Human life has been extended in the last decades thanks to modern medicine and general improvement in human life quality. However, the increase in our life expectancy has resulted in a new major challenge for human health: neurodegenerative diseases (ND) in the elderly. Diseases involving the loss of nerve cells are currently a main problem in the aging population worldwide, which will go worse since a further rise in human longevity is expected, especially in developed countries. For instance, Alzheimer’s disease (AD) is believed to affect around 40 million people worldwide and its prevalence is expected to reach 135 million people by 2050 [1]. In addition, NDs truly disable patients causing great suffering to them, their relatives, and their caretakers affecting seriously not only the physical but also the mental health of all people involved.

Despite the huge amounts of funds, human effort, and the years passed, NDs do not have an effective treatment. Few drugs for treating symptoms at the beginning of the disease course have been approved. For example, the L-DOPA-based treatment of Parkinson’s disease (PD) therapies substitutes dopamine precursors due to the death of dopaminergic neurons, but its efficiency is temporal and limited [2]. The lack of success in the search for ND’s treatment is partially influenced by the incubation time of NDs. The molecular events that trigger neuron death start years (even decades) before symptom appearance. Central nervous system (CNS) plasticity, which allows neuronal web reconnection and reconfiguration when cells die to maintain the system function, disguises the disease. Thus, NDs might be already irreversible at symptom onset. A possible option to improve the chances of successful therapies will be to direct them to the preclinical phase of these diseases, although treatment after the onset of clinical symptoms is the most common scenario in NDs. Such situations point out the need for models for NDs, both to understand the basis of the underlying pathology and to test potential drugs.

Most common NDs share in the center of their pathology the aggregation of a misfolded protein, as well as an unknown and complicated etiology in the majority of the patients. The first animal models for the study of NDs were knock-out (KO) mice, aimed to recapitulate the loss-of-function of the genes coding the proteins that get aggregated. Mice’s advantages concerning other animals include their small size, short lifespan and generation time, easy manipulation, and availability of well-established molecular biology techniques that allow genetic manipulation. In addition, many tests aimed to check several aspects of human behavior have been established and normalized in rodents. Thus, transgenic mice have been implemented as the preferred animal model for the study of NDs. The later development of homologous recombination-mediated genetic engineering further facilitated transgenic mice creation [3] by targeting particular cell populations and allowing spatiotemporal expression of the transgenes. The existence of genetically inherited forms of certain NDs also facilitated the generation of transgenic mice models for disease study and therapy assessment. However, a general problem with transgenic mice models of NDs is that they often do not completely recapitulate the whole human phenotype, making the available models just partially useful. The very recent development of the simple and powerful CRISPR-Cas9 for gene editing is called to revolutionize and improve transgenic animal creation [4]. New models for NDs are already being produced by this technique and their utility shall therefore be compared to that of the currently existing ones in the future.

This review aims to compile the most relevant transgenic mice models for several NDs, their success and pitfalls, and discuss the currently unsolved problems in NDs modeling.

Synucleinopathies are a group of NDs which are molecularly characterized by the CNS accumulation of $\alpha$-synuclein ($\alpha$-syn) intracytoplasmic inclusions. Synucleinopathies include Parkinson’s disease (PD), dementia with Lewy Bodies (DLB), and multiple system atrophy (MSA). Although $\alpha$-syn is involved in all cases, the most affected brain areas and specific cell types differ between diseases, influencing the clinical signs associated with each pathology.

Tauopathies are NDs characterized by the pathological accumulation of microtubule-associated protein tau (MAPT) in neurofibrillary tangles (NFTs) and paired helical filaments (PHFs) that cause the death of affected neurons and glial cells. Tauopathies include frontotemporal dementia (FTD), progressive supranuclear palsy (PSP), corticobasal degeneration syndrome (CBS), chronic traumatic encephalopathy (CTE), Pick’s disease, and sporadic forms of AD. Some of them are inherited by mutations in the MAPT gene [79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93]. Given that AD has other molecular hallmarks and is the most common form of ND, the models for its study will address separately, being this part focused on other non-Alzheimer tauopathies.

