香京julia种子在线播放

Inbreeding consists of mating closely related individuals (sisters and brothers) taken from random bred populations for about 20 generations to produce a subpopulation whose members are homozygous, that is, have the same genotype. Inbred mice were initially generated by physiopathologists who identified the advantage of having individuals ruling out genetic variance and showing homogeneous traits to better circumscribe the nature and inheritance of several pathologies. For example, criteria of selection to start the production of inbred lines were “predisposition to develop neoplasia” to determine if cancer was inherited (Little, 1915), or “immunohistocompatibility response” to estimate the best immunological markers for tissue transplantation (Hellstrom, 1963). Incidentally, Bagg (1920) tested various inbred strains of mice in several multiple-choice mazes and found that the learning performance strongly varied between strains, whereas it was remarkably homogeneous within each strain. Later on, Vicari (1929) compared the time spent by DBA/2 and BALBc/ inbred mice, as well as by “Japanese Walzer” and “myencephalic bleb” mutant mice, to run a three-unit maze and observed that maze running times were strain-dependent. Some decades later, two seminal studies focusing on interstrain differences in learning and memory (Dennenberg, 1959; Bovet et al., 1969) have established the bases for behavioral genetics. Clearly, one advantage of inbreeding for neuroscientist interested in the genetic control of learning abilities is that no learning criterion is involved in the selection process. Thus, possible confounders like those evoked in the case of bidirectional selection where, for example, low and high learners in active avoidance might simply those having low and high pain thresholds or being less or more anxious (Río-Ȧlamos et al., 2015), were excluded. Another issue to be considered is that if inbred individuals are like homozygous twins, they all exhibit the phenotype of one single random bred individual. This means that their behavioral and neural traits are not distributed according to a normal curve, so an inbred population is in no way representative of a natural outbred population. As many strains have been accurately characterized, any neuroscientist interested in analyzing a particular behavioral or neural phenotype can select a priori either one strain expressing the behavioral or neural trait of interest, or several strains to be compared for their difference relating to this specific trait. Among the most commonly used strains, pure inbred (C57BL/6J, DBA/2J, BALB/c, FVB/NJ, 129SvEvTac) or mixed backgrounds (B6J/SJLJ) are predominant (Sultana et al., 2019). The DBA/2J mouse strain is often used to contrast the C57BL/6J strain, given that their genotype (Bottomly et al., 2011) and phenotype (Ingram and Corfman, 1980) are opposed in several aspects.

At the dawn of neuroscience, the most popular tasks performed to investigate learning in rodents were those designed by experimental animal behaviorists, which prevalently required to form motor habits or stimulus–response associations, and in which C57BL/6J mice (C57) were identified as poor learners. Specifically, C57 performed worse than BALBc, or DBA/2J (DBA), in the Lashley maze (Oliverio et al., 1972), the active avoidance (Bovet et al., 1969; Oliverio et al., 1972; Weinberger et al., 1992), and in situations of operant or instrumental conditioning (Renzi and Sansone, 1971) in which an elemental stimulus was used to initiate or stop responding. When O'Keefe and Nadel (1978) and Olton et al. (1978) identified the neural basis of spatial cognition in the hippocampal place cells, spatial tasks like the radial arm maze and the water maze were the golden standards to investigate cognitive functions and their alterations in rodents, especially in view of data showing that Alzheimer's disease (AD) patients with temporal lobe neurodegeneration were selectively impaired in episodic/declarative/spatial memory. We started testing C57 and DBA mice in a radial maze (Ammassari-Teule and Caprioli, 1985) and observed that C57 outperformed DBA, thereby reverting their previous status of bad learners. These findings were confirmed in other spatial protocols (Upchurch and Wehner, 1988, 1989; Passino et al., 2002) and extended to tasks which strongly rely on the hippocampus like reactivity to spatial novelty (Thinus-Blanc et al., 1996), contextual fear conditioning (Stiedl et al., 1999; Ammassari-Teule et al., 2000; Restivo et al., 2002), cross-maze place learning (Middei et al., 2004), and, very recently, pattern separation (Dickson and Mittleman, 2022). The C57 predisposition to do well in hippocampus-dependent tasks then prompted several groups to examine the structural and functional properties of their hippocampus. Compared to other strains, C57 were found to show a higher density of mossy fibers in the region inferior to the hippocampus (Barber et al., 1974; Lipp et al., 1988; Schwegler et al., 1988), an increased activity of hippocampal, but not cortical, protein kinase C (Wehner et al., 1990), and long-term stronger hippocampal potentiation (Matsuyama et al., 1997; Nguyen et al., 2000; Gerlai, 2001a; Jones et al., 2001).

