Hyperosmolar blood–brain barrier opening using intra-arterial injection of hyperosmotic mannitol in mice under real-time MRI guidance

Abstract

The blood–brain barrier (BBB) is the main obstacle to the effective delivery of therapeutic agents to the brain, compromising treatment efficacy for a variety of neurological disorders. Intra-arterial (IA) injection of hyperosmotic mannitol has been used to permeabilize the BBB and improve parenchymal entry of therapeutic agents following IA delivery in preclinical and clinical studies. However, the reproducibility of IA BBB manipulation is low and therapeutic outcomes are variable. We demonstrated that this variability could be highly reduced or eliminated when the procedure of osmotic BBB opening is performed under the guidance of interventional MRI. Studies have reported the utility and applicability of this technique in several species. Here we describe a protocol to open the BBB by IA injection of hyperosmotic mannitol under the guidance of MRI in mice. The procedures (from preoperative preparation to postoperative care) can be completed within ~1.5 h, and the skill level required is on par with the induction of middle cerebral artery occlusion in small animals. This MRI-guided BBB opening technique in mice can be utilized to study the biology of the BBB and improve the delivery of various therapeutic agents to the brain.

Introduction

The treatment efficacy of many central nervous system (CNS) diseases is considerably hindered by the limited accessibility of therapeutic agents to the brain. Poor drug penetration is mainly caused by the blood–brain barrier (BBB). In the presence of an intact BBB, molecules and compounds larger than 180 Da do not typically enter the CNS¹. Therefore, chemotherapeutics and therapeutic biologics such as trophic factors do not cross the BBB as their molecular weights usually are between 200 and 1,200 Da². As a consequence, >98% of the existing pharmaceutical agents are unable to enter the brain tissue after intravascular delivery because of the BBB^3,4. Thus, strategies to safely and efficiently open the BBB are needed. Various methods and drugs have been developed to induce transient permeabilization of the BBB, and intra-arterial (IA) mannitol-mediated osmotic opening most frequently has been used in both preclinical and clinical studies^5,6. Brightman et al.⁷ pioneered this approach in the 1970s, demonstrating that IA injection of hyperosmotic mannitol induced endothelial shrinkage and opening of endothelial tight junctions, allowing the passage of agents into the brain. In the following decades, IA mannitol has been the primary method used in preclinical models and clinical studies to open the BBB before infusion of therapeutic agents, including chemotherapeutics, vectors and stem cells^5,6,8,9.

However, osmotic BBB opening (BBBO) is highly variable and inconsistent¹⁰ due to multiple factors, including not only mannitol dose and injection rate but also vascular anatomy and cerebral hemodynamics of individual patients or experimental animals¹¹. As a result, the published protocols show high variability (Table 1). For instance, in different studies using rats, the infusion rate of mannitol into the carotid artery ranged from 2.6 ml/min¹² to as high as 7.2 ml/min¹³. Notwithstanding the wide variance of these parameters, this spectrum greatly exceeds the physiological perfusion rate in the internal carotid artery (ICA), risking damage. Additionally, several studies surprisingly reported that intravenous injection of mannitol breaks the BBB, facilitating brain accumulation of therapeutics including stem cells^14–16.

Table 1 |.

Hyperosmotic mannitol-induced BBBO in rodent studies

Animal species (weight or age)	Injection route	Injection rate	Injection duration or dose	Reference
Rat (200–240 g)	IA	7.2 ml/min	30 s	13,30,51,52
Rat (350–450 g)	IA	7.2 ml/min	5 ml/kg	53
Rat (170–340 g)	IA	15 ml/kg/min	30 s	12,54
Rat (~325 g)	IA	5.4 ml/min	25 s	55,56
Rat (300–350 g)	IA	4.98 ml/min	30 s	57
Rat (250–300 g)	Intravenous	N.D	10 min	14
Mouse (22–24 g)	IA	1 ml/min	0.75 ml	8
Mouse (8–12 weeks)	Intravenous	N.D	<2 min	16
Mouse (N.D)	Intravenous	N.D	5 ml/kg	58

BBBO; osmotic blood–brain barrier opening; N.D. not determined.

In our prior rat studies, we have shown that IA infusion into the ICA at rates >1 ml/min is damaging and results in multiple white matter hyperintensities on T2-weighted images¹⁷. Similarly, in a reported mouse BBBO study, the procedure was performed with IA infusion of mannitol at a very high rate of 1 ml/min⁸, which is likely to have affected the BBB in addition to the mannitol effect. The excessive pressure likely affected BBB status by directly damaging the brain cells rather than by safely opening endothelial tight junctions. Our IA injection experiments in several species including rat, rabbit and pig^11,17,18, as well as a clinical investigation¹⁹, showed the value and importance of monitoring local transcatheter perfusion with real-time MRI to facilitate BBBO at physiologically relevant, nondamaging infusion rates. Dynamic susceptibility contrast MRI before BBBO enables precise fine-tuning of the infusion rate, which is a critical factor to determine the perfusion territory in the brain. Due to extensive collateralization of cerebral circulation, a low infusion rate cannot reach the brain or reaches only a very limited region, while an excessive infusion rate induces brain damage. Additionally, dynamic susceptibility contrast MRI allows users to estimate the BBBO area based on the perfusion territory of a contrast agent in each individual animal prior to mannitol injection. Thus, the use of dynamic susceptibility contrast MRI ensures a safe, effective and predictable BBBO with high reproducibility. Moreover, combining the use of advanced imaging with the IA route ensures the administration of therapeutics to the correct location following BBBO, and thereby is expected to improve therapeutic effects and reduce complications in the clinical setting.

Development of the protocol

The majority of previously reported preclinical studies on BBBO used rats or large animals, likely because of the technical challenges when working with mice and damaging experimental conditions. An important motivation to develop an effective BBBO protocol for mice is the access to an abundance of transgenic and knockout animals that can be used to gain deep insights into a variety of disorders^20–22. Moreover, it is necessary to establish an effective BBBO protocol that determines the optimized injection parameters for a robust BBBO effect without excessive brain damage. We recently developed a safe and reproducible technique to open the local BBB in mice under the guidance of interventional MRI with outcomes validated by both MRI (Gd-enhancement on T1-weighted imaging) and histology (Evans blue dye extravasation)^11,23,24. This experimental platform was then exploited for targeted IA delivery to the mouse brain of three major classes of therapeutics: monoclonal antibody, nanoantibody and polyamidoamine dendrimer^24–26. We found that IA delivery, even in the setting of the intact brain, resulted in faster and higher uptake of therapeutics in the brain compared with intravenous injection. More importantly, osmotic BBBO further potentiated their uptake in the brain. Our approach has enhanced drug delivery to the brain in mice, and it may also help researchers reestablish the technique as a promising method facilitating effective drug delivery in other species.

