RETRACTED ARTICLE: Increased neuroinflammatory and arachidonic acid cascade markers, and reduced synaptic proteins, in brain of HIV-1 transgenic rats

This protocol was approved by the Animal Care and Use Committee of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, and followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication 86-23). Seven- to 9-month-old male, specific pathogen-free, Fischer 344/Hsd HIV-1 Tg rats (n = 6) and age-matched parental wild-type inbred Fischer 344/Hsd non-Tg control rats (n = 6) were purchased from Harlan Laboratories (Indianapolis, IN) and housed in an animal facility with controlled temperature, humidity, and 12-h light/dark cycle. Food (Teklad global 18% protein diet, 2018S (sterilized) for controls and 2918 (irradiated) for HIV-1 Tg rats (Harlan) [9] and water were provided ad libitum. After three days of acclimation, rats were anaesthetized with an overdose of CO₂ and decapitated. Their brain was rapidly excised, sagittally cut into four sections from the left and right hemispheres, frozen in 2-methylbutane at -50°C, and stored at -80°C until studied. One section from the left hemisphere from each rat was used to isolate the cytosolic fraction, a corresponding section from the right hemisphere was used for total RNA extraction, and remaining sections from both hemispheres were used to prepare nuclear extracts.

Cytosolic brain fractions were prepared as reported [38]. One section from each brain was homogenized in a buffer containing 20 mM Tris-HCl (pH 7.4), 2 mM EGTA, 5 mM EDTA, 1.5 mM pepstatin, 2 mM leupeptin, 0.5 mM phenylmethylsulfonyl fluoride, 0.2 U/ml aprotinin, and 2 mM dithiothreitol, using a Polytron homogenizer. The homogenate was centrifuged at 100,000 g for 60 min at 4°C, and the resulting supernatant (cytosolic fraction) collected. Protein concentrations were determined using Bio-Rad Protein Reagent (Bio-Rad, Hercules, CA).

Brain tissue was homogenized in Qiagen^® lysis solution and total RNA was isolated by phenol-chloroform extraction using a RNeasy^® lipid tissue mini kit (Qiagen, Valencia, CA). Complementary DNA was prepared from total RNA using a high-capacity cDNA Archive kit (Applied Biosystems, Foster City, CA). mRNA levels (IL-1β, TNFα, GFAP, CD11b, cPLA₂-IVA, sPLA₂-IIA, iPLA₂-VIA, COX-1, COX-2, mPGES, cPGES, 5-, 12-, 15-LOX, TXS, cytochrome p450 epoxygenase, drebrin and neurofilament-L) were measured by quantitative RT-PCR, using an ABI PRISM 7000 sequence detection system (Applied Biosystems). Specific primers and probes for cPLA₂-IVA _B, sPLA₂-IIA _B, iPLA₂-VIA, COX-1, COX-2, mPGES, cPGES, 5-, 12-, 15-LOX, TXS, cytochrome p450 epoxygenase, drebrin and neurofilament-L were purchased from TaqMan^R gene expression assays (Applied Biosystems), and consisted of a 20× mix of unlabeled PCR primers and Taqman minor groove binder (MGB) probe (FAM dye-labeled). The fold-change in gene expression was determined by the ΔΔC_T method [39]. Data are expressed as the relative level of the target gene (IL-1β, TNFα, GFAP, CD11b, cPLA₂-IVA _B, sPLA₂-IIA, iPLA₂-VIA, COX-1, COX-2, mPGES, cPGES, 5-, 12-, and 15-LOX, TXS, cytochrome p450 epoxygenase, drebrin or neurofilament-L) in the brain of the HIV-1 Tg rat normalized to the endogenous control (β-globulin) and relative to the control (calibrator). All experiments were carried out in triplicate from each control and HIV-1 Tg rat brain (n = 6).

