Porous Silicon Nanoparticles Targeted to the Extracellular Matrix for Therapeutic Protein Delivery in Traumatic Brain Injury

Abstract

Traumatic brain injury (TBI) is a major cause of disability and death among children and young adults in the United States, yet there are currently no treatments that improve the long-term brain health of patients. One promising therapeutic for TBI is brain-derived neurotrophic factor (BDNF), a protein that promotes neurogenesis and neuron survival. However, outstanding challenges to the systemic delivery of BDNF are its instability in blood, poor transport into the brain, and short half-life in circulation and brain tissue. Here, BDNF is encapsulated into an engineered, biodegradable porous silicon nanoparticle (pSiNP) in order to deliver bioactive BDNF to injured brain tissue after TBI. The pSiNP carrier is modified with the targeting ligand CAQK, a peptide that binds to extracellular matrix components upregulated after TBI. The protein cargo retains bioactivity after release from the pSiNP carrier, and systemic administration of the CAQK-modified pSiNPs results in effective delivery of the protein cargo to injured brain regions in a mouse model of TBI. When administered after injury, the CAQK-targeted pSiNP delivery system for BDNF reduces lesion volumes compared to free BDNF, supporting the hypothesis that pSiNPs mediate therapeutic protein delivery after systemic administration to improve outcomes in TBI.

Graphical Abstract

INTRODUCTION

Traumatic brain injury (TBI) results in up to 2.8 million injury-related hospitalizations and deaths in the United States annually.¹ Currently available treatments for TBI are palliative and do not address the underlying cause of disease, leading to long-term physical, behavioral, and/or psychosocial impairments in the majority of survivors.^2,3 After the initial injury, a secondary injury progresses over a course of days to weeks and the pathophysiology involves a series of biochemical and cellular cascades, including reactive oxygen species (ROS) generation, inflammatory response, neurodegeneration, blood-brain barrier (BBB) breakdown, and cell death.⁴ While the primary injury can only be prevented, the secondary injury is an opportunity for therapeutic intervention in order to preserve brain tissue proximal to the primary injury.

Brain-derived neurotrophic factor (BDNF) is a promising neuroprotective therapeutic to mitigate the progressive deterioration of brain tissue that occurs during the secondary injury after TBI.⁵ BDNF is a protein secreted from neurons and glia that promotes neuronal survival, neural plasticity, and neurogenesis^6,7 and its cognate receptor is tropomyosin receptor kinase B (TrkB).⁸ TrkB activation initiates signaling cascades that regulate apoptosis, neuronal plasticity, and neurogenesis.⁹ The binding of BDNF to TrkB activates pro-survival pathways and enhances expression of anti-apoptotic proteins through the phosphoinositide 3-kinase and Akt signaling pathway in neurons.^5,10 Previous studies have demonstrated that sustained BDNF levels in the brain achieved through stem cells,^11,12 genetically engineered cells,^13-18 viral gene therapy, and direct brain infusions¹⁹ can protect brain tissue in the context of nervous system injury, neurodegenerative disease, and psychiatric disorders.⁵ In a meta-analysis of available clinical data, including cohort studies and randomized controlled trials, treatment of TBI patients with cerebrolysin, a porcine-derived peptide mixture that includes BDNF,²⁰ led to improved outcomes in functional tests such as the Glasgow Outcome Scale and modified Rankin Scale.²¹ The promise of BDNF as a potential therapeutic for neurodegenerative diseases is highlighted by the initiation of a first-in-human clinical trial to evaluate adeno-associated virus vector (AAV2)-based gene therapy of BDNF in Alzheimer’s Disease patients (NCT05040217).²²

While elevated BDNF expression, along with other neurotrophic factors, contributes to the functional preservation of injured tissue, a major barrier to the development of BDNF as a clinical therapeutic for TBI is successful delivery to the injured brain after systemic administration. Challenges to intravenous delivery of BDNF include short circulation half-life and restricted transport into the brain.^5,23,24 Nanoscale drug delivery vehicles are platform technologies with the potential to increase the bioavailability and stability of protein therapeutics; for example, vehicles can be engineered to target specific tissues to increase retention and labile cargos can be encapsulated within the interior of vehicles to protect against degradation in the biological milieu.^24,25 In addition, the pharmacokinetic profiles of nanoscale drug delivery vehicles can be tuned independent of their therapeutic cargo, allowing for more facile formulation of a variety of cargos. However, previous efforts to encapsulate BDNF into nanoparticles have resulted in low mass loading of less than 1%.^26-28

Porous silicon nanoparticles (pSiNPs) are an attractive candidate nano-carrier for therapeutics due to their tunable pore sizes and versatile loading chemistries, which allow for the accommodation of therapeutic cargo with a wide range of sizes and chemical properties, including small molecule drugs,^29,30 nucleic acids,^31-33 and peptides/proteins.^34,35 In particular, sequestration of protein-based biologics within the porous nanostructure of pSiNPs has been shown to protect them from degradation in the blood, leading to extended in vivo half-life and bioactivity.^29,32,34-44 In addition, pSiNPs have the advantage of linear degradation profiles, and thus can achieve linear drug release profiles, due to their anisotropic degradation mechanism.⁴⁵ This strategy has been successfully employed to encapsulate nerve growth factor (NGF) into pSiNPs, demonstrating the local release of growth factors from a polymeric scaffold.^34,38 In addition, Segal and Shefi et al. demonstrated neurotrophic factor release from nanoporous silicon microparticles after local delivery to the brain, achieved through the implantation of chips or biolistic bombardment using a pneumatic gene gun through an opening in the skull.⁴⁶ However, so far, protein delivery from pSiNPs after systemic delivery has yet to be demonstrated.

