Brain BLAQ: Post-hoc thick-section histochemistry for localizing optogenetic constructs in neurons and their distal terminals

Optogenetic constructs have revolutionized modern neuroscience, but the ability to accurately and efficiently assess their expression in the brain and associate it with prior functional measures remains a challenge. High-resolution imaging of thick, fixed brain sections would make such post-hoc assessment and association possible; however, thick sections often display autofluorescence that limits their compatibility with fluorescence microscopy. We describe and evaluate a method we call “Brain BLAQ” (Block Lipids and Aldehyde Quench) to rapidly reduce autofluorescence in thick brain sections, enabling efficient axon-level imaging of neurons and their processes in conventional tissue preparations using standard epifluorescence microscopy. Following viral-mediated transduction of optogenetic constructs and fluorescent proteins in mouse cortical pyramidal and dopaminergic neurons, we used BLAQ to assess innervation patterns in the striatum, a region in which autofluorescence often obscures the imaging of fine neural processes. After BLAQ treatment of 250–350 μm-thick brain sections, axons and puncta of labeled afferents were visible throughout the striatum using a standard epifluorescence stereomicroscope. BLAQ histochemistry confirmed that motor cortex (M1) projections preferentially innervated the matrix component of lateral striatum, whereas medial prefrontal cortex projections terminated largely in dorsal striosomes and distinct nucleus accumbens subregions. Ventral tegmental area dopaminergic projections terminated in a similarly heterogeneous pattern within nucleus accumbens and ventral striatum. Using a minimal number of easily manipulated and visualized sections, and microscopes available in most neuroscience laboratories, BLAQ enables simple, high-resolution assessment of virally transduced optogenetic construct expression, and post-hoc association of this expression with molecular markers, physiology and behavior.


INTRODUCTION
Viral-mediated expression of genetically encoded fluorescent proteins, light-activated channels and pumps, designer receptors, and signaling indicators allows for specific detection, manipulation, and interrogation of select neurons within heterogeneous populations both in vivo and ex vivo. However, viral expression of these constructs is highly variable, owing to factors such as viral titer, spread and transport, infection efficiency, toxicity, postinfection time interval, and promoter strength and specificity, in addition to the variability associated with anatomical differences and intracranial injection procedures. Given that small differences in somatic and terminal field expression of optogenetic constructs (e.g., in distinct but neighboring striatal subcompartments; Crittenden and Graybiel, 2011) can have important functional consequences, it seems essential to assess with high resolution the specificity and extent of their expression, and to associate this expression with molecular, physiological and behavioral measures in individual brain slices and animals. However, simple and efficient means to do so using conventional techniques are limited.
High-resolution imaging of thick brain slices of the sort commonly used in slice physiology experiments (>250 μm) would allow for efficient post-hoc assessment and association. However, the autofluorescence seen in thick sections often impedes imaging of the fluorescent signal of interest. This is particularly true in brain regions like the striatum due to its high lipid content and aldehyde-induced catecholamine autofluorescence (Carlsson et al., 1961;Falck et al., 1982;Clancy and Cauller, 1998). Several techniques and post-hoc transformations have been developed to address this problem. For example, autofluorescence can be separated based on spectrum, quenched using chemical treatments or reduced by imaging with multiphoton microscopy (Corrodi et al., 1964;Oliveira et al., 2010). Recent approaches such as Scale (Hama et al., 2011), ClearT (Kuwajima et al., 2013), CLARITY (Chung et al., 2013;Tomer et al., 2014) and SeeDB (Ke et al., 2013) achieve high-resolution imaging of deep tissue by clearing lipids in ways that retain the fluorescent signal. The ease, efficiency and wide-scale applicability of these techniques, however, are limited by the fact that they are relatively laborious, can result in soft and fragile tissue, can take several days to weeks to conduct, and can require costly and sophisticated imaging equipment (e.g., multi-photon confocal, serial array tomography, light sheet microscopy). Furthermore, any technique that requires high-resolution reconstruction to image whole brain or very thick tissue is limited by the requisite computing power (Osten and Margrie, 2013).
