Nanoparticles for Diagnosis and/or Treatment of Alzheimer’s Disease
4.3 Physiological Factors Related with Brain-Located Pathologies: Focus on AD
4.3.3 In vitro and in vivo Models for BBB Permeability and AD Diagnostic/Th erapeutic Approach Assesment
Methods used to investigate BBB transport or for screening of potential delivery systems for their capability to facilitate drug or imaging agent transport across the BBB, (in vitro systems, cellular models, and animals [wild type (WT) or transgenic (TR)]), as well as animal models which have been developed up-to-date for assessment of AD diagnostic and/or thera- peutic approaches, are mentioned and briefl y described.
4.3.3.1 In vitro Methods
Th e in vitro methods can be categorized in two main groups: non-cell and cell-based permeability assays. Non-Cell-Based Permeability Assays: About 30 years ago, eff orts were initiated toward the development of meth- odologies for molecule membrane permeability screening. Various tech- niques, such as high-performance liquid chromatography (HPLC) (by development of affi nity columns), “immobilized artifi cial membranes”
(IAMs) that mimicked the properties of biological membranes, etc., have been identifi ed. Some of the later were moderately successful and capable of ranking compounds according to BBB permeability [95], but they were not found suitable for medium- to high-throughput operation. A more promising technology is the parallel artifi cial membrane permeability assay (PAMPA), which is suitable for the study of passively permeating compounds, showing moderately good correlation with data derived from the human colonic epithelial cell line (Caco-2) and from in vivo perme- ability studies. By modifying the lipid composition of the artifi cial mem- brane, which is the basic component of the PAMPA system, it appears to be capable of predicting CNS permeability with reasonable accuracy [96].
Cell-Based Permeability Assays: Several cell culture models (summarized in Table 4.2) have been developed and used for BBB permeability prediction studies. Most in vitro BBB cellular models proposed so far are based on co- cultures of brain endothelial cells (ECs) and astrocytes (or glial cells) [97],
in two-chamber cell culture systems. In particular, whereas bovine brain endothelial cells alone only partly recapitulate BBB properties, a co-culture system with rat glial cells has been extensively validated as a reference BBB model [98]. Because these cells express tight junctions as well as numerous membrane transporters, they constitute a valuable alternative or comple- ment to the epithelial cell lines Caco-2 and MDCK, currently used for drug screening by pharmaceutical industries because of their very high perme- ability restriction [99]. Alternative BBB models are also available, using pig, mouse, rat or human brain ECs [100–106]. In addition, stable immortalized
Table 4.2 Cellular models of the BBB.
BRAIN ENDOTHELIAL CELLS PRIMARY ENDOTHELIAL CELLS
Origin Ref
Bovine [98]
Porcine [102]
Rat [103]
Mouse [101]
Human [104]
IMMORTALIZED BRAIN ENDOTHELIAL CELL LINES
Bovine SV-BEC [111]
Porcine PBMEC [113]
Rat RBE4 [108]
Mouse bEND5 [112]
Human hCMEC/D3 [116]
CO-CULTURES
Glial Cells/Astrocytes [98]
Pericytes [106]
Astrocytes and Pericytes [97]
Neurons [105]
rat EC lines were produced and validated as in vitro models of brain endo- thelium: fi rst the RBE4 cell line [107–109], followed by a number of other cell lines [110– 113], some of which have been widely used for biochemical, immu- nological and toxicological studies. More recently, the human hCMEC/D3 brain EC line, which retains most of the morphological and functional char- acteristics of brain endothelial cells, was developed. Th is cell line expresses TJs and multiple active transporters, receptors and adhesion molecules, even without co-culture with glial cells, thus appearing as a unique and easy to maintain in vitro model of the human BBB [114–116]. Recently drug per- meability across hCMEC/D3 monolayers was signifi cantly decreased when the system was used under three-dimensional fl ow conditions [117, 118]
(compared to the non-fl ow setup), improving the predictability of the sys- tem. All together, these in vitro BBB models are helpful for screening of new drugs as well as for unraveling the molecular mechanisms of BBB-control under physiological conditions and in various CNS diseases.
