A brain microvasculature endothelial cell-specific viral vector with the potential to treat neurovascular and neurological diseases

Abstract

Gene therapy critically relies on vectors that combine high transduction efficiency with a high degree of target specificity and that can be administered through a safe intravenous route. The lack of suitable vectors, especially for gene therapy of brain disorders, represents a major obstacle. Therefore, we applied an in vivo screening system of random ligand libraries displayed on adeno-associated viral capsids to select brain-targeted vectors for the treatment of neurovascular diseases. We identified a capsid variant showing an unprecedented degree of specificity and long-lasting transduction efficiency for brain microvasculature endothelial cells as the primary target of selection. A therapeutic vector based on this selected viral capsid was used to markedly attenuate the severe cerebrovascular pathology of mice with incontinentia pigmenti after a single intravenous injection. Furthermore, the versatility of this selection system will make it possible to select ligands for additional in vivo targets without requiring previous identification of potential target-specific receptors.

Keywords: adeno‐associated virus; brain microvascular endothelial cells; gene therapy; neurovascular diseases.

Figure 1. Capsid variants of a random AAV display peptide library during five rounds of in vivo selection enriching for brain‐homing library particles

1. Randomly chosen library peptide inserts (shown in single‐letter code) recovered from the brain. Ten clones were sequenced after each round of selection. Intravenously injected library particles were allowed to circulate for two days in each selection round (n = 1 animal/selection round, age 8–12 weeks). Five selection rounds were performed. Amino acids characterizing the emerging motifs are highlighted in different colors. The consensus motif of each selection round is shown at the bottom of each column. Sequences were considered to show a consensus if at least four different clones displayed the same amino acids at the same position or at two adjacent positions.

2. In vivo transduction profile of luciferase reporter vectors displaying variants of the enriched library capsid sequence motifs or unmodified wild‐type AAV2 control capsid. Four weeks after i.v. injection of 5 × 10¹⁰ genomic particles/mouse containing a CMV‐luciferase reporter gene, luminescence was measured in the brain and control organs. Data are shown as bars (mean) with plotted individual data points (n = 3 animals/group, age 8–12 weeks).

Figure EV1. Luminescence mediated by recombinant AAV2 vectors displaying brain‐targeted peptides that were enriched by in vivo screening of a random AAV display peptide library

Tissues were analyzed 28 days after tail vein injection of vectors harboring the luciferase gene under the control of the CMV promoter (5 × 10¹⁰ genomic particles/mouse, age 8–12 weeks). Luminescence was normalized to total protein content in the tissue lysates. Data are shown as bars (mean) with plotted individual data points (n = 3 animals/group).

Figure 2. In vivo luminescence imaging of mice after intravenous injection of brain‐enriched AAV2 vectors carrying a luciferase reporter gene under the control of the CAG promoter

1. Luciferase reporter gene vectors displaying the brain‐targeted peptide NRGTEWD (“BR1”), unmodified wild‐type AAV2 capsid, or the previously reported brain‐targeting peptide DSPAHPS (PPS). Vectors were intravenously injected into mice (5 × 10¹⁰ genomic particles/mouse). Panels show representative examples of n = 5 animals, age 8–12 weeks per group. Animals were imaged in dorsal (left panel), ventral (second from left panel), and lateral (second from right and right panel) positions, 14 days after vector injection.

2. Close‐up imaging of AAV‐BR1‐treated mice. Mice were imaged in dorsal and lateral positions in a different color scheme allowing detailed visualization of the transduced brain in living animals, even as late as at day 264 after vector injection.

3. Virtual sections: sagittal, coronal, and transaxial (left panels) and three‐dimensional reconstruction (right panel) of the luminescence images of a mouse injected with BR1 vector (as in A). Images were obtained by measuring different wavelengths of the emitted light (Living Image software), confirming the brain as exclusive source of luminescence.

