Biomimetic Remodeling of Microglial Riboflavin Metabolism Ameliorates Cognitive Impairment by Modulating Neuroinflammation

Neuroinflammation, for which microglia are the predominant contributors, is a significant risk factor for cognitive dysfunction. Riboflavin (also known as vitamin B2) ameliorates cognitive impairment via anti‐oxidative stress and anti‐inflammation properties; however, the underlying mechanisms linking riboflavin metabolism and microglial function in cognitive impairment remain unclear. Here, it is demonstrated that riboflavin kinase (RFK), a critical enzyme in riboflavin metabolism, is specifically expressed in microglia. An intermediate product of riboflavin, flavin mononucleotide (FMN), inhibited RFK expression via regulation of lysine‐specific methyltransferase 2B (KMT2B). FMN supplementation attenuated the pro‐inflammatory TNFR1/NF‐κB signaling pathway, and this effect is abolished by KMT2B overexpression. To improve the limited anti‐inflammatory efficiency of free FMN, a biomimetic microglial nanoparticle strategy (designated as MNPs@FMN) is established, which penetrated the blood brain barrier with enhanced microglial‐targeted delivery efficiency. Notably, MNPs@FMN ameliorated cognitive impairment and dysfunctional synaptic plasticity in a lipopolysaccharide‐induced inflammatory mouse model and in a 5xFAD mouse model of Alzheimer's disease. Taken together, biomimetic microglial delivery of FMN may serve as a potential therapeutic approach for inflammation‐dependent cognitive decline.


Introduction
As resident immune cells in the central nervous system (CNS), microglia play key roles in modulating cognitive function. Microglia stimulate learning-related synapse formation in the healthy adult brain and maintain neuronal connectivity and synaptic homeostasis for learning and memory via synaptic pruning. [1] They also preserve cognitive function in neurodegenerative diseases by engulfing cellular debris and misfolded proteins, such as amyloid (A ), tau and -synuclein. [2] Additionally, microglia are major players in neuroinflammation, with their overactivation substantially increasing the production of cytokines and reactive oxygen species (ROS). [3,4] Neuroinflammation drives the progression of cognitive impairmentrelated diseases, including neurodegenerative disorders such as Alzheimer's disease (AD) and Lewy body dementia [3,5] ; and neurological disorders such as traumatic brain injury or stroke-associated cognitive dysfunction. [6] Thus, the regulation of microglia provides a promising approach to improve inflammation-related cognitive dysfunction. Metabolism shapes microglial function in stimulating learning and memory as well as inflammation-related cognitive decline. [7] Notably, riboflavin metabolism plays a vital role in regulating cognitive function; for example, low levels of vitamin B have been observed in patients with dementia and AD, and supplementation of vitamin B has proven beneficial in treating patients with cognitive impairment. [8] Among the beneficial effects, riboflavin (vitamin B2) protects cells from oxidative stress by enhancing antioxidant enzyme activities and the glutathione redox cycle and reducing pro-inflammatory responses in the brain. [9] Low levels of vitamin B expression in the brains of AD patients may be attributed in part to low levels of a key enzyme in riboflavin metabolism, riboflavin kinase (RFK), and supplementation of a metabolic product of RFK, flavin mononucleotide (FMN), suppresses A toxicity by regulating redox status. [10] RFK also mediates the activity of tumor necrosis factor (TNF)-activating nicotinamide adenine dinucleotide phosphate (NADPH) oxidase via coupling with TNF receptor-1 (TNFR1). [11] Thus, we hypothesized that RFK may contribute to the inflammatory state associated with cognitive impairment.
The objective of this study was to examine the role of microglial riboflavin metabolism in cognitive dysfunction. Here, we report that RFK is specifically expressed in microglia, and that its expression is enhanced by pro-inflammatory events. As an intermediate product of riboflavin metabolism, FMN inhibits RFK expression, while the ultimate downstream metabolite, flavin adenine dinucleotide (FAD), has no effect. Genetic knockdown of Rfk or supplementation of FMN attenuates the proinflammatory TNFR1/NF-B signaling pathway. Moreover, the effects of FMN on RFK expression rely on the regulation of lysinespecific methyltransferase 2B (KMT2B). Because intraperitoneal injection of FMN showed limited anti-inflammatory effects in a lipopolysaccharide (LPS)-induced mouse model, we establish a biomimetic microglial nanoparticle (BMNP) platform to specifically deliver FMN to the microglia. The BMNP strategy (termed MNPs@FMN) was designed with an encapsulated FMN-coated human serum albumin (HSA) nanoparticles core and a microglial BV2 cell membrane shell (Scheme 1A). Our results demonstrate that MNPs@FMN can successfully penetrate the blood-brain barrier (BBB) and facilitate microglial targeted www.advancedsciencenews.com www.advancedscience.com delivery (Scheme 1B). Importantly, MNPs@FMN ameliorated cognitive dysfunction, impaired synaptic plasticity, and inflammatory response in the LPS-induced mouse model and in a 5xFAD mouse model of AD. Taken together, our results suggest that microglial supplementation of FMN may serve as a novel therapeutic intervention for inflammation-based cognitive decline.