MAPT is a neuronal protein involved in the regulation of microtubule stability, microtubule dynamics, and axonal transport [94, 95]. Many MAPT mutations cause the inheritance of genetic forms of tauopathies supposedly by increasing tau’s propensity for aggregation and toxicity [96]. Nevertheless, as reported with other NDs, the majority of tauopathies are sporadic, and variable clinical and pathological presentations have been described in patients [79]. The mechanisms of tau aggregation as well as the disruption of molecular pathways that ultimately cause cell death are still poorly understood. Evidence indicates that native tau is soluble, contains charged and hydrophilic residues, and shows little tendency for aggregation [97, 98, 99]. Thus, to aggregate, tau must undergo conformational and post-translational modifications like phosphorylation [100, 101, 102, 103]. It has been reported that phosphorylation at certain residues like Ser422 is rarely detected in healthy adults [104] but is present in AD patients and related to loss of cholinergic neurons and cognitive impairment [105, 106].

AD is the most prevalent form of dementia and contributes to 60–70% of all dementia cases [137]. Patients show progressive symptoms that firstly include deficits in short-term memory that led to later cognitive impairment and neuropsychiatric symptoms that severely disable patients to the extent of being unable normal life activities [138].

AD is mainly characterized by the presence in diseased brains of amyloid plaques composed of amyloid-$\beta$ (A$\beta$) peptides derived from the processing of the amyloid precursor protein (APP) and NFTs composed of hyper-phosphorylated tau [139]. Such hallmarks were first reported in 1906 by German doctor Alois Alzheimer [140]. Apart from amyloid plaques and NFTs, AD brains are further characterized by synaptic and neuronal loss and reactive astrogliosis and microgliosis [141].

The discovery that amyloid plaques were composed of A$\beta$ peptides pointed to A$\beta$ as being the potential causal factor for AD. A$\beta$ peptides are formed during the cleaving of the transmembrane APP protein by proteases BACE1 [142] and gamma-secretase complex, releasing peptides A$\beta$40 and A$\beta$42 to the extracellular space [143]. These peptides are extremely hydrophobic and prone to aggregate, forming insoluble fibrils which turn into plaques. Initially, A$\beta$ deposits were thought to be neurotoxic and promote the formation of NFTs by a similar cascade of polymerization of phosphorylated tau molecules into progressively bigger and insoluble fibrils that end up forming NFTs [144, 145]. However, cumulative evidence showed that AD has a more complex and multifactorial pathogenesis, in which cognitive decline seems to be more linked to NFTs accumulation than to A$\beta$ deposition [146]. In addition, it seems that soluble A$\beta$ and tau oligomers that precede amyloid plaque and NFTs are the causative agents of synaptic damage and neuronal death [147, 148, 149].

More than 90% of AD cases are late-onset and of unknown etiology, which are known as sporadic AD cases (SAD) [150]. By contrast, the rest of the AD cases are caused by dominant autosomal inheritance of mutations in genes related to A$\beta$ generation like APP, PSEN1, and PSEN2. These last two genes codify proteins acting in the gamma-secretase complex. The familial forms of AD (FAD) are early-onset and high penetrant. Transgenic mouse modeling FAD aimed to gain insights into molecular mechanisms that will later be applied to SAD cases. However, there is no single mouse model that completely recapitulates all pathological and behavioral phenotypes of AD. Indeed, wild-type rodents do not develop A$\beta$ plaques or NFTs in normal conditions possibly because of their lifespan, which may not allow a long pre-symptomatic phase as happens in humans [151]. The most relevant transgenic mouse models generated to model AD will be discussed and their main advantages and caveats analyzed (Table 4, Ref. [144, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174]). It is important to note that before FAD modeling, models aimed to mimic the disruption in the cholinergic system in rodents by mechanical, electrical, or chemical lesions [175]. More detailed information about the AD transgenic mouse models available can be found at https://www.alzforum.org/.

