we extended the brain PBM approach to evaluate the improvement of the cognitive function in the chronic phase poststroke. The mice were subjected to middle cerebral artery occlusion (MCAO)/reperfusion for 45 min to produce consistent selective lesions in the cortex, striatum, and hippocampus, resulting in low mortality, as described in Figure 5a,b. Because the hippocampus largely contributes to Figure 5. Effect of 630 nm brain PBM using multi-LED array on cognitive dysfunction following ischemic brain injury. a) Cartoon of the MCAO/reperfusion mouse model. b) The microscopic appearance of a silicon-coated 7-0 monofilament insertion from the CCA to the ICA to occlude the MCA. c) Schematic diagram of experimental PBM procedure using a multi-LED array in a PSCI model. Mice are treated with 630-nm PBM (≈17 mW cm −2 , 20 min) once a day for 7 days from next day after the ischemic insult. Histological analysis is performed at 7 days after MCAO and the cognitive functional tests are performed at 28 days after MCAO. d) Anxiety is measured using the elevated plus maze test. e) Total time spent in open arms and f) total traveled distance in the elevated plus maze for 10 min. g) Spatial learning and memory are assessed using Morris water maze test. On days 2-6, the mouse is placed in an opaque water-filled pool (hidden platform) for learning trial tests, and on day 7, the mouse is allowed to swim freely without the platform for 90 s for the probe trial test. h) Average latency to find the visible platform on day 1 and hidden platform in the target quadrant on days 2-6 as a learning trial. i) Representative swimming traces during the probe trial session. j) Time spent in the target quadrant and k) target number of crossings where the hidden platform was previously placed during probe trial. All data are represented as mean ± SEM. N = 8-12 each. Statistical significance was determined via a one-way ANOVA with a e,f,g,k) Tukey post hoc test and a two-way ANOVA with a h) Tukey post hoc test. *p < 0.05 and ***p < 0.001 versus sham/control group; #p < 0.05 and ##p < 0.01 versus stroke/control group. www.advancedsciencenews.com www.advancedscience.com long-term memory deficit after stroke, the efficacy of brain PBM is clearly accessible. As shown in Figure 5c, the multi-LED array (630 nm) was implanted and periodic PBM therapy was performed daily for 7 days, and all behavioral tests were performed until 28 days after ischemic brain injury.
In this experiment, the elevated plus maze was used to study anxiety levels, as shown in Figure 5d and Figure S6b (Supporting Information). Mice generally prefer to stay in the closed arms, and they demonstrate fear of the height of the maze. However, the stroke/control mice expressed a significantly reduced anxiety-phenotype, which was characterized by the observation that they were spending more time in open arms, compared to the sham/control groups (1.56 ± 0.69% vs 4.92 ± 1% min in the sham/control and stroke/control groups, respectively, p < 0.05 [p = 0.023]; Figure 5e); this phenotype was not reversed by PBM treatment. In addition, no other significant differences in the total traveled distance were observed among all groups as shown in Figure 5f, suggesting that the brain PBM does not affect the mice's mobility. We further demonstrated whether brain PBM using a 630-nm multi-LED array could improve spatial learning and memory deficits in the stroke model using the Morris water maze test (Figure 5g). In the trial for testing spatial learning ability, the escape latency was defined as the time required for the mice to swim to a submerged platform ( Figure S6c, Supporting Information). As shown in Figure 5h, stroke/control mice took longer to find the platform compared to the sham/control groups (p < 0.001), while the PBM-treated stroke mice demonstrated an apparently shorter latency to find the hidden platform (p < 0.01 [p = 0.008]). In the probe trial test on the following day (day 7), the stroke/control mice spent less time in the target zone (p < 0.05 [p = 0.023]; Figure 5i,j) and crossed to the target zone significantly fewer times (p < 0.05 [p = 0.026]; Figure 5k) after the removal of the platform. PBM-treated stroke mice showed a significantly increased time spent in the target zone (p < 0.05 [p = 0.016]) and crossed to the target significantly more frequently (p < 0.05 [p = 0.011]) than stroke/control mice (Figure 5i-k). These results indicate that brain PBM (i.e., 630-nm wavelength) performed in the acute phase of stroke improved the spatial learning and memory, but did not recover the anxiety behavior in the chronic phase of stroke.

