Cognitive dysfunction in HIV encephalitic SCID mice correlates with levels of Interferon-[alpha] in the brain

Background: Interferon alpha (IFNα) is an antiviral cytokine produced in response to viral infection. IFNα also acts as a neuromodulatory molecule in the central nervous system (CNS). Elevated IFNα in the CNS causes cognitive deficits.

Objective: To determine if elevated levels of IFNα in an HIV encephalitis mouse model correlate with cognitive deficits.

Methods: C57BL/6J SCID mice were inoculated intracerebrally (i.c.) with HIV infected or uninfected (control) macrophages and cognitively tested in a water escape radial arm maze. After behavioral testing was completed, immunohistochemistry and ELISA were used to examine brain pathology and IFNα expression.

Results: Mice injected i.c. with HIV infected macrophages exhibited significantly more working memory errors, particularly in trials with the highest memory load. Immunohistochemistry indicated increased mouse IFNα staining prevalent on neurons and glial cells in the brains of mice with HIV infected macrophages compared to mice with uninfected control macrophages. In addition, IFNα levels in the brain correlated directly with working memory errors for mice with HIV infected macrophages.

Conclusions: These data suggest that the cognitive deficit noted for the C57BL/6J SCID mice with HIV infected macrophages is mediated by the infection induced increase in IFNα.

Introduction

HIV associated dementia (HAD) is a devastating neurological disease affecting approximately 15% of HIV-positive patients in the USA and affects as many as 30% of HIV patients worldwide ^[1-3]. HAD is a subcortical dementia characterized by cognitive slowing and memory impairment progressing to dementia and eventually death ^[2,4]. Brain pathology reveals HIV-positive mononuclear phagocytes (MP), increased MP, multinucleated giant cells, astrogliosis, loss of neuronal dendrites, and release of putative neurotoxins ^[5,6]. Despite the abundance of information characterizing HIV encephalitis (HIVE) and the clinical expression of HAD, relatively little is known about the specific pathological features that correlate with HAD ^[6]. Because neurons are not infected, the prevailing hypothesis of neuron dysfunction in HAD centers on the production of neurotoxins by macrophages and glia ^[7]. Putative neurotoxins include cytokines such as tumour necrosis factor α and interferon α (IFNα), chemokines such as macrophage induced protein-1, HIV proteins gp120 and tat, and cellular metabolites including arachidonic acid and nitric oxide ^[7]. Establishing specific pathological features of HIVE that correlate with the cognitive abnormalities of HAD and determining which putative neurotoxins are actually involved in HAD pathogenesis may lead to better treatments for patients ^[8,9].

IFNα is a pleomorphic cytokine that is increased during HIV infection in the brain ^[10,11]. IFNα interferes with virus replication and profoundly increases the immune response to viral infection ^[12]. IFNα is secreted by a wide variety of cells including macrophages, T lymphocytes, glia, and neurons ^[13]. Because of its antiviral and immunoregulatory effects, IFNα is used as a therapy in a variety of viral infections and cancers ^[14,15]. During prolonged treatment with high doses of IFNα patients develop central nervous system (CNS) side effects including cognitive slowing, amnesia, and impaired executive function ^[15], key symptoms consistent with subcortical dementia. Importantly, IFNα levels in the cerebrospinal fluid (CSF) of HIV positive patients with dementia are significantly higher than those that show no signs of dementia ^[11,16].

A SCID mouse model of HIVE has been developed which recapitulates many of the pathological and behavioral features seen in human HAD ^[17-20]. Analysis of the HIVE mouse model using the Morris water maze, a reference memory task, has consistently shown cognitive deficits in HIVE mice that are comparable to behavioral abnormalities associated with HAD in humans ^[19-22]. Pathologically, HIVE SCID mice contain HIV-positive human MP, multinucleated giant cells, increased mouse MP in the CNS, astrogliosis, loss of dendritic arborization, and increased expression of proinflammatory cytokines ^{[17,19,22-24]}. In this report, data are presented that C57BL/6J SCID mice produce IFNα in response to intracranial (i.c.) inoculation with HIV-infected human macrophages. The increased IFNα expression in the brains of HIVE mice directly correlated with behavioral abnormalities. We hypothesize that IFNα mediates early cognitive deficits in human HAD.

Materials and methods

Animals

Four-week old C57BL/6J SCID male mice were obtained from Jackson Laboratory (Bar Harbor, Maine, USA). Mice were single-housed in micro-isolator cages (biosafety level-3 equivalent) and given 1 week to acclimate before experimentation. The animal room was set on a 12-h light cycle. Cages, bedding, food, and water were sterilized before use. Animal protocols were approved by the Medical University of South Carolina's Institutional Animal Care and Use Committee.

