Patient Material

Formalin-fixed, paraffin-embedded sections from thirty-one brain specimens were acquired from several centers in the National NeuroAIDS Tissue Consortium (NNTC): twenty-two cases were from the Manhattan HIV-1 Brain Bank in New York, NY; five cases were from the National Neurological AIDS Bank in Los Angeles, CA; and four cases were from the Texas NeuroAIDS Research Center, in Galveston, TX. Brain regions included cerebral cortex (frontal, temporal or parietal) and associated white matter.

Table 1 summarizes the cases included in this study. Cases were divided into groups according to their HIV serostatus and the brain pathology: HIV-seronegative (HIV−), HIV-seropositive without encephalitis (HIV+), and HIV encephalitis (HIVE). Several HIV+/HIVE cases were also found to have CNS opportunistic infections, lymphoma or stroke and these were grouped together as comorbidities. The average age of individuals in the four groups was not significantly different. HIV− cases were 66.7% male but the percent of males in the HIV+ and HIVE groups was higher. 84% of all HIV+ patients had a history of combination antiretroviral therapy (cART) at some point in the past based on information from the NNTC database and retrospective chart analysis. Case information for a subset of MHBB cases has been published (22;23). Information on substance abuse obtained from patients’ urine toxicology and retrospective chart analysis is provided in Table 1.

Table 1

Human brain specimens used for CB1 and CB2 analyses.

Immunohistochemistry for ECS Proteins

Brain sections were subjected to immunohistochemistry (IHC) for ECS proteins. IHC was performed for CB1, CB2 and FAAH. Staining for CB2 was performed with two different antibodies: one chosen because of its use in a previous report (13) and a second selected for comparison.

For IHC, brain sections were deparaffinized and subjected to antigen retrieval in 95 °C in modified citrate buffer at pH 6.1 (Target Retrieval Solution: DAKO, Carpinteria, CA) for twenty minutes. They were then treated with 3% H₂O₂, blocked with 5–10% normal horse serum in PBS for 30 minutes to one hour, incubated with the primary antibody and subjected to a polymer-based detection method (ImmPRESS) with diaminobenzidine (DAB) as the chromogen. All washes were done with phosphate-buffered saline. Single-stained slides were counterstained with hematoxylin, but double stained slides were not counterstained. For double-labeled IHC, IHC for ECS proteins was performed first and IHC for von Willebrand factor second as described (24). Table 2 summarizes the antibodies, reagents and methods for immunostaining.

Table 2

List of antibodies and methods employed for immunohistochemistry, immunofluorescence and immunoblotting.

Analysis of Single-stained Sections

Brain sections were analyzed for the distribution and location of positively-stained cells. Cells were identified as perivascular macrophages (PVC), astrocytes or microglia based on morphology and location. When parenchymal cells could not be positively identified, they are indicated as glia (see Table 1).

Immunofluorescence for ECS Proteins

Immunofluorescence for cannabinoid receptors and cellular markers was performed. Brain sections were deparaffinized and subjected to antigen retrieval in 95 °C sodium citrate buffer for twenty minutes. They were then blocked with 10% normal goat serum for 30 minutes to one hour. For most antibody combinations, double immunofluorescence was performed by combining the primaries into a cocktail and incubating the sections in the primary solution overnight at 4 °C. Extended incubations with the primary antibodies at 4 °C (two-three nights) were performed. Since the antibodies for mCB2 and HIV-1 p24 were both mouse clones, staining for these markers was performed sequentially. Immunofluorescence was completed with either the application of a primary and then a conjugated secondary (the two-step method, most antibodies), or via an avidin-biotin complex method (CB1 only). After the secondary step was completed, slides were not dehydrated, but were treated with 0.3% Sudan black (MP Biomedicals, LLC.; Solon, OH) in 70% ethanol for five minutes. This step is useful for quenching autofluorescence from lipofuscin (25). Slides were washed and mounted using Vectashield HardSet Mounting Medium containing 4′,6-diamidino-2-phenylindole (DAPI) nuclear counterstain (Vector Laboratories, Ltd.). Tonsil was used as the positive control for CB2 staining (26). Details regarding the antibodies, regents and methods used for immunofluorescence are provided in Table 2.