Effects of Brain PBM on AIM2 Inflammasome Activation and Pyroptosis in Stroke Mice
Focusing on the specific wavelength of 630 nm, we investigated the beneficial effect of brain PBM based on the expression of AIM2, caspase-1, and gasdermin D (GSDMD), a pyroptosis marker, in the hippocampus and cortex. Compared to the sham/control group, a higher intensity of AIM2 + /caspase-1 + staining in the hippocampus and cortex was observed in the stroke/control group (7 days post stroke). Notably, as displayed in Figure 6a,b, consistent with the results of the cognitive function tests, PBM-treated stroke mice showed an apparent reduction of AIM2 + /caspase-1 + expression in the hippocampus and cortex (hippocampus: 452.82 ± 85.75% vs 634.41 ± 118.56%, stroke/PBM and stroke/control group, respectively; cortex: 938.96 ± 209.33% vs 4295.28 ± 634.35%, stroke/PBM and stroke/control group, respectively, p < 0.001). In addition, simi-lar results were observed with GSDMD staining. Stroke/control mice showed robust staining for GSDMD in both the hippocampus and cortex, which was significantly reduced in the PBMtreated group at 7 days post stroke compared to the stroke/control group (hippocampus: 184.81 ± 16.95% versus 275.12 ± 19.25%, stroke/PBM and stroke/control group, respectively, p < 0.01 (p = 0.007); cortex: 129.04 ± 18.41% versus 328.40 ± 30.13%, stroke/PBM and stroke/control, respectively, p < 0.01 (p = 0.001)), as presented in Figure 6c,d. Overall, the implanted multi-LED array equipped with a 630-nm wavelength was obviously effective in the acute phase of stroke by attenuating AIM2 inflammasome activation and mediating pyroptosis at the injured site at 7 days post stroke, as schematically demonstrated in Figure 6e.

Brain PBM-Derived Modulation of Microglia Polarization to an Anti-Inflammatory Phenotype in Stroke Mice
The next approach for the implantable brain PBM (i.e., 630 nm) was to examine microglia polarization in the ischemic mice. In this experiment, Iba-1 (microglia marker) was double-stained with CD86 (M1 surface marker) or CD206 (M2 surface marker) for the hippocampus and cortex. As presented, the visualized images substantiated clearly different intensity levels for the Iba-1 + , CD86+/Iba-1 + , and CD206 + /Iba-1 + cells in the stroke/control groups compared to the sham/control groups. The PBM treatment suppressed the expression of Iba-1 + (105.41 ± 6.10% vs 130.41 ± 8.24%, stroke/PBM and stroke/control, respectively, p < 0.05 (p = 0.012); Figure 7c). The double-stained cells showed different trends in the intensity levels for CD86 + /Iba-1 + (5960.95 ± 1466.21% vs 11 520.42 ± 1279.07%, stroke/PBM and stroke/control, respectively, p < 0.001, Figure 7a,d). It was associated with stroke/control mice, but had no effect on cortical CD206+/Iba-1+ (Figure 7b,e). The collective set of data suggested that reduced Iba-1 + microglia by brain PBM was the result of decreasing M1 phenotype microglia, not the changing M2 phenotype microglia in the cortex. Therefore, we inferred possible serial processes in the inflammatory system, as schematically depicted in Figure 8a. Finally, as shown in Figure 8b-e, the microglia polarization induced by brain PBM was confirmed by the increased production of IL-1 (i.e., proinflammatory cytokine) and reduced arginase-1 and IL-10 (i.e., anti-inflammatory cytokines), which were markedly reversed in the PBM-treated group at 7 days post stroke. The conclusive results demonstrated that effective PBM at a wavelength of 630 nm in the acute phase of stroke appropriately modulated microglia polarization to produce anti-inflammatory cytokines, which may facilitate long-term cognitive recovery.

Discussion
In this study, we developed a lightweight and compact system of miniaturized electronic devices that incorporates flexible substrate and uses a simple packaged LED design to conduct in vivo experiments of brain PBM for a small animal. Using this device platform, we evaluated the beneficial effects of brain PBM on tissue and functional recovery following ischemic stroke. Among the various wavelength ranges, the wavelength of 630 nm was Figure 6. Effect of 630-nm brain PBM using a multi-LED array on AIM2 inflammasome and pyroptosis in the hippocampus and cortex following ischemic brain injury. a,b) Representative photographs and quantification graphs of AIM2 + /Cas-1 + and c,d) GSDMD (pyroptosis marker). AIM2 and GSDMD were labeled with red fluorescence, Cas-1 was labeled with green fluorescence and DAPI was labeled with blue fluorescence. Quantification of AIM2 + /Cas-1 + and GSDMD fluorescence integrated optical density in the hippocampus and cortex. Scale bar = 50 μm. All data are represented as mean ± SEM. N = 3 each. Statistical significance was determined using a one-way ANOVA with a Tukey post hoc test. ***p < 0.001 versus sham/control group; ##p < 0.01 and ###p < 0.001 versus stroke/control group. e) Schematic illustration of the effect of 630 nm brain PBM on AIM2 inflammasome signaling pathway. Brain PBM may prevent the neuroinflammation and pyroptotic cell death via inhibiting the activation of AIM2 inflammasome and related signaling pathway.