Huntington’s disease (HD) is an inherited ND mainly characterized by motor, psychiatric and cognitive impairments [179]. Unstable CAG repeat expansion in the huntingtin gene (HTT) is the cause of the disease, which codifies for a polyglutamine domain in the N-terminal part of the protein. HD is inherited in an autosomal dominant pattern. Healthy HTT alleles contain from 6 to 35 repeats of the CAG triplet in the (CAG)${}_{\text{n}}$CAACAG region of the HTT gene. A higher risk of disease development has been established from 36 to 39 repeats, while 40 or more repeats cause fully penetrant HD. The age of onset of motor symptoms negatively correlates with the CAG repeat number, the more repeats the sooner the symptoms appear. The main pathological hallmark of HD is the extended neuronal loss in the striatum and cerebral cortex areas as well as extensive brain atrophy [180].

Once the HTT gene was discovered in 1993 [181] a great variety of HD animal models have been produced including transgenic mice. These models have been of utility for unraveling the pathological mechanisms of the disease and for the evaluation of therapeutic interventions in preclinical studies (Table 5, Ref. [182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195]).

HTT gene ablation in mice results in early embryonic lethality. Thus, the normal function of huntingtin in adult mice relies on the Cre/loxP site-specific recombination strategy to produce conditional ablation, finally resulting in a progressive degenerative neuronal phenotype [182]. In contrast with the experience of other NDs, HTT transgenic mice do recapitulate human HD phenotype including brain HTT aggregates and inclusions, motor and cognitive impairment, synaptic plasticity deficits, electrophysiological alterations, and neuron loss (Table 5). Motor symptoms include tremors, hypokinesia, and lack of coordination. Interestingly, transgenic mice expressing only the mutant N-terminal part of the HTT gene codifying only for exon 1 develop the disease earlier and with more pronounced symptoms than transgenic mice expressing several copies of the full mutant HTT gene or knock-in mice [183, 184, 185, 186]. Despite this disadvantage of full-length HTT models, they are more suitable for the study of certain HTT therapeutic interventions like HTT lowering strategies [187, 188, 189, 190].

Transmissible spongiform encephalopathies (TSEs) or prion diseases are fatal neurodegenerative diseases that affect humans and other mammal species, some of them included in the human food chain. Humans are affected by Creutzfeldt-Jakob disease (CJD), kuru, Gerstmann-Sträussler-Scheinker syndrome (GSS), and familial fatal insomnia (FFI). TSEs produce long incubation times and their main symptoms are neurological behavior abnormalities, but motor dysfunction, cognitive impairment, and cerebral ataxia can also appear. In individuals affected by a TSE the normal form of the prion protein, also known as cellular form or PrP${}^{\text{C}}$, is converted into a disease-associated form known as PrP${}^{\text{Sc}}$ [196]. PrP${}^{\text{C}}$ transformation into PrP${}^{\text{Sc}}$ causes a change in the protein tridimensional structure characterized by an increase in the $\beta$-sheet content [197]. While PrP${}^{\text{C}}$ is monomeric, soluble in nonionic detergents, and sensitive to protease action, PrP${}^{\text{Sc}}$ tends to aggregate, is not soluble in non-ionic detergents, and is partially resistant to proteases [198].

One of the main differences between prion diseases with the rest of NDs is that they have a wider range of different etiologies. They can be infectious, iatrogenic, sporadic, or genetic.

Infectious and iatrogenic prion diseases are caused by the entry of an external PrP${}^{\text{Sc}}$ source that starts transforming host PrP${}^{\text{C}}$ into new, ascent PrP${}^{\text{Sc}}$. Thus, variant CJD (vCJD) is an infectious TSE caused by the consumption of Bovine Spongiform Encephalopathy (BSE)-contaminated meat products while iatrogenic CJD, for example, may be caused by a cornea transplant from a CJD-infected donor [199]. In the particular case of infectious TSEs, certain prion agents can be transmitted from one species to a different one. This is known as interspecies prion transmission and is affected by the homology between the primary sequences of inoculum PrP${}^{\text{Sc}}$ and host PrP${}^{\text{C}}$ [200]. Mismatches between sequences also influence TSE progression [200]. Thus, the study of prion transmission is of crucial importance from the point of view of human health and food safety. For that purpose, transgenic mouse lines expressing the prion protein from different species of interest would provide useful models for the study of how certain TSEs may jump from one species to another.

