Neuroprotective activity of a virus‐safe nanofiltered human platelet lysate depleted of extracellular vesicles in Parkinson's disease and traumatic brain injury models

Abstract Brain administration of human platelet lysates (HPL) is a potential emerging biotherapy of neurodegenerative and traumatic diseases of the central nervous system. HPLs being prepared from pooled platelet concentrates, thereby increasing viral risks, manufacturing processes should incorporate robust virus‐reduction treatments. We evaluated a 19 ± 2‐nm virus removal nanofiltration process using hydrophilic regenerated cellulose hollow fibers on the properties of a neuroprotective heat‐treated HPL (HPPL). Spiking experiments demonstrated >5.30 log removal of 20–22‐nm non‐enveloped minute virus of mice‐mock particles using an immuno‐quantitative polymerase chain reaction assay. The nanofiltered HPPL (NHPPL) contained a range of neurotrophic factors like HPPL. There was >2 log removal of extracellular vesicles (EVs), associated with decreased expression of pro‐thrombogenic phosphatidylserine and procoagulant activity. LC‐MS/MS proteomics showed that ca. 80% of HPPL proteins, including neurotrophins, cytokines, and antioxidants, were still found in NHPPL, whereas proteins associated with some infections and cancer‐associated pathways, pro‐coagulation and EVs, were removed. NHPPL maintained intact neuroprotective activity in Lund human mesencephalic dopaminergic neuron model of Parkinson's disease (PD), stimulated the differentiation of SH‐SY5Y neuronal cells and showed preserved anti‐inflammatory function upon intranasal administration in a mouse model of traumatic brain injury (TBI). Therefore, nanofiltration of HPL is feasible, lowers the viral, prothrombotic and procoagulant risks, and preserves the neuroprotective and anti‐inflammatory properties in neuronal pre‐clinical models of PD and TBI.

proteins, including neurotrophins, cytokines, and antioxidants, were still found in NHPPL, whereas proteins associated with some infections and cancer-associated pathways, pro-coagulation and EVs, were removed. NHPPL maintained intact neuroprotective activity in Lund human mesencephalic dopaminergic neuron model of Parkinson's disease (PD), stimulated the differentiation of SH-SY5Y neuronal cells and showed preserved anti-inflammatory function upon intranasal administration in a mouse model of traumatic brain injury (TBI). Therefore, nanofiltration of HPL is feasible, lowers the viral, prothrombotic and procoagulant risks, and preserves the neuroprotective and anti-inflammatory properties in neuronal pre-clinical models of PD and TBI.

K E Y W O R D S
human platelet lysate, nanofiltration, neuroprotection, prion, virus

