To investigate the neuropathogenic mechanism of HAND-related pathologies observed in human patients, relevant animal models are essential. Several animal models develop specific aspects of cognitive defects and neuropathological key features of HAND.
Non-human primate models
The simian immunodeficiency virus (SIV)-infected macaque is an established relevant model for studying the pathogenesis of HAND. In monkeys, SIV can enter the brain shortly after infection and causes brain abnormalities. SIV infection recapitulates the main features of immune response of HIV infection [
49‐
53]. Additionally, HIV-associated neuropathologies in the brains of HIV patients are also developed in the SIV-infected macaque. For example, pre-synaptic damage was reported in SIV-infected macaques, as indicated by elevated levels of neuronal damage marker 14-3-3 protein in the CSF [
54,
55]. SIV-infected macaques developed various types of behavioral impairments, similar to those observed in HIV patients, as shown by a number of behavioral and neurophysiological testing modalities [
56‐
60]. This model is particularly useful to study the pathogenesis of HAND in the era of HAART, because the infected macaque can be treated with HAART regimens to mimic the clinical settings [
61]. It is also very helpful for the investigation of the synergized effects of drug abuse and HIV infection during neuropathogenesis [
62‐
65]. In addition, because of the multi-time accessibility of CSF, plasma and CNS samples during the progression of infection, this model allows the investigation of the development of HAND through the progressive stages.
Although studies with SIV-infected macaques provide valuable insights into the pathogenesis of HIV infection, it is important to keep in mind that SIV and HIV are not the same. For example, CCR5-preferred HIV can gain the ability to use CXCR4 to enter into monocyte-derived macrophages [
66,
67], while CCR5-preferred SIV uses other co-receptors such as CXCR6, GPR15 and GPR but not CXCR4 to enter host cells [
68]. To address these limitations, simian-human immunodeficiency virus (SHIV) was constructed, in which the env gene of SIV was replaced by HIV-1 env. Therefore, the hybrid viruses are biologically more similar to HIV than SIV. Macaques infected with SHIV89.6P (CXCR4/CCR5 virus) developed encephalitis characterized by multinucleated giant cells, astrogliosis, microglial nodules, activated macrophages and astrocytes, and perivascular cuffing with mononuclear cells in the white matter [
69]. CCR5 (R5)-tropic SHIVSF162P3N virus caused giant cell SIV encephalitis in approximately 30% of infected rhesus macaques that developed AIDS [
70]. Giant cell SIV encephalitis lesions included white matter damage, necrosis, and astroglial and microglial activation [
70]. SHIVKU, a CXCR4 virus, also could productively replicate in the CNS of rhesus macaques and caused pathological changes [
71‐
73]. Despite the significant contributions of non-human primate models to understanding HIV-1-associated neuropathogenesis, these models are limited by their availability and high cost of maintenance.
Rodent models
For reasons that are not completely defined, rodents cannot be productively infected by HIV-1. To circumvent this drawback, transgenic mice are generated to express HIV-1 proteins such as the envelope protein gp120 and the transactivator of transcription (Tat), both of which are neurotoxic. In a gp120 transgenic mouse (gp120Tg) model, the gp120 transgene is controlled by the glial fibrillary acidic protein promoter, and thus gp120 is restricted to astrocytes [
74]. The release of astrocytically expressed gp120 protein can affect nearby neurons. Confocal imaging of brain sections labeled with dendritic and synaptic markers revealed the dendritic vacuolization, loss of dendritic spines and presynaptic termini in the neocortex and the hippocampus [
74]. This gp120Tg mouse also showed reaction of glial cells [
74] and impaired proliferation and differentiation of neuronal progenitor cells [
75,
76]. Additionally, aging (12 months) gp120Tg mice developed deficits in motor and cognitive performance [
74,
77].
In another transgenic mouse model, the Tat transgene is expressed in astrocytes in a Dox-regulated manner [
78]. The inducible expression of Tat provides the ability to study the temporal effect of Tat released from astrocytes. This transgenic mouse displays degeneration of neuronal dendrites, neuron death, astrocytosis and enhanced infiltration of activated monocytes and T lymphocytes, and these alterations are largely observed in the cerebellum and cortex [
78]. Other studies described more subtle neuronal injuries such as spine loss and synaptic degeneration in hippocampal pyramidal CA1 neurons and striatal neurons [
79‐
81]. The Tat transgenic mice develop impairments in spatial memory and novel object recognition memory [
78,
81,
82].
Transgenic mice with full-length [
83,
84] or
gag-pol-deleted HIV-1 genomic DNA [
85] have been reported. The integrated HIV-1 genome in the transgenic mouse somewhat resembles HIV-1 provirus. In addition, the transgenic HIV-1 genome has the potential to express multiple HIV-1 proteins. These strengths of this transgenic strategy, however, also complicate the result interpretation for determining the causal relationship between specific HIV-1 proteins and observed phenotypes. Despite low levels of viral protein expression, the full-length transgenic mouse model shows impaired nerve conduction, axonal degeneration and decreased nerve fiber density in the peripheral nervous system. They are also impaired in motor function [
83], and show hyper-reaction of microglia and astrocytes [
84,
86].