Genetic TSEs are due to mutations in the PRNP gene. In humans, approximately 10–15% of TSEs are genetic [201]. PRNP mutations are thought to spontaneously promote the misfolding of PrP${}^{\text{C}}$ into PrP${}^{\text{Sc}}$ and/or stabilize the PrP${}^{\text{Sc}}$ molecules once formed [201]. There are three different human genetic prion diseases based on their clinical and pathological features: familial CJD (fCJD), GSS, and FFI [202]. fCJD is rapidly progressive dementia in which patients present cerebral spongiform degeneration. GSS is a slowly progressive disease in which patients show ataxia, few spongiform degeneration, and abundant PrP${}^{\text{Sc}}$ amyloid plaques. Finally, the principal FFI symptom is progressive insomnia that derives in hallucinations and dementia, as well as high spongiosis in the thalamus.

The prion protein amino acid sequence is well conserved throughout evolution in mammal species [202]. This fact suggests that the prion protein in its cellular form has an important function. However, KO mice for the PRNP gene were fully developed till the end of their lifespan without any obvious detriment phenotypes [203, 204]. Different functions were inferred for the prion protein, but there is controversy on which of these inferred functions are real and the role that the different genetic backgrounds of the animal models may have played in the observed phenotypes [205] like synaptic and electrophysiological deficits, hematopoietic stem cell renewal, circadian rhythm regulation, processing of sensory information in the olfactory system or neural stem cell proliferation in adult neurogenesis [203, 204]. However, the most important and validated effect of PRNP depletion is resistance to prion infection [203, 204, 206].

With a focus on human health, transgenic mice expressing human PrP have been developed to study genetic, sporadic, and infectious prion diseases (Table 6, Ref. [203, 204, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225]). Different human PrP transgenic lines have been produced and found susceptible to kuru and sCJD prions [208, 209, 210, 211, 226, 227]. Human PrP transgenic models are also useful for the study of human susceptibility to animal prion strains. To date, the only recognized zoonotic prion disease is vCJD which first appeared in the United Kingdom in 1996 [228] and was soon proved to be caused by BSE-contaminated meat consumption [229]. The M129V polymorphism of human PrP is of special importance for human prion susceptibility. All vCJD definitely diagnosed cases are homozygous for the methionine allele, except for one recently found heterozygous case [230]. Transgenic mouse lines overexpressing human PrP harboring the M129 allele (Table 6) are successfully infected with BSE prions with a high resistance as reflected in the long survival times and partial attack rates that animals present [211, 212, 227, 231]. These models mimic perfectly the real situation since virtually all of the UK population was exposed to BSE-contaminated meat, although just 232 cases of vCJD have been reported [232]. Transgenic mice overexpressing human PrP harboring the heterozygous M129V and homozygous V129 alleles (Table 2) have not been infected with BSE but vCJD transmission is successful, thus pointing to the risk that human to human secondary vCJD infection would pose for the population [233]. Sheep-passaged BSE can easier infect M129 human PrP transgenic mouse lines more when compared to cattle BSE [212, 234]. For other prions not proved to be zoonotic, animal bioassays using transgenic models showed certain zoonotic potential for several agents like cattle atypical BSE prions and small ruminant classical scrapie [209, 235, 236]. More studies are needed for other emerging strains like small ruminant atypical scrapie and cervid chronic wasting disease prions [237, 238]. In fact, chronic wasting disease prions have proved to have certain zoonotic potential [239].

Both for the study of the onset, progression, and molecular mechanisms involved in these diseases as well as for the test of possible treatments, transgenic mouse models of genetic TSEs would be of great usefulness. Something important to note is that practically all attempts to produce transgenic mouse models for human genetic TSEs using the human PrP sequence had been unsuccessful, except for one recent model for GSS [224]. The rest of the successful transgenic mouse models for genetic TSEs have been made in the mouse, bank vole, or cattle PrP sequences or chimeric mouse/human PrP molecules.