| INTRODUCTION
No efficient pharmacological treatment to balance the multifaceted cognitive and motor functional deterioration associated with neurodegeneration and brain trauma is developed. [1][2][3] Such relative failures contrast with substantial therapeutic achievements made in treating other pathologies. Therefore, a safe, practical, accessible, and affordable therapy of brain pathologies is still needed. Accumulating preclinical data now suggest that the human platelet lysate (HPL) proteome, made of a physiological combination of trophic factors, could emerge as a novel multifaceted biotherapy to treat diseases affecting the central nervous system (CNS). [4][5][6][7] Intense neuroprotective actions of HPL administration to the brain by intracranial/cerebroventricular or intranasal (i.n.) routes are found in animals models of stroke, 8 Parkinson's disease (PD), 9,10 Alzheimer's disease (AD), 11,12 traumatic brain injury (TBI), 13 or amyotrophic lateral sclerosis (ALS). 14 These in vivo results confirmed the convincing protective effects previously observed in cellular models of neurological diseases. 15,16 Most studies have linked the neuroprotective benefits of HPLs to their unique physiological mix of neurotrophic growth factors, cytokines, antioxidants, antiinflammatory molecules, and neurotransmitters, present in a soluble form or possibly loaded in extracellular vesicles (EVs). [17][18][19][20][21] The complex platelet proteome can synergistically activate complementary protective biological pathways and counterbalance pathological gene and protein expressions resulting from brain disorders and trauma. 9,12,13 For example, in vitro and in vivo studies showed that platelet lysate treatment can (a) protect against progressive or acute loss of synapses, (b) restore neuronal integrity, and (c) counterbalance neuroinflammation, oxidative stress, and defects in cognitive and/or motor functions. [4][5][6]15,22 Translational applications of HPL made from pooling platelet concentrates (PCs) from blood donors are now being considered, with clinical trials planned in patients with ALS. 14,23 HPLs for brain administration should have optimal quality and safety and meet established specifications. Safety requirements led us to develop a dedicated HPL, termed HPPL (for heat-treated human platelet pellet lysate), that is depleted of plasma, of relatively low protein content to avoid overloading the cerebrospinal fluid, essentially free of neurotoxic or clottable proteins, and with low pro-thrombogenic, proteolytic, and proinflammatory activities. 10,13 As HPPL is made from human blood, a vital concern is avoiding the risk of transfusion-transmitted infections.
Virus safety is particularly critical for HPLs manufactured from multiple allogeneic PCs to ensure quality consistency since pooling increases contamination risks. [24][25][26] It is crucial to prevent contamination by highly pathogenic blood-borne viruses like human immunodeficiency virus (HIV) and hepatitis B virus (HBV) and various blood-borne viruses exerting brain neurotoxicity such as hepatitis C virus (HCV), Dengue virus (DENV), Zika virus (ZIKV), West Nile virus (WNV), or, potentially, Human simplex virus (HSV) or severe acute respiratory syndrome coronavirus (SARS-CoV)-2 (which does not seem transmissible by blood). 27,28 Preventative measures include careful screening of healthy candidate blood donors and viral testing of individual donations by serological and nucleic acid tests. 29 However, the optimal viral safety margin requires implementing dedicated robust virus-inactivation or -removal treatments that do not affect the products' therapeutic safety and efficacy, 28,30,31 and can be applied to pooled platelet biomaterials. 24,25 We hypothesized that one virus-reduction technology ideally suited for HPPLs should be "nanofiltration". Nanofiltration is a bioprocessing procedure of virus removal used in the plasma product industry; 30,31 a purified protein solution is filtered through a device made of multiple, cuprammonium-regenerated cellulose hollow fibers, which have a cutoff of a few nanometers that is small enough to retain viruses, but large enough to let proteins flow through the nanosized membranes. To ensure a good compromise between the risks of clogging the filter by large proteins and virus-removal efficiency, we thought that a nanofilter with a cutoff of 19 ± 2 nm should remove, by membrane entrapment, blood-borne viruses, including the ca. 35-45-nm neurotoxic flaviviruses (HCV, DENV, ZIKV, and WNV) and coronaviruses, as well as the ca. 150-nm HSV. 30,31 However, we were uncertain as to how this virus-removal step would affect the HPPL proteome and EV content and, as a result, its neuroprotective activities. To answer these questions, we first developed conditions to nanofilter HPPL and then verified, by spiking experiments using 20-22-nm minute virus of mice-mock virus particles (MVM-MVPs), the extent of virus removal by an immuno-quantitative polymerase chain reaction (qPCR) assay. We then characterized the impact of this nanofiltration step on the HPPL composition, including neurotrophic factors, EVs content, prothrombotic and procoagulant qualities, using various proteomic and biophysical assays. Finally, we used validated cellular and in vivo models of PD and TBI to assess the neuroprotective and anti-inflammatory functions of the nanofiltered HPPL (NHPPL). This experimental design is illustrated in Figure 1.

| Characterization of starting PCs
Pooled PC donations had a mean platelet number of 587 Â 10 3 cells/ mm 3 . Residual red blood cells (RBCs at <0.7 Â 10 6 cells/mm 3 ) and white blood cells (WBCs at <0.8 Â 10 3 cells/mm 3 ) were undetectable, meeting the standard specifications for clinical applications in transfusion and our requirements for preparing HPPL.

| Feasibility and virus-removal experiments
HPPL made from pooled PCs following our established methods 10,13 was pre-filtered using 0.2-and 0.1-μm filters. We could readily filter close to 20 ml of this HPPL through 0.001 m 2 Planova-20N within approximately 3 h at a constant flow-rate of 0.1 ml/min and without reaching the maximum pressure of 0.098 MPa fixed by the supplier. The transmembrane pressure recorded during Planova 20N filtration is shown in Figure S1. Therefore, the nanofiltration of the HPPL on Planova 20N was achieved without clogging. The capacity of a specially manufactured Planova-20N with a filtration area of 0.0001 m 2 to effectively remove small viruses was assessed by spiking MVPs into 4 ml of pre-filtered 0.2-0.1-μm HPPL, to reach an expected final concentration of 10 10 MVPs/ml. MVP concentrations in spiked HPPL were determined by immuno-qPCR to be 1.63 Â 10 9 ± 8.2 Â 10 8 and 1.83 Â 10 9 ± 1.77 Â 10 8 , respectively, when immediately frozen after spiking or after being kept at room temperature during the duration of the nanofiltration experiment. This indicated no loss in detectable MVM associated with spiking into HPPL or storage at room temperature followed by freezing, demonstrating the absence of detrimental F I G U R E 1 Overall experimental design.
interference of the test material and the processing steps of the immuno-qPCR assay. Ct values obtained by immuno-qPCR in the NHPPL were the same as the baseline value of the unspiked control, thereby confirming absence of MVP, indicating a log reduction value (LRV) ≥5.39 log by nanofiltration (Table 1).