Overview of the procedure

The overall BBBO procedures in this protocol are illustrated in Fig. 1. Briefly, as BBBO agent we used 25% mannitol (wt/vol), which is administered using an IA route. The first key procedure is gaining access to the cerebral circulation through catheterization of the common carotid artery (CCA) in a mouse with ligation of extracerebral branches (Steps 11–20). The next critical step is securing the catheter and transition to the MRI suite for MRI-guided IA infusion (Steps 21–26). Animal setup and acquisition of MR images are routine and can be easily performed by a trained technician. IA infusion should be performed using an MRI-compatible infusion pump with dynamic imaging. Pre-injection of contrast agent is used to visualize the area of brain perfusion, and the infusion rate is modulated to achieve optimal targeting area. At that point, mannitol is infused using the same parameters and BBBO territory is confirmed by T1 MRI. We have used this approach in several studies as reported^23,25,26, and here we detail the protocol, including procedural videos, to help other laboratories use and adapt the method.

Fig. 1 |. BBBO under real-time MRI guidance in mice.

a, Schematic illustration of the protocol. b, Procedural steps. The step numbers correspond to the Procedure. BBBO, blood–brain barrier opening; CCA, common carotid artery; ECA, external carotid artery; ICA, internal carotid artery; MCA, middle cerebral artery; MRI, magnetic resonance imaging; OA, occipital artery; PPA, pterygopalatine artery.

Applications and adaptations of the approach

Neurooncology, neurodegenerative and cerebrovascular diseases are the best examples of research areas where this technique will be of the most interest and application. For example, a growing number of antitumor agents have proven high potency against cancer cells such as glioblastoma cells in vitro, but the BBB limits their access to tumor cells in vivo. Of note, IA delivery of a gene therapeutic with BBBO exhibited marked inhibition of brain tumor in mice; our approach could be used to further improve the IA delivery of anticancer therapeutics²⁷. Also, a number of neuroprotective agents that were promising for the treatment of neurodegenerative disorders such as Alzheimer’s disease and Parkinson’s disease^28,29 failed in clinical trials, and the value of these agents could be revisited once a safe and effective method for their targeting to the brain is available. Some reports also indicate the benefit of infusing stem cells synergized with BBBO to enhance their homing to brain lesions and amplify therapeutic effects^15,30.

Additionally, the BBBO model presented here focuses on a mouse model, which is easily accessible for studying brain tumors and neurodegenerative disorders. For example, the xenograft model of human brain tumor including glioblastoma (cell lines U87, U251, T98G and A172) is well established in immunodeficient mice such as nude mice, NOD/SCID mice³¹. In a recent study, we successfully modeled human pediatric (pediatric diffuse intrinsic pontine glioma) and adult gliomas (cell lines GBM1, GBM551) in immunocompetent mice through costimulatory blockade. Mice are the most common species used for studying neurodegenerative disorders³², including Alzheimer’s disease³³ or Parkinson’s disease³⁴. Therefore, the target audience for this BBBO protocol in mice is quite broad, given that it can be used by researchers interested in a variety of CNS disorders.

While MRI may be considered complex and expensive, the majority of research centers have access to animal scanners and the necessary expertise. We strongly encourage researchers to routinely perform the BBBO under MRI guidance; however, to reduce cost and time, once optimal infusion parameters, including rate and duration of IA mannitol, are established for individual experimental setups, studies may be continued under consistent experimental conditions without MRI.

Our protocol for opening the BBB can be adapted for studies in large animals such as rabbit, dog and pig; however, there are important considerations, including the costs required for access to specialized infrastructure and personnel such as clinical imaging equipment and a trained neurointerventionalist. In fact, we have already utilized an MRI-guided approach to estimate and optimize the BBBO territory in a rabbit model¹¹ and a canine model (P.W., M.J., M.P., unpublished data). Also, we implemented this approach clinically in a human patient with a brain tumor, using IA delivery of bevacizumab under real-time MRI following BBBO, which yielded a clinical benefit¹⁹.

Therefore, based on our extensive experience, we describe below the procedures necessary to adapt the protocol to larger animals and humans (Fig. 2). The main difference of BBBO between small animals and large animals/humans is in gaining vascular access. While in mice the procedure is invasive and requires surgery, in large animals and humans it is performed with a minimally invasive transcutaneous approach. In large subjects, under guidance of x-ray fluoroscopy, a microcatheter is advanced via transfemoral access for selectively catheterizing the artery in the brain. For adjustment of parenchymal perfusion territory, an MRI contrast agent (Feraheme, dissolved in saline at 1:100; 0.3 mg/ml) is injected using a standard infusion pump starting from the lowest rate for 20–30 s (0.06 ml/min in our studies) under real-time GE-echo planar imaging (GE-EPI) (echo time (TE) 36 ms, repetition time (TR) 3,000 ms, field of view 1,080 × 1,080 mm², matrix 128 × 128, slice thickness 0.7 mm and temporal resolution 3 s). The microcatheter tip position can be modulated if necessary. Notably, our study on eight rabbits¹¹ and three dogs revealed that any given IA infusion rate resulted in different brain perfusion territories in different animals within the same species, probably due to flow dynamics within a specific vascular territory and the individual hemodynamic variability of animals (P.W., M.J., M.P., unpublished data). Therefore, the transcatheter titration is strongly encouraged for each animal. Once the optimal infusion rate for reaching a desired perfusion territory is determined, 25% (wt/vol) mannitol is administrated using the same rate over 30 s. Intravenous gadolinium (Gd) is administered 5 min after mannitol injection with T1-weighted images (spin echo rapid acquisition with relaxation enhancement (RARE) sequence) acquired to assess BBB integrity. Gd enhancement in the perfusion territory observed on T1-weighted images indicates a successful BBBO. In the case of unsuccessful BBBO, an additional bolus of mannitol is immediately implemented for 30 s until the BBBO is observed on T1-weighted images.

Fig. 2 |. Schematic illustration of BBBO under real-time MRI guidance in large animals and humans.