While a subset of TBI patients may undergo surgical intervention to remove blood clots and relieve intracranial pressure,⁴⁷ the brain is generally not readily accessible to therapeutics. Although intravenously delivered materials can accumulate in the injured brain through the damaged BBB that is a hallmark of TBI,^{48-51, 52} access to the brain is transient, with a majority of materials excluded from the brain 6 hours after injury as the BBB rapidly re-establishes.^33,48,49 This highlights the need for active targeting strategies, in order to increase penetration and retention of potential therapeutics. We recently identified a peptide ligand, CAQK, that binds to extracellular matrix (ECM) components upregulated after brain injury.³³ CAQK-modified pSiNPs were able to deliver siRNA to brains in a penetrating TBI model after intravenous administration and mediated significant silencing of a reporter protein compared to control targeted pSiNPs.³³ This approach exploited the transient damage to the BBB caused by the injury to allow passive accumulation of the nanoparticles into brain tissue where they selectively bound to, and were retained by, accessible brain ECM.

This work combines the above two approaches for protein loading and peptide-mediated targeting of ECM in the injured brain. Here we load BDNF into pSiNPs with a loading efficiency of 13% by mass of the pSiNP-protein construct, a substantial improvement in mass loading compared to encapsulation of BDNF by other systems (<1%),^26-28 and then modify the exterior of the nanoparticles with polyethylene glycol (PEG) and the brain injury-targeting peptide CAQK. We confirm that the BDNF retains its bioactivity after loading, we track the CAQK-modified pSiNPs into injured brain tissue after systemic administration in a mouse model of TBI, and we demonstrate a substantial reduction of brain lesion volumes relative to free BDNF or PBS controls. This work therefore represents the first time chemically-targeted nanoporous silicon has achieved the delivery of protein payloads after systemic administration and improved outcomes in an animal model of TBI.

RESULTS AND DISCUSSION

Loading of pSiNPs with Protein Cargo and Surface Modification with Targeting Peptide.

The pSiNPs used in this study were synthesized by electrochemical etching of mesopores into single crystalline silicon wafers, followed by ultrasonic fragmentation to form mesoporous nanoparticles. The electrochemical etch parameters (48% hydrofluoric acid, 46 mA/cm²) were set to achieve a nominal pore size in the range of 10-20 nm in diameter and an average size of ~130 nm in hydrodynamic diameter (Figure 1).⁴⁵ The pore diameter was chosen based on previous studies to be sufficient to accommodate protein cargos.⁴⁵ The proteins were loaded into the pores of pSiNPs using an oxidative trapping method (Figure 1a,S1),^34,38 which involves partial conversion of the Si skeleton in pSiNPs to SiO₂ by mild oxidation in deionized water. The SiO₂ shell then re-structures in the presence of the aqueous protein solution. This combination of conversion of Si to SiO₂ and aqueous restructuring of SiO₂ results in a volume expansion of the skeleton and sealing of the protein within a hydrated silicate framework, effectively trapping the protein and protecting it from degradation in the blood upon systemic delivery. We studied two protein cargoes in this work: the therapeutic protein BDNF, and the model protein lysozyme, which was labeled with Rhodamine B. Lysozyme has a molecular weight of 14.5 kDa and an isoelectric point of 11.35⁵³ and BDNF has a molecular weight of 13.5 kDa and an isoelectric point between 9 and 10,⁵⁴ making their size and charge properties similar at physiologic pH, which are important parameters to consider for pSiNP protein loading. Rhodamine B-labeled lysozyme provides an established functional readout of protein activity in release assays and allows tracking of the payload by image analysis in the biodistribution studies. BDNF was used in efficacy studies in SH-SY5Y cultures and the animal model of TBI. The SiO₂ shell in the pSiNPs is negatively charged at physiologic pH, with a zeta potential ranging from −24 to −28 mV,³² which enhanced the loading of the positively charged protein cargoes via electrostatic interactions (pI of lysozyme = 11.35⁵³ and pI of BDNF ~ 9-10⁵⁴).

Figure 1: Synthesis and characterization of CAQK-targeted pSiNPs encapsulating a protein cargo.

(a) Schematic of protein loading, surface modification, and CAQK-functionalization of pSiNPs. (b) FTIR analysis of pSiNPs during synthesis steps. Nanoparticle diameters as measured by (c) DLS and (d) TEM (scale bar in inset = 100 nm).