| Platelet extracellular contents and related procoagulant functional activities
We next assessed the size distribution and number of platelet-EVs (PEVs) in HPPL before and after nanofiltration using dynamic light scattering (DLS) and nanoparticle tracking analysis (NTA). DLS evidenced a significant lowering of the PEV mean size distribution, which decreased from ca. 171.5 nm to ca. 10.5 nm after nanofiltration ( Figure 2a), and the NTA showed ca. 90% reduction in the PEV concentration from 6.20 Â 10 10 ± 3.48 Â 10 8 to 6.21 Â 10 9 ± 3.12 Â 10 8 ( Figure S2). Moreover, the STA-procoagulant-phospholipid assay, that specifically measures the impact of PEVs as a contributor to blood coagulation, showed a significant prolongation in the coagulation time by the NHPPL (of ca. 110 s) compared to the non-nanofiltered HPPL (of ca. 24 s), consistent with removal of PEVs contributing to blood coagulation. Furthermore, the content of phosphatidylserine (PS)-expressing EVs (Figure 2b) was significantly less in NHPPL compared to the crude (p < 0.0001) and the 0.2-0.1-μm-filtered (p < 0.01) HPPL.
Thus, this nanofiltration process contributed to a substantial removal of EVs and to a significant decrease in its procoagulant effect.

| Proteomics analysis
Totals of 1117, 1011, and 897 proteins (with a false discovery rate

≥6.15
Note: (a) Immediately frozen at À80 C after spiking; (b) frozen at À80 C after being kept at room temperature during the nanofiltration duration. N = 2 for each condition. Abbreviation: MVP, mock virus particle.  (**p < 0.01, ****p < 0.0001), compared to HPPL. Statistical evaluation was performed by a one-way ANOVA followed by Fisher's least significant difference test.

| In vivo anti-inflammatory activity in a TBI model
The controlled cortical impact (CCI) injury performed in the right hemisphere of mice was applied to induce gene expressions of several proinflammatory markers. 13  NHPPL treatment induced a similar same trend as with HPPL, with significant downregulation (p < 0.05) of the Tlr4, Cd68, and Gfap proinflammatory genes, and relative decreases in Ccl3, Ccl4, Ccl5, Tlr2, and Trem2. There was no significant statistical difference in any of these markers between HPPL and NHPPL treatment. We concluded that NHPPL had intact functional activity to modulate proinflammatory markers post-CCI injury.