Proteins from the cytosolic fraction (65 μg) were separated on 4-20% SDS-polyacrylamide gels (PAGE) (Bio-Rad), and electrophoretically transferred to a nitrocellulose membrane (Bio-Rad). Cytosolic protein blots were incubated overnight in Tris-buffered-saline containing 5% nonfat dried milk and 0.1% Tween-20, with specific primary antibodies for proinflammatory cytokines: IL-1β (1:500), TNFα (1:500); astrocytes: glial fibrillary acidic protein (GFAP) (1:1000); CD11b (1:1000); AA cascade proteins: cPLA₂-IVA, sPLA₂-IIA, iPLA₂-VIA, COX-1 (1:1000), COX-2 (1:1000), cytochrome p450 epoxygenase, TXS, 5-, 12-, 15-LOX, mPGES, cPGES (1:1000); gp120 (1:100); tat (1:100); drebrin (1:1000), BDNF (1:1000) (Santa Cruz, Santa Cruz, CA);); neurofilament-L (1:500) (Cell Signaling Technology, Danvers, MA) and β-actin (1:10,000) (Sigma Aldrich, St. Louis, MO). The cytosolic blots were incubated with appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (Bio-Rad), and were visualized using a chemiluminescence reaction (Amersham, Piscataway, NJ). Optical densities of immunoblot bands were measured using Alpha Innotech Software (Alpha Innotech, San Leandro, CA) and were normalized to β-actin. All experiments were conducted on 6 independent samples. Values are expressed as percent of control.

Nuclear extracts were prepared as reported [40, 41] and protein concentrations were determined using Bio-Rad Protein Reagent (Bio-Rad). NF-κBp50 and NF-κBp65 activities were measured according to the manufacturer's instructions (Panomics, Freemont, CA), using nuclear extracts obtained from the control and HIV-1 Tg rats. Briefly, 10 μg of nuclear extract from each sample was preincubated with biotin-labeled NF-κBp50 or p65 oligonucleotides in a separate vial for 60 min. The labeled oligonucleotide-nuclear protein complexes were immobilized on a streptavidin-coated 96-well plate. The bound oligonucleotide nuclear protein complex was detected by adding NF-κBp50 or p65 antibody to the respective NF-κBp50 or p65 complex, followed by addition of secondary antibody conjugated to HRP. Color was developed with tetramethylbenzidine substrate and optical densities were measured at 450 nm. Values are expressed as percent of control. All experiments were conducted on 6 independent samples.

Active caspase-3 protein was measured according to the manufacturer's instructions (Cell Signaling, Danvers, MA), using cytosolic brain fractions from the control and HIV-1 Tg rats Briefly, 100 μl (100 μg) of cytosolic fraction was incubated with pre-coated capture antibody in a microwell plate overnight at 4°C. After incubation, the target protein was captured by coated antibody. Following extensive washing, an HRP-linked secondary antibody was added to recognize the bound antibody complex. Color was developed with tetramethylbenzidine substrate and optical densities were measured at 450 nm. Values are expressed as percent of control. All experiments were conducted on 6 independent samples.

In a separate cohort of animals, astrocyte and microglia morphology was analyzed by immunohistochemistry. Following CO₂ anesthesia, the brain (n = 4) was rapidly excised, cut in the midsagittal plane, and the individual hemispheres immersion-fixed in 4% paraformaldehyde/phosphate buffer (pH 7.2) for 18 h, followed by cryoprotection. Fifty- μm free-floating coronal serial cryosections of the forebrain were stored in solution (FD Neurotechnologies, Baltimore, MD) at -20°C. Sections (between +1.0 and 0.4 mm from bregma) were washed with phosphate buffered saline (PBS), equilibrated to room temperature (RT), transferred to 10 mM citrate buffer containing 0.05% Tween-20 and incubated 30 min at 80°C. Sections were then rinsed in PBS and incubated 2 h in blocking solution (2% goat serum, 1% bovine serum albumin, 0.1% Triton X-100 in automation buffer (Biomedia, Foster City, CA). Sections were incubated with anti-GFAP or ionized calcium binding adopter molecule 1 (Iba-1, 1:500, Dako, Glostrup, Denmark) in blocking solution for 18 h at 4°C, re-equilibrated to RT, washed with PBS, and incubated with Alexa Fluor antibody conjugates (1:250, Invitrogen, Carlsbad, CA) in blocking solution without Triton X-100 for 2 h at RT. Digital images of immunostaining in the somatosensory cortex and the dentate gyrus of the hippocampus were collected using a LSM 410 inverted confocal laser-scanning microscope (Carl Zeiss, Oberkochen, Germany) equipped with argon, HeNe, and iFlex 2000 PSU lasers. Image stacks were collected at 1.5 mm steps (20×) or 1.0 mm steps (63×) and displayed as a single image using 3D maximum projection.

Data are expressed as mean ± SEM. We used t-tests for independent samples for group comparisons. We further tested significance using the false discovery rate (FDR) to correct for multiple comparisons. We set alpha = 0.01 to reduce type one error risk. An alpha = 0.01 and an n of 15 markers per mRNA and protein assays would give a 14% chance of at least one false positive for each mRNA and protein assay using the following formula 1-(1-.01)e¹⁵. A p value less than 0.01 and 0.001 is represented by ** and *** respectively.