In order to verify surface modification and protein loading, we analyzed the pSiNPs by Fourier-transform infrared spectroscopy (FTIR), dynamic light scattering (DLS), and transmission electron microscopy (TEM) at each stage of synthesis. After oxidation and protein loading, the FTIR spectrum revealed characteristic amide I and II bands (at 1650 cm⁻¹ and 1530 cm⁻¹, respectively), denoting the presence of protein (Figure 1b).⁵⁵ Size measurements of unloaded pSiNPs and protein-loaded pSiNPs taken by DLS and TEM indicated that the loading process resulted in a diameter increase of ~5-10 nm, likely due to protein adsorption onto the pSiNP surface (Figure 1c,d,S2). To increase in vivo half-life and stability in systemic circulation,^56,57 the pSiNP surface was then aminated and modified with polyethylene glycol using NHS chemistry. PEG modification stabilized the nanoparticles, allowing them to be administered intravenously. To increase the retention of nanoparticles in the injured brain tissue, the pendant PEG groups were then conjugated to CAQK, a peptide previously discovered to bind upregulated proteoglycan complexes in injured brain tissue.³³ CAQK was attached to the distal end of PEG by reacting the free thiol of CAQK with a maleimide on the PEG. For some of the fluorescence imaging experiments, a FAM-labeled CAQK peptide was used. The surface chemistry was confirmed by FTIR; in particular, the appearance of a C-H stretching peak (2869 cm⁻¹) confirmed successful PEG conjugation.⁵⁸ The absolute diameter of the nanoparticles increased by ~10-20 nm after PEG and CAQK surface modification, while hydrodynamic diameter measurements increased by ~20-30 nm, likely due to the size increase in the hydrated shell around the nanoparticle after PEG modification (Figure 1c,d,S2). This size increase is consistent with the ~10 nm Flory radius of 5 kDa PEG and peptide.⁵⁹ Approximate number of peptides conjugated per pSiNP was found to be ~40,000-50,000 peptides per nanoparticle, as evaluated by quantifying the amount of FAM-labeled CAQK by absorbance and estimation of nanoparticle number by Nanoparticle Tracking Analysis. The zeta potential after surface modification of the protein-loaded pSiNPs containing the pendant PEG and CAQK groups was −1 ± 1.5 mV in PBS.

Model Protein Cargo is Released Linearly from pSiNPs and Retains Activity.

Due to the high cost of BDNF, protein loading and performance parameters of the nanoparticle system were initially optimized using lysozyme as a model protein cargo. The degradation of pSiNPs is anisotropic due to the vertical pore orientation created by electrochemical etching that results in an increased surface area in the horizontal plane.⁴⁵ Silicon dioxide hydrolyses and dissolves in aqueous conditions to form biocompatible orthosilicic acid, releasing the protein payload during this process.⁴⁵ pSiNP-Lysozyme-PEG-CAQK were degraded in PBS and imaged with TEM over 48 hours, revealing a reduction in total nanoparticle size over time (Figure 2a). Lysozyme loading was quantified to be 14.5% by mass relative to the pSiNP-protein construct, measured by fully degrading pSiNPs in PBS at 37°C and measuring protein content with a BCA assay (Figure 2b). Next, we quantified the activity of the protein released from degraded pSiNPs. Lysozyme activity after release from pSiNPs was assayed in a Micrococcus lysodeikticus cell assay (Figure 2c). Lysozyme hydrolyzes the cell wall of M. lysodeikticus, which can be monitored by decreased absorbance measured at 450 nm.⁶⁰ The combination of total protein measurements (Figure 2b) and protein activity (Figure 2c) confirmed that >95% of the protein released maintained its activity.

Figure 2: Loading and biodistribution of model protein in pSiNPs.

(a) Analysis of pSiNP-Lysozyme-PEG-CAQK size by TEM image analysis after degradation in PBS at 37 °C after 0, 12, 24, and 48 hours (scale bar = 100 nm). (b) Time-dependent release of model protein lysozyme from pSiNPs in PBS at 37 °C, measured by BCA assay. The mass percentage loading of lysozyme in the pSiNP constructs was 14.5% by mass relative to the pSiNP-protein construct. (c) Activity of lysozyme after release from pSiNPs. Lysozyme activity was assayed through the hydrolysis of Micrococcus lysodeikticus, measured by loss of absorbance at 450 nm. (d) Schematic depiction of the protocol followed in the biodistribution study. The right hemisphere was injured, followed by intravenous administration of pSiNP-Lysozyme-PEG-CAQK 2 hours post-CCI, and brains collection for downstream imaging and analysis 2 hours post-injection. (e) Time-gated image of pSiNPs in whole brains and (f) confocal image of lysozyme model protein (red), CAQK (green), and nuclei (blue) from injured brain sections (scale bar = 200 μm).

CAQK Peptide-Targeted pSiNPs Mediate Protein Delivery into the Injured Brain after Systemic Administration.