Following anesthesia, a microcatheter is positioned via a transfemoral access for selectively catheterizing the artery in the brain under guidance of x-ray fluoroscopy. MRI contrast agent (Feraheme) is infused via the microcatheter for visualization and optimization of a desired parenchymal perfusion territory and the subsequent BBBO.

Alternative BBBO technologies and limitations of the approach

Recently, focused ultrasound (FUS) has emerged as a noninvasive targeted technique for BBBO and neurotherapeutics delivery. This new approach has been shown to be safe in animals and humans, and is currently being used in clinical trials for the treatment of many diseases, including Alzheimer’s disease and brain tumor^35–38. However, FUS may not be widely available in the clinic for many years. Additionally, FUS is a relatively expensive approach that requires more specialized equipment and technicians, whereas our catheter-based osmotic technique is available to a wider base of users. In the FUS setup, the agents (microbubbles and therapeutics) are typically administrated systemically, potentially exposing the patient to systemic adverse effects of the drug. On the contrary, in our model, the IA catheter is placed proximally to the targeted structure. After mannitol-mediated BBBO under MRI guidance, the same catheter is used for the infusion of therapeutic agents as a one-stop-shop procedure. This approach ensures both targeted disruption of BBB and highly localized deposition of therapeutic molecules. In the future, IA drug delivery could be used in combination with FUS BBBO to exploit the synergy between these two methods.

One limitation of our method is that in mice IA catheterization is a rather invasive procedure and requires permanently ligating the ipsilateral CCA, which was reported to result in a consistent reduction (17–19%) in blood flow in the frontoparietal cerebrocortical regions in the ligated hemisphere³⁹. Although we did not observe any ischemia on MRI images, such chronic cerebral hypoperfusion was shown to cause delayed white matter lesions and cognitive impairment⁴⁰. Additionally, the occipital artery (OA) is cauterized in our approach, which may result in ischemic damage in regions such as the external cranium and muscles of the sternomastoid region⁴¹. However, in large animals and humans, these potential effects are much less of a concern because neurointerventional surgery is minimally invasive and all arteries remain patent. A second limitation of our method is that we are not able to achieve superselective BBBO in the mouse brain because the size of cerebral vasculature prohibits the distal placement of the catheter. The relatively wide size of the BBBO region is also a limitation. Additionally, we found that our approach with intracarotid mannitol infusion at a safe rate is primarily routed to the deep brain structures without perfusion through the cerebral cortex, thus resulting in BBBO limited to deep structures. Consequently, BBBO does not consistently involve the cerebral cortex. This phenomenon is likely due to the specifics of blood supply and collateralization^42,43. Despite this phenomenon, a robust BBBO in the cortex is still achievable by following the optimized procedures that we recently developed²⁴. The last limitation relates to MRI image processing. In our protocol, for the adjustment of the BBBO area, based on contrast pre-injection via a catheter, we used the fast, real-time MRI sequence (dynamic GE-EPI). While rapid and sensitive, this sequence is subject to susceptibility artifacts and image distortion, especially during IA infusion. As a result, coregistration of EPI hypointensity (used for estimating BBBO territory) and T1 post-Gd (used to verify BBBO area) is challenging. Thus, we calculated the area percentage for each scan. This limitation is much less of a concern in large animals and in the clinical setting as larger brain size and low magnetic field help to reduce the artifacts and the use of a strong contrast agent will compensate for any loss of the signal to noise.

Experimental design

Controls

For the purpose of this protocol on image-guided BBBO, no treatment arm was involved; thus, a control animal group was not needed. The untreated contralateral hemisphere of the brain was deemed a sufficient and appropriate control to determine mannitol-induced local BBBO in the ipsilateral hemisphere by histology and MR imaging. To rule out a direct effect of the infusion on the BBB, we also performed a control treatment with IA saline at the same rate as mannitol. MR images and Evans Blue staining were performed afterwards to assess the BBB status.

Preparation of catheter

A 1- to 1.5-m-long micro-polyethylene catheter (PE-8–100, inner diameter (ID) 0.20 mm, outer diameter (OD) 0.36 mm) was used for each animal in our study. The length of catheter may vary depending on the distance between the infusion pump and the animal located in the isocenter of the scanner. To ensure a proper seal and successful delivery of the solution to the ICA, catheters are immersed in a droplet of a Krazy Glue at one end. The other end is coupled with a 30-Gauge half-inch needle connected to a syringe. The distal catheter tip with applied glue is then inserted into the ICA as described below.

Immunodeficient mice

While any strain of mice potentially may be suitable for this protocol, our overarching goal is to study the chemotherapy of brain tumors in combination with BBBO; hence, we use male immunodeficient (SCID/NCr) mice (6–8 weeks old, 20–25 g) to avoid rejection of human tumor cells within the mouse host. Immunodeficient mice must be insulated from pathogens, housed with temperature- and humidity-controlled conditions, and fed with sterilized food and water to maintain optimal health. All surgical instruments should be sterilized. Following IA mannitol-induced BBBO, the mice are observed daily for 7 d for signs of infection, loss of body weight, and overall health. We successfully used this protocol for adult mice, both male and female, without noticeable difference between sexes (C.C., P.W., M.J., M.P., unpublished data). While it should be possible to perform this procedure in younger mice, smaller size of arteries may prove challenging.

MR imaging

Baseline T2- and T1-weighted images (before infusion of any substance) were performed for anatomical reference. T1 scans provide the best contrast visible as hyperintensity for paramagnetic contrast agents such as Gd-containing compounds. T2 scans in particular provide excellent soft tissue contrast, and damage, edema or inflammation can be easily appreciated. Infusion of an MRI contrast (superparamagnetic iron oxide (SPIO) or Gd) into cerebral arteries results in a reduction of a signal on dynamic gradient echo T2* sequence sensitive to magnetic field heterogeneity (contrast visible as hypointensity), and the signal void immediately clears once the infusion is stopped. GE-EPI can sample this T2* hypointensity in the brain at a temporal resolution of 2 s per volume. Such real-time MRI allows precise spatiotemporal visualization of the parenchymal perfusion territory. Before and after mannitol treatment, T1-weighted images were acquired to verify BBB status. T2-weighted scans were repeated to evaluate the brain injury. Each imaging protocol is described in detail under Step 23. Of note, 11.7 T Bruker MRI was used for our protocol establishment, although 9.4 T and 7 T Bruker MRI are also suitable providing sufficient resolution and image quality in mice using our proposed MRI sequences.

