Methods for mechanical delivery of viral vectors into rhesus monkey brain

BACKGROUND
Modern molecular tools make it possible to manipulate neural activity in a reversible and cell-type specific manner. For rhesus monkey research, molecular tools are generally introduced via viral vectors. New instruments designed specifically for use in monkey research are needed to enhance the efficiency and reliability of vector delivery.


NEW METHOD
A suite of multi-channel injection devices was developed to permit efficient and uniform vector delivery to cortical regions of the monkey brain. Manganese was co-infused with virus to allow rapid post-surgical confirmation of targeting accuracy using MRI. A needle guide was designed to increase the accuracy of sub-cortical targeting using stereotaxic coordinates.


RESULTS
The multi-channel injection devices produced dense, uniform coverage of dorsal surface cortex, ventral surface cortex, and intra-sulcal cortex, respectively. Co-infusion of manganese with the viral vector allowed for immediate verification of injection accuracy. The needle guide improved accuracy of targeting sub-cortical structures by preventing needle deflection.


COMPARISON WITH EXISTING METHOD(S)
The current methods, hand-held injections or single slow mechanical injection, for surface cortex transduction do not, in our hands, produce the density and uniformity of coverage provided by the injector arrays and associated infusion protocol.


CONCLUSIONS
The efficiency and reliability of vector delivery has been considerably improved by the development of new methods and instruments. This development should facilitate the translation of chemo- and optogenetic studies performed in smaller animals to larger animals such as rhesus monkeys.

Here, we address the first issue: the rhesus monkey brain is significantly larger than that of a mouse/rat (Chareyron et al., 2011). Thus, to achieve the desired vector penetrance and uniformity of coverage across a whole brain area requires the delivery of large quantities of virus in a precisely targeted fashion. Solving this issue requires the development of suitable surgical approaches to (i) achieve dense, uniform receptor expression across large brain structures, (ii) verify accuracy of virus delivery, and (iii) reduce the impact of mechanical failure at the site of injection (i.e., needle deflection or blockage). We focus on optimizing the delivery of one commonly used chemogenetic toolthe hM 4 Di receptor (a Designer Receptor Exclusively Activated by a Designer Drug, DREADD) -into deep subcortical structures and large cortical areas of the monkey brain. To obtain immediate verification of injection location on the day of surgery, the contrast reagent manganese MnCl 2 •4H 2 O (Mn 2+ ) was co-infused with virus. To achieve widespread and uniform viral expression in cortical areas, a set of multi-channel injection devices was constructed that allow slow infusion of large volumes of viral vectors across the targeted tissue(s). To reduce needle deflections or bending during stereotactically guided injections to subcortical structures, a needle guide that attaches to the infusion apparatus was produced to improve mechanical stability. We show that combining these three innovations produces a more uniform level of expression of hM 4 Di across larger target areas within monkey brain that is comparable to expression levels previously achieved in rodent studies. All of these developments should be applicable to any viral construct.

Surgical procedure
All experimental procedures conformed to the Institute for Laboratory Animal Research Guide for the Care and Use of Laboratory Animals and were performed under an Animal Study Proposal approved by the Animal Care and Use Committee of the National Institute of Mental Health. For subcortical injections, targets were derived from a pre-operative T1-weighted scan using a 3-Telsa MR scanner (Philips, Achieva DStream), (Saunders et al., 1990). All scans were acquired using the coronal orientation with a field of view of 100 × 100 × 60 mm 3 and a matrix of 256 × 256 pixels in plane. The slice thickness for each scan was 1 mm, the bandwidth was 260 Hz/pixel, the repetition time was 2720 ms, and the echo time was 4.3 ms using a modified driven equilibrium Fourier transform, and a flip angle of 11.6 degrees. For scanning, each rhesus monkey was placed in a MRI-safe stereotaxic frame (Jerry Rig USA, NJ) for the pre-operative scan. A human knee coil was placed on top of the head. For the MR scan a tooth-marker (Walbridge et al., 2006) mounted on a stereotaxic arm was used to measure the position of the tip of the left canine in 3D space (i.e. mediolateral, dorsoventral, and anterior-posterior co-ordinates were recorded). For surgery, the monkey was placed in the stereotaxic frame so that the tooth-marker measurements of the position of the tip of the left canine were the same as the measurements recorded at the time of the MRI. All cortical injections were visualized under an operating microscope (Zeiss, OPMI Vario S88 Surgical Microscope). All surgical procedures were conducted using aseptic techniques in a dedicated operating suite. At least 12 h before surgery, the subjects were treated with an antibiotic (cefazolin; 25 mg/kg i.m.) and steroidal anti-inflammatory (dexamethasone; 0.5-4.0 mg i.m., SID to TID) to reduce the risk of post-operative infection, edema, and inflammation. Antibiotic and steroid treatment continued for 3-7 days post-surgery. On the day of surgery, a bone flap was removed and an incision made in the dura to provide access to the target region. Needles were inserted into the target region (cortical or subcortical) and the virus plus contrast reagent (see below) was infused at the rate of 0.5 μL/min, to a total volume of 10 μL of virus per injection (unless specifically stated otherwise). At the end of injections a 10-min wait time allowed pressure to dissipate, and virus to diffuse into the target tissue. All injections in this study used a lentivirus expressing an hM 4 Di-CFP fusion protein under a human synapsin promoter. The virus was made in-house at a titer of 1 × 10 9 i.u./mL (Lerchner et al., 2014). After the needles were removed, dura was sutured closed, the bone flap was sewn into position, and the soft tissue was closed in anatomical layers. Immediately following surgery, injections were visualized with a T1-weighted MRI scan. All constructs were allowed to express for a minimum of 6 weeks prior to tissue harvest, to allow protein expression to reach maximum stable levels (Nagai et al., 2016).