In this work we employed the peptide CAQK as a targeting ligand for the nanoparticles. CAQK has been shown to bind to the ECM upregulated in injured brain tissue, and to selectively enhance nanoparticle accumulation in the damaged regions of the mouse brain.³³ We and others have demonstrated that intravenous delivery of targeted pSiNPs successfully led to accumulation and retention of cargo in target tissues, such as siRNA against STAT3 in breast cancer tumors⁶¹ and siRNA against peptidylprolyl isomerase B⁶² or green fluorescence protein in the injured brain.³³ Concerning biodegradation of the material, prior work has demonstrated that pSiNPs accumulate mostly within the liver and spleen after intravenous administration and degrade into orthosilicic acid in organs within 1-4 weeks.²⁹ Orthosilicic acid is naturally found in many tissues and excreted renally.⁶³ Previous studies of in vivo degradation of pSiNPs after intravenous administration found that degradation profiles approximate zero-order release,^64,65 similar to the linear release profile we observe (Figure 2b). Intravenously delivered pSiNPs are well-tolerated in mice up to 20 mg/kg.²⁹ In the present studies, the protein-loaded pSiNP formulations were administered at 5 mg/kg of pSiNP.

Controlled cortical impact (CCI) is a well-studied model of TBI in mice and the injury is created by performing a craniotomy and impacting the exposed dura with an electromagnetically-driven piston.⁶⁶ To determine if the targeted pSiNPs could deliver a protein cargo to the injured brain tissue after CCI, pSiNPs were synthesized with the model protein, Rhodamine B-labeled lysozyme. CAQK-targeted pSiNPs carrying this payload were administered intravenously 2 hours post-injury in mice given a CCI on the right hemisphere and brains were harvested 2 hours after injection (Figure 2d). The intrinsic luminescence from the quantum-confined silicon domains in the pSiNPs was imaged with time-gated imaging, which highlights the long-lived (microseconds) excited state of silicon quantum dots while suppressing short-lived (nanoseconds) tissue autofluorescence.⁶⁷ The time-gated images revealed the presence of intact pSiNPs in the injured hemisphere of the brain after systemic administration (Figure 2e). Brains were then sectioned and imaged by confocal microscopy to determine distribution of injected materials within the tissue (Figure 2f). We observe co-localization of fluorescent signal from Rhodamine B-labeled lysozyme and FAM-CAQK in the injured brain, surrounded by fluorescent signal from FAM-CAQK in adjacent areas. We believe that the FAM-CAQK fluorescent signal not co-localized with Rhodamine B-labeled lysozyme is due to surface degradation of pSiNPs, thus liberating FAM-CAQK that can subsequently diffuse into the surrounding tissue. These results support the hypothesis that CAQK-targeted pSiNPs can accumulate in the injured brain and deliver protein cargos after systemic administration. This result is consistent with previous studies of CAQK-targeted pSiNPs carrying siRNA cargo in a penetrating brain injury model, where CAQK- but not CGGK control-modified pSiNPs mediated gene silencing in the injured brain tissue.³³

BDNF is Released Linearly from pSiNPs over 72 Hours and Maintains Activity.

BDNF loading was determined by ELISA to be 13.3 ± 0.7% by mass relative to the total construct and complete degradation of the pSiNP-BDNF construct occurred over 72 hours in PBS at 37 °C (Figure 3a). To confirm that BDNF released from pSiNPs maintained activity, we performed a functional assay of the pSiNP-BDNF constructs in SH-SY5Y cultures, a human neuroblastoma cell line, which were differentiated with retinoic acid to induce the expression of TrkB, the receptor for BDNF.^68,69 Improved cell viability and neurite extension are well-established responses to BDNF treatment in SH-SY5Y cultures expressing TrkB.^69-71 The pSiNP-BDNF and free BDNF-treated cultures exhibited significantly increased cell viability with treatment in a dose-dependent manner compared to untreated and empty pSiNP-treated cultures, which had no effect on cell viability even at their highest dose (Figure 3b). Free BDNF-treated cultures displayed higher cell viability at most doses compared to pSiNP-BDNF-treated cultures, likely due to the immediate availability of free BDNF in in vitro conditions (Figure 3a,b). Imaging of cultures for morphological changes revealed that free BDNF and pSiNP-BDNF treatments increased presence of fine and complex neurites, with multiple crossing points, and a lower density of cytoskeletal actin surrounding the nucleus compared to control and pSiNP treatments (Figure 3c). The similar release profiles of lysozyme and BDNF (Figure 2b, 3a) and the preservation of activity observed with the released lysozyme (Figure 2c) and the released BDNF (Figure 3b,c) support the hypothesis that the protein cargos retain their biological activity in the pSiNP formulations.

Figure 3: BDNF loading and activity in differentiated SH-SY5Y cultures.