Despite the huge efforts done by the scientific community, NDs do not have any effective treatment these days. Numerous models have been generated based on the assumption that modeling the genetic “simpler” versions of NDs will provide useful insights to be later applied to the more common sporadic NDs. Unfortunately, in many cases, the generated models do not even fully reproduce the human phenotypes. Usually, the pathological and behavioral phenotypes displayed by the mice are milder, especially regarding behavioral alterations and neuron loss even though they were created using the same causative mutations (reviewed by [240]). This suggests the possible existence of unknown compensatory mechanisms in mice that may explain late-onset disorders and provide targets for novel strategies designed to extend neuronal function and survival [241]. In addition, it is also possible that this genetic “simpler” version of NDs is not only caused by mutations. In this sense, environmental factors might be the key to producing more reliable models. For instance, a high fat diet suministered to a transgenic mouse model of AD generated early prediabetic hyperinsulinemia that exacerbate AD pathology [242]. Other interesting option to generate models might be to produce transgenic mice with enhanced neuroprotective abilities. An example of such an approach is the overexpression of TREM2 (gene encoding receptor expressed on myeloid cells 2) in Tau transgenic mice, who showed that higher levels of TREM2 prevented neuronal loss and attenuated Tau pathology [243].

The partially manifested phenotypes are usually “cured” by candidate drugs in preclinical trials, but this success does not translate to the following human clinical trials. For instance, the Tg2576 mice model for AD has been cured or their health has been improved by treatments evaluated in preclinical trials more than 300 times [244]. Therefore, projects aimed to “cure mice” by simply reducing the misfolded protein that is accumulating seem not to be working fine [245]. Indeed, it seems that misfolded protein accumulation in big aggregates is a protective mechanism given that shorter oligomers seem to be the most toxic species (reviewed in [148]).

Researchers from different NDs agree that although having a distinct molecular basis (in terms of the protein that gets misfolded and accumulated), behavioral phenotypes tend to overlap and the same happens at the molecular level regarding the pathways that led to cell death. Thus, a holistic and comprehensive approach aimed to unravel the neural death/survival circuits and their implication in different NDs may result in more successful candidates for the treatment of NDs. General pathways such as protein homeostasis, mitochondrial impairment, inflammation, oxidative stress, and cell death routes like apoptosis and autophagy are common to several NDs [246, 247]. Along the same line, the mentioned above mice modeling also environmental factors, with enhanced neuron survival as well as mice aimed to model all these general pathways will provide therapeutic targets not necessarily restricted to one ND. These newer models will probably be generated using the CRISPR-Cas9 system [4]. CRISPR systems are adaptable immune mechanisms used by bacteria to protect themselves from foreign nucleic acids. Their combination with the Cas9 nuclease and a guide RNA allows directing Cas9 to a specific target DNA site using standard RNA-DNA complementarity base pairing rules (reviewed in [4]). This results in facile, rapid, and efficient modification of endogenous genes in a wide variety of cell types and novel organisms that have traditionally been difficult to manipulate genetically. In addition, the CRISPR-Cas9 system avoids random integration in the host DNA typically obtained in classic transgenesis but it may cause off-target edition.

Despite the lack of success in generating treatment options, transgenic mouse models have undoubtedly extended our understanding of NDs in the past years. Those models that at least show the progressive nature of the disease may serve to decipher the molecular mechanisms that end up producing neurodegeneration, thus generating potential options for early diagnosis and treatment before the neurons are irrevocably lost. In this way, combining the information provided by the “genetic” models with that provided by future “pathway-focused” ones may provide opportunities to create new strategies for drug development. Even little progress may have a direct impact on the aged population healthcare given that the prevalence of neurodegenerative diseases is rapidly growing.

All authors (AMM, SC, NFB, JCE, and JMT) have contributed to the conception and design of the manuscript. AMM has been involved in drafting the manuscript and all authors (AMM, SC, NFB, JCE, and JMT) have been involved in revising it critically for important intellectual content. All authors (AMM, SC, NFB, JCE, and JMT) give final approval of the version to be published. All authors (AMM, SC, NFB, JCE, and JMT) have participated sufficiently in the work to take public responsibility for appropriate portions of the content and agreed to be accountable for all aspects of the work in ensuring that questions related to its accuracy or integrity.

Not applicable.

Not applicable.

This research received no external funding.

The authors declare no conflict of interest.