Microglial RFK Contributes to the LPS-Induced Inflammatory Response
To investigate inflammation-based cognitive impairment, we first established an LPS-induced mouse model. Behavioral performances of LPS-treated mice in the open field test (OFT), Y maze and Morris water maze suggest that the exploratory ability, working memory, and spatial memory were impaired ( Figure S2A and B, Supporting Information), and it also increased the hippocampal protein expression of IL-1 (FC = 1.369, p = 0.0021), IL-6 (FC = 1.386, p = 0.0024) and TNF-(FC = 1.382, p = 0.0026) ( Figure S2C-E, Supporting Information). Consistently, the volume of ionized calcium binding adapter molecule (Iba1)-positive cells was increased (FC = 1.527, p < 0.0001), while the process complexity and endpoint voxels were reduced (process complexity: FC = 0.463, p < 0.0001; endpoint voxels: FC = 0.596, p = 0.0004) ( Figure S2F,G, Supporting Information), which suggests that LPS promotes a change in the microglia from a "resting" phenotype to an "activated" phenotype.
To address how riboflavin metabolism acts on microglial function in cognitive decline, we examined the expression pattern of RFK in neural cells. RFK expression was detected in Iba1-positive cells, but not NeuN or glial fibrillary acidic protein (GFAP)positive cells in the hippocampus and cortex ( Figure 1A, Figures S3 and S4, Supporting Information), suggesting that RFK in the brain is mainly expressed in the microglia. As additional verification, RFK co-localized with GFP expressed under the control of a Cx3cr1 (classic microglial marker) promoter in Cx3cr1-GFP reporter genetic mice ( Figure 1B). Furthermore, RFK expression was increased in LPS-treated microglial BV2 cells (FC = 2.540, p = 0.0093) and in the cortex and hippocampus of LPS-induced mice (cortex: FC = 4.550, p = 0.0006; hippocampus: FC = 2.016, p = 0.0086) and 5xFAD mice (cortex: FC = 3.787, p = 0.0012; hippocampus: FC = 2.210, p = 0.0002) ( Figure 1C-H). To evaluate the function of RFK in the LPS-induced inflammatory response, we designed small interference RNAs (siRNAs) to suppress RFK expression ( Figure 1I,J). Our results demonstrate that Rfk knockdown with siRNA-3 (hereafter referred to as "Rfk siRNA") decreased the gene expression of Tnfa and Ifng (Tnfa: FC = 0.844, p = 0.0035; Ifng: FC = 0.488, p = 0.0166) ( Figure 1K), further indicating that RFK may play a role in TNF--associated inflammation.
RFK has previously been reported to couple TNFR1 to active NADPH oxidase. [11] Therefore, we assessed the role of RFK in the LPS-induced TNFR1/NF-B signaling pathway. Rfk knockdown decreased RFK and TNFR1 expression (RFK: FC = 0.326, p < 0.0001; TNFR1: FC = 0.380, p = 0.0040), as well as the phosphorylation of NF-B and I B in LPS-treated microglial BV2 cells (p-NF-B: FC = 0.639, p = 0.0478; p-I B : FC = 0.544, p = 0.0452) (Figure 2A-C). We also verified our results by staining TNFR1 with cellular membrane probes (DIL) ( Figure 2D). Because TNFR1/NF-B is associated with LPS-induced inflammation and NF-B is responsible for the transcription of proinflammatory cytokines, [12] we also examined the impact of Rfk knockdown on the homeostatic and inflammatory genes. The results demonstrate that Rfk siRNA suppressed Il-1b (FC = 0.758, p < 0.0001), Il-6 (FC = 0.598, p < 0.  Figure S5, Supporting Information). These results indicate that RFK expression is increased in LPS-induced in vitro and in vivo brain inflammation models, and that the increased RFK expression contributes to LPS-induced TNFR1/NF-B signaling activation.