Gp120 and tat protein levels were detectable in cytosolic brain fractions of HIV-1 Tg but not of control rats (Figure 1A). Brain mRNA and protein levels for the astrocyte structural protein, GFAP, were not significantly altered in the HIV-1 Tg rats compared to controls (Figure 1B-C). As a molecular marker for activated microglia and macrophages [42], mRNA level of the CD11b was elevated significantly by 7.1-fold in the HIV-1 Tg compared with control brain (Figure 1D), corresponding to a significant 190% elevation in CD11b protein (Figure 1E) (p < 0.001).

To characterize regional specificity of the changes, we measured GFAP immunoreactivity and microglial markers in the hippocampus and somatosensory cortex. In contrast to the initial report on the HIV-1 Tg rat [7], histological examination of the somatosensory cortex (Figures 2 A-B) and of the hippocampus (Figures 2 C-D) did not indicate increased GFAP immunoreactivity in the HIV-1 Tg rats, as their cells maintained a normal thin process-bearing morphology and there was no evidence of astrocyte hypertrophy (Figures 2 E-H). When we examined the morphological phenotype of microglia within various brain regions using Iba-1+ to label diverse phenotypes, minimal differences from control were noted in the HIV-1 Tg rats (Figures 3 A-H). In the somatosensory cortex, microglia maintained a normal appearance with fine ramified processes and had no prominent evidence of activation or of a phagocytic phenotype (Figures 3 A-B). When these immunopositive cells were examined at higher magnification (Figures 3 E-F), the Iba-1+ cells displayed decreased arbor complexity. Given previous reports of deficits in a hippocampal-dependent spatial memory task in HIV-1 Tg rats, we further examined the morphological phenotype of microglia within the dentate gyrus of the hippocampus. Overall labeling of Iba-1+ microglia was not significantly different in the HIV-1 Tg compared to control rats, with no evidence of overt microglia activation or amoeboid phenotype (Figures 3 C-D). At higher magnification, Iba-1+ microglia displayed fine processes and complicated arborization in the control brain (Figure 3G). A distinct difference was noted in the Iba-1+ microglia in the HIV-1 Tg rat hippocampus, with the cells displaying diminished arbor complexity and approximately 50% shortened processes (p < 0.05 by t-test) as determined by a modified Sholl analysis (Figure 3H), but with no evidence of amoeboid phagocytic microglia.

HIV-1 Tg rats showed significantly increased mRNA levels of inflammatory cytokines IL-1β (9.6-fold) (p < 0.001) and TNFα (3.5-fold) (p < 0.01) respectively (Figures 4A, B). These elevations corresponded to elevated brain protein levels of IL-1β (59%) and TNFα (45%) as compared to controls (Figures 4C, D) (p < 0.01). There was a 73% increase in NF-κBp50 activity in HIV-1 Tg compared to control rat brain (Figure 4E) (p < 0.01). However, NF-κBp65 activity did not differ significantly between groups (Figure 4F).

Brain protein and mRNA levels of a number of AA cascade markers were elevated significantly in HIV-1 Tg rats relative to controls. Mean mRNA levels of cPLA₂-IVA, sPLA₂-IIA and COX-2 were increased (p < 0.01) in HIV-1 Tg compared to control rats by 5-fold, 9-fold and 4.5 fold respectively (Figures 5A-C), but the iPLA₂-VIA mRNA level did not differ between groups (HIV-1 Tg 0.92 ± 0.12 vs. control 1.00 ± 0.30). Mean mRNA levels of mPGES (Figure 5D), COX-1 (HIV-1 Tg 0.87 ± 0.20 vs. control 1.00 ± 0.20) and cPGES (HIV-1 Tg 0.97 ± 0.20 vs. control 1.00 ± 0.20) were not significantly different between groups.

The mean protein level of cPLA₂-VIA was increased by 119% (p < 0.01), whereas sPLA₂ IIA protein was not changed significantly, as the increase was only at p < 0.05 (Figures 5E, F). The mean iPLA_2--VIA protein level also did not differ significantly between groups (HIV-1 Tg 114 ± 6.8 vs. control 100 ± 15). The mean protein level of COX-2 was increased significantly by 42% (p < 0.01) (Figure 5G), but the mean mPGES protein level was not (Figure 5H). COX-1 and cPGES protein levels did not differ significantly between groups (COX-1, HIV-1 Tg 118 ± 17 vs. control 100 ± 15; cPGES, HIV-1 Tg 101 ± 11.2 vs. control 100 ± 11).