Digital imaging analysis for immunoreactivity of CB1 and CB2 in white matter

Digital image analysis was performed on samples with colorimetric staining and statistical analyses were performed as previously described (24;27). Briefly, 400X brightfield images were captured randomly using a Leica Leitz DMRB microscope and an Olympus DP12 digital camera. Twenty images were captured per case. After image capture, the immunoreactivity in the images was subjected to analysis using NIH ImageJ (28). Because of our interest in glial cells, and the presence of CB1 in neurons of all cases, images from white matter were selected for quantification of differential glial expression of cannabinoid receptors (27). A single threshold for background staining was used to process all images. Analysis of the complete image was performed except for, in rare cases, where intravascular immunoreactivity or staining artifact was present. If this was the case, these items were removed and the remainder of the image was processed. The data are expressed as the percent of pixels in the image that are positive for DAB (brown color). Several cases (indicated in Table 1) that were immunostained had high levels of chromogen background or staining levels below the average threshold used and these cases were excluded from image analysis.

Digital imaging of immunofluorescence staining was performed at the Albert Einstein College of Medicine Analytic Imaging Facility (Bronx, NY) using an Olympus IX81 electronically motorized microscope (Olympus America, Inc.; Center Valley, PA) fitted with a Sensicam QE cooled CCD camera (Cooke Corporation, Auburn Hills, MI). The microscope was connected to a computer executing the IP Lab imaging software for Windows (BD Biosciences, Rockville, MD) through which images were captured. Images were processed (fluorophores merged) with NIH ImageJ.

Statistical analyses

Image analysis data were imported into GraphPad Prism Version 4.0 (GraphPad Software, La Jolla, CA) and the group data (HIV−, HIV+, HIVE and comorbidities) were compared using one-way analysis of variance (ANOVA) with Bonferroni multiple testing correction. To determine if CB1 staining was correlated with the level of CB2 staining, Spearman correlation was performed when data for both markers was available for a single case.

Analysis of cannabinoid receptor proteins by immunoblot

We assessed the cannabinoid receptor antibodies’ ability to detect cellular protein by immunoblot. Highly enriched primary human fetal microglia and astrocyte cultures were prepared as previously described (29). Cell lysates were prepared with lysis buffer (10 mM Tris–HCl [pH 8.8], 50 mM NaCl, 0.5 mM Na₃VO₄, 30 mM Na₄P₂O₇, 50 mM NaF, 2 mM EDTA, 1% Triton X-100). Fifty μg of protein was separated by 10% SDS-polyacrylamide gel electrophoresis and then transferred to polyvinylidene fluoride membrane. Blots were blocked in a 1:1 ratio of 5% milk and 5% BSA in TTBS (Tris-buffered saline with 0.1% Tween-20). Primary antibodies were diluted in 5% BSA in TTBS at various concentrations (see Table 2) overnight at 4 °C. After incubation and washes in TTBS, the HRP-conjugated anti-mouse or anti-rabbit IgG secondary antibody (Pierce Biotechnology, Rockford, IL) was applied at a concentration of 1:1000 at room temperature for two hours. Signals were developed using SuperSignal Pico or Femto enhanced chemiluminescence (Pierce Biotechnology). Blots were stripped with Restore PLUS Western Blot stripping buffer (Thermo Scientific, Rockford, IL) for 15 minutes. Blots were washed three times before re-blocking and re-probing with an additional antibody. Blots were probed for CB1, then re-probed for both CB2 antibodies, FAAH and β-actin as a loading control. See Table 2 for a summary of reagents, their sources and the concentrations used for immunoblotting.