(a) Time-dependent release of BDNF from pSiNPs in PBS buffer at 37 °C, quantified by ELISA. The loading of BDNF in the pSiNPs was 13.3% by mass relative to the pSiNP-protein construct. (b) Cell viability in retinoic acid-differentiated SH-SY5Y cultures treated with BDNF, pSiNP, and pSiNP-BDNF. The “pSiNP" control trace corresponds to a concentration of pSiNPs that is the same amount of Si by mass as was used in the pSiNP-BDNF formulation; i.e., each point in the pSiNP control trace corresponds to a mass/volume of empty pSiNPs that is ~7.5x the ng/mL value indicated on the x-axis. (Gray line represents untreated cells; n=6, mean ± SD, **** p ≤ 0.0001 Two-way ANOVA with Dunnett's post-test compared to pSiNP control). (c) Representative images of SH-SY5Y cells treated for 72 hours with matched concentrations of 300 ng/mL BDNF and stained with NF200 (red), phalloidin (green), and Hoechst (blue) (scale bar =100 μm).

BDNF is a growth factor with pleiotropic functions known to improve neuron survival, synaptic function, and cell signaling in animal models of injury and neurodegeneration.⁵ Previous studies have demonstrated that elevated BDNF concentrations in the brain achieved through stem cell engraftment increased the expression of synaptic proteins and improved neurological scores in murine models of TBI.^11,72-74 Due to the numerous challenges with cell therapy, alternative approaches that enable the systemic delivery of recombinant BDNF protein are desired.

The half-life of BDNF is reported to be <10 minutes in plasma^75,76 and ~3 hours in brain tissue.⁷⁷ This very short time window illustrates the challenge of developing BDNF, and protein-based biologics in general, as systemically administered therapeutics. The use of a nanoparticle carrier in order to protect and increase bioavailability of BDNF in the brain after systemic administration has therefore been considered by several researchers as a potential solution to this problem. Of particular relevance to the present work, a previous report of nanoparticle-mediated delivery of BDNF employed poly(lactic-co-glycolic acid) (PLGA) nanoparticles.⁷⁸ This PLGA nanoparticle formulation of BDNF improved neurological behavior scores over free BDNF, supporting the potential benefit of BDNF delivered by nanoparticles.⁷⁸ In this prior work, BDNF was adsorbed onto the surface of the PLGA nanoparticles, likely due to the known challenge of maintaining protein activity when encapsulating protein drugs into the core of PLGA nanoparticles. In addition to the denaturing property of polymers and polymer surfaces, surface adsorption has a limited loading capacity, it usually displays burst release kinetics, and it leaves the biologic exposed to potential degradation by proteolytic enzymes. By comparison, the present approach preserves protein activity (Figures 2c, 3b,c), can achieve high mass loading (~13 wt%) to enhance potency, and it yields zero-order release kinetics (Figures 2b, 3a).

Treatment with pSiNP-BDNF Reduces Lesion Volumes in a Mouse Model of TBI.

After CCI, there is progressive loss of tissue due to secondary injury mechanisms that results in a lesion,^79-81 including neuronal cell death which peaks 24-72 hours after the primary injury.^82,83 The volume of this lesion correlates with functional deficits such as motor activity and memory.^84,85 In order to evaluate the therapeutic potential of pSiNP-BDNF after systemic delivery, we administered pSiNPs via the tail-vein in a mouse CCI model of TBI and measured lesion volume three days post-injury (Figure 4a). Mice were injured on the right hemisphere of the brain with a CCI, and CAQK-targeted pSiNPs carrying BDNF were administered through the tail-vein 2 hours post-injury (5 mg/kg pSiNP and 0.65 mg/kg BDNF). Controls included animals that received PBS, empty pSiNPs with PEG-CAQK (5 mg/kg), or free BDNF (1 mg/kg). While our previous work demonstrated the improved accumulation of CAQK-targeted pSiNPs compared to control-targeted pSiNPs,³³ a limitation of the current study is that we do not evaluate the impact of CAQK-targeting on pSiNP-BDNF accumulation. Three days after injury, brains were harvested and serial coronal sections were collected every 0.5 mm across the injury lesion (11-12 sections per brain). Lesion area was measured from each section and the lesion volume calculated by the trapezoidal rule, as described previously.^81,86,87 Nissl staining of neuronal cell bodies supported the presence of neurons in the area surrounding the injury in all groups (Figure S3), suggesting that any neurons that died during primary or secondary injury were included in the lesion area. We observed a 24.0% reduction in lesion volume in pSiNP-BDNF treated mice compared to the PBS treated control group (Figure 4b,c). In comparison, lesion volumes in free BDNF treated mice were equivalent to lesion volumes in the PBS control. We hypothesize that the reduction in lesion volume in pSiNP-BDNF treated mice is due to the ability of pSiNPs to localize to injured brain tissue (Figure 2e,f) and release active BDNF over a period of time (Figure 3) that is consistent with peak neuronal apoptosis.^82,83

Figure 4: Lesion volume decreases after pSiNP-BDNF treatment.

(a) Schematic and timeline of injury, treatment, and lesion volume analysis. (b) Changes in lesion volumes relative to PBS-treated controls. (n = 7 per group, mean ± SEM, # p = 0.13 One-way ANOVA with Dunnett's post-test compared to PBS control). Both pSiNP and pSiNP-BDNF formulations contained PEG-CAQK surface chemistry as described in the text. (c) Representative images of H&E-stained coronal brain sections at 1.5 mm caudal from bregma for each treatment group with measured lesion area filled in gray (scale bar = 1 mm).