the most effective in reducing the direct infarction volume and neurological and motor impairment after ischemic brain injury. In addition, the implanted brain PBM with a 630-nm LED array in the acute phase of stroke showed remarkably improved capability of spatial learning and memory in the chronic phase of stroke. Furthermore, PBM treatment with 630-nm LEDs attenuated AIM2 inflammasome activation and inflammasomemediated pyroptosis and modulated microglial polarization in the hippocampus and cortex 7 days following ischemic stroke. Therefore, the ipsilesional cortical application of the classified PBM device at a certain wavelength (i.e., 630 nm) prevented tissue and functional damage in the acute ischemic brain and obviously attenuated the development of cognitive impairment in poststroke chronic phase through the combinatorial regulation of AIM2 inflammasome-mediated pyroptosis and microglia polarization.
To date, extensive studies have suggested that brain PBM therapy may have positive effects on stroke lesions, and in most cases, the approaches have been focused on the scale of the wavelength, focused irradiation, energy density, and periodic duration. [44,45] As a more advanced and practical approach, detailed experimental parameters were chosen, and the relevant mechanisms were further elucidated by in vivo studies with LLLT using a multi-LED stimulator as follows. First, to determine the suitable wavelength for the prevention or treatment of ischemic stroke, mice in the in vivo studies were exposed to three different wavelengths (630, 850, and 940 nm) on a fixed frequency (i.e., 20 min, twice a day for 3 days, before or after ischemic brain injury), as presented in Figure 3b,c. Notably, among the prepared sets of multi-LEDs, PBM at a specific wavelength (i.e., 630 nm) represented the best preventive effect on the tissue and functional outcomes after ischemic brain damage (Figure 3c,d); furthermore, PBM with 630nm LLLT was the most therapeutically effective for functional recovery in stroke rehabilitation (Figure 3f,g). Recently, it has been reported that PBM with wavelengths in the ≈600 nm range has protective or therapeutic effects in rabbits, rats, and mice in stroke models. [26][27][28][29]46,47] However, pre-and post-treatment of PBM with 940-nm LLLT also showed functional improvement after ischemic stroke as presented in Figure 3, but the effect was weaker than that of 630-nm PBM. On the other hand, the 850-nm LLLT was not sufficiently effective in our model system although there have been several case studies with progressed results for stroke patients, based on 808-nm PBM treatment, as reported previously. [31] This is surprising by the fact that even slight wave-www.advancedsciencenews.com www.advancedscience.com length differences have unexpected results. As the most studied wavelength range for stroke research is 808-nm PBM, another evaluation using those wavelength ranges will be follow-up investigations related to appropriate light sources (i.e., micro LED chips) and irradiation parameters. Indeed, it is worth noting that the optimized optical power input of PBM with 630-nm-based LLLT facilitated improvements in ischemic brain deficit on the supplied current of 20 mA in the operation of multi-LED array for power density of ≈17 mW cm −2 , indicating the most remarkable improvement in the tissue and functional outcomes. For comparison, it was also conducted with other inputs, such as 10 or 30 mA, in the in vivo PBM processes (data not shown). Therefore, we chose to apply a multi-LED array at a 630-nm wavelength (≈17 mW cm −2 ) for the subsequent PBM experimental schemes. [48] As reported previously, the preservation of cerebrovascular blood vessels and reduction of neuroinflammation through glial cells are important recovery strategies after ischemic stroke. [49] To this end, the brain PBM therapy increases the cerebral blood flow and metabolic capacity of the neurons and promotes neuronal protection and survival through anti-inflammatory, antiapoptotic, and antioxidant responses. [26][27][28][29][30][31] Figure 4 shows that PBM with 630-nm LLLT increased endothelial cells and decreased astrocytes and microglia in stroke-damaged lesions. Our findings are consistent with a previous report on 610-nm-based PBM that showed protection of the blood-brain barrier with reduced neuroinflammation by decreasing astrocytes and microglia, and reducing the expression of inflammatory mediators after ischemic brain injury. [29] Overall, brain PBM using an implanted multi-LED array with a wavelength of 630 nm alleviates tissue damage, as well as minimizes functional deficit by preserving blood vessels and attenuating glial cell activation after focal cerebral ischemia.