We describe and evaluate a method adapted from conventional histochemical techniques called Brain BLAQ (Block Lipids and Aldehyde Quench) to rapidly reduce autofluorescence in thick brain sections, enabling efficient axon-level imaging of neurons and their processes using standard epifluorescence microscopy. Unlike many methods, BLAQ can be performed post-hoc on thick sections commonly used in electrophysiological and optogenetic experiments with minimal time investment, inexpensive reagents, and simple optics. BLAQ can also be readily applied for mesoscale reconstruction after in vivo experiments using few sections that are easily manipulated, imaged, and archived. We demonstrate the utility of BLAQ by assessing the complex innervation patterns within striatum following viral-mediated transduction of various optogenetic constructs and fluorescent proteins in cortical pyramidal and midbrain dopaminergic neurons. The experiments we describe also revealed caveats and experimental factors that should be considered when using viral transduction and BLAQ histochemistry.

VIRUS INJECTIONS
Under isoflurane anesthesia (1-5% in 0 2 ), mice were stereotaxically injected with AAVs into various brain regions (Table 1; Figure 1). AAVs were manually injected using a 1 μL Hamilton syringe at 20-30 nL/min, and allowed to diffuse for 5 min following injection before syringe removal. Mice were at least 30 days of age at time of injection, and brains were analyzed 3-8 weeks post infection.

BRAIN BLAQ PROCEDURE
A list of required materials and a concise protocol for the Brain BLAQ procedure can be found in the Appendix (see Supplemental Data). Thick brain slices were fixed with 4% formaldehyde in 0.1 M PB or PBS. Fixative was initially at room temperature to preserve microtubules, which can be sensitive to cold fixation (Wallin and Stromberg, 1995); after 30 min, sections were transferred to 4 • C for 12-24 h. All subsequent washes/incubations were performed in multi-well tissue culture dishes on an orbital shaker set to low speed. Sections were combined within wells to conserve antibody since the thick slices accommodated by this protocol could be easily ordered according to anatomical position following treatment. Sections were washed for 1 h in PBST (PBS, 0.2% Triton X-100), and rinsed twice for 1 min in deionized water (diH 2 O) to remove ions and detergent. Slices were then incubated twice for 10 min in freshly prepared sodium borohydride (NaBH 4 ; 5 mg/mL in diH 2 O) to quench aldehyde-induced catecholamine fluorescence and reduce free aldehyde groups. Following two 1min rinses in diH 2 O, slices were incubated twice for 15 min at room temperature in filtered Sudan Black B solution (0.2% in 70% ethanol). Slices were washed twice for 30 min in room temperature PBS to mordant the stain and slow subsequent destaining. Additional quenching and blocking was occasionally performed if background fluorescence returned after extended washing.
Quenched and blocked brain sections were blue-black in color, rigid and easily manipulated with a paintbrush. The direct fluorescent signal was significantly attenuated (70-80%), and could be partially recovered by washing in PBST for 12-16 h. This attenuation was often advantageous since the signal from fluorescent reporters is frequently saturated near the injection site, distorting cellular-level resolution.
Immunostaining was used to amplify and stabilize the fluorescent signal in projections that are not easily detected because of the high background fluorescence or quenching by the preceding treatments. Slices were blocked using 5% BSA in PBST for 4 h, and incubated in primary antibody for 72 h at 4 • C. Following four washes in PBST over a total of 16-24 h, slices were incubated in the appropriate Alexa-conjugated secondary antibody for 48 h Frontiers in Neuroanatomy www.frontiersin.org February 2015 | Volume 9 | Article 6 | 3 at 4 • C. Slices were then washed four times in PBST over a total of 16-24 h. DAPI (100 ng/mL) was added in some instances during the first of these PBST washes. Finally, slices were washed for 1 h in PBS to remove detergent and prevent destaining with storage.