FMN, but not FAD, Inhibits LPS-Induced Inflammatory Response
Riboflavin (vitamin B2) is metabolized to FMN and FAD via reactions catalyzed consecutively by RFK and FAD synthetase (FADS) (Figure 3A). FMN and FAD then bind tightly or covalently to flavoenzymes and participate in the regulation of a range of redox reactions. [13] To examine the effects of FMN and FAD on RFK expression, we injected FMN or FAD into untreated mice for five weeks ( Figure S6A, Supporting Information). The results suggest that FMN, but not FAD, decreases RFK protein and mRNA expression in the hippocampus ( Figure S6B-D, Supporting Information). FMN also reduced Il-1b, Tnfa, Ifng, and Csf1r mRNA expression in the hippocampus, while FAD showed no obvious effects on their expression levels ( Figure S6E-H, Supporting Information).
Next, we evaluated the effects of FMN and FAD on the LPSinduced inflammatory reaction in BV2 cells and primary microglia. Different doses of FMN inhibited the LPS-induced expression of Il-1b, Il-6, Tnfa, and Ifng; in contrast, FAD had no detectable anti-inflammatory effects in LPS-treated BV2 cells   L) Immunofluorescence staining and quantification of RFK with Iba1 in LPS-treated primary microglia with Control siRNA or Rfk siRNA n = 8-11. Scale bars, 40 μm. Magnified images are shown in the right columns. Scale bars, 13 μm. Results are expressed as mean ± SEM. ** p < 0.01, *p < 0.05 versus Ctrl siRNA; ## p < 0.01, # p < 0.05 versus LPS + Ctrl siRNA. Statistical significance was determined using One-way ANOVA followed by Tukey's post-hoc test.

The Inhibitory Effects of FMN on RFK are Mediated by Regulation of KMT2B
We further explored the anti-inflammatory effects of FMN, including whether these effects rely on regulation of RFK. First, we employed RNA-sequencing (RNA-seq) to identify the transcriptional profiles in the hippocampus of FMN treatment (40 mg kg −1 dosage was chosen because it dramatically reduced RFK expression). Consistent with our in vivo results ( Figure S6E-H), RNA-seq data uncovered 154 downregulated differentially ex-pressed genes (DEGs), most of which enriched in inflammationrelated pathways, such as "Cytokine-cytokine receptor interaction", "Chemokine signaling pathway", and "NF-B signaling pathway" (Figure 4A-C, Figure S8, Supporting Information). Representative DEGs are listed in Figure 4D. These results confirmed the anti-inflammatory effects of FMN, which we also observed in vitro.
Next, we sought to determine the mechanism by which FMN could inhibit RFK expression. Recently, RFK was shown to be epigenetically regulated at the transcriptional level by a histone H3 lysine 4 (H3K4) methyltransferase, KMT2B. [14] We, therefore, examined whether FMN regulates RFK via KMT2B. Consistent with this possibility, we found increased expression levels of  Figure S9E-H, Supporting Information). Importantly, we found that recombinant KMT2B protein successfully prevented FMN's effects on RFK, TNFR1/NF-B signaling, and pro-inflammatory cytokines in LPS-treated primary microglia ( Figure 4N-Q, Figure S9I,J, Supporting Information).
Thus, these results indicate that FMN inhibits the LPSinduced inflammatory response via regulation of RFK/KMT2B, suggesting that FMN may comprise a novel and promising candidate for therapeutic anti-inflammation in neuroinflammatory diseases.

Hippocampal Rfk Knockdown Rescues Cognitive Impairment and the Pro-Inflammatory Response
We further compared the effects of Rfk knockdown with FMN supplementation in LPS-treated mice ( Figure 5A). Here, we found that Rfk knockdown promoted recovery of the spontaneous alterations in the Y maze, as compared with the LPS and LPS +  Figure 5M,N). These data suggest that Rfk knockdown, but not direct FMN supplementation, attenuates LPS-induced pro-inflammatory effects and cognitive impairment.