We recently reported that AIM2 inflammasome-mediated inflammation and pyroptosis possibly contribute to PSCI. [35] Based on the results, we hypothesized that the PBM approach could improve PSCI with a proven protective effect on acute brain injury at a 630-nm wavelength. As expected, brain PBM in the acute phase of ischemic stroke induced beneficial effects on spatial learning and memory in the chronic phase of stroke, as presented in Figure 5. Moreover, 7 days after ischemic brain injury, activated-AIM2 inflammasome and pyroptosis were inhibited by our PBM treatment (Figure 6a-d). To date, the pronounced effect of brain PBM on improving cognitive function has been reported in several studies. For example, the PBM approach based on a wavelength of 610 nm at an early stage of disease onset has been shown to improve learning and spatial memory by increasing amyloid-degradation in an Alzheimer's disease animal model (i.e., 5XFAD mice). [50] In addition, reports suggest that PBM with 627 and 660 nm wavelengths improved spatial memory in an A 25-35 peptide-induced mouse model [51] and in 18-month-old BALB/c mice, [52] respectively. Similar to these results, our PBM with 630-nm LED arrays, especially in the acute phase of stroke, attenuated AIM2 inflammasome activation and mediated pyroptosis in the injured regions, improving spatial learning and memory.
Postischemic inflammation is a hallmark of ischemic stroke pathology that plays a critical role in acute brain damage and has profound implications in long-term recovery. [53] The inflammatory response is regulated by M1 and M2 polarization of microglia, and the failure of the M1/M2 phenotype balance in inflammation-associated injury can lead to chronic inflammation and secondary damage. [54] As presented in Figure 7, we observed that our PBM treatment with a 630-nm wavelength clearly reduced microglia (Iba-1 + ) in the ischemic cortex. A detailed observation of microglia suppressed by PBM revealed reduced M1-phenotype microglia (CD86 + /Iba-1 + ) without altering M2phenotype microglia (CD206 + /Iba-1 + ). Moreover, the PBM approach inhibited proinflammatory cytokine IL-1 and increased the anti-inflammatory cytokines, Arg-1 and IL-10 ( Figure 8). Similar to these results, previous studies showed that PBM treatment at 610-nm wavelength reduced IL-1 and IL-18 levels at 72 h post stroke; we also recently reported that AIM2 activation and other inflammasome signaling, including IL-1 , could be involved in the development of PSCI during the subacute phase post stroke. [26,35] Therefore, the collective set of results indicates that inhibiting AIM2 inflammasome-mediated proinflammatory response and promoting anti-inflammatory response using brain PBM with a wavelength of 630 nm may contribute to recovering cognitive impairment after ischemic stroke.

Conclusion
In conclusion, ipsilesional cortical application of brain PBM using an implantable multi-LED array may prevent tissue and functional damage in acute ischemic stroke injury and attenuate the development of cognitive impairment after stroke through regulation of AIM2 inflammasome-mediated pyroptosis and microglia polarization. Therefore, brain PBM therapy has the potential to develop into a safe and effective neuroprotective treatment for patients with acute stroke damage and PSCI. Based on this basic research, clinic or home-based PBM therapy using scalable manufacturing strategies for LED devices could be a promising strategy for neurorehabilitation in the future.