BBBO

Hyperosmotic mannitol (25%; wt/vol) is used in our method as BBBO agent, and this same product is routinely and safely used clinically, administered via an intravenous route, to decrease cerebral edema or intracranial pressure^44,45. As preparation before the injection, the modified microcatheter is filled with heparinized saline (1,000 units per ml, 1:50), and particular attention is given to eliminating all the air from the injection line to prevent coagulation and stroke. The proximal CCA is permanently ligated, and the tip of the microcatheter is placed in the ICA via a small arteriotomy on CCA. Also, both the pterygopalatine artery (PPA) and external carotid artery (ECA) need to be temporarily ligated, to permit all the solution to enter the cerebrum. Based on our extensive experience, the mannitol infusion rate and duration are two other critical factors. Here, real-time MRI is used to determine a safe IA infusion rate starting from 0.05 ml/min up to a maximum safe infusion rate of 0.15 ml/min using an MRI contrast agent. Two classes of contrast agents were tested (SPIO and Gd). Both are suitable, and even though Gd is typically used as T1 agent, when injected intraarterially its higher concentration produces T2 contrast as in the case of SPIO. Pre-injection with a contrast agent allows one to estimate the territory of BBBO. The infusion rate was controlled by an MRI-compatible syringe pump. Once the perfusion territory was optimized, IA mannitol was administrated at the same rate. In our experimental setup, we found that the rate of 0.15 ml/min over 30 s duration is not sufficient to induce the opening, while 1 min duration of IA mannitol induced BBBO consistently without complications as evaluated by MRI and histopathology²³. IA saline administered at the same rate did not result in BBBO, thereby excluding the effect of infusion itself on the BBB status (Supplementary Fig. 1). Finally, the temporary ties at ECA and PPA are carefully unfastened, and the distal CCA near the ICA–ECA bifurcation is permanently ligated to close the arteriotomy after the microcatheter is removed.

In our protocol, infusion rate is the only adjustable parameter to safely achieve a desired brain perfusion region, whereas in large animals or in humans there is more space to control the territory such as with distal catheter tip placement. The advantage of this, however, is that MRI shows which region is targeted and the downstream results can be appropriately interpreted. In agreement previous studies^46–48, we also found that the BBB remains opened for 4 h. Importantly, the BBBO region correlates well with the brain territory highlighted by pre-injection of the contrast agent, indicating that the BBBO region is predictable. Therefore, the use of real-time MRI is essential when a precise BBBO region is required or detailed information about the exposed brain regions is important. Studies in which the precise BBBO location is not critical, such as those investigating only the effect of BBBO on uptake of a therapeutic in the brain, may be continued without MRI once optimal infusion parameters are established for the individual experimental setup.

Materials

Biological materials

• 6- to 8-week-old male SCID/NCr mice (CB17/Icr-Prkdc^scid/IcrCr, Charles River, cat. no. 561) ! CAUTION All experiments using live rodents must conform to relevant institutional and national regulations. All experimental protocols described here were approved by the Institutional Animal Care and Use Committee (IACUC) of the Johns Hopkins University ▲ CRITICAL For xenografting, immunocompromised hosts are essential. Mice should be housed under temperature- and humidity-controlled conditions with free access to sterilized food and water.

Reagents

• Krazy Glue (cat. no. KG82048SN)

• Hair-removing lotion (Nair; CVS, cat. no. 339818)

• Ethanol, 70% (vol/vol) (VWR, cat. no. 97064–768)

• SPIO nanoparticle formulation (AMAG Pharmaceuticals, Ferumoxytol, Feraheme, cat. no. NDC 59338–775-10)

• 25% (wt/vol) mannitol injection, solution (Hospira, cat. no. NDC 0409–4031-16)

• Gd: gadoteridol injection (BRACCO, NDC 0270–1111-16)

• Ketoprofen (Fort Dodge Animal Health, NDC 0856–4396-02)

• Enrofloxacin antibacterial injectable solution (PUTNEY, NDC 26637–721-02)

• 0.9% (wt/vol) sodium chloride injectable (Hospira, cat. no. NDC 0409–4888-10)

• Heparin sodium injection, solution 1,000 units per 1 ml (Upjohn, NDC 0009–0268-01)

• Isoflurane (Isoflurane USP, cat. no. NDC 66794–017-25) ! CAUTION Isoflurane is detrimental if not properly scavenged; long-term exposure may cause adverse health effects on the operator, including dizziness, headache, vomiting and nausea. An exhaust system is necessary to eliminate excessive isoflurane during the surgery.

• Oxygen for vaporized anesthetic (Airgas Healthcare, tank grade 1072)

Equipment

• Micro-polyethylene tubing 0.20 mm ID × 0.36 mm OD (SAI Infusion Technologies, PE-8-100)

• 30 gauge half-inch needle (PrecisionGlide IM; BD, cat. no. 305106)

• Small cotton-tipped applicator (Fisher Scientific, cat. no. 23-400-118)

• 15 ml Falcon tubes (Thermo Fisher Scientific, cat. no. 339650)

• 50 ml Falcon tubes (Thermo Fisher Scientific, cat. no. 339652)

• 1.5 ml Eppendorf tubes (Thermo Fisher Scientific, cat. no. 280175)

• 1 ml tuberculin syringe with 26 gauge needle (BD, cat. no. 309625)

• Surgical microscope (Leica, M320)

• Bruker Biospec 11.7 T scanner

• MRI operation system (Bruker ParaVision 6.0.1)

• Physiological monitoring system (Small Animal Instruments, Model 1030)

• Isoflurane vaporizer (Atlantic Biomedical, cat. no. ISOTEC W322941)

• MRI compatible programmable syringe pump (Harvard Apparatus, cat. no. PHD2000)

• Gemini cautery system (Braintree Scientific, cat. no. GEM 5917) ! CAUTION Average temperature of the cautery unit can reach 1,204 °C; be cautious when using the tool to avoid burns, and do not use in the presence of flammable materials.

Surgical tools

▲ CRITICAL All the surgical tools listed below must be sterilized before surgery.