For measurement of BBB-permeability, cell monolayers are generated by growing the cells on permeable fi lters, mounted in a device that separates the apical (luminal) compartment and the basal (abluminal) compartment (Ussing chamber, Transwell system, etc.). Th e drug is added to one com- partment (the donor) and the amount appearing in the other compartment (receiver) is determined over time. For measurement of Uptake by Cells: the study typically involves incubation of the cells with a tracer molecule, stop- ping the uptake process at set time-points (e.g., by washing with cold buff er solution, or using a stopping solution containing specifi c inhibitors to block transport), lysing the cells and analyzing the contents for presence of the tracer (by radioactive counting, quantifi cation of fl uorescence, HPLC or liq- uid chromatography-mass spectrometry [LC/MS]), as well as total protein content. In order to better mimic the in vivo situation, some three-dimen- sional “dynamic” in vitro BBB models (DIV–BBB) have been developed. In these, intraluminal fl ow is incorporated, providing an optimal combination of conditions, which encourage BBB diff erentiation. For this, endothelial cells are seeded inside porous tubes with astrocytes on the outside, and medium fl ow is maintained via the lumen. Such systems provide high TEER values while numerous transporters and receptors (characteristic of the BBB) are also upregulated [119, 120]. Although diffi cult to use for kinetic studies, such models have been used to test drug permeability.
4.3.3.2 In vivo (and in situ) Methods
For BBB permeability: Th ere are two methods that are used for assess- ing BBB permeability in vivo [121]; these include determinations of
brain:plasma ratio (log BB); and measurement of the permeability surface area product (PS or log PS), from which permeability (P) can be derived provided that the vessel surface area (S) can be estimated. Most pharma- ceutical companies generate log BB data in animals (oft en rat) as part of standard pharmacokinetic profi ling of compounds [122]. Th ese measure- ments are generally made over several hours, with a number of animals required per data point, making the studies costly and labor intensive.
Moreover, a number of factors, including metabolism and binding, aff ect the brain distribution and therefore log BB may not be an accurate mea- sure of BBB permeability. Determining the unidirectional infl ux coeffi cient (Kin) by using the in situ saline-based perfusion method more accurately refl ects the BBB permeation step, eff ectively isolating this “kinetic” ele- ment of drug penetration [121, 122]. Because of the accurate quantitation, the Kin (or PS) measurements are considered as “gold standard” references for other methods [123]. For drugs acting on the basic CNS target sites (membrane receptors, transporters), the critical concentration is the free concentration in brain interstitial fl uid (ISF). Brain/plasma ratio measured especially at longer times and for more lipophilic agents will be aff ected by drug distribution into brain lipids and nonspecifi c binding. Measuring free concentration with a microdialysis probe is possible but technically diffi cult, and there is a particular problem of recovery of more lipophilic agents [124, 125].
For assessment of proposed AD diagnostic and therapeutic approaches—
AD mouse models: Transgenic mice models that mimic a range of Alzheimer’s disease–related pathologies are currently available (summa- rized in Table 4.3). Th ey have been widely used in the preclinical testing of potential therapeutic modalities and have played a pivotal role in the development of immunotherapies for Alzheimer’s disease, which are cur- rently in clinical trials. Th e most common approach to create genetically manipulated mice is to microinject a complementary DNA (cDNA) trans- gene (usually with a pathogenic mutation). However, some transgenic ani- mals have been created with a whole, wild-type human genomic fragment that includes promoter, introns and fl anking sequence (the whole APP gene [126], PS-1 gene [127], or tau gene [128]). Other approaches target the endogenous mouse gene with pathogenic mutations or human gene sequence (the knock-in approach) [129]. Th ere have been problems thus far with the uniformity of results due to the diff erent mouse strains and methods used.
Amyloid-forming APP mice: Th ere have been many attempts to create models of amyloidogenesis, some of which replicate the amyloid pathology rather well. Th e fi rst mouse model with extensive AD-like neuropathology
Table 4.3 Transgenic Mice Models that Mimic Alzheimer’s Disease or Related Pathologies.