There were statistically significant increases in mean brain mRNA levels of 5-LOX, 15-LOX and cytochrome p450 expoxygenase in HIV-1 Tg relative to control rats by 2.9-fold (Figure 6A) (p < 0.001), 4.6-fold (Figure 6B) (p < 0.01) and 4.4-fold (Figure 6C), respectively. Upregulation of these was unaccompanied by significant elevations in the respective mean protein levels, whose increases in each case were only at p < 0.05 (Figures 6D-F). Further, there was no significant difference in 12-LOX or TXS protein between groups (data not shown).

The active caspase 3 protein level (Figure 7A), and levels of neurofilament-L mRNA and protein (Figures 7C, D) did not differ significantly between HIV-1 Tg and control rats, as the former mean decreased at p < 0.05 and the values for neurofilament-L increased only at p < 0.05. BDNF protein (Figure 7B) and drebrin mRNA and protein (Figures 7E, F) were significantly less in HIV Tg than control rats (p < 0.01).

We have reported increased cPLA₂-IV and sPLA₂-IIA activities in the brain of 7- to 9-month-old HIV-1 Tg rats [9]. Consistent with these findings, HIV-1 Tg rat brain in the present study showed elevated protein and mRNA levels (p < 0.01) of cPLA₂-IVA and an elevated mRNA level of sPLA₂-IIA, without a significant change in iPLA₂-VIA or sPLA₂-IIA protein levels. COX-2 mRNA and protein levels were significantly higher in HIV-1 Tg rats than controls, whereas COX-1, cPGES or TXS did not differ significantly, consistent with our report of an increased brain concentration of PGE₂ but not of TXB₂ in HIV-1 Tg rat brain [9]. mPGES protein and mRNA levels were also were not increased in HIV-1 Tg rats.

Our earlier study also showed elevated levels of leukotriene B₄, a product of 5-LOX and leukotriene A₄ hydrolase, in the brain of HIV-1 Tg rats [9]. Consistent with that report, HIV-1 Tg brain in the present study showed significantly increased 5-LOX mRNA without a significant change in 12-LOX expression. This change was accompanied by increased mRNA levels of cytochrome p450 expoxygenase and 5-LOX. Given that epoxyeicosatrienoic acid produced by cytochrome p450 expoxygenase can be neuroprotective [43, 44], the elevated brain mRNA level of cytochrome p450 epoxygenase in HIV-1 Tg may reflect a compensatory neuroprotective process. While elevations in protein levels of 5-LOX, 15-LOX and cytochrome p450 expoxygenase did not reach significance because of our requirement for multiple comparisons, in each case changes were in the same direction as elevations of the respective mRNA at p < 0.05.

The changes in the AA cascade markers noted in HIV-1 Tg rat brain may be related to microglial activation, with release of proinflammatory cytokines and activation of the NF-κB transcription factor. NF-κB binding sites are present on the promoter region of the gene transcripts of the AA cascade markers, cPLA₂-IVA, sPLA₂-IIA and COX-2 [45‐47]. Cell culture studies have shown that IL-1β or TNFα can induce transcription of cPLA₂, sPLA₂ and COX-2 genes in an NF-κB-dependent manner [14‐17, 48]. NF-κBp50 is known to regulate transcription of pro-inflammatory genes [49, 50] and can influence HIV-1 gene expression [51]. Elevated DNA binding activity of NF-κBp50 in the HIV-1 Tg rat suggests that the elevated AA cascade markers in the current study may be related to increased levels of IL-1β and TNFα and increased NF-κBp50 DNA binding activity, but are independent of NF-κBp65.

In the absence of HIV-1 replication, the presence of gp120 and tat proteins in HIV-1 Tg rat brain likely account for microglial activation and the increased level of CD11b. In vitro, gp120 directly stimulates microglia and increases expression of CD11b [52]. However, microglia did show retraction of their processes and diminished complexity of arborization, which suggests an early reactive response. As we did not examine animals younger than 7 months, we cannot conclude that these changes were age-related. The altered microglial morphology in the hippocampus is of interest, given the role of the hippocampus in spatial learning tasks and the proposed involvement of microglia during synapse stripping and remodeling [53]. CD11b and Iba-1 cannot be used to distinguish between resident microglia and infiltrating blood borne monocytes. Thus, while we did not observe amoeboid brain macrophages, we cannot rule out a contribution of monocytes from the circulation, especially since gp120 can compromise the blood-brain barrier [54].