Lesion volumes in mice treated with empty pSiNPs (CAQK-targeted without BDNF) were reduced by 12.8% compared to the control PBS treatment (p = 0.58). pSiNPs and their degradation product silicic acid are known to interact with calcium ions, forming calcium silicate,³¹ and ROS, which catalyzes the oxidation of porous silicon during its degradation process.⁶⁵ Calcium and ROS levels are elevated in the brain microenvironment after TBI as part of the damaging biochemical pathways that make up secondary injury.^88-90 However, more extensive studies on the interaction of pSiNPs with the injured brain microenvironment are needed to be able to draw conclusions about the extent or nature of the effect observed in animals administered empty pSiNPs.

CONCLUSIONS

While therapeutic proteins are attractive candidates to address the complicated disease biology of TBI, the systemic delivery of proteins for treatment of TBI is challenging due to the instability of proteins in the blood and limited transport into the brain.^5,23,24,91 This work demonstrated a peptide-targeted pSiNP platform that addressed these challenges. We found that up to 13% by mass of the BDNF protein could be loaded into the pores of pSiNPs, and measurement of cell viability in retinoic acid differentiated SH-SY5Y cultures confirmed that BDNF subsequently released from this carrier retained its bioactivity. Accumulation of intravenously administered CAQK-targeted pSiNPs and their protein cargo in injured brain tissue was confirmed using a fluorescently tagged model protein in a CCI mouse model of TBI. When formulated with BDNF and administered systemically 2 hours post-injury, the CAQK-modified BDNF-loaded pSiNPs reduced brain lesion volumes by ~24% compared to treatment with PBS or free BDNF. This work demonstrates that ligand-targeted pSiNPs can be used to deliver a therapeutic protein cargo to injured brains after systemic administration, and that this treatment leads to phenotypic improvements in a TBI animal model.

The positive results seen in this study for treatment of TBI suggest that systemic and targeted delivery of BDNF using nanoparticles may also have potential in other central nervous system injuries. For example, BDNF is a promising therapeutic for stroke^92-95 and spinal cord injury.^17,18,96,97 Additionally, the progressive deterioration that occurs after TBI is caused by multiple disease pathways, such as ROS generation, inflammation, and vascular dysfunction.⁴ It has been hypothesized that monotherapies are insufficient to address the multi-factor disease pathology of TBI and combination therapy may be a key to overcome the decades of failed TBI clinical trials.⁹⁸ Due to their versatile chemistry and tunable pore size, pSiNPs provide a promising platform technology for multiple therapeutic cargos in combination therapies that might better address the multi-factorial causes of TBI disease pathology.

EXPERIMENTAL PROCEDURES

Materials.

CAQK peptide and fluorescein-conjugated peptide FAM-CAQK were purchased from CPC Scientific, Inc. (San Jose, CA). Highly boron-doped p-type silicon wafers, 1.2 mΩ-cm resistivity, single-side polished on the (100) face, were obtained from Virginia Semiconductor (USA). Concentrated hydrofluoric acid (HF, 48% aqueous, ACS grade) was obtained from Fisher Scientific (USA). Succinimidyl valerate-polyethylene glycol-maleimide (SVA-PEG-MAL, MW 5 kDa) was purchased from Laysan Bio (Arab, AL). Absolute ethanol, methanol, and 3-(Ethoxydimethylsilyl)propyl-amine (APDMES) were obtained from Sigma-Aldrich (St. Louis, MO). BDNF was obtained from R&D Systems (USA).

Synthesis of pSiNPs.

The preparation followed a published "perforated etch" procedure.⁹⁹ Single-crystalline highly doped p-type silicon wafers were anodically etched in an electrolyte consisting of hydrofluoric acid (48%) and ethanol in a 3:1 ratio by volume. CAUTION: HF is highly toxic and corrosive and contact with skin should be avoided. Procedures involving HF should always be carried out in a fume hood configured to handle HF and the operator should wear appropriate protective gloves, gown, and face shield. Etching was carried out in a Teflon etch cell using a platinum coil counter electrode. Prior to preparation of the porous silicon layers, the wafer surface was cleaned using a sacrificial etch consisting of electrochemical anodization (60 sec, 46 mA/cm²) in the HF electrolyte, followed by ethanol rinse, then dissolution of the resulting porous film with aqueous KOH (1 M). The wafer was then rinsed with water (1x) and ethanol (2x). An etching waveform consisting of a square wave of 200 cycles, in which a lower value of current density of 46 mA/cm² was applied for 1.2 s, followed by an upper value of current density of 365 mA/cm² applied for 0.363 s (Keithley 2651A Sourcemeter power supply). The multilayered porous nanostructure was removed from the crystalline silicon substrate by application of current pulse of 3.7 mA/cm² for 250 s in an electrolyte consisting of 1:29 (v:v) of 48% aqueous HF:absolute ethanol. The freestanding porous silicon sheets were placed in a sealed vial containing 1 mL of deionized water per mg of porous silicon and subjected to ultrasonic fracture in an ultrasonic bath (model 97043-960, VWR International) operating at a frequency of 35KHz and power of 48 Watts with a 1.9 L capacity overnight. The resulting ~130 nm-diameter nanoparticles were collected using centrifugation (15,000 rpm, 10 min, Eppendorf Centrifuge Model 5424R) and washed 3 times with ethanol and then isolated by centrifugation. Nanoparticles prepared in this manner are mesoporous, consisting of a crystalline silicon core skeleton coated with a surface layer of silicon dioxide.

