Development of a method for isolating brain capillaries from a single neonatal mouse brain and comparison of proteomic profiles between neonatal and adult brain capillaries

C57BL/6 J pregnant mice were purchased from Japan Clea (Tokyo, Japan). All animals were bred in the Kumamoto University Faculty of Pharmaceutical Sciences Animal House and were housed in 12-h light/dark environment. All animal experiments were approved by the Institutional Animal Care and Use Committee at Kumamoto University and followed the Fundamental Guidelines for Proper Conduct of Animal Experiments and Related Activities in Academic Research Institutions under the jurisdiction of the Ministry of Education, Culture, Sports, Science, and Technology as well as the Animal Research: Reporting in Vivo Experiments guidelines.

Newborn mice were dissected at 7 days of age. A single neonatal mouse brain was frozen in liquid nitrogen in a 1.5-mL tube and stored at − 80 °C. For preparation, a single mouse brain was transferred to a 2-mL screw-cap tube (Watson, Tokyo, Japan) after thawing, and 1 mL of a homogenizing buffer (101 mM NaCl, 4.6 mM KCl, 2.5 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 15 mM HEPES, pH 7.4) was added. The brain was subsequently homogenized by a bead homogenizer (Bead Mill 4, Thermo Fisher Scientific, Waltham, MA, USA) for 30 s at 1 m/s without beads to reduce homogenizing power. Thereafter, the homogenate was uniformly homogenized by pipetting and transferred to a new 2-mL tube. Moreover, 50 μL of the brain homogenate was dispensed in another tube as whole brain lysate. The homogenate was centrifuged (1000 × g, 10 min, 4 °C), and the supernatant was carefully removed. Up to 1 mL of the homogenizing buffer was added to the pellet and suspended by pipetting gently. After suspension, an equal volume of 32 w/v% dextran/homogenize buffer was added into the tube and mixed by inverting. The samples were immediately centrifuged (4500 × g, 15 min, 4℃), and the supernatant was collected into a new 2-mL tube. The pellets were kept on ice. The supernatant was centrifuged (4500 × g, 15 min, 4℃) again, and the supernatant was discarded. After removing the fat content adhered at the wall of the tube using Kimwipe (Nippon Paper Crecia, Tokyo Japan), pellets were suspended in the suspension buffer (homogenized in buffer containing 25 mM of NaHCO₃, 10 mM of glucose, 1.2 mM of pyruvate, and 5 g/L of bovine serum albumin, 200 μL × 2), and the samples of the two tubes were combined into one tube. A cell strainer (pluriStrainer-Mini, 70 μm mesh, pluriSelect, Leipzig, Germany) was filled with 800 mg of glass beads (0.35–0.5 mm, AS ONE, Osaka, Japan) and washed 4 times with 500 µL of suspension buffer. Then, the combined suspension sample was added to a cell strainer with glass beads and washed 10 times with 500 µL of suspension buffer. Glass beads were transferred to a new tube using a spatula. Thereafter, 1 mL of suspension buffer was added to glass beads and mixed by inverting. The supernatant was then quickly transferred into a new tube. The suspension buffer of 500 µL was re-added on glass beads and mixed by inverting; the supernatant was quickly transferred into the previous tube. The tube was centrifuged (3300 × g, 5 min, 4℃), and the supernatant was removed. The pellet was suspended using 100 µL of a homogenization buffer. Part of the isolated brain capillary fraction was used for microscopy. Isolated brain capillary fractions were lysed in a hypotonic buffer (10 mM NaCl, 1.5 mM MgCl₂, 10 mM Tris–HCl, pH 7.4) by sonication, and protein levels were measured using a Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific).