Considering strain-specific levels of performance, it became rapidly evident that a majority of tasks in which C57 performed poorly were those in which DBA performed well, and vice versa. Beyond their aforementioned superiority in the Lashley maze and active avoidance, DBA mice were found to score better than C57 in cue-based fear conditioning (Paylor et al., 1994; Ammassari-Teule et al., 2000) and ethanol-induced conditioned place preference (Cunningham and Shields, 2018). On the one hand, these findings pointed out the remarkable ability of DBA to detect relevant elemental sensory stimuli to rapidly form stimulus–response associations in pavlovian or instrumental conditioning paradigms, or to implement egocentric orientation in spatial tasks. On the other hand, their inability to form configural representations, either of the aversive context in conditioning tasks or of distal environmental cues for allocentric orientation in spatial tasks, was interpreted as the consequence of the poorly functional morphological, biochemical, and plastic properties of their hippocampus, which led to consider them as a natural model of hippocampal dysfunction (Paylor et al., 1993). Supporting this view, DBA mice behave randomly in the hippocampus-dependent simultaneous olfactory discrimination task and do not form training-induced dendritic spines in the hippocampus (Restivo et al., 2006).

The notion of memory systems arises from observations initially carried out in human subjects (Cohen and Squire, 1980) and later in rodents (Packard et al., 1989; Packard and White, 1991; McDonald and White, 1993) that different types of memory are governed by dissociable brain substrates. For example, episodic, declarative, spatial, or context-based memory are supported by temporal lobe regions, among which the hippocampus plays a central role, whereas stimulus–response associations and procedural memory, including motor habits, are overall controlled by the striatum. The independence of memory systems was demonstrated by data showing that disrupting the neural support of one system leaves unaltered the operations supported by the other system or even improves the preserved system by suppressing conflictual responding (McDonald and White, 1995).

In a majority of individuals, memory systems can be activated separately, concurrently, or sequentially, depending on the situation to copy with. For example, in the plus maze task (Packard and McGaugh, 1996), rodents were first trained to turn left to find a food reward in the unique baited arm and then exposed to probe trials in which they were released from the opposite starting arm. In those trials, they could either reproduce the motor response reinforced during training (turning left) and do not go to the baited arm or invert it (turning right) and go to the baited arm. Interestingly, the rats were found to turn right after a short training duration (1 week), consistent with hippocampus-based place learning, but to turn left after long training duration (3 weeks), consistent with striatum-based response learning. Interestingly, inactivation of the hippocampus after short training made the rats show striatum-based motor learning, whereas inactivation of the dorsolateral striatum after long training made them show hippocampal-dependent place learning. The point is that when C57 and DBA were trained in the same plus maze task, C57 showed place learning and predominant hippocampal activation after both short and long training, whereas DBA never relied on a particular system even they prevalently activated the dorsolateral striatum (Passino et al., 2002). Indeed, the consequence of the C57 inability to disengage the hippocampus is that inactivation of this region at any probe trial indeed disrupts place learning but does not promote response learning. Similarly, the consequence of the DBA/2 inability to rely on a particular system is that the inactivation of any region at any probe trial does not promote neither place nor response learning (Middei et al., 2004). Thus, contrary to what happens in rats, disrupting the neural substrate of one memory system but does not promote the utilization of another system in these genotypes. Indeed, the C57 propensity to engage the hippocampus in any situation they face is also observed in fear conditioning (FC) paradigms. Specifically, studies dissecting the neural bases of FC in outbred populations of rats have identified (i) tone fear conditioning (TFC) as an elemental associative learning system involving the basolateral amygdala (BLA) but not the dorsal hippocampus (Phillips and LeDoux, 1992; Paré et al., 2004) and (ii) contextual fear conditioning (CFC) as a configural learning system involving by both regions (Selden et al., 1991; Phillips and LeDoux, 1992). Different from that observed in these populations, C57 mice were found to concurrently activate the BLA and the dorsal hippocampus in both CTC and CFC, although they showed considerably less freezing in TFC than in CFC (Pignataro et al., 2013). In line with the view that recruiting the hippocampus in TFC is an obstacle to implement elemental stimulus–response associations, lesions to the dorsal hippocampus were found to enhance C57 TFC performance (Ammassari-Teule et al., 2002).