| DISCUSSION
There is a pressing need to develop safe neuroprotective and neurore- inflammatory and antioxidative biomolecules. 13 We have obtained preclinical evidence that a tailor-made purified heat-treated HPL (termed HPPL) prepared from platelets isolated from clinical-grade PCs is strongly neuroprotective in an MPTP mice model of PD. 10 We found that it modulates immune responses, promotes wound healing, and improves cognitive function in mouse models of mild and moderate/severe TBI. 13 These in vivo models confirmed the in vitro neuroprotective and neuroregenerative activities of HPPLs in primary neuronal cells and neuronal cell cultures. 15,16 To ensure safe brain administration, this HPPL is purposely depleted of the plasma protein compartment and is heat-treated; such processing removes bulk proteins to avoid protein overload of the CSF, neurotoxic fibrinogen and prothrombotic factors and proteolytic enzymes. 10 It is also vital to engineer HPLs that meet virus safety standards for clinical translation.
As platelet lysates are made from human blood, virus safety is mandatory considering the history of massive viral transmissions by pooled blood products, and as standardized HPLs require the mixing of ca. 50-250 human PCs, 26,39,40 inevitably increasing the risk of virus contamination. 25,26 Virus safety is especially vital as these platelet materials will be administered to new patient populations who, by contrast to hemophiliacs and immune-deficient individuals who need life-long transfusions, are not typically exposed to human blood products. 25 16 However, that treatment cannot inactivate all possible blood-borne viruses, especially some non-enveloped viruses. 43 Our study here identified for the first time that HPPL could be subjected to a dedicated nanofiltration process using 19-nm pore-size cartridges.
Low-protein content-HPPL was prepared from PCs centrifuged to pelletize the platelets and remove the plasma. HPPL was pre-filtered using The virus-removal capacity, assessed by immuno-qPCR detection of MVP was greater than 5 log units, thus demonstrating the efficiency of nanofiltration for eliminating even small viruses. Nanofiltration can remove a wide range of viruses and has been instrumental in improving the safety of industrial plasma-derived products thanks to a removal mechanism that is determined by the size of the virus relative to that of the membrane. 31,45 In the biological product industry, including blood plasma-derived products, nanofiltration has indeed sequence. In addition, we have restricted the ELISA determination to these three growth factors considering that a more extensive characterization of the impacts of the nanofiltration process has been performed by proteomics.
The proteomic analysis identified that the heat-treatment previously found to improve or normalize the safety and efficacy of HPPL 10,16 led to even more pronounced removal of fibrinogen and thrombogenic factors. A decrease in total protein number was detectable after nanofiltration, but over 80% of them were common. Proteins contributing to essential functions of HPPL for brain therapy, infections or during cancer progression. 50 Also, platelets in the blood circulation are known to contribute to defense mechanisms against various pathogens and can undergo activation that triggers multiple signaling pathways through its secretome products and the recruitment of immune cells. 46,51 Furthermore, the KEGG pathway analysis revealed decreasing abundances of some proteins involved in coagulation through the filtrations. This observation supports that the NHPPL should have an even lower risk of thrombogenicity than HPPL when administered to the brain.
Another informative aspect was the impact of nanofiltration on the removal of EVs. Most proteins removed were associated with extracellular exosomes. EVs can be classified as "small," "medium," or "large" according to their size. 52 In platelet lysates, EVs have a size ranging between approximately 50 and 300 nm. 21  NTA. In addition, the MP-activity functional assay, a capture assay that quantifies functional PS-expressing EVs, was used to evaluate the impact of nanofiltration on these specific EVs. A previous study showed F I G U R E 7 Modulation of neuroinflammatory markers in a mouse model of traumatic brain injury. A controlled cortical impact (CCI) was applied, and mice received either 60 μl heated platelet pellet lysate (HPPL), nanofiltered HPPL (NHPPL), or PBS, on three consecutive days by intranasal administration. Mice were sacrificed on day 7 post-injury, the ipsilateral cortex was dissected out, and cytokine and glial marker mRNA levels were quantified by an RT-qPCR. (n = 5-7 mice per group). Data are reported as the mean ± SEM.; *p < 0.05; **p < 0.01; ***p < 0.001 for CCI vs. Sham; # p < 0.05; ## p < 0.01 for CCI-PBS vs. CCI-HPPL or CCI-NHPPL by a one-way ANOVA followed by Fisher's least significant difference test.
that treatment of HPPL at 56 C for 30 min lowers the content in functional PS-expressing EVs. 53 We found here that the amount of EVs bearing PS in NHPPL was significantly decreased compared to HPPL.
The STA-procoagulant-PPL assay, which is a phospholipid procoagulantdependent clotting time assay, suggested removal of procoagulant PEVs, as indicated by a robust prolongation in the coagulation time from the NHPPL compared to the HPPL. The decrease in PS-expressing EVs afforded by nanofiltration should limit even more the risks of coagulation and therefore further improve the safety of NHPPL for brain administration.
An in vitro study was performed using LUHMES cells as a PD model to unveil any impact of nanofiltration on the neuroprotective activity of HPPL. NHPPL at a dose of 5% (v/v) was not toxic and maintained a capacity to protect cells from the erastin neurotoxic drug. Cells remained viable 24 h post-treatment, similar to that achieved with the HPPL in our previous studies. 10,16 We also evaluated the ability of NHPPL to support cell maturation, as was observed previously with HPPL. 16,22 , Our study using the SH-SY5Y neuroblastoma cell line revealed that after 1 week of treatment with 2% NHPPL, cells strongly expressed the β-III tubulin differentiation marker, similar to what was observed with the HPPL and RA-positive controls. Thus, NHPPL maintained a capacity to promote cell differentiation and maturation, which is essential to counterbalance progressive neuronal degeneration.
We performed CCI, an in vivo model of mild TBI, 54 to assess the capacity of the NHPPL to modulate inflammatory markers after a concussion, as found previously with the non-nanofiltered HPPL. 13 HPPL and NHPPL were delivered through an i.n. route over three consecutive days and their anti-inflammatory actions in ipsilateral cortical tissues were assessed 7 days after injury. Both were administered at a dose of 60 μl/ day for a total dose of 180 μl over 3 days. This daily dosing was selected based on our previous in vivo studies using HPPL in PD 10 and TBI 13 mouse models. Also, the in vitro functional activity of the HPPL and NHHPL batches used here was consistent with that observed with HPPL in our previous studies. 10,16,22 Gene expression data showed that the overexpression of inflammatory markers in non-treated TBI mice was significantly downregulated in mice treated with both the HPPL and NHPPL.
This result suggests that, after nanofiltration, the HPPL retained the antiinflammatory potential in TBI models. In addition, compared to our previous study, where both topical and intranasal administrations were performed, 13 our current results demonstrate that in this TBI mouse model, i.n. administration of the NHPPL alone was effective. This is clinically vital for the treatment of TBI where there is no brain access and intranasal administration would be the only feasible delivery option. Our data indicate that removing PEVs from HPPL by nanofiltration does not affect the neuroprotective activity, suggesting that these EVs are not essential to the neuroprotective and anti-inflammatory activity of this biomaterial. However, EVs from other HPLs promote cell growth and migration of neuronal cells and stimulate network formation in primary neuronal cultures. 21 Our in vivo evaluation has two limitations. First it was conducted, for comparative purposes with our previous work, 13