A western blot analysis of the isolated brain capillary fraction was performed as previously described [16]. Samples were separated on a sodium dodecylsulfate (SDS) polyacrylamide gel and blotted onto a polyvinylidene fluoride (PVDF) membrane. The following primary antibodies were used: Claudin-5, 1/4000 dilution (35–2500; Thermo Fisher Scientific); β-actin, 1/10000 (8H10D10; Cell Signaling Technology, Danvers, MA, USA); and horseradish peroxidase (HRP) conjugated secondary antibodies, 1/10000 dilution (Goat anti mouse IgG HRP, 7076S, Cell Signaling Technology). Band intensity was quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

The peptide samples of brain capillary fractions and whole brain lysates for quantitative proteomics were prepared by trypsin digestion using phase transfer surfactant. Proteomic analysis was conducted as described previously [17, 18]. Briefly, each sample was analyzed by data independent acquisition (DIA/SWATH) on the TripleTOF 6600 (SCIEX, Framingham, MA, USA) interfaced with the Eksigent nanoLC 400 (SCIEX). Mass calibration and checking of intensities, retention times, and peaks were done every 5–6 samples by trypsin-digested β-galactosidase with the auto-calibration function in Analyst TF 1.7.1 (SCIEX). Protein identification and quantification were conducted using the library-free search function in DIA-NN 1.8 with the UniProt mouse reference proteome allowing one miss-cleavage [19]. Default settings were used for amino acid modifications and the properties of the precursors and fragments. The intensities of the precursors were normalized using retention time-dependent cross-run normalization, and the concentration of each protein was calculated from specific peptides using the MaxLFQ algorithm [20], which was integrated into the DIA-NN. Peptides and proteins were filtered at a false discovery rate of less than 1% for identification and quantification. To validate the quality of the present proteomic data, the distribution of intensities and coefficients of variance (CV) of the identified proteins are shown in Additional file 2: Fig. S1. The score plot of principal component analysis is shown in Additional file 2: Fig. S2. The intensity distributions were not significantly different among the samples according to a one-way ANOVA (Additional file 2: Fig. S1A and B). The %CV histograms overlapped between the two or four groups, and the averages and medians were less than 21.0% and 17.7%, respectively (Additional file 2: Fig. S1C and D). The replicate samples from each group formed clusters in the score plots (Additional file 2: Fig. S2). The protein concentrations of all quantified proteins are listed in Additional file 1: Tables S1 and S2. Raw data files of the liquid chromatography with tandem mass spectrometry analysis have been deposited in jPOST (jPOST ID: JPST001978/PXD039296 for neonatal data and JPST002139/PXD041770 for adult data).

Numerical data are expressed as mean ± standard deviation. Between-two group comparisons were performed using the paired-samples t-test or Welch t-test in Microsoft Excel (Microsoft Corp.; Redmond, WA, USA). Network analysis of the differentially expressed proteins was conducted using STRING 11.5 (https://​string-db.​org/​) [21]. The UniProt accession number list of differentially expressed proteins was set to search for multiple proteins, and the organism was set as Mus musculus. Clustering was performed using MCL with an inflation parameter of 3 (Fig. 5A and Additional file 2: Fig. S3). Cluster 1, which contained 60 proteins, was further analyzed using functional enrichment analysis (Fig. 5B and Additional file 2: Fig. S4, Additional file 1: Table S4). Principal component analysis was performed using SIMCA 14 (Sartorius, Gottingen, Germany) with centering and scaling variables set to Pareto Variance. Graphs and heat maps were created using GraphPad PRISM7 (GraphPad, Boston, MA, USA).

First, we prepared brain capillary fractions from neonatal mice brains using the previously reported isolation method from the single brain of an adult mouse [15]. However, no brain capillary was observed in the recovered fraction. To adapt to the fragility of the neonatal brain, the isolation method was modified by reducing homogenizing power, and filtration was omitted to improve recovery (Fig. 1A). Consequently, the fraction enriching neonatal brain capillaries (nBC fraction) was obtained from a single neonatal brain (postnatal day 7), as shown in Fig. 1B and C. The diameter of the isolated capillaries was 5–10 μm as previously reported [22]. The enrichment of brain capillaries was examined by Claudin-5, which is specifically expressed in BCECs (Fig. 1D). Claudin-5 was detected in the nBC fraction, but not in the whole brain lysate, suggesting that brain capillaries were concentrated in the nBC fraction.