Injection visualization
To allow visualization of the injected area up to 24 h after the injection procedure, the contrast reagent manganese MnCl 2 •4H 2 O (Mn 2+ ) was co-infused with virus. Mn 2+ is dissolved in double-distilled water to produce a 10 mM stock solution. The stock solution is sterile filtered through a 0.22 μm filter (Corning, NY). Once the virus is thawed to room temperature, the Mn 2+ stock solution is added to the virus aliquot in the volume required to produce the desired final concentration. Concentrations of 0.1 and 5 mM Mn 2+ were tested. The effect of Mn 2+ on the viability of the virus was assessed using an in vitro assay; one Hamilton syringe was filled with virus and a second was filled with virus plus 10 mM Mn 2+ . The solutions from each syringe were injected into separate wells of cultured HEK cells at a series of time points (0.5, 1, 2, 4, 6, & 8 h) after initial loading. Time point 0 was pipetted directly into the well without loading in the Hamilton syringe to serve as a baseline.

Virus injection apparatus
At the core of all three injector arrays is a 3D-printed manifold (3D Systems Projet 3510, using Visijet M3 Crystal material). A Harvard Apparatus Pump 11 Elite Nanomite is used to infuse virus through the channels in the array at a controlled rate. Each array is attached to the infusion apparatus (Supplemental Fig. 1) via Tygon ND100-80 microbore silicone tubing that has an inner diameter of 0.01 mm, and an outer diameter of 0.03 mm (Component Supply Company, TN). In surgery, the syringes are back loaded with the virus and Mn 2+ solution, loaded into custom housing (Supplemental Fig. 1), and attached to the pump. The pump is set to infuse at 1 μL/min until flow of virus is visible from all channels under the operating microscope; at high magnification (10X) it is possible to see the small droplets that form on the needle tips as the viral solution is effluxed. Once flow is confirmed, the infusion rate is lowered to the target rate (0.5 μL/min) and the apparatus is inserted into the target tissue under visual guidance.

Ventral array assembly
To target the ventral surface of the cortex a manifold of 4 × 4.5 x 1.3 mm (see Supplemental Fig. 2 for additional detail) was used. Four surface holes are printed with a diameter of 0.3 mm to allow the insertion of four 31-guage needles of 2.5 mm in length, countersunk by 0.5 mm to produce a 2 mm length needle for injections. The holes are arranged such that a single array placement will result in four separate injections at 2 mm spacing. The rear of the array contains four channels, with a diameter of 0.3 mm, into which 30-gauge hypodermic tubing of 30 mm length is inserted (Fig. 1). All components are stabilized within the printed manifold with 'Krazy Glue Maximum Bond' (ethyl cyanoacrylate > 95%, and 2,2′-Methylenebis(4-methyl-6-tert-butylphenol), 0.1%) with a viscosity of 85 cps at room temperature (Elmer's Products, Inc., OH).