Because the brain sections were robust and easily manipulated, they were often temporarily placed on a slide, imaged, and returned to the staining tray for subsequent processing or antibody stripping. To permit imaging from both sides, thick slices were either coverslipped in glass-bottom, multi-well plates, or mounted between two large coverslips separated by a spacer gasket made from a border of surgical tape (3M Transpore). Sudan Black leached into some mounting media with prolonged storage, leading to a darkening of the mounting medium and background fluorescence in the red channel. If this occurred, the background fluorescence could be easily reversed by removing the coverslip and rinsing the mounting medium away with PBS. Fluoromount Aqueous Mounting Medium (Sigma, F4680) did not cause significant leaching of the dye. Alternatively, slices could be stored in PBS at 4 • C for at least 6 months with little apparent loss of signal.
Images were captured using Zeiss AxioVision LE 4.1 software with a Zeiss Axiocam on a Zeiss SteREO Lumar microscope equipped with standard DAPI, GFP and Cy3 filters, and a 1.5X telescoping zoom objective (12.5-150X, resolution 0.6 μm, z-precision 30 μm, field of view 23 mm, FWD 30 mm). This objective collected sufficient signal to image whole sections without the need to montage. Single axons were easily visible by direct GFP/RFP fluorescence at the maximum zoom with this objective. Background was subtracted in each channel and contrast enhanced for presentation. Digital gain was doubled to capture direct GFP fluorescence in striatum following blocking and quenching ( Figure 2B ).

RESULTS
A typical AAV2/9-FLEX-GCaMP5 injection into M1 cortex of an Emx1 Cre mouse before Brain BLAQ exemplifies the poor fluorescence resolution in 350 μm-thick tissue commonly used for ex vivo recordings (Figure 2A). After quenching and blocking with sodium borohydride and Sudan Black, cells near the injection site were easier to resolve and many processes were visible without immunostaining (Figure 2A ); however, the remaining haze represents pyramidal cell collaterals that were no longer resolvable due to signal attenuation. After anti-GFP antibody amplification, individual deep and superficial pyramidal cells were clearly visible, as were many of their processes, even at low magnification (Figures 2A ,A ).
As with the injection site, M1 projections to the striatum were poorly resolved before BLAQ (Figure 2B), weakly detected after blocking and quenching ( Figure 2B ), but readily visualized after the full BLAQ protocol (Figure 2B ). The same section was flipped and imaged (Figure 2B ), demonstrating the anatomical differences seen across a single 350 μm-thick section. At 150X magnification, individual processes were clearly visible within the matrix but very sparse in striosomes ( Figure 2C). Striosomes avoided by M1 projections were most apparent in lateral and ventrolateral quadrants where sparse fibers contrasted with the black fiber bundles after BLAQ (Figure 2C). This pattern was observed in all 13 mice examined with little variability between subjects or the three viral constructs injected into M1. Our findings replicate earlier reports that motor cortex efferents preferentially target the matrix (Gerfen, 1984;Crittenden and Graybiel, 2011) and images from the Allen Institute mapping project showing sparse M1 projections to striosome-like regions within the striatum (http:// connectivity.brain-map.org/). Individual axons within the corpus callosum and external capsule were also easily visualized after BLAQ (Figure 2D). Furthermore, sparse cells were clearly resolved in temporal somatosensory cortex (Figure 2E), suggesting infection of somatosensory axon terminals or viral transport.
Conventional fluorescence microscopy typically uses thin tissue sections to achieve sufficient antibody penetration and high-resolution imaging. Notably, the resolution and expression patterns of fluorescently labeled corticostriatal processes observed in thin (50 μm) GFP-immunostained, BLAQ-untreated sections from an Emx1 Cre mouse injected with GCaMP6s into M1 ( Figure 2F) mimicked those seen following BLAQ treatment of thick (350 μm) sections (Figures 2A ,B -E,G ,H ). Parallel processing of thick, GCaMP6s-expressing slices after GFP immunolabeling alone (Figure 2G; exposure time: 528 ms), relative to adjacent slices treated with the full BLAQ protocol (Figure 2G ; exposure time: 2.8 s), also showed the efficacy of BLAQ to reduce autofluorescence and signal saturation in thick tissue, and thereby allow higher resolution of corticostriatal afferents.