| Virus-removal assessment during nanofiltration
An immuno-qPCR-based assay using an MVM-MVP kit with a MOCK-V solution (Cygnus Technologies) was used to evaluate the virus-removal efficiency of nanofiltration (Supporting Information S1).

| Evaluation of total protein, trophic factors, proteomic, and bioinformatics analyses
We determined the total protein concentration and growth factor content and performed proteomics analysis by LC-MS/MS as described before 13,56 and in Supporting Information S1. (Custom Design & Fabrication) exactly as before. 13 The treatment using HPPL, NHPPL, and PBS (vehicle control) was administrated approximately 2 h post-injury by intranasal (i.n.) administration. Test materials (60 μl) were delivered using a pipette by alternating the nostrils and maintaining 5-min intervals between each 20-μl administration. This treatment was repeated on three consecutive days, with each mouse receiving 180 μl in total. On day 7, mice were sacrificed by cervical dislocation, the brains were quickly taken and rinsed in cold PBS, and the injured area of the ipsilateral cortex was collected using a 4.0-mm biopsy punch. Samples were then frozen in liquid nitrogen until further gene expression analysis by qPCR (Supporting Information S1).

| Statistical analysis
Statistical analyses were performed using GraphPad Prism software vers. 6.0, and data are expressed as a mean ± standard deviation (SD) or standard error of the mean (SEM). A one-way analysis of variance (ANOVA) followed by Fisher's least significant difference (LSD) test was performed for comparison, and differences were considered significant at p < 0.05. and NHPPL by intranasal delivery alone was effective to exert an antiinflammatory activity in the TBI model. In our previous studies in TBI models with created brain access, 13 the HPPL was first applied topically 1 h after injury onto the wound, followed by 6 days of intranasal delivery. The current study reveals that i.n. alone could be an option for the anti-inflammatory treatment of TBI without brain access, an information that is vital for clinical translation.

| CONCLUSION
The limitations of our study include the fact that the antiinflammatory effects of HPPL and NHPPL has not been evaluated in a TBI model using female mice, nor the protection of dopaminergic neurons in a PD mice model. Besides, we did not establish yet the optimum delivery schedule of the NHPPL to the brain, considering also that it will likely depend upon the mode of delivery (intracranial, intranasal, topical and/or intracerebroventricular) that can affect the effective dose delivered to the pathological site. Finally, although we show an improvement in safety parameters, such as due to the removal of pro-coagulant factors, associated with nanofiltration, only long-term administration in pre-clinical models will allow assessing the safety of NHPPL. However, jointly, the cumulative data unveiled in our study, are vital to support future translational developments of this biotherapy of brain disorders.

DATA AVAILABILITY STATEMENT
The datasets generated during and/or analysed during the current study are available from the corresponding author upon reasonable request.