Evaluation of Aav Capsids and Delivery Approaches for Hereditary Hemorrhagic Telangiectasia Gene Therapy

Nosebleeds and intracranial hemorrhage from brain arteriovenous malformations (bAVMs) are among the most devastating symptoms of patients with hereditary hemorrhagic telangiectasis (HHT). All available managements have limitations. We showed that intravenous delivery of soluble FMS-related tyrosine kinase 1 using an adeno-associated viral vector (AAV9-sFLT1) reduced bAVM severity of endoglin deficient mice. However, minor liver inflammation and growth arrest in young mice were observed. To identify AAV variants and delivery methods that can best transduce brain and nasal tissue with an optimal transduction profile, we compared 3 engineered AAV capsids (AAV.cc47, AAV.cc84 and AAV1RX) with AAV9. A single-stranded CBA promoter driven tdTomato transgene was packaged in these capsids and delivered intravenously (i.v.) or intranasally (i.n.) to wild-type mice. A CMV promoter driven Alk1 transgene was packaged into AAV.cc84 and delivered to PdgfbiCre;Alk1f/f mice through i.v. injection followed by brain AVM induction. Transduced cells in different organs, vessel density and abnormal vessels in the bAVMs, and liver inflammation were analyzed histologically. Liver and kidney function were measured enzymatically. Compared to other viral vectors, AAV.cc84, after i.v. delivery, transduced a high percentage of brain ECs and few hepatocytes; whereas after i.n. delivery, AAV.cc84 transduced ECs and perivascular cells in the brain, and ECs, epithelial cells, and skeletal muscles in the nose with minimum hepatocyte transduction. No changes to liver or kidney function were detected. Delivery of AAV.cc84-Alk1 through i.v. to PdgfbiCre;Alk1f/f mice reduced bAVM severity. In summary, we propose that AAV.cc84-Alk1 is a promising candidate for developing gene therapy in HHT patients.


Introduction
Hereditary Hemorrhagic Telangiectasia (HHT), also known as Osler-Weber-Rendu Syndrome, is an autosomal dominant disorder characterized by the development of telangiectasia (small arteriovenous malformations) on the mucosa and arteriovenous malformations (AVMs) in various organs including brain.HHT is the second most common genetic bleeding disorder affecting 1 in 5,000 individuals or approximately 60,000-70,000 Americans [1].There is currently no FDA approved treatment for HHT patients [1].
Although several angiogenic proteins are dysregulated in HHT [2], elevated levels of vascular endothelial growth factor (VEGF) [3,4] is of clinical interest due to the availability of targeted anti-VEGF agents.VEGF plays an important role in AVM pathophysiology [3,5,6].VEGF stimulation is necessary for AVM formation and progression in the brain and skin of adult mice [5,[7][8][9].Therefore, anti-VEGF could be a therapeutic option for HHT patients.However, systemic therapies with anti-VEGF agents, such as antibodies and kinase inhibitors, have many drawbacks [10][11][12] and need long-term repeated administration, because telangiectasias and bAVMs are chronic lesions, in which angiogenesis could be activated from time to time [13,14].Adeno-associated viral vector (AAV) mediated gene transfer has many advantages over direct systemic delivery of anti-angiogenic agents because it can mediate long-term transgene expression [15,16], which makes repeated dosing unlikely to be needed.In addition, there are many engineered AAV capsids that make targeted delivery of therapeutic genes possible.We previously showed that in mice, intravenous (i.v.) delivery of AAV9-sFLT1, that has a recombinant AAV serotype 9 capsid and expressing the soluble FMS-related tyrosine kinase 1 (sFLT1), reversed bAVM phenotypes [17].However, systemic sFLT1 expression caused minor liver in ammation and growth arrest of young mice [17].Severe toxicity in nonhuman primates (NHPs) and piglets following high dose i.v.administration of AAV9-like vectors has also been reported [18,19].Thus, AAV9 is not an optimal vector for development of gene therapy for HHT patients.
To identify an AAV capsid(s) that can restrict therapeutic gene expression in the brain and nose with minimum hepatocyte transduction, we screened several engineered AAV capsids and discovered that one of the engineered AAV capsids, AAV.cc84, can predominantly transduce brain endothelial cells (ECs) after i.v delivery.More importantly, AAV.cc84 can transduce brain ECs and peri-vascular cells after intranasal (i.n.) delivery with minimal transduction of hepatocytes.Intra-nasal delivery of AAV.cc84 also transduced nasal epithelial cells and skeletal muscle cells.
The therapeutic e cacy of AAV.cc84 was tested in PdgfbiCre;Alk1 f/f mice that develop bAVMs upon induction of brain focal angiogenesis and Alk1 deletion in ECs [8].Delivery of AAV.cc84-Alk1 to PdgfbiCre;Alk1 f/f mice through i.v.mediated Alk1 expression in brain ECs, reduced vessel density and abnormal vessels, and increased vascular pericyte coverage in bAVMs.Therefore, AAV.cc84 is a promising vector for developing gene therapy strategy for treating HHT patients.