Method of infection

Primary human macrophages were obtained from H. Gendelman (University of Nebraska Medical Center, Omaha, Nebraska). Cells were cultivated in Dulbecco's modified Eagle's medium (DMEM; Gibco, Grand Island, New York, USA) with 10% human serum, glutamine (Gibco, Grand Island, New York, USA), penicillin-streptomycin (Sigma, St. Louis, Missouri, USA), and monocyte-colony stimulating factor (Sigma) at 37°C with 5% CO₂ in Teflon-coated culture flasks (CoStar, Cambridge, MA, USA). After 7 days, macrophages were divided into two cultures; resuspended in 9 ml media consisting of DMEM with 10% human serum, glutamine supplement, and penicillin-streptomycin. One culture was infected with 1 ml HIV-1ADA (multiplicity of infection, 0.1; from H. Gendelman), an R5 strain of HIV that uses the chemokine receptor CCR5 as a co-receptor. The alternative control culture had 1 ml DMEM added to the flask. After 1 h of infection, virus was washed off and both cultures were resuspended in 25 ml media. Cells were collected 2 weeks after HIV infection. Some human macrophages were put on slides and stained for HIV p24 antigen. Approximately 40% of macrophages expressed viral antigen. The macrophages were resuspended in phosphate-buffered saline (PBS) for inoculation into mice. Under xylazine (5 mg/kg) and ketamine (95 mg/kg) anesthesia, HIV-infected or uninfected macrophages (10⁵) were injected i.c. into the right frontal lobe of 5-week old SCID mice (n = 10 in each group for behavioral testing and pathology analysis, n = 5 in each group for protein assays) ^[18].

Apparatus and testing procedure

Animals were tested on the water radial-arm maze (RAM), a spatial working and reference memory task (Fig. 1), 6 days after inoculation. The black maze was filled with black-dyed water (18-21°C). Hidden platforms were placed in four of the eight arms. Each subject kept the same platform locations throughout testing.

Fig. 1

A subject was released in the start arm and had 2 min to locate a platform. If the allotted time expired, the mouse was guided to the nearest platform. Once the platform was found, the mouse remained on it for 15 s, and then was returned to its heated cage for 30 s until its next trial. During this inter-trial interval, the just-located platform was removed. The mouse was placed back in the start arm to locate remaining platforms. A daily session repeated this process until all four platforms were found. Thus, for each animal a daily test consisted of four trials. Behavioral testing took place between 9: 00 am and 12: 00. Each mouse (n = 10 per group) was tested for 12 days. Day 1 was a maze habituation day and did not count towards scoring.

Behavioral scoring

An arm entry was counted when the entire body of the mouse, excluding the tail, passed into an arm (see insert, Fig. 1). Errors were quantified for each daily session using orthogonal measures of working and reference memory errors, as carried out previously using the water RAM ^[25-28]. A mouse could make three types of errors: (i) reference memory errors (RM)-the first entry into an arm that never had a platform; (ii) repeated RM errors (RRM)-repeated entries into arms that have never had a platform; (iii) working memory correct errors (WMC); an entry into an arm that once had a platform but which was removed once the animal located it.

Tissue sectioning and immunocytochemistry

After behavioral testing was completed, the mice were sacrificed (18 days after i.c. inoculation of the cells), and their brains were removed for pathological analysis. Tissue sectioning and immunocytochemistry were performed as described previously ^[19]. Mouse brains were snap frozen in tissue-freezing medium, and 5 μm coronal sections were taken starting at the frontal lobe through the occipital lobe. Approximately 70 sections were taken per mouse. Tissue sections were then stained using an immunoperoxidase method ^[19] for human macrophages (1: 50 EBM11; DAKO, Glostrup, Denmark), HIV (1: 20 p24; DAKO), human IFNα (1: 200 PBL, Piscataway, NJ, USA), astrocytes (1: 750 glial fibrillary acidic protein; Chemicon, Temecula, California, USA), microglia (1: 20 F4/80; Caltag, Burlingame, California, USA), mouse IFNα (1: 200 PBL), and neuronal dendrites [1: 200 microtubule associated protein-2 (MAP2), Chemicon], providing 10 slides for each antibody per brain. Slides were then reviewed using light microscopy (Olympus microscope: Melville, New York, USA), and pathology was quantified using densitometry analysis. The percentage of HIV-infected cells in the brains was calculated as the number of p24 positive human monocytes/the total number of human monocytes × 100.