Lateral array assembly
To target the lateral surface of the cortex a square manifold of 4 × 4 x 3 mm (see Supplemental Fig. 2 for additional detail) is used. Four surface holes with a diameter of 0.46 mm accommodate four 26gauge hypodermic tubes of 8 mm length. Four 31-gauge needles of 5.5 mm length are inserted into the hypodermic tubing and glued in place. A central hole of diameter 0.65 mm accommodates a 23-gauge wire that attaches to the injection apparatus (Supplemental Fig. 1). The needles are arranged such that a single array placement will result in four separate injections at 2 mm square spacing (Fig. 2). All components are secured to the array using Krazy glue.

Linear-lateral array assembly
To target cortex in sulci on the lateral surface of the brain, a rectangular manifold of 6 x 2 x 3 mm (see Supplemental Fig. 2 for additional detail) is used. Four holes with a diameter of 0.46 mm accommodate four 26-gauge hypodermic tubes of 8 mm length. Four 31-gauge needles of 5.5 mm length are inserted into the hypodermic tubing. A center surface hole with a diameter of 0.65 mm accommodates a 23-gauge wire that secures into the injection apparatus (Supplemental Fig. 1). The holes are arranged such that a single array placement will result in four separate injections at 2 mm linear spacing (Fig. 3). All components are secured to the array using Krazy glue.

Needle guide for subcortical targeting
A detachable needle guide made of anodized aluminum was built to mount onto the Nanomite pump. It is compatible with any system using 5/16″ diameter stainless steel dowels. The arm can be sterilized with EtO prior to use. Once mounted, the aperture in the base of the needle guide is aligned to the trajectory of the needle. As the needle is lowered to target, the needle guide sits just touching the skull and the needle    passes through the aperture as it is lowered (Fig. 4C). The syringes were Hamilton (Cole-Parmer, IL) 81008 Gastight Syringes, 100 μL, cemented needle, 30-gauge, 45-degree bevel. Needles were sheathed with TSP polyimide coating (product TSP320450; Molex, IL) to create a reflux resistant step (see Supplemental Fig. 4 for details).

Injection visualization
The contrast reagent, Mn 2+ , was co-infused with the virus, lenti-hSyn-hM 4 Di-CFP (titer of 1 × 10 9 i.u./mL), to provide post-operative confirmation that the virus was infused, and that it was targeted to the correct location. At each injection site, where 10 μL was injected at a rate of 0.5 μL/min, a MR signal was visible at the lowest concentration tested, 0.1 mM Mn 2+ (Fig. 5B). Higher concentrations produced a higher contrast (as expected), but also made it more difficult to localize the focus of the injection (Fig. 5C). The addition of up to 10 mM Mn 2+ did not interfere with the viability of the virus (Supplemental Fig. 6); the toxicity of 10 mM Mn 2+ was not assessed in vivo.

Virus injection apparatus
Cortical injections were performed using a 2 × 2 multi-channel injector for targeting ventral cortex (Fig. 6AI), a 2 × 2 injector for targeting dorsal cortex (Fig. 6BI), and a 4 × 1 injector for targeting intrasulcal cortex (Fig. 6CI). 5 μL per channel was injected into the ventral surface, and 10 μL per channel was injected into the dorsal surface. Accuracy of injections and approximate coverage was assessed with post-operative MRI (Fig. 6, column II). Coverage was assessed at higher resolution (5X magnification) post-mortem with a Diaminobenzidine (DAB) stain; bright-field images reveal the somatic and dendritic expression profiles of each injection type (Fig. 6III). Somatic expression at high resolution (5X magnification) was examined with fluorescence immunohistochemistry under a confocal microscope (Fig. 6IV). A GFP antibody was used to detect CFP expression (Abcam, MA).

Needle guide
Sub-cortical injections were performed using a single sheathed syringe. The custom-built needle guide (Fig. 4) constrained the path of the needle to maximize accuracy of targeting. MR imaging was used to plan the location of the injection targets (Fig. 7A, B). Virus (lenti-hSyn-hM 4 Di-CFP) was co-infused with 0.1 mM Mn 2+ to allow immediate post-op visualization of needle patency and accuracy of placement (Fig. 7C). Post-mortem visualization of hM4Di-CFP expression using bright field (Fig. 7D) and confocal (Fig. 7E) microscopy confirmed that injections accurately targeted the tail of the caudate nucleus. Contrasting this result with that of a surgery performed in the absence of the needle guide (Supplemental Fig. 5) illustrates the utility of the needle guide.