• Vannas scissors (ROBOZ Surgical Instruments, RS-5608)

• Dumont #7 forceps (Fine Science Tools, cat. no. 11297–10)

• Dumont #5 forceps (Fine Science Tools, cat. no. 11251–20)

• Micro-Adson forceps (Fine Science Tools, cat. no. 11018–12)

• Surgical scissors, sharp (Fine Science Tools, cat. no. 14002–12)

• Fine scissors, sharp (Fine Science Tools, cat. no. 14060–10)

• Student halsted-mosquito hemostats (Fine Science Tools, cat. no. 91308–12)

• Crile-wood needle holders (Fine Science Tools, cat. no. 12003–15)

• Braided silk suture 4–0 for ligation (Surgical Specialties, cat. no. SP-114)

• Braided silk suture 3–0 attached needle for wound closure (Ethicon, cat. no. 632G)

Procedure

Preparation of catheters ● Timing 10 min for one catheter + drying time 12 h

• 1To prepare a 1- to 1.5-m-long catheter, first, using small scissors, cut one distal end of the catheter with a 45° angle (Fig. 3a). Insert a 30 gauge half-inch needle into the proximal end of the catheter (Fig. 3a).

  ▲ CRITICAL STEP To minimize damage to the artery wall, use a soft catheter. Therefore, a sharp tip is necessary for easier cannulation.

• 2Add a droplet of a Krazy Glue to the catheter as a bead ~5 mm from the distal tip (Fig. 3b). Keep the catheter secured with tape at room temperature (22–25 °C) on a trestle overnight (~12 h) to let it dry (Fig. 3c). A ~5-mm-long section of the catheter from the bead is left free of any glue to allow insertion into the artery lumen (Fig. 3d).

  ▲ CRITICAL STEP To shape a round bead (2.0 mm L × 1.0 mm W), rotate the catheter along its axis for 5–6 min until the glue thickens. The bead is critical to prevent the catheter from slipping out from the artery when the animal is transferred into MRI and during IA injection.

• 3Seat the 30 gauge half-inch needle firmly on a 1 ml syringe and flush the catheter with 100 μl of 70% (vol/vol) ethanol first, then wash with sterile dH₂O (three to four times, 100 μl per time).

  ▲ CRITICAL STEP Check that the distal tip of the catheter is entirely smooth and that all the couplings are not leaky. Do not force too much pressure during flushing, especially when the catheter is quite long because it may damage the coupling between the needle and the catheter.

  ? TROUBLESHOOTING

• 4Sterilize the rinsed catheter with UV light for 1–2 h in a sterile laminar flow hood. The catheter is now ready to be used.

  ■ PAUSE POINT Catheters can be preserved in plastic Ziplock bags for several weeks. It is practical to prepare larger number of custom catheters at this step and stock them for future use. The catheters should be tested for leaks, collapse and clogging prior to use.

Fig. 3 |. Schematic diagram of the catheter assembly.

a, Prepare a 1– 1.5-m-long catheter for MRI. The distal end is cut with a 45° angle; the proximal end is connected with a 30 gauge half-inch needle. b, Apply a droplet of Krazy Glue to the distal end of the catheter. c, Place the catheter, suspended on a clean trestle, and allow it to dry for 12 h. The bead displays an almost round shape. d, The portion between the tip and bead is ~5 mm long.

Preoperative preparation ● Timing 10–20 min

• 5Place 25% (wt/vol) mannitol solution in the heating incubator set at a temperature between 60 and 70 °C over 15 min to dissolve any precipitates. Cool mannitol to room temperature before use.

  ▲ CRITICAL STEP 25% (wt/vol) mannitol is prone to form crystals. Heating to 60–70 °C and mixing by hand (two to three times) is sufficient to dissolve the solution

• 6Dilute heparin as 1:50 in saline. Dissolve Feraheme (SPIO, ferumoxytol) in saline at 1:30 (0.3 mg/ml) and Gd (gadoteridol) in saline at 1:50 (5.59 mg/ml).
• 7Connect the catheter from Step 4 to a 1 ml syringe, rinse with sterile saline, and then fill with heparin from Step 6. The catheter is placed on the surgical table and ready for cannulation (in Step 20).
• 8Clean the operative table with 70% (vol/vol) ethanol and cover with a sterile drape to create a sterile environment for placing sterile tools.

  ▲ CRITICAL STEP Maintain sterility throughout the entire surgical procedure.

• 9Put the mouse in a chamber with 4% isoflurane to induce anesthesia (Fig. 4).

  ▲ CRITICAL STEP Monitor the animal during induction of anesthesia and during surgery; an excess of isoflurane can cause irregularities in breathing and death.

  ? TROUBLESHOOTING

• 10After confirmation of complete anesthesia by testing withdrawal reflexes to toe and tail pinch, transfer the mouse to a surgical bed with a nose cone delivering 1.5–2.0% isoflurane with oxygen flow at 0.5–1.0 L/min for surgery (Fig. 4).

  ? TROUBLESHOOTING

Fig. 4 |. Schematic diagram of the preoperative preparation of the mouse.

The mouse is placed in a chamber with 4% isoflurane to induce anesthesia, and 1.5–2% isoflurane is used to maintain the anesthesia. The hair in the operative area is removed using depilation cream (Nair). 0.5% iodophor and 70% ethanol are subsequently applied for disinfection of the operative area.

Catheterization procedure ● Timing 20–30 min

• 11Place the mouse in a supine position on a foam tap. Tape the forelimbs laterally to expose the ventral neck area and allow for a broad surgical space (Fig. 4).
• 12Apply depilation cream (Nair) using a small cotton-tipped applicator to the hair around the trachea. A few minutes later, remove the excess Nair and hair with a piece of gauze (Fig. 4).
• 13Dip a small cotton-tipped applicator in 0.5% iodophor (vol/vol), and use to disinfect the operative area on the mouse. Then disinfect the same area with 70% (vol/vol) ethanol (Fig. 4).
• 14Make a 20–30 mm midline skin incision. Gently separate sternohyoid muscles along the midline with a cotton swab, and retract the right sternohyoid tissue laterally to expose the trachea (Fig. 5a).
• 15Under an operating microscope, bluntly dissect the right CCA, which is exposed under the sternohyoid muscle and situated laterally to the trachea (Fig. 5b).
• 16Isolate the CCA along the carotid sheath via blunt dissection until the ECA and ICA bifurcation is exposed. Carefully separate the vagus nerve from the carotid artery without damaging it (Fig. 5c).

  ▲ CRITICAL STEP The vagus nerve and carotid artery are close. Extra attention must be paid to avoid damage to the vagus nerve during dissection. Be cautious not to compress or tug excessively the carotid arteries, which are fragile.