Characterization, Brain Targeting, and Biodistribution of MNPs@FMN
Because direct FMN supplementation did not recapture the antiinflammatory effects of RFK, we speculated that the BBB, drug duration or cell targeting specificity may hinder its effects. Therefore, we developed an effective strategy for targeted FMN delivery (termed MNPs@FMN) via the following steps: first, we extracted microglial BV2 cell membranes; second, we prepared FMN-encapsulated HSA nanoparticles as the core; lastly, we bioorthogonally attached isolated microglial membranes onto the surface of the HSA-FMN nanoparticles ( Figure 6A). We used the ultraviolet absorption at 450 nm to draw a standard curve of FMN ( Figure 6B, Figure S10A, Supporting Information). FMN, NPs@FMN, and MNPs@FMN have similar absorption peaks   Figure 6C). Furthermore, the average hydrodynamic diameter of the NPs@FMN increased from 78 nm to 106 nm with a polydispersity index of 0.285 and 0.229 after they were coated with the microglial membranes ( Figure 6D,E). The results frompotential, transmission electron microscopy images and sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) demonstrated that the MNPs@FMN nanoparticles had been successfully engulfed by the BV2 microglial membrane ( Figure 6F-H). The drug encapsulation efficiency (EE)%, loading efficiency (LE) and release behaviors suggest that MNPs@FMN is released more slowly and thus may be long-lasting ( Figure 6I,J). Flow cytometry results showed that MNPs@FMN exhibited a greater efficiency in cellular uptake than NPs@FMN ( Figure 6K,L). Moreover, these results were verified by costaining with cytoskeletal -Tubulin ( Figure 6M,N). Upon uptake, MNPs@FMN accumulated in the lysosomes as indicated by costaining with lysosomal tracker (Lyso-tracker) ( Figure 6O,P). Co-staining of Cy5.5 with Iba1 or Lyso-tracker, and evaluation of cellular Cy5.5 uptake in primary microglia further confirmed these results (Figure 6Q,T, Figure S10B,C, Supporting Information), suggesting that MNPs@FMN was degraded in the lysosome and then FMN was released.
The in vitro BBB permeability of MNPs@FMN results show that the fluorescence signals in the bEND.3 cell monolayer (insert) and in the lower chamber were higher after incubation with MNPs@ FMN-Cy5.5 than after incubation with NPs@FMN-Cy5.5, suggesting that MNPs@FMN has greater ability to cross the BBB in vitro (Figure 7A-D). In vivo fluorescence imaging and ex vivo results also indicated that the brain distribution of MNPs@FMN compared with NPs@FMN-Cy5.5 was earlier and prolonged ( Figure 7E-H), providing additional physiological relevance to these findings. MNPs@FMN was obviously accumulated in the microglia rather than in neuron or astrocytes in the CA1, CA3, and DG regions of the hippocampus ( Figure 7I-K). We also found much higher FMN concentration in the hippocampus after intravenous delivery of MNPs@FMN than intraperitoneal and intravenous delivery of free FMN (FC = 1.337, p = 0.0007 and FC = 1.295, p = 0.0015) purified using the ultraperformance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) method ( Figure 7L,M). These results suggest that MNPs@FMN has better FMN delivery efficiency.

MNPs@FMN Ameliorates LPS-Induced Cognitive Deficits, Dysfunctional Synaptic Plasticity, and Inflammatory Response
To evaluate whether microglial targeted delivery of FMN improves inflammation-based cognitive dysfunction, we evaluated MNPs@FMN intervention in the LPS-induced mouse model. Hippocampal synaptic plasticity is essential to learning and memory. [15] Therefore, we examined the effects of MNPs@FMN on the long-term potentiation (LTP) in the hippocampus. We first detected the synaptic function in the Schaffer collateral pathway (SC-CA1) in hippocampal slices and recorded fEPSPs in the CA1 stratum radiatum by stimulating the SC/commissural pathway at various intensities. MNPs@FMN increased fEPSP slopes in CA1 hippocampal slices over those associated with LPS-PBS, LPS-MNPs, and LPS-NPs@FMN ( Figure 9A)  Because our observations suggest that MNPs@FMN also localizes, in part, to the liver, lung, and kidney of mice after intravenous administration, we further examined the potential adverse effects of MNPs@FMN on non-targeted organs. However, we did not detect pathological changes in the liver, heart, kidney, or lung as evaluated by H&E staining after MNPs@FMN administration ( Figure S12, Supporting Information). Moreover, there were no detectable alterations in serum biochemical indicators of liver or kidney function after NPs@FMN and MNPs@FMN administration (Table S1, Supporting Information). These results suggest that MNPs@FMN restores cognitive function and synaptic plasticity in LPS-induced mice by suppressing the inflammatory response, with no obvious side effects.