Multi-LED Array Preparation and Characterization:
The flexible substrate used in this study was PI that has sputtered Cu layers on both sides. Primary and secondary metallization was performed via holes. The initial size of the PI substrate was 10 × 10 cm, and the thickness of the PI and Cu patterns were 25 and 1 μm, respectively. A conventional photolithography/etching process was used to pattern the Cu layers. Next, packaged LED chips (Wurth Electronics, Niedernhall, Germany) were manually Figure 7. Effect of 630 nm brain PBM using multi-LED array on microglia polarization in the hippocampus and cortex following ischemic brain injury. a) Double immunostaining of Iba-1 (green) with CD86 (red) or b) CD206 (red) in the hippocampus and cortex. DAPI is labeled with blue fluorescence. Scale bar = 50 μm. c-e) Quantification of Iba-1 + , CD86 + /Iba-1 + , and CD206 + /Iba-1 + fluorescence integrated optical density in the hippocampus and cortex. All data are represented as mean ± SEM. N = 3 each. Statistical significance was determined using a one-way ANOVA with a Tukey post hoc test. **p < 0.01 and ***p < 0.001 versus sham/control group; #p < 0.05 and ###p < 0.001 versus stroke/control group. Figure 8. Effect of 630-nm brain PBM using a multi-LED array on proinflammatory and anti-inflammatory cytokine production following ischemic brain injury. Protein levels of ischemic brains are examined using western blot analysis the next day after the last PBM treatment. a) Schematic illustration of the effect of 630-nm brain PBM on microglia polarization. The resident microglia activate two types of polarization by stroke, and PBM treatment may inhibit the production of M1 phenotype microglia and reduce proinflammatory cytokine (IL-1 ). b) Representative western blot images of IL-1 and anti-inflammatory cytokines (Arg-1 and IL-10) in the ipsilateral lesion after stroke. c-e) Quantitative immunoblots of IL-1 , Arg-1, and IL-10. Internal protein is normalized to -tubulin. All data are represented as mean ± SEM. N = 3-4 each. Statistical significance was determined via a one-way ANOVA with a Tukey post hoc test. *p < 0.05 and ***p < 0.001 versus sham/control group; #p < 0.05 and ##p < 0.01 versus stroke/control group. placed on the contact pads after applying solder paste (i.e., Sn3.0Ag0.5Cu [Kester, Itasca, IL, USA]). The reflow process was set at a temperature of ≈250°C with a melt interval duration of ≈1 min. Then, the thin layers of an overlay (12 μm, PI with epoxy adhesive) were conformed and bonded to package the top/bottom planes, protecting the exposed surface of the electrodes on the circuit, except for the LED components. Finally, the individual multi-LED chips were separated into a desired design using an ultraviolet laser cutting machine. A water-proof PDMS encapsulant was then applied using a dip-coating process on the surface-mount LEDs (SYL-GARD 184; Dow, Midland, MI, USA; 10:1 mixing ratio of prepolymer and curing agent), which were cured in an oven at 80°C for 1 h. This encapsulation can protect the solder joints, resulting in a reliable evaluation before implantation. The performance of the LEDs and interconnected stabilities were measured by a semiconductor parameter analyzer (Keithley 4200A; Keithley, Beaverton, OR, USA), and the mechanical durability of the multi-LED array was repeatably tested by loading pressure in a folded state (CT3 analyzer; Brookfield, Middleboro, MA, USA).
Multi-LED Array Implantation: For the epicranial implant, the scalp and underlying tissues were removed, and the multi-LED array was implanted using general dental cement (DurelonTM; 3M ESPE, Seefeld, Germany). The multi-LED array was positioned over the sensorimotor cortex (Figure 1a-c; Figure S7, Supporting Information) of the ipsilesional region.
Animal Experiments and Experimental Procedures: Male C57BL/6 mice were purchased from Hana Biotech (Ansan, Korea). Mice were housed under a 12-h light/dark cycle and allowed ad libitum access to food and water. The animal protocol used in this study was reviewed and approved by the Pusan National University -Institutional Animal Care and Use Com-mittee (PNU-IACUC) as per their ethical procedures and scientific care (PNU-2019-2182, PNU-2019-2390). For the effect of brain PBM on acute brain injury, 630, 850, or 940 nm PBM was repeated twice a day before or after focal cerebral ischemia for 3 days over the ipsilateral cortical region (Figure 3b,e). Two types of brain PBM were used depending on whether it was for prevention or treatment. One type of PBM was performed twice a day from 3 days before ischemic brain injury to determine the preventive effect ( Figure 3b) and the other type was performed immediately after focal cerebral ischemia to assess the therapeutic effect of PBM (Figure 3e). For evaluating the effect of PBM on PSCI, brain PBM was performed once daily for 7 days over the ipsilateral cortical region (Figure 5c). PBM was performed using a 630-nm LED array, which emits ≈17 mW cm −2 light. PBM current was applied continuously for 20 min using a bench power supply (PST-3202; GW INSTEK, New Taipei City, Taiwan) under anesthesia. The same set of experimental conditions were provided to minimize the comparability issues with repetitive anesthesia, in the efficacy study of PBM. While the stimulated groups were anesthetized, the control groups were also equally anesthetized for 20 min without LED irradiation. Anesthesia was achieved by a nose cone-delivered isoflurane (1.5% maintenance in 80% N 2 O and 20% O 2 ). Rectal temperature was maintained at 36.5-35.5°C using a Panlab thermostatically controlled heating mat (Harvard Apparatus, Holliston, MA, USA).