Double fluorescence immunohistochemistry

Five-μm frozen tissue sections from brains with control and HIV-infected macrophages were fixed in 95% EtOH for 10 min. Sections were blocked with 2% serum in PBS (1% horse serum, 1% goat serum; Sigma) for 1 h, and then incubated in primary antibody NeuN; 1: 200 (Neuron cell bodies; Chemicon) and mouse IFNα; 1: 200 (PBL). Tissue was incubated for 30 min in the dark with secondary antibody 1: 100 (FITC labeled goat antirabbit; Vector Labs, Burlingame, California, USA) and 1: 75 (Texas Red horse antimouse; Vector Labs). Slides were viewed under an Olympus BH-2 microscope with fluorescence.

Densitometry scoring

Immunoperoxidase stained slides were imaged at 20 × using an Olympus microscope with a DP11 digital camera. Images were analyzed with NIH Image 1.63.

Densitometry readings for mouse IFNα, microgliosis and astrogliosis

The standard for positive staining for each antibody was the section from mouse brain with HIV-infected macrophages that was most intensely positive for mouse IFNα, mouse microgliosis, or mouse astrogliosis. The intensity of the immunoperoxidase stain was measured using NIH image. The measured range of intensity for this positive control section set the range for positive staining of both control and HIV brain sections to determine the quantification of positive staining.

Densitometry for MAP2

The left hemisphere (noninjected hemisphere) served as the control for a section. The intensity of immunoperoxidase staining was measured. This value was set at 100% for the tissue section. The range of positive staining on the left hemisphere was applied to the region directly around the human cells in the right hemisphere of the section, which displayed decreased MAP2 staining. The measured value of the right hemisphere was compared to the left hemisphere of the section to determine the relative difference in positive MAP2 staining.

Protein extraction

Mice (n = 5 per group) were sacrificed 18 days after i.c. cell inoculation and their brains removed for protein extraction. Brains were snap frozen in tissue-freezing medium; frontal cortex was removed and weighed. The tissue was placed in a 5-ml glass conical tube with Tissue Protein Extraction Reagent (T-PER, Pierce, Rockford, Illinois, USA) at 10 ml reagent/g tissue and Halt Protease Inhibitor Cocktail (Pierce) at 10 ul/ml T-PER. Tissue was homogenized, and the protein concentration measured using a Coomassie Plus Blue Assay (Pierce) on a Spectronic Genesys 5 spectrophotometer (Spectronic Instruments, Leeds, United Kingdom) at 595 nm.

Mouse IFNα ELISA

Mouse IFNα quantification from brain tissue was determined by sandwich immunoassay ELISA (PBL). Protein samples were standardized to equal concentrations and measured in triplicate. The ELISA plate was read at 450 nm on a Versamax microplate reader (Molecular Devices, Sunnyvale, California, USA) and Softmax Pro 3.1.2 software (Sunnyvale, California, USA).

Statistical analysis

StatView 5 for MacIntosh computers (SAS Institute, Cary, North Carolina, USA) was used for all statistical analyses. The 11 testing days were blocked into two phases: block 1, days 2-7; block 2, days 8-12. For WMC and RRM, data for each block were analyzed using a 1-between (HIV status), 2-within (days and trials) repeated measures analysis of variance (ANOVA). Trial 1 was not included in the WMC analyses because it is not possible to make a WMC error on the first trial. For RM, each phase was analyzed using a 1 between (HIV status), 1 within (days) repeated measures ANOVA. Analysis of trials was done with post hoc t tests. Analysis of densitometry and ELISA were competed using a two-way ANOVA. Pearson r correlations were used to evaluate relationships between the neurobiological and behavioral variables. Significance was set at P < 0.05 for all analyses.

Results

C57/B6 SCID mice injected with HIV-infected macrophages exhibit increased memory deficits

Mice injected i.c. with HIV-infected macrophages made more WMC errors than controls during block 2 of RAM testing (P < .01)(Fig. 2a), and exhibited disproportionately more errors as working memory load increased (P < 0.05 for trials 3 and 4) (Fig. 2b). Mice with HIV-infected macrophages performed similar to controls when memory load was low, indicating that they were able to perform the task when cognitive demand was low. However, as the number of items the mouse needed to remember increased, mice with HIV-infected macrophages were less able to successfully maintain performance. These findings indicate that the motivator (water-escape) was not responsible for the performance deficit, as mice with HIV-infected macrophages did not differ from controls on early trials when the procedural and motivational requirements of the task were identical to later trials, but working memory load was low.