Transcriptome Mechanism of MNPs@FMN in Improving Inflammation-Based Cognitive Dysfunction
Because our results indicated that FMN may reduce the proinflammatory response via feedback regulation of RFK, we further evaluated the effect of MNPs@FMN on hippocampal RFK levels. Western blotting and PCR assays indicate that MNPs@FMN decreased LPS-induced hippocampal KMT2B and RFK expression (Figure 10A,B, Figure S13, Supporting Information). We further examined the underlying mechanisms of MNPs@FMN in improving LPS-induced cognitive deficits via RNA-seq ( Figure 10C). PCA score plots revealed a distinct separation of components in the WT, LPS, and LPS + MNPs@FMN groups ( Figure S14A, Supporting Information), for which the gene expression and FPKM distributions were similar (Figure S14B-D, Supporting Information). Furthermore, volcano plot revealed 519 DEGs between the LPS and WT groups, and the downregulated DEGs were enriched in biologically meaningful Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, such as "Ribosome," "Oxidative phosphorylation," "Huntington disease," "Alzheimer's disease," and "Parkinson's disease" (Figure 10D,E). The downregulated DEGs that were enriched in the "Alzheimer's disease" pathway are shown in Figure 10F.
We further identified 38 DEGs that were decreased by LPS and reversed by MNPs@FMN treatment, and indicated their enriched signaling pathways ( Figure 10G-J, Figure S15, Supporting Information). Among the MNPs@FMN-altered DEGs, ubiquinol cytochrome c reductase binding protein (Uqcrb), cytochrome C oxidase subunit 8b (Cox8b), and calmodulin-like protein 4 (Calml4) were enriched in the "Alzheimer's disease" pathway ( Figure 10K). We verified the expression of these three genes in the hippocampus by qRT-PCR ( Figure 10L). Furthermore, we confirmed that FMN reversed the effect of LPS in decreasing Uqcrb, Cox8b, and Calml4 expression in microglial BV2 cells, and that RFK overexpression abolished the effects of FMN (Figure 10M-P). Thus, these results reveal a molecular mechanism underlying MNPs@FMN-improved cognitive function in LPSinduced mice.

MNPs@FMN Improves Cognitive Function in the 5xFAD Mouse Model of AD
Given that our RNA-seq data revealed MNPs@FMN regulation of pathways related to AD, we compared its effects with that of intravenous delivery of free FMN in the 5xFAD mouse model of AD. MNPs@FMN increased the spontaneous alterations in the Y maze, as compared with the AD-PBS and AD-FMN groups (FC = 1.244, p = 0.0325; FC = 1.280, p = 0.0153) (Figure 11A,B). During five days' training in the Morris water maze, MNPs@FMN shortened the latency to the target ( Figure 11C Figure 11K-O). Additionally, the microglia assumed a less "activated" phenotype ( Figure 11K-N). These findings indicate that MNPs@FMN rescues cognitive function, pro-inflammatory response, and A plaque pathology in an AD mouse model.

Discussion
Low intake of riboflavin has been correlated with the development of cognitive impairment, and studies suggest that cognitive function in middle-aged and elderly people may be improved by dietary intake of riboflavin and other forms of vitamin B. [16] Notably, the cognitive protection of riboflavin may be associated with its anti-oxidative and anti-inflammatory effects.