The enrichment of brain capillaries in the nBC fraction was further evaluated by proteomics. The nBC fraction was prepared from a single neonatal brain by four biological replicates. The total protein amount of the fraction was 31.5 ± 9.7 μg (n = 4), which is less than half of the fraction from a single adult mouse brain (73.5 μg) [15] but sufficient for proteomic analysis. The analysis identified 4,362 proteins in the nBC fraction and 5,090 proteins in whole brain lysate; 3,782 proteins were identified in both fractions, and 2,739 proteins demonstrated significant differences (p < 0.05, Additional file 1: Table S1). The reproducibility among replicates was evaluated by the coefficient of variance (%CV) of the quantified values of the 4,362 proteins identified in the nBC fraction. The average %CV was 18.9%, and the 75^th percentile was 23.8% (Fig. 2A). The reproducibility of the enrichment process was assessed by the %CV of nBC-to-brain enrichment ratio (Fig. 2B). The average %CV of the enrichment ratio was 23.8%, and the 75^th percentile was 30.1%, suggesting sufficient reproducibility in the preparation of the nBC fraction for analyzing the protein expression profile.

The enrichment of proteins selectively expressed in brain capillaries was investigated to assess the enrichment of brain capillaries in the nBC fraction (Table 1). The nBC-to-brain enrichment ratios of the selective proteins in the brain capillary were greater than 5.23, with a maximum value of 20.8. The enrichment ratio of Claudin-5 was 17.5, and its concentration in the nBC fraction is consistent with the result of the western blot shown in Fig. 1D. Enrichment ratios were also calculated for the marker proteins of neurons, astrocytes, pericytes, oligodendrocytes, and microglia (Table 1). The enrichment ratios of markers for neurons, astrocytes, and oligodendrocytes were less than 1, suggesting the exclusion of those cells during the capillary isolation from the brain. In contrast, the enrichment ratios of the markers for pericytes and microglia were significantly greater than 1, indicating the inability to remove pericytes and microglia from the nBC fraction. The range of the ratio of pericyte markers was similar to that of brain capillary markers (5.51–16.2 vs. 5.93–20.8), while that of the microglia maker was lower (about 2). Therefore, pericytes were enriched in the fraction in a manner similar to that in brain capillaries, and the enrichment of microglia was lower than that of pericytes.

Contamination of the cerebral artery and venous vessels was assessed using proteins selectively expressing vascular smooth muscle cells that adhere to large cerebral blood vessels [23]. The enrichment ratios of aortic smooth muscle actin (Acta2), transgelin (Tagln), and myosin-11 (Myh11) ranged from 1.02 to 6.80 (Table 1), which were lower than those of the marker proteins for BCECs or pericytes. Therefore, the arteries and venous vessels were retained in the nBC fraction but were not as concentrated as the brain capillaries and pericytes.

To evaluate changes in protein expression in mouse brain capillaries between neonates and adults, we compared the proteomic data of brain capillary fractions isolated from the single brain of a neonatal mouse and those of an adult mouse (nBC and aBC fraction, respectively). The proteomic data of the aBC fraction reported in our previous study was re-analyzed in concomitance with the present nBC fraction data [15]. Consequently, 3,673 and 3,680 proteins were identified in the nBC and aBC fractions, respectively, and 2,816 proteins were identified in both fractions (Additional file 1: Table S2). The number of identified proteins decreased compared to that of the analysis in the last section as the mass spectrometry data obtained from different measurements were compared. To assess the reproducibility of preparation between the nBC and aBC fractions, the %CV values of the intensity of the common 2,816 proteins were compared. The average %CV value of the nBC fraction was significantly greater than that of the aBC fraction (19.3% vs 16.5%, p < 0.001, Fig. 3A). However, the distribution of %CV values were similar between both fractions, suggesting that the reproducibility of preparation was slightly lower in the nBC fraction than in the aBC fraction, but the variation was small for comparison of protein expression. Moreover, the brain capillary (BC)-to-brain enrichment ratios of 2,155 proteins, which were identified in all brain capillary fractions and all brain lysate samples, were compared. As shown in Fig. 3B, the enrichment ratios were significantly correlated between neonatal and adult brains (r = 0.832, p < 0.0001), suggesting that the enrichment manners are similar in the fractions prepared from neonatal and adult brains.