  ? TROUBLESHOOTING

• 17Temporarily ligate the ECA with a 4–0 silk suture, and cauterize the OA branching from ECA with the Gemini cautery system (Fig. 5d–g, Supplementary Video 1).

  ▲ CRITICAL STEP The cauterization of the OA allows for convenient dissection of the ICA to isolate the PPA. Cautiously cauterize the OA; be careful not to accidentally cauterize the ICA underneath the OA.

• 18Dissect the deep ICA until exposing the bifurcation of ICA and PPA. Cautiously isolate the PPA, and apply temporary ligation with a 4–0 silk suture (Fig. 5h,i, Supplementary Video 2).

  ▲ CRITICAL STEP It is critical to occlude the PPA; failure to do so will result in the majority of the solution entering extracerebral circulation.

• 19Permanently ligate the proximal CCA with a 4–0 silk suture to completely block blood flow. Use a 4–0 silk suture to temporarily ligate the trunk of CCA near the bifurcation, and place another 4–0 silk suture loosely around the CCA (Fig. 6, Supplementary Video 3).

  ? TROUBLESHOOTING

• 20Make a transverse incision on the CCA close to the permanent ligation using a surgical scissor, and then insert the tip of the catheter from Step 7 into the small arteriotomy. Tighten the loose suture around the CCA, and untie the ligated one. The blood backflow into tubing indicates successful catheterization. Then, manually advance the catheter toward the ICA until the bead on the catheter reaches the arteriotomy. Tighten the tubing with the 4–0 silk suture that has been used to permanently ligate the proximal CCA (Fig. 6, Supplementary Video 3).

  ▲ CRITICAL STEP The catheter must be filled with heparin to prevent coagulation. Any air bubbles must be eliminated before insertion to avoid stroke complications or imaging artifacts confounding imaging data. While advancing the catheter, you should align it with the natural route of the ICA.

  ? TROUBLESHOOTING

Fig. 5 |. Surgical procedures before catheter insertion.

a, Skin cut along the midline of the neck (Step 14). b, Dissection of superficial adipose tissue and the sternohyoid muscles to expose trachea (Step 15). c, Isolation of CCA (Step 16). d, Isolation of ECA (Step 17). e, Temporary ligation of ECA and exposure of OA and ICA (Step 17). f, The anatomical position of the arteries. g, Isolation of OA and subsequent cauterization (Step 17). h, Isolation of PPA (Step 18). i, Temporary ligation of PPA (Step 18).The IACUC at the Johns Hopkins University approved all experimental procedures shown in this figure.

Fig. 6 |. Schematic diagram of catheter cannulation.

Before arteriotomy and catheter insertion (lefthand picture), the proximal CCA is permanently ligated with a suture to block the blood flow and another suture around the distal CCA is temporarily ligated. A loose suture is then placed in between. Once the catheter is inserted (middle picture), the loose suture is tightened to secure the catheter, and at that point the temporary ligation is released. When the catheter is advanced into ICA (righthand picture), the suture released previously is tightened around the catheter to provide better stability. Finally, the suture around the proximal CCA is used to again tie the catheter. In total, the catheter is held by three sutures.

BBBO induction under MRI guidance ● Timing 20–30 min

• 21Quickly transfer the mouse with a secured catheter to a MRI scanner room and place the animal in an MRI animal bed in a prone position. The bed is equipped with a water heating system for homeothermic maintenance as well as a facemask connected to a veterinarian isoflurane vaporizer for delivery of gas anesthesia. Deliver isoflurane (1.5–2.0%) with oxygen flow at 1.5 L/min while the animal is undergoing imaging. Cover the mouse’s head with a dedicated surface coil (2 × 2 phased-array coil), and place a sensor pouch under the abdomen to monitor the respiration using a monitoring system. Slide the mouse into the magnet using automatic or manual drive (Fig. 7).

  ▲ CRITICAL STEP All the animal experiments were performed on an 11.7 T Bruker MRI. The 9.4 T and 7 T Bruker MRI can also provide sufficient resolution and image quality in mice using the proposed MRI sequences in our protocol, as confirmed by our recent studies on a 9.4 T Bruker MRI.

• 22Acquire baseline T2- and T1-weighted images as anatomical reference first.
• 23Then, connect the IA catheter to an MRI-compatible programmable syringe pump for infusion of contrast agent, SPIO (0.3 mg Fe/ml) or Gd (5.59 mg/ml). Acquire GE-EPI during infusion. Increase the infusion rate at increments of 0.05 ml/min until the contrast agent perfuses a desired size and location of the brain, as visualized on real-time MRI (Fig. 7, Supplementary Video 4). The imaging protocol for each sequence is presented in Table 2. The signal-to-noise ratio and contrast-to-noise ratio is presented in Supplementary Fig. 2.

  ▲ CRITICAL STEP When the syringe needs to be switched, care must be taken to eliminate air that may cause stroke.

  ▲ CRITICAL STEP The rate of 0.15 ml/min was sufficient to perfuse the desired brain region in our studies. The injection rate needed to reach desired perfusion territory may vary among mice.

  ▲ CRITICAL STEP While both SPIO and Gd can be used for visualization of transcatheter perfusion, Gd is superior to SPIO in this step as there is no need to inject extra Gd to verify the BBB status in Step 25. Either one can be used for brain perfusion prediction using GE-EPI sequence with the same parameters.

  ? TROUBLESHOOTING

• 24After the optimal infusion rate is determined using SPIO or Gd, immediately infuse mannitol at the same rate for 1 min via the IA catheter.
• 25After mannitol injection, administer Gd (0.75 mg/g for each mouse) intraperitoneally (i.p.) when SPIO is used as the perfusion contrast agent. Repeat T1-weighted images to visualize BBB breach. If Gd is used as the perfusion contrast agent, T1-weighted images are acquired upon completion of mannitol injection.

  ▲ CRITICAL STEP If there is no Gd enhancement in the perfusion region, indicating the BBBO is unsuccessful, immediately administer an additional bolus of mannitol for 30 s until the BBBO is observed on T1-weighted images.

• 26Upon the confirmation of BBBO, any potential therapeutic agents should be administered immediately using the IA catheter at the same speed as mannitol injection.

Fig. 7 |. Mouse MRI setup.

The mouse with a secured catheter connected to a syringe is transferred to a MRI scanner room and placed in a small MRI animal bed in a prone position. The syringe is connected to the MRI-compatible pump, which can be controlled outside the MRI scanner room. On the console, the images are acquired using a computer with installed Bruker ParaVision 6.0.1 system and the respiration is monitored.