Materials and Methods
Engineered AAV capsids.
Three engineered AAV capsids (AAV.cc47,AAV.cc84 and AAV1RX) were produced using methods published in previous papers [20][21][22][23].Their transduction e ciency and cell speci city were compared with AAV9.A single-stranded CBA promoter driven tdTomato transgene was packaged in these capsids to mediate ectopic expression of a red uorescent protein (RFP).A CMV promoter driven human Alk1 cDNA was packaged into AAV.cc84 to test the therapeutic e cacy.

Animals
All protocols and experimental procedures for using laboratory animals were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of California, San Francisco (UCSF).Animal husbandry was provided by the staff of the IACUC, under the guidance of supervisors who are certi ed Animal Technologists, and by the staff of the Animal Core Facility.Veterinary care was provided by IACUC faculty members and veterinary residents located on the San Francisco General Hospital campus.All mice were housed in a pathogen-free area in 421×316 cm 2 cages with 12 hours light/dark cycle.
Wild type (WT) C57BL/6 mice were purchased from the Jackson Laboratory (Bar harbor, ME) and used for testing viral transduction e ciency and speci city.PdgfbiCreER;Alk1 f/f mice that express an ECspeci c TM inducible Cre recombinase and have their Alk1 gene exons 4-6 oxed [24] were used to test the therapeutic e cacy of AAV.cc84-Alk1 vector.

BAVM model induction and viral vector delivery
For testing viral transduction e ciency and speci city, AAV vectors were i.v. or i.n.delivered to WT mice.
Samples were collected 4 weeks after vector delivery.For i.v.delivery, 1X10 11 viral genomes (vgs) were suspended in 100 ml of AAV formulation buffer combining 500 ml of 1X dPBS with 500 µl of 1 M MgCl2 and 50 µl of 10% (wt/vol) Pluronic F-68 [23] and were injected to each mouse through tail vein.For i.n.delivery, 20 µl of viral suspension containing 2X10 10 vgs was dripped into the mouse nostrils, 10 µl each side.
For injection of AAV-VEGF into the brain, mice were anesthetized by 4% iso urane inhalation and were placed in a stereotactic frame with a holder (David Kopf Instruments, Tujunga, CA).A burr hole was drilled in the pericranium, 2 mm lateral to the sagittal suture and 1 mm posterior to the coronal suture.
Two µl of viral suspension containing 2X10 9 vgs of AAV-VEGF was stereotactically injected into the right basal ganglia, 3 mm underneath the brain surface, at a rate of 0.2 µl per minute using a Hamilton syringe.
The needle was withdrawn after 10 minutes, and the wound was closed with a 4 − 0 suture.