Discussion & conclusions
Achieving efficient and reliable viral vector delivery is critical for increasing the adoption of modern molecular tools in research with rhesus monkeys; we have described three new methods to this end. Coinfusion of the contrast reagent, Mn 2+ , permits a verification process that is advantageous in primate research. Being off target by even small distances can result in vastly different behavioral and physiological effects. Being able to verify injection accuracy immediately post-operatively provides the means to proceed with assurance that targeting of the injection was as intended. Thus, the likelihood of missing targets due to needle deflection, or slight differences in animal positioning can be minimized.
The use of Mn 2+ as a reporter for virus injection and localization allows rapid verification of injection accuracy. Intraoperative MRI is superior, in the sense that it provides feedback during the virus injection, thus permitting adjustments of targeting in real-time (Szerlip et al., 2007). However, the use of a contrast reagent for post-operative imaging overcomes the difficulty of verifying injection accuracy when surgical procedures can only be completed with instruments that are not MR compatible (such as those involving the injector arrays described herein). The longevity of the contrast signal produced by Mn 2+ allows injection accuracy to be verified several hours after the end of the procedure; an option that is not reliable when using contrast reagents with a shorter half-life, such as gadolinium. It is known that Mn 2+ is toxic at high concentrations (120 mM) (Simmons et al., 2008). However, the concentrations of Mn 2+ reported here (< =5 mM) produced no detectable cell loss, in vivo. We have used 0.1 mM Mn 2+ to visualize injections into other brain regions not reported in this study, including ventral tegmental area, substantia nigra pars compacta, superior colliculus, amygdala, and head of caudate nucleus; we have never observed any reduction in transfection efficiency, cell loss, or other markers of tissue inflammation when using the Mn +2 . Furthermore, concentrations up to 10 mM caused no loss of virus viability in vitro (Supplemental Fig. 6).
Intracerebral virus injections in monkeys are often performed via visually guided hand-held injections for surface cortex (Eldridge et al., 2015;Upright et al., 2018) or stereotaxic injections of deeper cortical structures (Saunders et al., 1990). Hand-held injections of viral vectors can, in our experience, produce patchy expression in the target region (Eldridge et al., 2015). The use of the multi-channel injector array allowed the needles to remain stable within cortex, while virus was infused slowly. The needles remained situated for several minutes after the end of the injection, to allow residual pressure to dissipate. This approach appears to result in more uniform coverage of the target regions (Fig. 6), particularly for the ventral injector (Fig. 6A). The higher concentration of Mn 2+ tested, 5 mM, produced an over-estimation of the area transfected ( Fig. 6A-II cf. 6A-IV), although this higher concentration more closely approximated the region of dendritic + somatic expression (Fig. 6A-II cf. 6A-III). The lower concentration of Mn 2+ , 0.1 mM, more closely approximated the somatic expression profile of the virus (Fig. 6B-II and 6C-II cf. 6B-IV and 6C-IV). The difference in the density of staining between the DAB (Fig. 6III) and fluorescent images (Fig. 6IV) results from different staining protocols; we routinely visualize the fluorescent sections under high magnification to resolve individual somata, hence a lower density of staining is preferable for this application. As a result, dendritic expression tends to be less clearly visible in the fluorescent sections under lower magnifications (such as that used in Fig. 6).
Using stereotaxic co-ordinates to target subcortical structures is a method that has proven useful in past studies (Saunders et al., 1990). However, mistargeting can occur as a result of deflections, for example if the needle deviates following contact with a blood-vessel, or if the needle slides along the pial surface when inserted at a tangent. Needle deflections can be mitigated by using a guide tube. This approach is only convenient when the subject has a previously implanted chamber to provide access through a cranial window. The needle guide described herein seems to prevent the needle from deviating from the intended track during an acute surgery, providing the precision required to deliver virus to small subcortical structures.
In summary, co-infusion of manganese with the viral vector allowed for immediate post-operative verification of injection accuracy. The multi-channel injection devices increased the efficiency and uniformity of viral delivery to cortical regions of the monkey brain. The needle guide improved targeting accuracy for sub-cortical structures by preventing needle deflection. The combination of these three innovations allowed for the transduction of hM 4 Di in a reliable and reproducible manner across different regions of the monkey brain. These developments should facilitate the translation of molecular methods used in small animal research to studies with rhesus monkeys.