In addition to immunostaining for fluorescent proteins, we stained some sections to mark probe implant and virus injection sites. For example, to reveal the prior location of a fiber optic probe stereotaxically implanted into the dorsal striatum of a GCaMP6s-injected Emx1 Cre mouse, we co-stained thick (350 μm) sections for GFP and the microglial and perivascular marker, isolectin GS-IB 4 (IB 4 ; 1 μg/ml; Pennell et al., 1994). IB 4 staining revealed the implant site and adjacent vasculature in neighboring slices processed in parallel either without BLAQ treatment ( Figure 2H) or with BLAQ treatment (Figure 2H ). Both approaches effectively identified the implant site, but the full BLAQ protocol enabled higher resolution imaging of surrounding GCaMP6s expression. Similarly, staining for IB 4 revealed the site of GCaMP6s injection in M1 (Figure 2I), while staining for the activated microglial marker, Iba, illuminated a larger region ( Figure 2J). Immunostaining for phospho-ERK (pERK) was more sensitive at detecting activated microglia than isolectin GS-IB 4 or Iba immunofluorescence, and marked numerous microglia streaming from the ventricles and into the injection site ( Figure 2K, merged with GFP and DAPI in Figure 2K ; arrow in Figure 2K indicates streaming microglia). Immune responses are purported to be low with AAV constructs (Rogers et al., 2011) but may occur due to brain trauma, helper virus protein contamination, or endotoxin contamination (Nadeau and Rivest, 2000;Woodcock and Morganti-Kossmann, 2013), and are likely vastly under-reported in studies using intracranial surgery. BLAQ paired with microglia and pERK immunostaining can readily detect immune responses that may impact neuron function locally and downstream of the site of infection.

FIGURE 2 | Brain BLAQ detection of AAV-mediated GCaMP expression in M1 pyramidal cells and their projections in Emx1 Cre mice. (A-A")
Expression of GCaMP5 at M1 injection site in a 350-μm-thick brain slice before treatment ( striosome projections were assessed with PrL/IL AAV9-FLEX-tdTomato injections into Emx1 Cre ; Nr4a1-GFP mice, which express GFP in MSNs localized preferentially in the striosomes (Davis and Puhl, 2011). Despite a small injection volume, viral spread was seen in some mice into adjacent orbital cortex ( Figure 3A). Massive innervation of the claustrum and medial and lateral striosomes in the dorsal striatum was observed (Figures 3B-B , arrow in Figure 3B indicates claustrum). Axons were clearly visible at 150X within these striosomes, adjacent to brightly fluorescent axon bundles in the dorsomedial striatum Tomato-expressing mPFC projections were also seen in the medial cerebral peduncles, interdigitated with GFP-positive fibers arising from MSNs of Emx1 Cre ;Nr4a1-GFP mice (Figure 3E), but no somata were detected in the mesencephalon. While innervation of the hippocampal cornu ammonis (CA) regions and the dentate gyrus was faint, dense projections were observed in the fasciola cinerea, a structure contiguous with the dentate gyrus ( Figure 3F, red arrow). Significant tomato expression was also seen in soma within CA3 (Figure 3F, white arrow) and occasionally in the amygdala, suggestive of viral transport. The amygdala also received dense mPFC projections (Figures 2G,H) that clearly avoided the central nucleus of the amygdala (Figure 3G).