Table 2 |.

Imaging protocol for each sequence

Parameter	T2-weighted image	T1-weighted image	GE-EPI
TE	10.0 ms	6.7 ms	9.7 ms
TR	2,500 ms	350 ms	1,250 ms
Number of repetitions	1	1	24
Scan time	240,000 ms	179,436 ms	60,000 ms
Bandwidth	37,500 Hz	100,000 Hz	454,545 Hz
Flip angle	90°	90°	90°
Number of slices	15	15	15
Slice orientation	Axial	Axial	Axial
Slice thickness	0.7 mm	0.7 mm	0.7 mm
Matrix	256 × 256	128 × 256	80 × 160
Field of view	14 × 14 mm	14 × 14 mm	14 × 14 mm
Readout orientation	Anteroposterior	Anteroposterior	Anteroposterior
Spatial resolution	0.05 × 0.05 mm2	0.11 × 0.11 mm2	0.09 × 0.16 mm2
Temporal resolution	2,500 ms	350 ms	1,250 ms

General postoperative care ● Timing 10 min per animal

• 27Retract the catheter until its tip reaches the ICA-ECA bifurcation, re-ligate the distal CCA with a 4–0 suture and then withdraw the entire catheter. Tie off the distal CCA trunk permanently with a 4–0 silk suture to ligate the arteriotomy (Supplementary Video 5).
• 28Carefully untie the suture on the ECA and PPA, which is important for preserving proper blood supply (Supplementary Video 5).

  ▲ CRITICAL STEP Restoring perfusion in ECA and PPA substantially improves animal recovery and general condition after surgery.

• 29Verify that there is no bleeding. Irrigate and close the incision, and switch off the anesthesia (Supplementary Video 5).
• 30Inject 300 μl of 0.9% (wt/vol) saline, i.p.; 5 mg/kg of body weight of ketoprofen (or a similar analgesic) and antibiotics (e.g., enrofloxacin) i.p. for relieving pain and preventing infection. Provide analgesia and antibiotics for 3 d post-surgery.

  ▲ CRITICAL STEP Reducing the pain and distress level is essential as it contributes to animal recovery after surgery and experimental outcomes.

• 31Observe animals for 7 d post-operation, including inspecting surgical wounds and signs of sickness or infection, and monitoring body weight.

Troubleshooting

Troubleshooting advice can be found in Table 3.

Table 3 |.

Troubleshooting table

Step	Problem	Possible reason	Solution
3	Catheter is leaky	The catheter is punctured at the connection with needle	Withdraw the needle, and cut off the end portion of the catheter and reconnect
9	Withdraw the needle, and cut off the end portion of the catheter and reconnect	Isoflurane causes respiratory depression	Immediately remove the animal from the chamber; if no respiratory movements are detected, compress the chest to stimulate breathing. Regularly stop compressions to check whether the animal can breathe spontaneously
10–29	Low respiratory rate	Isoflurane concentration is too low or high	Adjust isoflurane as needed
16	The animal becomes dyspneic or has sudden sporadic breaths	Vagus nerve is tugged excessively by forceps or ligated	Stop dissecting until the respiration recovers Verify whether the nerve is ligated with the carotid artery. If yes, remove sutures and retie the carotid artery only
20	The catheter is stuck at the bifurcation of ICA and ECA	The tip of the catheter goes into the trunk of the ECA	Retract the catheter to the CCA, and align and advance it toward the ICA
	No blood backflow	The catheter goes into the PPA	Retract the catheter to the ECA–ICA bifurcation site, and align and advance it along with the natural position of the ICA
23	Severe susceptibility artifact and image distortion during acquisition of GE-EPI scan	Magnetic field inhomogeneity	Repeat shimming making sure entire brain volume is included in the field of view. Reduce TE

Timing

Steps 1–4, preparation of catheters: 10 min for one catheter; drying time 12 h

Steps 5–10, preoperative preparation: 10–20 min

Steps 11–20, catheterization procedure: 20–30 min

Steps 21–26, BBBO induction under MRI guidance: 20–30 min

Steps 27–31, general postoperative care: 10 min per animal

Anticipated results

Real-time MRI estimates the perfusion territory in the mouse brain

We tested a broad range of transcatheter contrast infusion speeds ranging from 0.05 ml/min to 0.2 ml/min. At the speed of 0.05 ml/min, there was no drop of T2* intensity on dynamic scans, and the increase of speed from 0.05 ml/min to 0.1 min/ml resulted in a realtively limited perfusion area (Supplementary Fig. 3). An infusion rate of 0.15 ml/min resulted in a broader brain perfusion territory, which we considered as sufficient. (Fig. 8a), as visualized by a marked reduction in signal intensity for the duration of the injection bolus (Fig. 8b). Importantly, further increasing the infusion rate resulted in delayed brain injury (in three out of four mice) at the infusion rate of 0.20 ml/min, shown as T2 hyperintensity and further characterized by neuroinflammation on histology²³ (Supplementary Fig. 4).

Fig. 8 |. Use of real-time MRI to ensure an effective infusion rate via IA injection to predict perfusion territory in a mouse brain.

a, Injecting an MRI contrast (SPIO) at 0.15 ml/min results in cerebral perfusion, as marked by a decrease in pixel intensity on T2*-weighted scans (red square indicates region of interest (ROI), quantified in b), with no distinct difference in pixel intensity in untreated hemisphere as a baseline (blue square indicates ROI, quantified in b). b, Signal intensity is normalized by setting the maximal value of ROIs as 1. Start represents the beginning of IA infusion. Stop represents the end of the infusion. The IACUC at the Johns Hopkins University approved all experimental procedures shown in this figure. Figure adapted with permission from ref. ²³.

Real-time MRI shows the variability of transcatheter perfusion

During the protocol development, 26 mice were subjected to IA infusion of an MRI contrast agent (SPIO or Gd) under MRI guidance. The infusion rate was adjusted until a desired perfusion territory was achieved as we described in the protocol (Step 23). The rate of 0.15 ml/min was determined to be optimal in our setting. We observed variability in perfusion territory with the same infusion rate among mice²⁴. For instance, IA infusion of contrast agent under dynamic GE-EPI imaging revealed T2* hypointensity in the cerebral cortex (Fig. 9a) at a frequency of 23.07%. The lack of cortical perfusion (Fig. 9b) using this delivery route was found at much higher rates (76.93%), highlighting the variability of clinical outcomes, necessitating the application of real-time MRI to facilitate precise treatment delivery to the target region²⁴.