The HIV-1 transgenic rat has been studied by multiple groups as a model of HIV-associated neurological diseases. It contains a
gag-
pol-deleted HIV-1 genome that is controlled by the viral promoter. Since without
gag and
pol genes that are responsible for viral replication, it cannot produce infectious virions [
87]. This rat model expresses multiple viral proteins. In particular, the expression of Tat, gp120, nef and vif RNAs show age-dependent profiles, shifting from peripheral immune organs to the CNS at 10–11 months of age. These features of HIV-1 gene expression indicate that the HIV-1 transgenic rats can model specific aspects of HIV-1-infected individuals on HAART [
88]. The 7-to-9-month-old animals show up-regulated expression of neuroinflammation markers such as interleukin-1β (IL-1β), tumor necrosis factor α (TNF-α) and microglial/macrophage marker CD11b [
89], which may contribute to the observed synapto-dendritic injury [
89]. The transgenic rats develop spatial learning deficits [
90,
91] and are impaired in motor performance [
92].
The HIV-1 transgenic rodent models described above provide useful tools to study the contribution of viral proteins to the pathogenesis of HAND. However, they have significant limitations. Foremost, they do not acquire HIV-1 infection and thus cannot faithfully model the initial infection stages or the AIDS progression, which are key events associating with HAND development. Understandably, efforts continue to create additional rodent models to mimic HIV infection. One strategy is to introduce human HIV-1 receptors and co-receptors in transgenic rodents [
93]. However, it appears that HIV-1 replication was defective in CD4 or CCR5 transgenic rodents [
94,
95].
Potash et al. designed a creative approach to generate a novel mouse model of HIV-1 infection. They constructed a chimeric HIV-1 virus by replacing the HIV-gp120 coding region with the gp80 envelope gene from the ectotropic murine leukemia virus. This chimeric virus, called EcoHIV, can enter to the host cells by binding to cationic amino acid transpoter-1 (mCAT) [
96]. Despite the widespread expression of mCAT in the mouse tissues, persistent infection seems to be restricted to splenic lymphocytes, peritoneal macrophages and brain [
96,
97]. EcoHIV infection by stereotactic inoculation into the mouse basal ganglia caused pre-clinical brain pathology such as microglia and astrocyte activation [
96,
98]. However, the lack of gp120 in the chimeric virus presents specific limitations in this model. First, it is unclear to what degree the chimeric virus mimics the HIV-1 infection. For example, it may not target the same populations of cells as HIV-1. In addition, because gp120 is a major HIV-1 neurotoxic protein, this model may not recapitulate some of the neuropathological phenotypes related to HAND.
HIV-infected humanized mice are the exciting new rodent models. One strategy is to generate humanized mice with CNS HIV infection by direct injection of infected human cells. HIV-infected human monocyte-derived macrophages or HIV-infected human microglia cells are injected into the brain of severe combined immunodeficiency deficient (SCID) mice [
99,
100] or reconstituted SCID mice with human peripheral blood leukocytes (PBLs) (huPBL/SCID) [
101,
102]. SCID and huPBL/SCID mice with the infected human cells recapitulate the several neurological pathologies observed in HIV patients with HIVE, including multinucleated giant cells, astrogliosis, microglial activation and neuronal damage [
99‐
102]. The SCID-HIVE mouse model also develops cognitive deficits. Morris water maze tests revealed their learning and memory impairments, regardless of HAART treatment [
103]. Using these models, isolate-specific cognitive deficits and neuropathology were reported. Intracranial injection of macrophages infected with a clade B HIV-1 isolate (HIV-1(ADA)) into SCID mice caused worse performance in cognitive tests and more severe pathological changes than a clade C HIV-1 isolate (HIV-1(Indie-C1)) [
104].
Another strategy to generate humanized mice is systemic transplantation of human hematopoietic stem cells (CD34
+ cells) or adult human peripheral blood mononuclear cells into various immunodeficient mice so that the mice host the human target cells for HIV-1 infection [
105‐
110]. Various neuropathologies were reported in HIV-infected humanized mouse models. For example, NOD/SCID-IL-2Rγ
c
nul (NSG) mice with engrafted human CD34+ stem cells (NSG-hCD34+) developed a functional human immune system containing T lymphocytes, monocytes and macrophages could be efficiently infected with HIV [
111‐
114]. Neuronal and synaptic damages were detected by immunohistochemical staining of various neuronal and synaptic markers such as microtubule associated protein-2, neurofilament and synaptophysin. The neuropathologies appeared to correlate with glial cell activation [
112,
113]. The animals also showed memory deficits and persistent anxiety [
112,
113]. Although less used for CNS infection, other humanized mouse models (e.g. humanized bone marrow/liver/thymus mouse models) have been used for studies on HIV pathogenesis, transmission, replication and prevention.