We next examined the efficacy of BLAQ in detecting dopaminergic fibers by tracing VTA projections to striatum labeled following AAV5-FLEX-ChR2-mCherry infusion into the VTA of DAT Cre mice. Horizontal sections showed discrete subregions of VTA projections to the nucleus accumbens (NAc) and ventral striatum that diminished caudally and laterally ( Figure 4A). Adenosine A 2A receptors are expressed by dopamine D2 receptor-expressing, indirect pathway MSNs (Svenningsson et al., 1999), but also show a heterogeneous matrix-like expression pattern in the ventral striatum ( Figure 4A ). This matrix-like distribution was supported by A 2A staining in the Nr4a1-GFP mouse (M. I. Davis, unpublished observations). Some regions of segregation of A 2A receptor expression and VTA innervation were seen within the NAc and ventral striatum in all five mice examined (Figure 4A , DAPI counterstain; higher magnification, Figures 4B-B ; more dorsal plane, Figure 4C). The Islands of Calleja, known to express the striosome marker FoxP2 (Takahashi et al., 2008), also received dense VTA projections (Figure 4B, arrow). Interestingly, putative striosome-like structures with dense VTA innervation (Figure 4C, arrows) localized to the same general region as those presented as a trio of islands in the ventrolateral striatum in the coronal plane that were nearly devoid of M1 innervation ( Figure 2C). Higher (150X) magnification imaging of the medial forebrain bundle ( Figure 4D) and single-cell detection at the injection site ( Figure 4E) reiterate the utility of BLAQ for high-resolution fluorescence imaging of densely myelinated structures in thick tissue preparations.

DISCUSSION
We report a method adapted from conventional histochemical procedures called Brain BLAQ that enables efficient, highresolution assessment of optogenetic construct expression in thick brain sections commonly used in electrophysiological and optogenetic experiments. Using standard epifluorescence microscopy of easily manipulated and visualized sections, and requiring minimal time investment and inexpensive reagents, Brain BLAQ provides investigators with a simple approach to reduce thick-tissue autofluorescence, enabling efficient post-hoc Frontiers in Neuroanatomy www.frontiersin.org February 2015 | Volume 9 | Article 6 | 7 association of viral transduction with molecular markers and functional measures. In addition to improved detection of virally transduced fluorescent constructs, BLAQ enables immunolabeling of many different native epitopes. We report here successful staining for adenosine A 2A receptors, the microglial marker Iba, and pERK, and have successfully stained (data not shown) for calbindin, cholecystokinin, parvalbulmin, tyrosine hydroxylase, and dopamine D1 receptors. Importantly, however, some antigens can be compromised by the BLAQ protocol, most likely by the brief 70% ethanol treatment. We found immunolabeling of cannabinoid-1 and mu-opioid receptors to be partially compromised by the BLAQ protocol. Notably, reducing the ethanol concentration or the duration of Sudan Black treatment can help retain antigenicity of certain epitopes (e.g., cannabinoid-1 receptors). However, the presence of ethanol in the Sudan Black B solution aids the penetration of Sudan Black B and antibodies in subsequent steps of the BLAQ protocol (Llewellyn-Smith and Minson, 1992); therefore, efforts to retain antigenicity may come at the expense of complete autofluorescence blocking and antibody penetration.
Brain BLAQ is not only suitable for use with standard epifluorescence microscopy, but is also amenable to laser scanning confocal microscopy and serial reconstruction. Confocal microscopy of BLAQ-treated tissue can resolve fluorescent signals over 150-200 μm deep with negligible loss of power (W. Shain and K. Trett, unpublished observations), so thick slices can be imaged from both sides to enable full reconstruction of the fluorescent signal. One important consideration noted during the development of BLAQ is tissue shrinkage. Brain sections shrunk by less than 5% when incubated for 5 min in the Sudan Black/ethanol solution, but closer to 30% when incubated for 30 min. As mentioned above for retaining antigenicity, using graded ethanol incubations or decreasing the duration of Sudan Black treatment will reduce shrinkage, but at the expense of antibody penetration and autofluorescence blocking. Although tissue shrinkage may be a concern for some applications where it is essential to retain precise three-dimensional morphology, it will be of little consequence for most applications (e.g., rapid assessment of overall viral transduction) if treated sections are assessed with reference to a rodent brain atlas (Simmons and Swanson, 2009).