[9] As  a key enzyme in riboflavin metabolism, RFK has been shown to couple with TNFR1 to active NADPH oxidase, [11] suggesting that it may play a role in mediating the anti-inflammatory effects of riboflavin. Here, we reported for the first time that RFK is expressed in the microglia and that it is upregulated in the cortex and hippocampus of the LPS-induced mouse model and 5xFAD mouse model of AD, suggesting that it may be related to inflammation-based cognitive impairment. Functionally, TNFR1 induces RFK membrane recruitment and subsequent phosphorylation of NF-B, which leads to the transcription of pro-inflammatory cytokines, such as IL-1 , IL-6, and TNF-. In vitro and in vivo Rfk knockdown attenuates the proinflammatory response mediated by the TNFR1/NF-B signaling pathway, and hippocampal Rfk knockdown improves the cognitive function in the LPS-treated mouse model. Rfk mRNA expression has previously been demonstrated to be decreased in AD patients and the A -based yeast model. [10] However, in our study, we observed increased RFK protein expression in the cortex and hippocampus of the LPS-induced mouse model and 5xFAD mouse model of AD. These disparate findings may result from differences in the model systems, given that we mainly focused on murine RFK function. It is worth noting that the latter study limited its evaluation of AD patients to the cerebellum, prefrontal cortex, and visual cortex. [10] Therefore, additional experimentation will be needed to further characterize the RFK regulatory patterns in different regions of the brain. Importantly, our results provide a mechanism by which FMN may have therapeutic efficacy for neurological disorders. FMN and FAD, which are catalyzed by RFK and FADS, serve as critical mediators of riboflavin metabolism, and as cofactors for flavoenzymes, they contribute to the maintenance of normal flavoprotein functionality, with impacts on redox reactions, DNA repair, protein folding, and other physiological processes. FMN was previously reported to prevent A toxicity by increasing ATP and NADH generation. [10] Moreover, our lab demonstrated that FMN protects DA neurons from oxidative stress in the PD mouse model. [17] In the present study, we report that FMN, but not FAD, improves inflammation-based cognitive function by regulating the TNFR1/NF-B pathway. While H3K4 methyltransferase KMT2B induces RFK transcription to activate the TNF-/NOX2 pathway, we also found that KMT2B is the upstream regulator of RFK, and interestingly, KMT2B is increased in the hippocampus of the LPS-induced mouse model and 5xFAD mouse model, further supporting its relationship with inflammationbased cognitive dysfunction. Mechanistically, FMN regulates RFK via epigenetic regulation of KMT2B, as KMT2B overexpression conversely abolishes the effects of FMN on inflammation. Because RFK expression is dependent on the AD pathology, [10] FMN may provide an alternative intervention approach for cognitive dysfunction via activation of the riboflavin pathway. While FAD showed little therapeutic efficacy in our experimental sys-tem, its activity has been associated with metabolic diseases, such as multiple acyl-coenzyme A dehydrogenase deficiency and hepatic steatosis. [18] On the other hand, FMN has additional roles in generating ATP and NADH, which are novel regulators of calcium homeostasis, mitochondrial functions, and aging. [19] Therefore, additional research is needed to clarify the distinct roles of FMN cofactors in neurodegenerative diseases. Estrogen plays a vital role in regulating memory, [20] and it also could modulate riboflavin pathways (such as riboflavin carrier protein). [21] In this study, we wanted to preclude the effects of estrogen, so we performed the studies only in male mice.
Though microglia are key players in inflammation-related cognitive dysfunction, they are resistant to manipulation by recombinant viruses such as lentiviruses and adeno-associated viruses. [22] Furthermore, the BBB deters pharmacological treatments. Consequently, a variety of engineered technologies have emerged for neurodegenerative and neurological diseases, such as the photoresponsive vaccine-like chimeric antigen receptor and stem cell-derived extracellular vesicle systems designed to targeting microglia in inflammation-related depression and cerebral ischemia. [23] In this study, we establish a biomimetic microglial nanoparticle system (MNPs@FMN) to specifically deliver FMN to microglia. In vitro, ex vivo, and in vivo models confirmed the efficacy of delivering MNP to microglia using this approach. Intriguingly, the in vivo results are suggestive of slow release of FMN from the MNPs@FMN, providing a potential of long-lasting and controlled release of this system. Our results demonstrate that MNPs@FMN significantly ameliorates cognitive deficits and dysfunctional synaptic plasticity and attenuates inflammatory response in the LPS-induced mouse model and 5xFAD mouse model of dementia. As the most abundant plasma protein in humans, HSA is natural and stable. Furthermore, HSA is a non-toxic, non-immunogenic endogenous protein with high biocompatibility, biodegradability, and good biosafety, so it provides an ideal carrier material. At present, HSA nanoparticles, such as Abraxane, have been approved by the US Food and Drug Administration (FDA) for clinical treatment. [24] Thus, we chose HSA as a carrier to target the BBB that is conducive to future clinical use. Astroglial activation is also a vital cause of neuroinflammation, [25] and although we did not observe colocalization between RFK and astrocytes, our results suggest that microglial FMN supplementation suppresses astroglial activation, indicating that it may affect the glial interplay, which needs further study.