Immuno uorescence staining and image quanti cation.
For testing AAV transduction e ciency and speci city, mice were anesthetized with iso urane inhalation and perfused with PBS through the left cardiac ventricle followed by 4% paraformaldehyde.The brains, hearts, lungs livers, intestines and noses were collected, incubated in 4% paraformaldehyde overnight, and then in 4% paraformaldehyde containing 30% sucrose for 1 week and then frozen in dry ice.Brains were sectioned into 20-µm-thick sections and the other organs were sectioned into 10-µm-thick sections using a Leica RM2155 microtome (Leica Microsystems, Wetzlar, Germany).RFP expression in the hearts, lungs and livers were directly observed on Vectashield (Cat #H-1500, Burlingame, CA,) mounted sections.RFP expression in the brains and noses were analyzed on sections co-stained with anti-RFP antibody (Cat # 632496, Takara Bio Inc., Japan) and anti-CD31 antibody (marker for ECs, AF3628, R&D Systems, Minneapolis, MN) under the Keyence microscope (model BZ-9000, Keyence Corporation of America, Itasca, IL).The distribution of RFP + cells was quanti ed using NIH Image J 1.63 software.
To analyze the effect of AAV.cc84-Alk1 on bAVMs, the brain samples were frozen on dry ice and cut into 20-µm-thick coronal sections using a Leica CM1950 Cryostat (Leica Microsystems).Two sections per brain within the AAV-VEGF injection site were chosen for co-staining with antibodies speci c to CD31 (1:100, Cat # SC-18916, Santa Cruz Biotechnology, CA) and Alk1 (1:50, Cat # AF770, R&D systems) or CD 31 and CD13 (marker for pericytes, 1:100, Cat # AF2335 R&D Systems) separately.Three images were taken from each section (right of, left of, and below the injection site) using the 20X objective lens of the Keyence microscope (model BZ-9000, Keyence Corporation of America).Vessel density and vascular pericyte coverage were quanti ed using NIH image J 1.63 software.Dysplastic vessels (vessels with lumen size larger than 15 µm) were quanti ed manually.For all immunostaining, sections were incubated at 4°C overnight with the primary antibodies indicated above.Alexa Fluor 594-conjugated (1:500, Invitrogen, Carlsbad, CA) or Alexa Fluor 488-conjugated IgG (1:500, Invitrogen, Carlsbad, CA) were used as secondary antibodies.

Hematoxylin and eosin staining
Liver sections were stained with hematoxylin and eosin using a standard protocol.Ten sections from each liver were examined and imaged using a Keyence microscope to detect in ammatory cells.
Colorimetric assays for alkaline phosphatase activity, alanine transaminase activity and creatinine levels Approximately 500 µl of blood was collected through the facial vein of each mouse, kept on ice for 30 minutes, and then centrifuged at 1,000 g for 10 minutes.Sera were collected for subsequent analyses.

Statistical analyses
All quanti cations were performed by at least two researchers who were blinded to the group assignment.Sample sizes for each experiment are shown in the gure legends.Data were analyzed using GraphPad Prism 9 software.Differences between two groups were analyzed using the unpaired two-tailed t test.Differences among multiple groups were analyzed by one-way ANOVA followed by multiple comparisons with Tukey's correction.Data are presented as mean ± standard deviation (SD).A p value of less than 0.05 was considered statistically signi cant.
AAV.cc84 transduced nasal epithelial cells, ECs, and muscle tissue after i.n.delivery.
Since epistaxis from nasal telangiectasias is one of the most devastating symptoms of HHT patients, to test if AAV.cc84 can be used to deliver therapeutic genes into the nose, we checked RFP expression in the nose after i.n.delivery of AAV.cc84-RFP.We found AAV.cc84 transduced nasal epithelial cells, ECs, and skeletal muscle cells (Fig. 3).Therefore, AAV.cc84 shows promise for treating nasal telangiectasias.
AAV.cc84 did not alter liver and kidney function.
To test if AAV.cc84 in uenced liver and kidney function after i.v. or i.n.delivery, ALP, ALT, and Cr levels in the sera were analyzed.Their levels in AAV.cc84 treated mice were similar to the untreated controls for both i.v. and i.n.delivery (Supplementary Fig. 6).