Fig. 2

HIVE SCID mouse histopathology

As typically seen in the HIVE SCID mouse model, similar amounts of human macrophages were seen in the right frontal lobes of both mice with HIV-infected and uninfected macrophages in the region surrounding the injection site. The number of HIV positive cells was counted and compared to the total number of human monocytes counted to determine percentage of HIV-infected cells in mice with HIV-infected macrophages. In this study 87.5% of human monocytes expressed p24 antigen determined by immunohistochemistry (IHC) analysis, indicating that viral proteins are produced in the HIV-infected human macrophages 18 days after i.c. injection. The higher percentage of HIV-infected human macrophages in brain compared to culture (approximately 40%) suggests that HIV spread to uninfected human macrophages during this time. The presence of activated astrocytes and microglia along with changes in MAP2 staining for neuronal dendrites for both groups was seen in the anterior portion of the frontal cortex extending throughout the right hemisphere, and concentrated near the injection site of the monocytes (Fig. 3a-f). Densitometric analysis of brain tissue sections from the mice showed increased astrogliosis and microgliosis surrounding the injection site in mice with HIV-infected macrophages compared to mice with control macrophages (P < 0.05, Fig. 3g and h). Densitometry analysis of MAP-2 staining showed decreased dendritic arborization of neurons surrounding the HIV-infected macrophages in mice with HIV-infected macrophages compared to mice with uninfected macrophages (P < 0.05, Fig. 3i).

Fig. 3

HIVE SCID mice express mouse IFNα in the brain

Immunohistochemical staining revealed increased mouse IFNα in brains of mice with HIV-infected macrophages compared to brains of mice with control macrophages (Fig. 4a,e). Densitometry showed increased intensity of staining for mouse IFNα in mice with HIV-infected macrophages compared to brains of mice with control macrophages (P < 0.05) (Fig. 4i). Mouse IFNα ELISA quantification confirmed more than a fourfold increase of IFNα in brains of mice with HIV-infected macrophages compared to mice with control macrophages (P < 0.05)(Fig. 4j). HIV-infected macrophages inside the mouse brain produced human IFNα (data not shown). The location of human IFNα was limited to human macrophages in the brains of mice injected with HIV-infected macrophages. Human IFNα was found neither on mouse cells nor on human macrophages in the brains of mice injected with uninfected human monocytes (data not shown). Conversely, mouse IFNα was not detected on human macrophages.

Fig. 4

To determine the location of mouse IFNα, double immunofluorescent staining for neurons (NeuN) and mouse IFNα was completed on 5-μm sections of brains from mice with HIV-infected and control macrophages (Fig. 4b,f,c,g). Brain tissue from mice with control macrophages showed little overall presence of mouse IFNα, and a small amount of colocalization of IFNα with neurons (Fig. 4c,d). Mice with HIV infected macrophages also showed increased mouse IFNα compared to controls, and co-localization of IFNα with neurons (arrows on Fig. 4h). IFNα was also present on NeuN-negative cell types; these cells are possibly glia (arrowheads on Fig. 4h).

Mouse IFNα expression correlates with abnormalities in behavioral testing

Densitometry measurements of IFNα in mice with HIV-infected macrophages correlated with mean behavioral performance on trial 4 of days 8-12 on the water RAM, when working memory load was highest. Mice with HIV-infected macrophages with higher levels of mouse IFNα tended to make more WMC errors (r = 0.69, P = 0.04, Fig. 5). These data strongly suggest that higher levels of IFNα are related to the cognitive dysfunction of HIVE SCID mice, accounting for just under 50% of the total variance.

Fig. 5

Discussion

Consistent with previous studies, in this report SCID mice with HIVE demonstrate cognitive deficits ^[19,20]. This study extends prior findings by utilizing a different strain of SCID mouse (C57BL/6J, previous strain CB-17) as well as evaluating cognitive status on a new behavioral paradigm that evaluates working and reference memory simultaneously. C57BL/6J SCID mice with HIV-infected macrophages committed more errors than control mice throughout testing, and the impairment was most profound when working memory load was the highest in later trials. The performance deficits seen in mice with HIV-infected macrophages were specific to working memory, a behavior that has been mapped to the hippocampus and frontal cortex ^[27-30], regions of the brain that are affected in HAD patients ^[4,6,31].