In vitro, gp120 can stimulate AA release and PGE₂ formation in glial cells [19‐21] and elevate levels of IL-1β in co-cultures of primary hippocampal neurons and astrocytes [19]. In the initial characterization of the HIV-1 Tg rat, a response of astrocytes was suggested by an increase in GFAP immunoreactivity [7]. In the current study, we did not find an astrocytic response. Consistent with no change in astroglial morphology, protein and mRNA levels of astroglial marker GFAP were not significantly altered. Further studies are needed to understand the role of astrocytes in HIV-1 infection. Similar to gp120, tat protein is also known to stimulate AA release and COX-2 expression in rat brain [55‐57]. The changes observed with neuroinflammatory and AA cascade markers in HIV-1Tg rats could be due to the presence of tat protein in the HIV-1Tg brain. Altogether, viral proteins can induce neuroinflammatory and AA cascade markers in brain. Despite altered protein levels of the AA cascade enzymes 5-LOX, 12-LOX and p450 epoxygenase a p < 0.05 in HIV-1 Tg brain, these changes did not reach statistical significance at p < 0.01. This may be due to the small sample size; further studies are required to understand changes in HIV-1 Tg brain.

Neuropathological features of human HIV-1 infection include cortical atrophy, altered dendritic arborization of neurons, and decreased synaptic density [10‐12, 58]. Neurons are vulnerable to both gp120 and the HIV-1 virus protein, tat [59, 60]. Gp120 and tat are reported to induce apoptosis of neurons in vitro and in vivo [61‐63] by activating caspases, particularly caspase-3 [59]. The current study did not show a statistically significant increase in the protein level of active caspase-3 in HIV-1 Tg rats. Damage and apoptosis of neurons would be manifest as a loss of neuronal and related markers. Within this framework, we now report a significantly lower mRNA and protein levels of the post-synaptic dendritic marker drebrin, and a reduced protein level of BDNF (p < 0.01). Protein and mRNA levels of neurofilament-L were reduced only at p < 0.05 in the HIV-1 Tg rats. Reduced BDNF is consistent with a report that gp120 reduces BDNF in rat brain in association with neuronal death [59]. A lifelong presence of gp120 in brain may impair neuronal development by reducing neurofilament and microtubule expression [32]. A significantly reduced level of drebrin suggests that altered synaptic structure contributes to cognitive-behavioral defects reported in the HIV-1 Tg rat [5, 6].

In brain, microglia are the primary source of TNFα [64, 65], and its release is implicated in neurotoxicity [66]. In HIV-1 Tg rats, elevated levels of IL-1β and TNFα and increased expression of AA cascade enzymes, have been implicated in neuronal damage [67] and cognitive-behavioral impairment [68‐73]. A recent study indicates that similar changes could contribute to cognitive impairment in HIV-1 infected patients despite antiretroviral therapy [74]. An association of increased expression of AA cascade enzymes with neurocognitive/neurodegeneration also has been suggested for Alzheimer disease and vascular dementia [75, 76]. In this regard, cPLA₂ inhibition or deletion improved learning and memory performance in a transgenic mouse model of Alzheimer disease [72]. Treatment with lithium or sodium valproate also was beneficial in HIV-1 associated dementia patients [77, 78], possibly by attenuating neuroinflammation and an upregulated brain AA cascade [79, 80]. Collectively, these observations suggest that neuroinflammation associated with increased AA metabolism can contribute to cognitive impairment, and that attenuation of AA release by inhibiting cPLA₂ may be beneficial.

The significant changes observed with AA cascade and neuroinflammation markers in HIV-1 Tg rats must be interpreted with caution because potential contamination of brain tissue by peripheral cells during cytosolic or total RNA isolation would give higher background levels for measured proteins, except for the neuron-specific marker drebrin. However, such changes are unlikely because we areported increased global brain AA incorporation from plasma in awake HIV-1 Tg rats [9]. Further examination is required to study the extent of activation of neuroinflammation and AA cascade markers in peripheral cells of HIV-1 Tg rats.

While differences between HIV-1 Tg and controls rates in several brain measures at the p < 0.05 level were not considered statistically significant because of the constraint of multiple comparisons (see Methods), they should be given some weight for several reasons, and might be reconsidered in the future with larger samples. This study was exploratory, and was focused on generating hypotheses that could be tested more discretely in the future. Importantly, many of the p < 0.05 changes in a protein occurred with a significant change at p < 0.01 in the respective mRNA, and vice versa, making the p < 0.05 change more credible.