The aforementioned experiments show that C57 predominantly form configural environmental representations in which elemental stimuli are embedded in and need to be disentangled to predict reinforcement and guide behavior. Thus, any manipulation of experimental factors that facilitates disentangling is expected to enhance cue-based performance in this mouse strain. This possibility was demonstrated in a study in which C57 and seven other mouse strains including DBA were trained to press a lever upon presentation of an elemental stimulus (tone or light) to avoid delivery of an electric footshock. DBA showed superior avoidance performance when the tone or the light was of short duration. However, a gradual increase in the stimulus duration was found to progressively abolish interstrain differences, thereby suggesting that C57 benefited for longer cue presentation to disentangle them from the context (Renzi and Sansone, 1971). Interestingly, Cunningham and Shields (2018) explored more recently the possibility that the most robust ethanol-induced conditioned place preference (CCP) shown by DBA compared to C57 might depend on strain differences in sensitivity to contextual cues. They found that compared to single cueing, multiple cueing increased CCP in both strains but that this effect disappeared more rapidly in DBA and was not sufficient to elevate the CCP performance of C57 to the level of DBA. They therefore concluded that CPP differences were due to a genotype-specific sensitivity to ethanol reward. Nevertheless, the fact that DBA mice outperform in amygdala-dependent conditioned taste aversion, that is, a task based on the association of a single gustatory stimulus with illness (Dudek and Fuller, 1978; Risinger and Cunningham, 1995), suggests a more global interpretation that DBA perform better due to the stronger BLA modulation of striatal-based elemental cue processing (Desmedt et al., 1998; Goode et al., 2016). Other examples which show that interstrain differences in fear conditioning vary or persist depending on whether freezing is recorded shortly or long after training (Nie and Abel, 2001; Balogh and Wehner, 2003) indicate that if the predominance of a memory system determines the nature of information each strain preferentially relies on, the time necessary to consolidate this information is a much important variable in the modulation of the behavioral responses. Indeed, manipulations like environmental enrichment (EE) or physical exercise modify cognitive abilities, attention, and anxiety in inbred mice but paradoxically exert improving or disrupting effects depending on the duration of and the age mice were exposed to these manipulations (Singhal et al., 2019). Among the principally observed strain-specific effects, EE was found to increase attention in C57 (van de Weerd et al., 2004), to either decrease (Chapillon et al., 1999) or increase (van de Weerd et al., 2004) anxiety in BALB/c, to reduce reactivity to novelty in both C57 and DBA (Dickson and Mittleman, 2021), and to accentuate C57 vs 129S6/SvEV differences in locomotor activity, anxiety, and social interactions (Abramov et al., 2008). Thus, no univocal, strain-independent beneficial effect of EE of behavior was observed.