1 mg of the pSiNPs resulting from the above procedure were washed once with DI water and then suspended in 1 mL of a solution consisting of recombinant human BDNF (200 μg/mL) dissolved in DI water. The nanoparticles were incubated under constant agitation at room temperature for ~18 hours. For the experiments involving the model protein lysozyme with conjugated Rhodamine B, the protein was loaded in a similar fashion. Protein mass loading was determined by measuring protein concentrations remaining in the supernatant after centrifugation by either BCA (lysozyme) or ELISA (BDNF; BDNF DuoSet ELISA, R&D Systems). The nanoparticles were washed once in DI water (via centrifugation/resuspension) to remove any unloaded protein from the solution, and then subsequently with 70% ethanol and finally with 100% ethanol. The loaded nanoparticles were aminated by incubation in an ethanol solution containing 12 μL/mL of APDMES for 3 hours under constant agitation. The nanoparticles were washed 3 times in 100% ethanol to remove any additional APDMES and incubated for 1 hour with 5 kDa SVA-PEG-MAL (500 μg in 1 mL ethanol) with constant agitation to achieve a PEGylated surface. After 3 more washes with 100% ethanol to remove any additional PEG, the nanoparticles were incubated with CAQK (100 μg/1 mg pSiNPs) overnight at room temperature. The nanoparticles were washed with 100% ethanol, 70% ethanol, and water and resuspended in water for characterization and use.

Characterization of pSiNPs.

A Malvern Zetasizer Nano (Malvern Panalytical Ltd.) was used to determine the hydrodynamic diameter and zeta potential of the nanoparticles. A Thermo Scientific Nicolet 6700 FTIR instrument fitted with a Smart iTR diamond ATR fixture was used to determine the FTIR spectra of the nanoparticles. A Genesys 150 UV/VIS Spectrophotometer (Thermo Fisher Scientific, Inc.) was used to evaluate the absorbance of FAM-peptide and a NanoSight LM10-HSB/GFT14 (Malvern Panalytical Ltd.) was used to determine the concentration of pSiNPs to calculate the number of peptides conjugated per pSiNP. For degradation studies, 0.5 mg/mL of lysozyme-loaded pSiNPs surface modified with PEG and CAQK were incubated in PBS at 37 °C. At 0, 12, 24, and 48 hours, aliquots of degraded nanoparticles were removed from the stock, pelleted, and resuspended into ethanol and loaded onto 42 μm formvar/carbon 400 mesh copper grids (Ted Pella, Inc.) and dried. The stock nanoparticles were similarly pelleted at each timepoint and resuspended in fresh PBS to ensure that degradation would not be stalled by the solubility of silicic acid in the supernatant.²⁹ TEM images were obtained on a JEOL 1400 plus electron microscope (JEOL USA, Inc.) operated at 80KeV and equipped with a Gatan Oneview camera (Gatan, Inc.).

Measurement of Protein Release from pSiNPs.

0.5 mg/mL of lysozyme or BDNF-loaded pSiNPs and an equivalent concentration of control nanoparticles were incubated in PBS at 37 °C. For release assays, supernatant was collected every 24 hours for up to 4 days and evaluated for protein concentration by BCA or ELISA. The percent protein release was calculated from the total cumulative release over the experiment. Cumulative released BDNF was calculated to be 51 ± 1.3% of BDNF calculated during loading. Lysozyme activity was determined by a Microroccus lysodeikticus cell assay as previously described with some modification.⁶⁰ Briefly, 0.5 mg/mL of lysozyme loaded pSiNP were incubated within a solution containing M. lysodeikticus cells. Lysozyme activity was evaluated by measuring the absorbance of the intact cells at 450 nm using a plate reader (Tecan) over 5 minutes.

Bioactivity of BDNF in Retinoic Acid Differentiated SH-SY5Y Cultures.