Discussion
In this study, through screening 3 engineered AAV capsids, we found that AAV.cc84 is a promising vector for developing gene therapies to treat HHT bAVMs and nasal telangiectasia.AAV.cc84 was previously obtained by cross-species evolution in mouse and pig central nervous system following IV dosing [23].
Here, we demonstrate that AAV.cc84 transduces brain ECs predominantly after i.v.delivery and i.n.delivery.AAV.cc84 also transduced nasal ECs, epithelial cells, and skeletal muscle cells after i.n.delivery.In addition, AAV.cc84 can transduce cultured HUVECs, indicating that this vector can delivery genes into human ECs.Although AAV1RX could transduce brain ECs and other cells in the brain without transducing hepatocytes after i.v. and i.n.delivery, it was not as effective as AAV.cc84 and was very di cult to produce in large quantities.Although AAV.cc84 transduced some hepatocytes, cardiomyocytes, and smooth muscle cells on the intestine walls after i.v.delivery, minimal numbers of hepatocytes and cardiomyocytes were transduced after i.n.delivery.
Nasal bleeding and bAVM rupture are the major devastating symptoms of HHT patients.Ninety % of HHT patients develop recurrent, spontaneous epistaxis due to rupture of nasal telangiectasias, which have a profound impact on quality of life due to chronic bleed, development of iron de ciency anemia, and requirement of treatment with regular iron infusions and blood transfusions [26,27].Both surgical and pharmacological approaches are available for control of hemorrhage from mucosal telangiectasias, and these are often used in combination [28][29][30][31].Anti-angiogenic approaches have the promise of reducing hemorrhage, modifying the disease process, and delaying development of telangiectasias in addition to potentially causing regression of existing vascular malformations [1].However, these manipulations have limited e ciency and many limitations.Currently, there is no FDA approved treatment for prevent bleeding from nasal telangiectasia.
The rupture of bAVMs is unpredictable and life-threatening and can cause long-term disability.All current treatments for bAVM, including surgery, radiation, or endovascular embolization, are associated with considerable risks.Despite successfully achieving obliteration of the AVM nidus, there is growing recognition of AVM recurrence [32].In addition, the treatment selection for patients with unruptured bAVMs has become increasingly controversial because the natural history for these patients may be less morbid than invasive therapies [33,34].As for other HHT phenotypes, no speci c medical therapy is currently available to bAVM patients.
AAV-based gene therapy can mediate long-term transgene expression [15,16], which makes repeated dosing unlikely to be needed.Previously, we showed in mice that i.v.injection of AAV9-sFLT1 reversed bAVM phenotypes [17].However, uncontrolled systemic sFLT1 expression caused minor liver in ammation and growth arrest of young mice [17].In this study, we identify an engineered AAV vector, AAV.cc84, that can limit sFLT1 expression in the brain and nose without signi cant hepatocyte transduction through i.n.delivery.Since sFLT1 is a secreted protein, all transduced cells can release sFLT1 to its surrounding tissue.Cell-speci c transduction is not crucial.The reduced hepatocyte transduction is key to reducing systemic side effects.
Mutant HHT causative genes in ECs is essential for AVM initiation.Naturally, correction of Eng or Alk1 gene expression can rescue AVM phenotype [43].Kim et al. reported that overexpression of Alk1 in ECs can rescue phenotype in mice with Alk1 or Eng knocked out in ECs, without untoward effects [44].This observation indicates that ECs of all subtypes can tolerate ectopic expression of Alk1.Here, we showed that AAV.cc84 can transduce brain ECs after i.v. and i.n.delivery.We also showed in this study that i.v.delivery of AAV.cc84-Alk1 induced Alk1 expression in brain ECs and reduced bAVM phenotype severity in mice with Alk1 deletion in ECs.Previously, we showed that the number of Alk1 de cient ECs is positively correlated with bAVM severity [25].In this study, although i.v.delivery of AAV.cc84-Alk1 restored Alk1 expression in only 45.8% brain ECs, it still reduced bAVM severity.Therefore, AAV.cc84 is a promising candidate for delivering an Alk1 transgene into brain ECs and the nose of HHT subjects.The fact that overexpression of Alk1 rescued the phenotypes in Eng mutant mice suggests that overexpression Alk1 in ECs may be effective for both HHT2 (caused by ALK1 mutation) and HHT1 (caused by ENG mutation) patients, which together make up more than 80% of the HHT patient population.
In summary, in this study, we have identi ed an engineered AAV capsid, AAV.cc84, that can transduce brain ECs after i.v. or i.n.delivery.This capsid also mediated transduction of nasal ECs, epithelial cells, and skeletal cells.More importantly, this vector can transduce cultured HUVECs.We have also demonstrated that AAV.cc84 mediated Alk1 overexpression in brain ECs rescued bAVM phenotypes in PdgfbiCreER;Alk1 f/f mice.Therefore, this capsid can not only be used to mediated secreted protein expression but also can transduce therapeutic genes into brain ECs speci cally.
AAV.cc84 transduced brain ECs after i.v.injection with some hepatocyte transduction.AAV transduced cells were detected by their RFP expression (red, stained by an RFP speci c antibody).Brain ECs were stained by an anti-CD31 antibody (green).The white square region in the AAV.cc84 merged picture is enlarged to show colocalization of RFP and CD31 staining.Scale bar: 50 mm.Figure 2 AAV.cc84 transduced brain ECs and perivascular cells after i.n.delivery.Transduced cells were detected by their RFP expression (red, stained by an RFP speci c antibody).Brain ECs were stained by an anti-CD31 antibody (green).The white square region in the AAV.cc84 merged picture is enlarged to show colocalization of RFP and CD31 staining (Arrow head) and RFP + perivascular cells (Arrow).Scale bar: 50 mm.