In this study, we identified three candidate genes (Uqcrb, Cox8b, and Calml4) that may be related to FMN's neuroprotection. Uqcrb is a subunit of mitochondrial complex III in the mitochondrial respiratory chain, [26] and its expression is reported to be correlated with colorectal cancer and glioblastoma. [27] Furthermore, Uqcrb inhibits hypoxia-induced mitochondrial ROS generation in tumor cells and displays anti-angiogenic activity. [28] in the CA1 of hippocampi of 5xFAD mice after intravenous delivery of PBS, free FMN, or MNPs@FMN. Scale bars, 100 μm for the left full view panel. Scale bars, 40 μm for Merge panel, and 10 μm for magnified images. L,M) Quantification of A intensity and the number of microglia engulfing A in the CA1 of the hippocampus n = 6-7. N,O) Immunofluorescence staining and quantification of RFK with Iba1 in the CA1 of hippocampi of 5xFAD mice after intravenous delivery of PBS, free FMN, or MNPs@FMN. Scale bars, 40 μm. Magnified images are shown in the right. Scale bars, 10 μm n = 7-10. Results are expressed as mean ± SEM. ** p < 0.01, *p < 0.05 versus WT; ## p < 0.01, # p < 0.05 versus AD; && p < 0.01, & p < 0.05 versus AD-FMN. Statistical significance was determined using one-way ANOVA and Tukey's tests for post hoc comparisons. 5 min on an ultrasonics processor at 30 W and 20 kHz to form a nanoparticle core (NPs@FMN-Cy5.5). Finally, NPs@FMN-Cy5.5 were coated with BV2 cell membrane to form the MNPs@FMN-Cy5.5.
In Vitro Evaluation of BBB Permeability of MNPs@FMN: In vitro BBB permeability evaluation was performed according to the methods of a previous study. [36] Approximately 1 × 10 5 -1 × 10 6 bEnd.3 cells were seeded in polyester transwell inserts (6 wells, pore diameter of 0.4 μm, 4.67 cm 2 , Corning, NY, USA), and a Millicell-ERS volt-ohmmeter (Millipore, Billerica, MA, USA) was used to detect the Trans-Epithelial Electrical Resistance (TEER) of the bEnd.3-cell monolayer to confirm the tightness of the monolayer to simulate the BBB. To evaluate the BBB permeability of MNPs@FMN across the monolayer, the culture media on the apical side was replaced with PBS, NPs@FMN or MNPs@FMN (0.5 μg/mL of Cy5.5 in 0.7 mL) for 3 h, and the lower side culture media was replaced with fresh DMEM. The fluorescence of the insert of the transwell and the bottom chamber of the bEnd.3 layer was imaged using a confocal laserscanning microscope (SP8; Leica, Hamburg, Germany).
Animals and Drug Treatments: Adult (8-week-old) male C57BL/6J mice were purchased from SPF Biotechnology Co., Ltd. (Beijing, China). Male 5xFAD mice (10-month-old) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). The age-and sex-matched wildtype (WT) mice were used as control. The mice were housed in groups of four under a 12/12 h light/dark cycle with ad libitum access to food and water. The animal experimental protocols were in compliance with the Institutional Animal Care and Use Committee of Guangzhou Medical University (Approval number: GY2020-041) and National Institute of Health guidelines on the care and use of animals (NIH Publications No. 8023, revised 1978).
To examine the effects of FMN and FAD on RFK expression, different doses of FMN and FAD (0, 5, 10, 20, 40 mg/kg) were intraperitoneally injected three times a week for 5 weeks. The mice were sacrificed, and hippocampal samples were collected and analyzed. The control group was injected with PBS.
To compare the effects of Rfk knockdown with free FMN on inflammation-based cognitive dysfunction, mice were injected with lentivirus (LV)-packaged Ctrl or Rfk shRNA for 3 weeks. Afterward, LPS (0.25 mg kg −1 ) was administered intraperitoneally for 7 consecutive days. Then, LV-Ctrl mice were intraperitoneally injected with PBS or FMN (40 mg kg −1 ) three times a week for 5 weeks. LV-Rfk mice were intraperitoneally injected with PBS.
To investigate the effects of MNPs@FMN on inflammation-based cognitive dysfunction, LPS (0.25 mg kg −1 ) was administered intraperitoneally for 7 consecutive days, and simultaneously, NPs@FMN, MNPs@FMN (equivalent dose of FMN), or unloaded MNPs vehicle were administered intravenously starting on the day of the first LPS injection once every other day for 11 days. After 3 days, behavioral tests were performed.
To study the effects of MNPs@FMN on the AD mouse model, 5xFAD mice were given intravenously free FMN or MNPs@FMN every other day for 11 days. The control group was given intravenously PBS. After 3 days, behavioral tests were performed.