We observed considerable variability in the extent and pattern of viral expression, particularly in downstream regions following AAV5-DIO-ChR2 targeting of VTA. For example, we noted sparse and highly variable VTA innervation of the basal amygdala across all DAT Cre mice. Dense projections to the striosome-like (Millhouse, 1986;Davis and Puhl, 2011) intercalated cells of the amygdala were seen in only one of two DAT Cre lines tested (second DAT Cre line: Bello et al., 2011; two of two mice analyzed showed projections, data not shown). The variability between mice in VTA target regions may be individual mouse-and/or strain-specific, but underscores the need for validation of the extent and pattern of expression in each experimental animal (not only those injected for histology purposes), particularly when attempting to target small regions.
Although rare and somewhat controllable by selecting appropriate capsid/genome combinations (Cearley et al., 2008;Salegio et al., 2013), anterograde and retrograde AAV transport events can occur, can easily go undetected using serial thin sections unless each section is examined, and are nearly undetectable against background autofluorescence using conventional epifluorescence microscopy of thick sections. While our aim was not to systematically compare expression patterns and transport characteristics of each AAV serotype, BLAQ revealed somatic expression of the AAV9-FLEX-tdTomato construct in neurons innervating, or innervated by, the mPFC in mPFC-injected Emx1 Cre mice. Emx1 Cre mice express Cre in the vast majority of excitatory cells in the neocortex and hippocampus, and in sparse subpopulations in some ventral pallial structures (Gorski et al., 2002). We also observed significant local spread and probable transport of AAV2/9-FLEX-GCaMP5, but not AAV1-FLEX-GCaMP6, to other cortical regions and rarely to distal targets (e.g., basolateral amygdala) after M1 infection. Somatic expression was not detected in striatum of Emx1 Cre mice following M1 infection with any viral construct, suggesting against significant viral spread (despite modest Cre expression in striatum of Emx1 Cre mice), and distal transport of the VTA-injected viruses in DAT Cre mice was not observed, although the latter may have been due to the absence of Cre-expressing cells in terminal regions. Significant local spread and transport can negatively impact experiments in which precise targeting of a single neuronal population is required, as with the expression of optogenetic constructs or designer receptors exclusively activated by designer drugs (DREADDs). Given that small differences in expression can result in meaningful functional differences, particularly in the striatum where subcompartments are believed to mediate very different aspects of movement, motivation, and pathology (Crittenden and Graybiel, 2011), unwanted viral transport and spread further exemplifies the need for a simple technique like BLAQ to evaluate the full extent of expression in each animal.
We demonstrated the utility of BLAQ using viral injections likely to differentially target striosome-matrix compartments within the striatum. Despite considerable evidence that striatal architecture is far more heterogeneous than the classical "direct" and "indirect" pathways, few groups examine the striosomematrix subdivisions of the striatum and NAc core, or the column/wedge sub-architecture of the NAc shell (Jongen-Relo et al., 1993;Härtig et al., 2003;Gangarossa et al., 2013). By further clarifying the anatomical segregation of these functionally distinct striatal subregions, this work exemplifies the need to routinely associate higher resolution imaging of viral expression with physiological and behavioral measures obtained using optogenetics in individual mice. Brain BLAQ can serve as a useful tool to this end, enabling researchers to efficiently assess viral expression patterns at the axon-level in thick brain sections using standard microscopy available in most neuroscience laboratories.
of BLAQ for high-resolution automated confocal imaging, and Amber Luo for assistance with mice genotyping. The Iba antibody was generously provided by Dr. Giorgi Kharebava, Laboratory of Molecular Signaling, NIAAA. This work was supported by the National Institute of Alcohol Abuse and Alcoholism Division of Intramural Clinical and Biological Research and is subject to the NIH Public Access Policy. The authors declare no competing financial interests.