Figure 3. Long‐term transgene expression in the brain mediated by AAV‐BR1 vector

AAV‐BR1 vector harboring the luciferase gene under the control of the CAG promoter was administered intravenously (5 × 10¹⁰ genomic particles/mouse, age 8 weeks). Long‐term transgene expression was analyzed by in vivo bioluminescence imaging at 16 time points during a 665‐day period (n = 2 animals). Original images of analyzed animals (above) and quantification of luminescence in the brain as the region of interest = ROI (bottom).

Figure 4. Luciferase transgene expression and vector biodistribution in tissue lysates

Luciferase activity and vector copy numbers were determined in tissue lysates 14 days after vector administration (5 × 10¹⁰ genomic particles/mouse, age 8–12 weeks).

1. Vector‐mediated luminescence. Transgene expression was measured in AAV‐BR1 harboring the luciferase gene under the control of the CAG promoter and control vectors (AAV‐PPS and wild‐type rAAV2) in brain and off‐target control organs (left panel). Comparison of luminescence in the brain mediated by AAV‐BR1 and control vectors (right panel). ****P < 0.0001 (BR1: brain vs. all), ***P = 0.0002 (PPS: heart vs. brain/muscle), ***P = 0.0001 (PPS: heart vs. lung/liver/spleen/kidney), *P = 0.0101 (WT: heart vs. lung), *P = 0.0120 (WT: heart vs. spleen), *P = 0.0148 (WT: heart vs. kidney), *P = 0.0113 (WT: heart vs. muscle), **P = 0.0011 (brain: BR1 vs. PPS), ***P = 0.0009 (brain: BR1 vs. WT).

2. Biodistribution of AAV‐BR1 and control vectors (AAV‐PPS and wild‐type rAAV2) in brain and off‐target control organs, excluding spleen (left panel). Genome copy numbers of AAV‐BR1 and control vectors (AAV‐PPS and wild‐type rAAV2) in the brain (right panel). ****P < 0.0001 (BR1: brain vs. liver/kidney/muscle), ***P = 0.0006 (BR1: brain vs. heart), ***P = 0.0003 (BR1: brain vs. lung), **P = 0.0025 (PPS: kidney vs. muscle), *P = 0.0378 (PPS: kidney vs. lung), *P = 0.0114 (PPS: kidney vs. liver), ****P < 0.0001 (WT: liver vs. all), **P = 0.0028 (brain: BR1 vs. PPS/WT).

Data information: Data are shown as mean + SEM, n = 5 mice per group. Data were analyzed by one‐way ANOVA, followed by Turkey's multiple comparison test. The mean of each column was compared to the mean of the column with the highest value.

Figure 5. AAV‐BR1‐mediated transduction of brain endothelial cells in vivo

C57BL/6 mice (age 8 weeks) were injected with AAV‐BR1 harboring an eGFP reporter gene under the control of the CAG promoter. Images show representative examples of n = 6 mice.

1. Representative images from cerebellum, olfactory bulb, striatum, cerebral cortex, and the spinal cord, 14 days after vector injection. BR1‐eGFP‐transduced cells (green) were positive for the endothelial marker CD31 (red). Scale bars represent 250 μm.

2. Higher magnification confocal images. The endothelial marker CD31 (red) colocalizes with vector‐mediated eGFP expression (green). Scale bars represent 50 μm (upper panel) or 10 μm (lower panel).

3. CD13 staining of pericytes. The vector‐mediated eGFP expression pattern (green) does not colocalize with CD13 (red). Scale bars represent 250 μm (upper panel) or 10 μm (lower panel).

4. Aquaporin 4 staining of astrocytic endfeet. The vector‐mediated eGFP expression (green) does not colocalize with aquaporin 4 (red). Scale bars represent 100 μm (upper panel) or 10 μm (lower panel).

5. Expression of eGFP in primary brain endothelial cells prepared from C57BL/6 mice, 14 days after injection with AAV‐BR1‐eGFP. Scale bars represent 500 μm (upper panel) or 100 μm (lower panel).

Figure EV2. Mouse primary brain endothelial cells transduced by brain‐targeted vectors

Vector‐mediated eGFP expression under the control of the CAG promoter (green), CD31 staining (red), and DAPI (blue; only right panel), 10 days after infection with 1 × 10¹⁰ genomic particles of recombinant vector per well. AAV‐BR1‐eGFP (upper panel), AAV‐PPS‐eGFP (middle panel), and AAV2‐WT‐eGFP (lower panel).