Expression of cannabinoid receptors in white matter in control and HIVE cases (Figure 1)

HIV− and HIV+ cases both express small amounts of CB1 and CB2 in the white matter. The staining was present in white matter glia and perivascular macrophages was punctate and delicate (Figure 1A–F). Often, there were many puncta in a region, which may indicate that staining is present in cross-sections of cellular processes. More significant staining was observed in subpial glia (not shown) and meningeal macrophages (Figure 1G, H). Benito and colleagues used a polyclonal anti-CB2 antibody (13). In our hands, this CB2 antibody stained vessels and perivascular regions effectively (see Figure 1C, and also Figure 2), but parenchymal glial staining was null to light. One exception was an HIVE case with confirmed lymphoma and a confounding bacterial infection. This case had high levels of glial staining in white matter (pCB2 < mCB2) (see below). In addition, some HIV+ cases had higher levels of mCB2 compared to HIV− cases. In addition to punctate staining, CB1 also showed cell staining, especially in HIVE or comorbidity cases. For example, serial sections of one HIVE case display the same perivascular cuff labeled with both CB1 (Figure 1I) and a macrophage marker, Iba1 (Figure 1J). CB2 labeling can be focal or diffuse. Serial sections from an HIVE case display mCB2 staining around a blood vessel (Figure 1K) that is positive for HIV-1 p24 in a stained serial section (Figure 1L). Based on morphology, cells staining with anti-CB1 are primarily macrophages and microglia, while those staining with the mCB2 are macrophages and microglia, multinucleated giant cells (K, left-hand inset) as well as reactive astrocytes (K, right-hand inset). Preliminary cellular analysis was performed on sections subjected to single immunohistochemistry for CB1 and mCB2. For single-stained cases, a description of cells positive for CB1 and mCB2 is provided in Table 1.

Figure 2

Endocannabinoid system (ECS) protein immunohistochemistry in HIVE. (A–D) In the gray matter, CB1 and FAAH are present in neurons in all brain cases (HIV− cases are shown), but CB2 is not present in neurons in the vast majority of cases …

The expression of cannabinoid receptor proteins in the white matter of comorbidity cases (Figure 1)

Many individuals with HIV/AIDS are immunosuppressed and they present with one or more opportunistic infections or neoplasms (comorbidities). Several such cases (most with HIVE) were included in our analysis (Figure 1M–T). CB1 was very highly expressed in brains of comorbidity cases. In most cases, CB1 expression was focal and observed in areas of pathology (Figure 1M, N, Q, and S, respectively), but diffusely stained microglia were present in some cases (indicated in Table 1). In other cases, such as coccidiomycosis or PML, CB1 was more diffusely expressed in glia (Figure 1O, P, respectively). Macrophages in an infarct were positive for both CB1 and CB2 (see the serial section of the stroke case, Figure 1Q, R). Microglia in a nodule were more strongly stained with CB1 than with CB2, as observed in serial sections (Figure 1S, T).

Cellular expression of CB1, CB2 and FAAH based on location and morphology (Figure 2)

In previous reports, in human and non-human primate CNS, CB1 has been observed in neurons, activated macrophages, astrocytes, astroglioma cells and in oligodendrocytes, while CB2 has been observed in endothelial cells, astrocytes, activated and perivascular macrophages and microglia (9–11;13;30–33). Figure 2 confirms the presence of CB2 in many of those cell types based on morphology. In order to compare our findings with the work done in SIVE (13), we performed immunohistochemistry using the same antibodies, a polyclonal CB1 and CB2. CB1 was present in neurons of all cases, as shown in Figure 2A (an HIV− case is shown). CB2 was present in neurons in some cases, but was generally not observed in neuronal cells (B, C). FAAH was present in neurons (Figure 2D), but not in glial cells to any significant degree (not shown). With the pCB2, CB2 labeling was restricted to vessel-associated areas (perivascular cells, intravascular cells, endothelial cells, and vascular smooth muscle: Figure 2E, G, I–L). Significant parenchymal staining was not detected with the pCB2 in most cases of HIVE with the exception of a case with two comorbidities (a bacterial infection and lymphoma, Figure 2H). To confirm that pCB2 labeled vascular smooth muscle cells (Figure 2I, J), endothelial cells and perivascular cells, (Figure 2I– L), double-label IHC with von Willebrand Factor, an endothelial cell marker, was performed (J, L). Endothelial cells (K) and perivascular macrophages (L) are also shown to be positive for pCB2 in these double-labeled slides. Two cases with lymphoma did display significant staining with pCB2 (G, H).