Trans/Line Pathology Refs
PD-APP Aβ deposits, neuritic plaques, synaptic loss, astrocytosis and microgliosis
[130]
Tg2576 Hamster Prp Behavioral, biochemical and pathological abnormalities
[131]
TgAPP23 Congophilic senile plaques that are
immunoreactive for hyperphosphorylated tau
[132]
TgCRND-8 Hamster PrP Aβ amyloid deposits, dense-cored plaques and neuritic pathology, early impairment in acquisition and learning reversal
[125]
TgR1.40YAC Aβ deposits, pathology accelerated inhomozygotes, or when bred to mutant PS-1YAC transgenic mice
[126]
PS/APP Fibrillar Aβ deposits in the cerebral cortex and hippocampus, reduced spontaneous alternation performance in a “Y” maze
[133]
PS-1 Elevated levels of the highly amyloidogenic 42- or 43-amino acid peptide Aβ42(43)
[134]
Alz27 Tau present in nerve cell bodies, axons and den- drites was phosphorylated at sites that are hyper- phosphorylated in paired helical fi laments
[135]
JNPL3 Murine PrP Motor and behavioral defi cits, with age- and gene-dose-dependent development of neurofi brillary tangles
[136]
hTau40 Axonal degeneration in brain and spinal cord, axonal dilations with accumulation of neurofi la- ments, mitochondria and vesicles
[137]
hTau Hyperphosphorylated tau accumulating as aggre- gated paired helical fi laments in the cell bodies and dendrites of neurons in a spatiotemporally relevant distribution
[138]
PS1/APP/Tau Development of plaques and tangles; synaptic dysfunction, including long term potentiation (LTP) defi cits, manifesting in an age-related manner (before plaque and tangle pathology)
[139]
was created by Exemplar_Athena Neuroscience in 1995 (the PD_APP line). Th is mouse line developed fi brillar and diff use amyloid pathology in the cortex and hippocampus [140]. A subsequent model (Tg2576) cre- ated by Hsiao and colleagues showed age-related cognitive impairment in addition to Aβ plaques [131]. Surprisingly no model showed overt cell loss [141] and amyloid associated cell loss has only been reported in one (APP- 23), although it was very marginal [134]. It is likely however, that neurode- generation is occurring in these models, as several of the mutant APP mice have now been shown to be cognitively impaired [125, 131, 133].
Presenilin mice: Both wild-type and mutant PS-1 mice have been created to determine the eff ects of PS-1 on APP processing. [128]. Studies on these mice indicated that PS-1 mutations had the eff ect of increasing Aβ 1–42.
Th e level of Aβ 42 generated in these mice was not suffi cient for amyloid to be formed. When mutant PS-1 mice were crossed to mutant APP mice, Aβ aggregation into plaques was greatly accelerated in the mouse line studied, indicating either that Aβ levels are critical for amyloid formation or that these AD genes were interacting synergistically [133]. PS-1 knockout (KO) mice have demonstrated decreased Aβ levels, further emphasizing the APP processing role of PS-1 [142], and PS-1 is now thought to be synonymous with the APP-cleaving enzyme, γ-secretase [143].
ApoE mice: ApoE KO, knock-in or cDNA mice have been created as tools to understand the role of this protein in AD [143]. When ApoE KO mice were crossed with PD_APP mice, the mice had signifi cantly reduced Aβ deposition. When ApoE4 mice were crossed with PD_APP mice, higher levels of amyloid were seen compared with an ApoE2/PD_APP cross, supporting the pathogenic role of ApoE4 in AD [144]. Th e mecha- nism by which ApoE exerts its eff ect is unknown, although some theories have been proposed.
Tau mice: Th e fi rst transgenic tau mouse was created in 1995 [145], but further neurofi brillary pathology was not apparent. Use of pathogenic mutations in tau cDNA transgenes led to the development of mice that form robust neurofi brillary (tangle) pathology of relevance to FTD-17 and AD [136]. Recent developments using a genomic, wild-type tau transgene has led to a mouse model that forms more AD-like tangles [138]. More recently created mouse models have cortical/hippocampal pathology and associated cell loss; these should be more informative for the study of pathogenic tau formation (including the contribution of hyperphosphory- lated or aggregated tau) and neurodegeneration/cell death. Unfortunately tangle formation in these models is not associated with Aβ aggregation, and the tau transgenic mice do not form amyloid plaques. Interestingly, exposing mutant tau mice to elevated Aβ (either through crossing to an
APP-overexpressing mouse or by injecting Aβ into the brain [145]) leads to increased tau pathology, suggesting that there is an interplay between Aβ and tau. Recently, a triple transgenic model (expressing mutant APP, PS-1 and tau) has been created, in which the APP and tau transgenes were co-injected (leading to co-integration) into a PS-1 knock-in line [139].
Progeny express all three transgenes in the same background strain. Th ey develop both plaques and tangles (plaques come before tangles in this model) and have shown synaptic dysfunction, although behavioral studies have not been reported.