Figure EV3. Immortalized human cerebral microvascular endothelial cells (hCMEC/D3) transduced by brain‐targeted vectors

Vector‐mediated eGFP expression under the control of the CAG promoter (green) and DAPI staining of nuclei (blue) are shown. AAV‐BR1‐eGFP (left panel), AAV2‐WT‐eGFP (middle panel), and AAV‐PPS‐eGFP (right panel), 4 days after infection with 1.6 × 10¹⁰ genomic particles of recombinant vector per well. The bar graph on the right shows infectivity, counted as GFP‐positive in eight randomly taken images from two wells per virus. Data are shown as bars (mean) with plotted individual data points.

Figure 6. AAV‐BR1‐iCre‐mediated gene recombination in Ai14 Cre reporter mice

Two weeks after vector injection (1.8 × 10¹¹ genomic particles/animal) into 16‐week‐old mice, Cre‐mediated gene recombination driven by the CAG promoter was observed mainly in brain endothelial cells (lower panel, red). No recombination was observed in control animals without BR1‐iCre‐virus injection (upper panel). CD31 (green) was used as a marker for brain endothelial cells. Panels show representative examples of n = 3 animals. Scale bars represent 250 μm.

Figure EV4. Off‐target transgene expression in mice injected with eGFP vectors

Tissue sections of different organs other than the brain analyzed 14 days after intravenous injection of 1.8 × 10¹¹ genomic particles/mouse, age 8 weeks. Tissues were analyzed for vector‐mediated eGFP expression under the control of the CAG promoter (green) and DAPI staining (blue). AAV‐PPS‐eGFP (left panel) or AAV‐BR1‐eGFP (right panel). Scale bars represent 200 μm.

Figure EV5. Effects of different injection routes on AAV‐BR1 targeting

1. Representative images of mouse brain cortices showing the extent of endothelial transduction by AAV‐BR1‐CAG‐eGFP injected through 4 different routes (intravenous: i.v., intraperitoneal: i.p., intramuscular: i.m., and subcutaneous: s.c.) 14 days after virus injection. Scale bars represent 250 μm.

2. Tissue sections of different organs (liver, heart, kidney, and muscle) analyzed 14 days after injection of 1.8 × 10¹¹ genomic particles through 4 different routes (as in A). Tissues were analyzed for vector‐mediated eGFP expression (green) and DAPI staining (blue). Scale bars represent 200 μm.

Figure 7. Therapeutic use of AAV‐BR1: normalizing endothelial cell survival and blood–brain barrier permeability in neonatal incontinentia pigmenti mice (Nemo ^−/+) with AAV‐BR1‐NEMO

BR1‐mediated expression of NEMO or eGFP was driven by the CAG promoter.

1. Immunostaining of cerebral microvessels. Treatment with AAV‐BR1‐NEMO normalized string vessels (white arrows, highlighted in white square inset) in Nemo ^−/+ mice as compared to Nemo ^−/+ mice treated with the AAV‐BR1‐eGFP control vector at postnatal day 8 (P8). The upper left panel shows the staining in untreated wild‐type control mice. String vessels were identified as capillaries that have lost CD31‐positive (green) endothelial cells but stain for the basement membrane protein collagen IV (red). Scale bars represent 200 μm. The lower right panel summarizes quantitative analysis of string vessels in Nemo ^−/+ and control mice (Nemo ^Fl or wild‐type) at P0 (n = 3 animals/group) and at P8 (n = 6 animals/group, *P = 0.0125). String vessels in the cerebral cortex were quantified as percentage of total vessel lengths.