Because the pCB2 did not display much parenchymal staining in our hands, we elected to use an additional antibody (mCB2) to confirm our findings. The mCB2 displays enhanced parenchymal immunoreactivity as compared to the pCB2 antibody as it is able to detect both perivascular as well as parenchymal staining (Figure 2F). Perivascular and meningeal macrophages stained intensely with mCB2 (arrow in 2F, for example). Peri-neuronal microglia stained for mCB2 in gray matter in one case (M), as do microglia-like cells (N) and hypertrophic astrocytes (O). Staining in astrocytes was light and punctuate, with puncta often affiliated with the cell membrane.

Image analysis of CB1 and CB2 immunoreactivity (Figure 3)

Digitized photographs of CB1 and CB2 immunostaining were subjected to analysis with NIH ImageJ. The number of cases (n = 20) that were stained effectively for image analysis for CB2 was lower than that for CB1 (n = 27). The mCB2-stained sections were used for analysis because they had significant parenchymal staining as compared to the pCB2-stained sections.

The average % area stained for CB1 ranged from 0.03% (for HIV−) to 0.16% (for comorbidities). There was a graded increase in CB1 immunoreactivity among the HIV−, HIV+, HIVE and comorbidity groups (Figure 3A). The mean (± SEM) percent areas stained were compared among groups by one-way ANOVA and were found to be significantly different (p < 0.001). Pairwise comparisons with the Bonferroni multiple comparison test indicated that while the average percent area stained for CB1 in HIVE was significantly different from HIV− (p < 0.01), CB1 in comorbidity group was significantly different from both HIV− or HIV+ (p < 0.001).

The mean CB2 immunoreactivity is displayed in Figure 3B. For CB2 analysis, the lowest group average was 0.18% (for HIV−) and the highest group average was 0.58% (HIVE). The average percent area stained for mCB2 among all groups was significantly different (p < 0.001). The average mCB2 stained area for the HIVE group was significantly higher than HIV−, HIV+ or comorbidity groups (p < 0.001, p < 000.1 and p < 0.01, respectively).

While there are statistically significant differences among groups for both CB1 and CB2 immunoreactivity, analysis of individual cases within each group (Figure 3C and D) demonstrates a considerable variability from case to case. Within a group, some cases may display very low reactivity, yet others display very high immunoreactivity. Interestingly, despite having no active viral replication detectable by immunohistochemistry in the brain parenchyma (22), the HIV+ group had some individuals with very high levels of receptor staining.

Visibly, parenchymal CB1 staining was more intense than CB2 staining in serial sections of the same case (see Figure 1S and T, for example), but CB2 staining was stronger than CB1 in perivascular or meningeal regions (see Figures 1G, H, for example). The individual values for CB1 and CB2 were compared via Spearman correlation, but no correlation was observed (data not shown). In some cases, there was very high CB1 staining with no CB2 staining (M10013, for example) or very high CB2 staining with little CB1 staining (M10074, for example). The scales for CB1 and CB2 cannot be compared because the staining was qualitatively different in terms of background levels (polyclonal vs. monoclonal) and because different threshold levels were used.

Immunofluorescence confirms the presence of CB1 in microglia and CB2 in both microglia and astrocytes (Figures 4 and ​and55)

The image analysis confirmed an increase in both CB1 and CB2 in HIVE. We next wished to confirm the cell types expressing CB1 and CB2 in the white matter. To accomplish this, immunofluorescence was performed on six cases. We observed that CB1 is expressed primarily in microglia and perivascular macrophages in the white matter. Staining with the macrophage marker CD68 and CB1 were often overlapped (Figure 4A, B, C). Perivascular cells (Figure 4A, B) and phagocytic parenchymal cells (C) were double-labeled. Only some HIV-1 p24+ cells expressed CB1 (Figure 4D). Although there were some identifiable CB1+ astrocytes visible on single-labeled sections, most GFAP+ astrocytes did not express CB1 (Figure 4E, F), providing additional confirmation that microglia and perivascular macrophages are the main cell types that express CB1 in HIVE cases. CB1 expression in neurons in the gray matter of an HIV− case (4G) and the white matter of an HIV− case and an HIV+ case (4H, I) are provided for comparison.