We compared the expression levels of proteins involved in the transport and cell junctions at the BBB between the nBC and aBC fractions (Table 2). Transporters, Abcb1a/Mdr1a, Abcc4/Mrp4, Abcg2/Bcrp, Slc2a1/Glut1, Slc3a2/4F2hc and Slco1a4/Oatp1a4 were significantly less expressed, while Slc1a4/Asct1, Slc7a1/Cat1, Slc16a1/Mct1 and Slco1c1/Oatp1c1 were significantly more expressed in the nBC fraction than in the aBC fraction. Among receptors, transferrin receptor were 4.11-fold more expressed in the nBC fraction than in the aBC fraction, while FcRn in the nBC fraction were expressed at 48.4% of that in the aBC fraction. As tight and adherens junction proteins, the expression of Claudin-5 was 1.49 higher in the nBC fraction than in the aBC fraction, whereas occludin, JAM-1 and Cadherin-5/VE-cadherin expression in the nBC fraction significantly decreased to 25.8%, 46.1%, and 35.1%, respectively, of that in the aBC fraction.

To perform a comprehensive comparison, we extracted the proteins with a BC-to-brain enrichment ratio greater than 5 in either the nBC or aBC fractions as the BC-enriched proteins, because the proteins selectively expressed in brain capillaries had a BC-to-brain enrichment ratio greater than 5.23 (Table 1). Consequently, 504 proteins were extracted, and 191 proteins showed a significantly different expression between fractions, with an nBC-to-aBC ratio > 2 or < 0.5, and a p-value < 0.01 (Fig. 4A, Additional file 1: Table S3). Among the proteins significantly more expressed in the nBC fraction than in the aBC fraction, Slc38a5/Snat5, a sodium-coupled neutral amino acid transporter 5, showed the greatest nBC-to-aBC ratio among transporters (9.83-fold, Fig. 4A). Therefore, by focusing on amino acid transporters, 9 proteins were detected in the nBC and aBC fractions, including non-BC-enriched proteins (Fig. 4B). Slc7a1/Cat1, Slc1a5/Asct2, Slc1a4/Asct1, and Slc38a5/Snat5 were > 1.8-fold more expressed in the nBC fraction than in the aBC fraction. The nBC-to-brain enrichment ratios of these transporters were all > 5.51, except for that of Slc1a4/Asct1. The expressions of Slc7a2/Cat2, Slc38a2/Snat3 and Slc7a5/Lat1 were not significantly different between the fractions. The expressions of Slc1a2/Eaat2 and Slc1a3/Eaat1 were significantly < 0.386-fold lower in the nBC fraction than in the aBC fraction, with nBC-to-brain enrichment ratios of < 0.506.

Network analysis was conducted with 191 proteins showing significantly different expression between the nBC and aBC fractions in Fig. 4A. Consequently, 60 proteins were classified into one large cluster (red symbols in Fig. 5A; details are shown in Additional file 2: Figs. S3 and S4). In this cluster, terms related to the extracellular matrix, laminin, integrin, and collagen were enriched (Fig. 5B). The protein expression levels of integrins and laminins were lower in the nBC fraction than in the aBC fraction (Fig. 5C). In contrast, the expressional changes were different among collagen subtypes: Col26a1 and Col3a1 were > 6.37-fold more expressed in the nBC fraction than in the aBC fraction, while Col4a1 and Col4a2 were < 0.330-fold less expressed in the nBC fraction than in the aBC fraction. These results suggest that extracellular components around brain capillaries are different between neonates and adults.