2. Quantification of active caspase‐3‐positive endothelial cells at P8 (n = 5 animals/group, *P = 0.0201).

3. Albumin in brain tissue as indicator for BBB leakage. In Nemo ^−/+ mice treated with AAV‐BR1‐NEMO, less albumin was found in brain tissue, indicating less BBB leakage. Representative Western blot with albumin from P8 Nemo ^−/+ mice treated with AAV‐BR1‐NEMO or AAV‐BR1‐eGFP control vector, respectively, as well as untreated wild‐type (WT) control mice. Right panel: quantitative analysis of the Western blots (n = 4 animals/group, *P = 0.0283).

4. Body weight of vector‐treated mice. WT control or Nemo ^−/+ mice at P8 treated with AAV‐BR1‐NEMO or AAV‐BR1‐eGFP (n = 18 animals/group, *P = 0.0038).

Data information: Data are shown as mean + SEM or as plotted individual points with bars representing the mean. Differences between vector‐treated Nemo ^−/+ mice were analyzed by unpaired t‐test. Source data are available online for this figure.

Figure 8. Normalizing endothelial cell survival and blood–brain barrier permeability by intravenous injection of AAV‐BR1‐NEMO in a conditional murine incontinentia pigmenti model (Nemo ^{be KO} mice)

BR1‐mediated expression of NEMO or eGFP was driven by the CAG promoter. All animals were at the age of 8–12 weeks.

1. Representative immunostainings of cerebral microvessels. String vessels (white arrows, highlighted in white square inset) were significantly reduced in Nemo ^beKO mice treated with AAV‐BR1‐NEMO compared to Nemo ^beKO mice treated with AAV‐BR1‐eGFP control vector. Nemo ^Fl mice served as control animals. Scale bars represent 200 μm.

2. Quantification of string vessel lengths in the cerebral cortex as percentage of total vessel lengths. Nemo ^beKO and Nemo ^Fl mice were treated with AAV‐BR1‐NEMO or AAV‐BR1‐eGFP control vector (n = 9 Nemo ^Fl animals + BR1‐eGFP, 10 Nemo ^beKO animals + BR1‐eGFP, 9 Nemo ^Fl animals + BR1‐NEMO, and 13 Nemo ^beKO animals + BR1‐NEMO), ****P < 0.0001.

3. Total vessel length measured as total CD31‐positive vessels. Vessels were restored in Nemo ^beKO mice treated with AAV‐BR1‐NEMO compared to the AAV‐BR1‐eGFP injected mice. Nemo^Fl mice served as a control (n = 9 Nemo ^Fl animals + BR1‐eGFP, 10 Nemo ^beKO animals + BR1‐eGFP, 9 Nemo ^Fl animals + BR1‐NEMO, and 13 Nemo ^beKO animals + BR1‐NEMO), ***P = 0.0007, **P = 0.0038 (NemoKO:eGFP vs. NemoFl:NEMO), **P = 0.0014 (NemoKO:NEMO vs. NemoKO:eGFP).

4. IgG and albumin Western blots of brain lysates. Less leakage in the BBB was detected in Nemo ^beKO mice treated with AAV‐BR1‐NEMO than in Nemo ^beKO mice injected with AAV‐BR1‐eGFP (the right panel indicates the quantified gel intensity under the various treatment conditions; n = 5 animals per group). Nemo ^Fl mice served as controls, ****P < 0.0001.

5. Quantitative immunoglobulin staining of coronal brain sections of Nemo ^Fl and Nemo ^beKO mice. Ig extravasation was significantly reduced in AAV‐BR1‐NEMO‐treated mice (n = 9 Nemo ^Fl animals + BR1‐eGFP, 10 Nemo ^beKO animals + BR1‐eGFP, 9 Nemo ^Fl animals + BR1‐NEMO, and 13 Nemo ^beKO animals + BR1‐NEMO), ****P < 0.0001.

6. Effect of AAV‐BR1 vector on BBB permeability. No vector (left) or empty AAV‐BR1 vector (right) was injected i.v. to wild‐type mice and BBB permeability was assessed by extravasation of the fluorescent tracer sodium fluorescein (n = 7 animals per group). No significant difference was detected (n.s.).

Data information: Data are shown as mean + SEM. Data were analyzed by two‐way ANOVA followed by Bonferroni's post‐test (A–E) or by Student's t‐test (F). Source data are available online for this figure.