Microglial cells and perivascular macrophages express CB1, but they also express CB2. A macrophage/microglial marker Iba1 highlights a microglial nodule that is CB2+ (Figure 5A). Iba1+/CB2+ cells are shown in Figure 5F–H. Some Iba1-negative cells have astrocytic morphology (red cells in merge panel of A). A subset of GFAP+ cells expresses CB2 (Figure 5B, C) and in some regions, CB2+ astrocytes are abundant (Figure 5D). Similar to CB1, only some HIV-infected cells express CB2 (Figure 5E). To confirm that our immunofluorescence with CB2 was specific, a section of tonsil with CB2+ lymphoid follicle is included (Figure 5I).

Analysis of the cannabinoid receptor antibodies by immunoblot (Figure 6)

Our findings from this study indicate that CB1 is expressed mainly in microglia in HIVE and comorbidity cases, that CB2 is expressed in both microglia and astrocytes in HIVE, and that there are discrepancies in the immunolabeling with the mCB2 and pCB2. Therefore, we tested all three antibodies for their ability to detect cellular proteins by immunoblots. We used protein lysates from human fetal astrocytes (n = 6) and microglia (n = 6) and probed for CB1, mCB2, pCB2, FAAH and β-actin as a loading control (Figure 6). In microglia, CB1 and pCB2 showed a main band around 40 kD, corresponding to the native forms of these proteins (34). Other bands may indicate glycosylated forms of the receptors. The mCB2 antibody also detected a 40 kD protein in microglia, but a more prominent band was identified at 72 kD. Levels of CB1 detected in astrocytes were lower than those for microglia and a 40 kD band was only faintly visible. The pCB2 did not detect any specific bands in astrocytes (not shown) and the mCB2 also displayed either the 40 kD or the 72 kD band, similar to cultured microglia. The FAAH antibody did not detect any specific bands on our immunoblots (not shown).

The authors thank the National NeuroAIDS Tissue Consortium, especially Ashley Wegel, Tissue Manager, and Lucas Barwick, Data Manager, of the NNTC and Kazuhiro Ando from the National Neurological AIDS Bank, for their efforts in compiling cases and data for this study. We appreciate the support of the Mt. Sinai Institute of NeuroAIDS Disparities and the Center for Multicultural and Community Affairs at the Mt. Sinai School of Medicine. We appreciate the assistance of Drs. Celia F. Brosnan and Sarah E. Lutz and Mrs. Christina Polumbo with immunofluorescence studies, and Drs. Leonid Tarassishin and Namjong Choi for assistance with tissue culture experiments. This study was supported by the NIH grants R25MH080663 (Mt. Sinai Institute for NeuroAIDS Disparities), RO1MH55477, and P30AI051519 (Einstein Center For AIDS Research). This publication was made possible from NIH funding through the NIMH and NINDS Institutes by the following grants: Manhattan HIV Brain Bank (U01MH083501, R24MH59724), Texas NeuroAIDS Research Center (U01MH083507, R24 NS45491), National Neurological AIDS Bank (5U01MH083500, NS 38841), California NeuroAIDS Tissue Network (U01MH083506, R24MH59745) and the Statistics and Data Coordinating Center (U01MH083545, N01MH32002). The contents of this study are solely the responsibility of the authors and do not necessarily represent the official view of the NNTC or NIH.

2-AG

2-arachidonoylglycerol

ECS

endocannabinoid system

FAAH

fatty acid amide hydrolase

HAND

HIV-associated neurocognitive disorders

IHC

immunohistochemistry

NNTC

National NeuroAIDS Tissue Consortium

mCB2

monoclonal CB2 antibody

pCB2

polyclonal CB2 antibody