Fig. 9 |. Variability of cortical involvement during IA infusion of a contrast agent in the mouse brain.

a,b, Representative T2* images during injection of a contrast agent (Gd) at a rate of 0.15 ml/min in which the cortex was (a) or was not (b) perfused as outlined by the red box indicating ROI. c, The constituent ratio of cortical involvement (% of mice this was seen in, n = 26 mice). ‘Cortex⁺’ represents contrast perfusion in the cortex, and ‘Cortex⁻’ represents the lack of perfusion. The IACUC at the Johns Hopkins University approved all experimental procedures. Figure adapted with permission from ref. ²⁴.

Prediction of mannitol-induced BBBO territory

The signal change maps of contrast (SPIO or Gd) perfusion on GE-EPI scan (Fig. 10a) and Gd enhancement on T1-weighted scan (Fig. 10d) were first calculated to present the MRI images clearly. Such an approach facilitated an effective BBBO as reflected by Gd enhancement on the T1-weighted scan (Fig. 10d), in which the hyperintensity region was previously highlighted by the contrast infusion (Fig. 10a). To determine the correlation between the contrast perfusion and Gd enhancement MRI, their histograms were drawn, and two Gaussian distributions that we previously described^23,24 were used to define the threshold that separated the pixels with a significant signal change (Fig. 10b,e). The areas with a substantial signal change are depicted (Fig. 10c,f). The contrast perfusion MRI showed an average signal change area of 26.55 ± 4.22% (n = 8), while Gd enhancement showed an average signal change area of 26.51 ± 4.52% (n = 8). No significant differences were observed between the regions (P = 0.960, Fig. 10g). Instead, we found an excellent correlation between the area of contrast perfusion and Gd enhancement MRI (r = 0.919, R² = 0.845, Fig. 10h). Taken together, the data indicated a successful BBBO by IA mannitol, as predicted by the perfusion pre-scan^23,24.

Fig. 10 |. Prediction of mannitol-induced BBBO territory.

a, Signal change map after contrast perfusion. b, Histogram analysis of pixel intensities in a, showing two Gaussian distributions (red lines). The blue arrow indicates the point where a cutoff of −53.9% was used to separate the two distributions. c, Segmented map shows the area where the relative signal change was smaller than −53.9%. d–f, Signal change map (d); histogram analysis (e); and segmented map (f; △S% > 31.4) at 5 min after i.p. injection of Gd. g,h, Bar graph (g) and correlation analysis (h) of the BBBO territory predicted by the contrast perfusion and that assessed using Gd (n = 8). Data shown as mean ± SD were compared by paired two-tailed t-test. The MRI images were processed by custom-written scripts in MATLAB (Supplementary Data 1 and Supplementary Methods). Figure adapted with permission from refs. ^23,24.

Safety and long-term consequences of mannitol-induced BBBO

To evaluate the safety of our BBBO protocol, mice were assessed for neuropathological sequelae using MRI and histology. MRI showed neither T2 nor T2* abnormalities 3 and 7 d post-BBBO, suggesting a lack of edema or inflammation and microhemorrhages (Fig. 11a). We also did not observe T1 Gd enhancement (Fig. 11a), indicating that the BBB reverted. All data revealed that the procedure was safe and did not cause permanent brain damage. Histological appearance (Supplementary Methods) further confirmed these observations. Glial fibrillary acidic protein (GFAP) and ionized calcium binding adaptor molecule 1 (IBA-1) staining 7 d post-BBBO showed no elevated astrocytic or microglia activation (GFAP⁺ astrocyte and IBA-1⁺ microglia) in the the BBBO region (Fig. 11b). There was no statistically significant difference in cell density between the BBBO region and the corresponding contralateral region (P = 0.208, P = 0.426; Fig. 11b,c)^23,24. Similarly, analysis of neuronal nuclei (NeuN) staining indicated no evidence of neuronal loss post-BBBO (P = 0.935, Fig. 11c)^23,24.

Fig. 11 |. MRI and histological assessment post-BBBO.

a, T2 and T2* images 3 and 7 d after BBBO showing no indication of brain damage. No Gd enhancement on T1-weighted images was observed in the brain, suggesting that the BBB was resealed. b, Fluorescent staining of the BBBO region and corresponding contralateral region with IBA-1, GFAP and NeuN. Scale bar 50 μm. c, Quantification of histological assessments showing comparable cell density for IBA1 (n = 8), GFAP (n = 8) and NeuN (n = 5) immunostaining between the ipsilateral and the contralateral hemisphere, indicating no inflammation and no neuronal loss after BBBO. Data shown as mean ± SD were compared by paired two-tailed t-test. Antibodies: GFAP (1:250, Dako), IBA-1 (1:250, Wako), NueN (1:100, Cell Signaling Technology). The detailed immunostaining protocol and analysis can be found in Supplementary Methods. The IACUC at the Johns Hopkins University approved all experimental procedures shown in the figure. Figure adapted with permission from refs. ^23,24.

In summary, we previously showed the importance of MRI for predictable and precise delivery of therapeutics to the target region both in large animal studies and in a clinical case^{11,18,19,49,50}. Here, we used real-time MRI to visualize the perfusion territory via IA infusion of an MRI contrast agent in mice. The imaging feedback, validated by histology, enabled us to determine a nondamaging IA infusion rate in each individual animal as well as to ensure a desired perfusion region in the brain. We also found that BBBO is predictable when the IA mannitol is infused at the predetermined rate, which means that BBBO can be performed in a predictable fashion. Thus, our results demonstrated a safe and reproducible BBBO under real-time MRI guidance in mice. The mouse BBBO model presented here has been shown to improve the delivery of therapeutic agents (monoclonal antibody, nanoantibody, polyamidoamine dendrimer) into brain parenchyma^24–26. Additionally, this safe and efficient model will contribute to achieving a more comprehensive understanding of the biology of BBB and BBBO. Furthermore, the method can be very valuable for drug development, especially in the case of brain tumors. Through bypassing the BBB obstacle, therapeutic delivery via our method can be utilized to treat various refractory CNS diseases and achieve high treatment efficacy.

Data availability

Source data are provided with this paper. The other data that support the findings of this study are available from the corresponding author upon reasonable request.