Stereotaxic Injection of LV-Rfk shRNA in the Hippocampus: LV-Ctrl and LV-Rfk were packaged by Sunbio Medical Biotechnology (Shanghai, China) and they were stereotaxically injected in the hippocampus according to our recent work. [17] Briefly, mice were anesthetized and placed in a stereotaxic frame. LV-Ctrl and LV-Rfk in 0.5 μL vol were delivered into the bilateral hippocampus at the target site (Bregma AP, −2.0 mm, ML, ±2.0 mm, DV, −2.0 mm). The syringe was left in place for 5 min before being slowly withdrawn from the brain.
Open Field Test: The OFT was performed according to the previous report. [38] In this study, mice were placed in a rectangular plastic box (40 × 40 × 40 cm). The movement was recorded for 15 min using a video tracking system (EthoVisione XT software, Beijing, China). The total distance, movement speed, number of entries to the center zone, and time spent in the central zone were analyzed.
Y-Maze Test: The Y-maze test was used to evaluate spontaneous alternations as a measure of the spatial working memory, according to previous methods. [39] The Y maze apparatus was constructed from gray plastic and consisted of three arms (30 cm long, 10 cm wide, and 20 cm high). Mice were placed within the center zone and were allowed to explore the Y-maze freely for 8 min. The time spent in each arm was calculated visually (EthoVisione XT software, Beijing, China). Non-overlapping entrance sequences were defined as spontaneous alternations (%).
Morris Water Maze (MWM): The MWM test was performed as previously described. [40] The test apparatus consisted of a circular pool with a diameter of 120cm and was filled with water (22 ± 1°C) made opaque white with bright white food coloring. The circular pool was divided into four quadrants, and different images (circles, squares, and triangles) were hung on the pool walls. During the 5-day training, the mice were placed in the water randomly in one of the four quadrants to look for the hidden platform for 60s, and they were allowed to stay on the platform for 20 s. If the mice failed to find the platform, they were guided towards it, where they remained for 20 s. The time to find the platform was recorded by a computerized tracking system (EthoVision XT; Beijing, China). On the sixth day of the probe test, the platform was removed, and the mice were allowed to swim freely for 60 s. The moving track and time spent in the target quadrant were recorded using a video tracking system (EthoVisione XT software, Beijing, China).
Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR): Cultured cells or hippocampal samples were harvested, and total RNA was isolated using Trizol reagent (Invitrogen, San Diego, CA, USA). The RNA was used to generate cDNA using a cDNA Reverse Transcription Kit (QIAGEN, Waltham, MA, USA). Realtime PCR was carried out with TB Green Premix Ex Taq (Takara, Shiga, Japan). Results were obtained using the 2 −ΔΔCT method as described previously. [41] Data were obtained from three separate experiments, each of which was performed in triplicate. Gene expression was normalized to GAPDH. Primer sequences are listed in Table S2, Supporting Information.
Western Blotting Assay: Total protein was extracted from cultured cells or hippocampal samples using RIPA Lysis Buffer (Beyotime Biotechnology, Shanghai, China). The total protein (30-40 μg) was separated on 6%-12% gel by SDS-PAGE and electrophoretically transferred onto polyvinylidene difluoride membranes. After blocking, the membranes were incubated with primary antibodies, washed, and incubated with HRP-conjugated secondary antibodies. The immunoblots were visualized using the GeneGnome XRQ Chemiluminescence imaging system (Gene Company, Hong Kong, China). Image J software was used to analyze the optical density of the bands.
Enzyme-linked Immunosorbent Assay (ELISA): ELISA was performed as previously described. [37] In brief, hippocampal tissues were homogenized in PBS buffer and centrifuged at 12 000 × g for 20 min at 4°C, and then the supernatants were collected. The concentrations of IL-1 , IL-6, and TNFin the hippocampus and cultured cellular supernatants were measured using ELISA kits (Enzyme-linked Biotechnology Co., Ltd., Shanghai, China) according to the manufacturer's instructions. OD values were measured by Multiscan Spectrum (PerkinElmer, MA, USA) at 450 nm, and the results were expressed as pg per mg protein (pg mg −1 protein) or pg mL −1 .
Immunofluorescence Assay: Mouse brains were fixed in 4% paraformaldehyde and dehydrated with 20%-30% sucrose buffer. The fixed brains were then sectioned using a freezing microtome (Leica, Hamburg, Germany). Afterwards, the brain sections or cultured cells were blocked with 5% BSA, and incubated with primary antibodies and then with fluorescent-labeled secondary antibodies. Images were captured