C57BL/6J SCID HIVE mice have similar histopathological findings to CB17 SCID mice, including the presence of HIV-infected human monocytes, increased gliosis, and neuronal abnormalities manifested by decreased MAP2 surrounding HIV infected cells ^[17,22,24]. Importantly, the results of this study demonstrate increased IFNα in the brains of mice injected with HIV-infected macrophages. There was a positive correlation between the amount of IFNα immunostaining on neurons and glia and cognitive abnormalities, with higher levels related to poorer cognitive performance. In addition, human IFNα was detected by IHC on HIV-infected human macrophages, but human IFNα was not detectable on mouse cells or on control human macrophages. The small amount of human IFNα is unlikely to contribute significantly to the cognitive deficits seen in this model compared to the relatively large amount of mouse IFNα produced in the brains of mice injected with HIV-infected macrophages. Additionally, IFNα is produced by neurons and glia and binds to the IFNα receptor (IFNAR) in a species specific manner ^[32-34]. Although mouse cells are not productively infected by HIV, IFNα induction is expected in the mice because the HIV proteins gp120 and tat are strong activators of IFNα in vitro ^[35-37]. The data in our study do not allow determination of whether the stained cells produced or merely bound the IFNα.

Increased IFNα in the CNS adversely affects cognitive performance in mice and humans. Repeated intra-peritoneal or intravenous injections of IFNα into mice results in increased cognitive dysfunction ^[38,39]. Behavioral analysis of transgenic mice that constitutively overexpress IFNα in astrocytes demonstrated cognitive abnormalities compared to wild type mice ^[40]. Similar to the transgenic mice, increased levels of IFNα in the brains of HIVE mice correlated with cognitive impairment in this study. Patients treated with IFNα therapy may develop neurological deficits including subcortical dementia. These deficits are dependent both on the dose and length of IFNα treatment ^[15,41,42].

IFNα has a direct impact on a variety of brain functions including altered physiological activity of neurons in the hippocampus ^[34,40,43]. IFNα inhibits long-term potentiation (LTP) and excitatory postsynaptic potentials (EPSP) in rat hippocampal neurons in vitro ^[13,43]. LTP activity plays a role in memory function, and EPSP are indicative of neuron signaling strength and health. Glutamate-mediated EPSP and LTP in the hippocampus are inhibited by human recombinant IFNα in a concentration-dependent manner ^[43]. In vivo electrophysiological examination of a transgenic mouse that overexpressed IFNα showed decreased EPSP and LTP in the hippocampus ^[40]. In vivo electrophysiology was also used to examine hippocampal function in HIVE SCID mice. These studies reported that EPSP and LTP in the hippocampus were partially inhibited in the HIVE SCID mouse ^[44-46]. Interestingly, the inhibition of EPSP and LTP in hippocampal neurons exposed to IFNα ^[40,43] is similar to the decreased signaling seen in the HIVE SCID mouse hippocampal slices ^[44-46]. It is therefore plausible that the excess IFNα produced in HIVE mice, as recorded in our study, produces detrimental effects on neurons through downstream targets of the IFNAR or interaction with other receptors such as opioid and/or glutamate receptors ^[34,43].

These cognitive and electrophysiological effects of IFNα may be reflected histopathologically by adverse effects on dendritic arborization ^[40]. MAP2 is found extensively in dendrites and long-branched dendritic arbors are associated with the health, viability, and connectivity of neurons ^[47]. Analysis of transgenic mice with astrocytes that constitutively produce IFNα showed a significant degeneration of axons and dendrites of neurons ^[40]. Similarly, our study demonstrates decreased MAP2 staining in HIVE mice compared to controls. The decreased dendritic arborizations noted in IFNα transgenic mice and HIVE mice may represent a histological correlate of the inhibition of LTP and EPSP described above. Furthermore, these histological and electrophysiological abnormalities for IFNα exposed or HIVE mice may represent elemental changes in neuron physiology and structure which relate to the cognitive deficits seen in the IFNα transgenic mice, HIVE mice, and humans with HAD.

In HIVE patients, IFNα expression is elevated as measured by increased expression of type I IFN genes, including IFNα and IFNα inducible genes in the frontal cortex ^[48]. IFNα was also found to be significantly higher in the CSF of HIV-infected patients with dementia compared to HIV patients without dementia or HIV-negative individuals ^[11,16]. It is known that patients with high levels of IFNα in the CNS have cognitive deficits, most importantly depression and frontal subcortical dementia ^[15,41,42], which are some of the same neurological problems seen in HIVE patients. Collectively, the data cited above strongly suggest that IFNα is involved in the pathogenesis of HAD in humans ^[6,11,16]. Over-expressed IFNα in the CNS may not only be a critical component in the pathogenesis of early cognitive dysfunction in HAD patients, but could be a factor in cognitive dysfunction in a wide range of infectious and neuroinflammatory illnesses.

Sponsorship: Supported by MUSC Institute of Neuroscience Grant, NIH RO1 MH62697-03, NIH C06 RR015455.