SH-SY5Y cells (ATCC) were plated at 32,000 cells/cm² in 96 well plates or at 16,000 cells/cm² on cover glasses coated sequentially in 0.05 mg/mL poly-D-lysine overnight and 0.05 mg/mL rat tail collagen 1 for 2 hours at 37°C. Cells were cultured in Minimum Essential Medium (MEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin for 24 hours before beginning the differentiation protocol. 10 μM retinoic acid in full media was introduced to the cells through a media change starting 24 hours after plating and exchanged every 48 hours. Cells were differentiated for 5 days to induce TrkB expression before beginning BDNF bioactivity studies.^68,69 For viability studies, doses ranging from 600 ng/mL (equivalent to 4.5 μg/mL pSiNP) to 0 ng/mL of free BDNF, pSiNP-BDNF surface modified with PEG, or pSiNP vehicles surface modified with PEG in serum-free MEM were introduced to the retinoic acid treated SH-SY5Y cells through a media change after washing the cells twice in serum-free MEM. Control wells representing 100% cell viability received full media with 10 μM retinoic acid. Cells were treated for 2 days and cell viability was evaluated with a CellTiter-Glo® Assay (Promega). For morphology studies, cells were cultured with treatments at 300 ng/mL for 3 days before being fixed and stained with anti-NF200 (Sigma), FITC-phalloidin (Millipore Sigma), and Hoechst with standard protocols. Fluorescent images were obtained on a Nikon Eclipse Ti2 (Nikon Instruments Inc.) of quadruplicate cover glasses.

Animal Injury Model.

All animal experiments were approved by the University of California, San Diego Institutional Animal Care and Use Committee (IACUC). A CCI model of TBI in mice was used for this study. 8-week-old C57BL/6J female mice (Jackson Labs) were placed in a stereotaxic frame under 2.5% isoflurane anesthesia and a 5 mm diameter craniotomy was performed over the right cortex, adjacent to the midline of the skull and midway between lambda and bregma. The right cortex was injured at a 2 mm depth with a 2 mm diameter stainless steel piston tip at a rate of 3 m/s using an ImpactOne (Leica Biosystems).

Protein Delivery from pSiNPs Targeting the Injured Brain in vivo.

Lysozyme-loaded pSiNPs surface modified with PEG and CAQK were injected via the tail-vein 2 hours after CCI (n=3) with control groups receiving pSiNP-Lysozyme-PEG-CAQK without injury or receiving an injury with a control PBS injection. Mice were perfused and brains were harvested 2 hours after injection. Time gated imaging was performed with an iSTAR 334T CCD camera (Andor Technology Ltd.) fitted with a Nikon AF micro lens (Nikko 105 mm) with laser excitation at 410 nm and a long-pass filter at 460 nm to quantify pSiNP localization. Brains were equilibrated in 30% w/v sucrose overnight and frozen in OCT (Tissue-Tek). Coronal sections were taken from the whole brain and counterstained with Hoechst to image the presence of Rhodamine B-tagged lysozyme being released from the pSiNPs which were tagged by FAM-labeled CAQK.

For BDNF treatments, 28 mice (n=7) were divided into 4 groups; PBS treated control, pSiNP-BDNF treatment surface modified with PEG-CAQK (13 μg BDNF equivalent, 100 μg pSiNPs), free BDNF treatment (20 μg), and a vehicle-only treatment (100 μg unloaded pSiNPs surface modified with PEG-CAQK) (Figure 4a).³³ 2 hours after CCI, treatments were administered intravenously through a tail-vein injection. Brains were collected 72 hours post-injury after perfusion with 10% formalin and frozen in OCT for tissue processing.

Lesion Volume Analysis and Histology.

10-μm thick coronal sections were taken every 0.5 mm from coordinates −4 mm to 1 mm from bregma for lesion analysis and processed with H&E staining to clearly visualize the lesion area with brightfield microscopy. Images of each section were blindly analyzed for lesion size by measuring the outlined lesion area using ImageJ software. Lesion volume was calculated with the trapezoidal rule. Sections taken from the center of each injury were Nissl stained using conventional protocols. Slides were imaged automatically using a Zeiss Axio Scan.Z1 and imaged with a color camera using a Nikon Eclipse Ti2 (Nikon Instruments Inc.).

Statistical Analysis.

Statistical analysis was performed on GraphPad Prism 8 software (9.1.2). In Figure 3b, cell viability was analyzed with a two-way ANOVA and Dunnett’s post hoc test conducted with p<0.05. In Figure 4b, lesion volume was analyzed with a one-way ANOVA with Dunnett’s post hoc test conducted with p<0.05.

ABBREVIATIONS

AAV2

Adeno-associated Virus Vector

APDMES

3-(Ethoxydimethylsilyl)propyl-amine

BBB

Blood-brain Barrier

BDNF

Brain-derived Neurotrophic Factor

CCI

Controlled Cortical Impact

DLS

Dynamic Light Scattering

ECM

Extracellular Matrix

FBS

Fetal Bovine Serum

FTIR

Fourier-transform Infrared Spectroscopy

HF

Hydrofluoric Acid

IACUC

Institutional Animal Care and Use Committee

MEM

Minimum Essential Media

NGF

Nerve Growth Factor

PBS

Phosphate-buffered Saline

PEG

Polyethylene Glycol

PLGA

Poly(lactic-co-glycolic Acid)

pSiNP

Porous Silicon Nanoparticle

ROS

Reactive Oxygen Species

SVA-PEG-MAL

Succinimidyl Valerate-Polyethylene glycol-Maleimide

TBI

Traumatic Brain Injury

TEM

Transmission Electron Microscopy

TrkB

Tropomyosin Receptor Kinase B