One of the first transgenic murine models of Alzheimer's disease is the Tg2576 mouse developed by Hsiao et al. (1996) which overexpresses human APP (isoform 695) containing the double mutation K670N, M671L (Swedish mutation) under the control of the hamster prion protein promoter. This hemizygous mutation was originally introduced in a C57(B6) × SJL F1 hybrid background and stabilized by repeatedly backcrossing mutant mice with B6 × SJL F1 hybrids. The main issue addressed at that time was confounding effects due to the insertion of the mutated gene and those due to the transgene per se. It was therefore proposed to backcross the transgenic lines to one or even more inbred strains for at least three generations before performing phenotypic characterization in F1 hybrids. This strategy, however, revealed to be inadequate as repeated backcrossing to inbred lines produced non-specific performance impairments, reduced fertility, and, in some cases (FVB/N or B6), was found to be lethal. This prompted Lassalle et al. (2008) to insert the HuAPP695-SWE transgene in three different backgrounds (homogeneous: C57; heterogeneous: CBAJ; and hybrid: B6J/SJLJ) and to perform phenotyping in F1 generations after only one generation backcrossing. Comparisons included evaluation of anxiety in the elevated plus maze, spatial learning in the water maze, and fear conditioning. The results showed that Tg (+) C57 mice were globally more active and less anxious than their Tg (-) counterparts, as well as than Tg (+) and Tg (-) in other backgrounds. The calculation of a spatial index in the water maze revealed that even though this index was rather comparable between Tg (-) B6/SJL, C57, and CBA, the genetic background significantly modulated the expression of the transgene, with the lower spatial index being found when the mutation was expressed in the C57 background. Fear conditioning data did not provide evidence of differences in CFC performance between the three Tg (+) and Tg (-) backgrounds possibly because the experiments were carried out in 17-month-old female mice and, hence, recruited additive effects of sex and age to those of background and mutation. Nevertheless, this study provided the first demonstration that the behavioral characteristic of the recipient mouse was defining the degree of mutation-induced cognitive impairment. After two years, Rustay et al. (2010) compared the effect of the Swe mutation in the B6/SJL and 129 backgrounds and reported more deleterious effects at late ages in the 129 backgrounds for parameters that were not properly cognitive (locomotor activity, spontaneous alternation) and for which Tg (-) 129 were scoring lower than Tg (-) B6/SJL. Interestingly, models in which three (APP, PS1, and tau) or five (two APP and three PS1) familiar AD mutations were concurrently inserted in mouse genomes were predominantly developed in a C57 background (Sterniczuk et al., 2010; Forner et al., 2021; Sil et al., 2022) or hybrid backgrounds with a B6 component. Confirming the inadequacy of the DBA background for AD mutations, insertion of the APPswe and PSEN1de9 mutated genes in DBA exacerbated lethal seizures and even lessened amyloid deposition (Jackson et al., 2015).

These models were generated using the App gene knock-in strategy to overproduce pathogenic Aβ without overexpressing APP with the objective to avoid artifacts due to APP overexpression per se. Murine Aβ sequences (Swedish, Beyreuther/Iberian, artic) were humanized by changing amino acids that differ between mice and humans and then introduced separately or concurrently to generate APP^NL, APP^NL−F, or APP^NL−G−F mice expressing Wt human Aβ under control of the mouse App locus (Nilsson et al., 2014; Saito et al., 2014). The point is that if humanized Aβ does increase the face validity of these models as far as amyloidosis is concerned, the consequences of App-KI manipulation on neural and behavioral parameters do not differ much from those observed in the first-generation transgenic models. App-KI mice show the same increased glutamate release probability and intense astrocytosis/gliosis around the Aβ plaques, discordant results regarding hippocampal LTP (decreased or intact), and late memory impairments (Nilsson et al., 2014; Baglietto-Vargas et al., 2021; Benitez et al., 2021). Furthermore, the fact that robust amyloidosis and its metabolic consequences took at least 18 months to emerge has led to the development of third-generation models, that is, double KI mutants obtained by crossing APP^NL−G−F mice with mice bearing PS1-KI (Sato et al., 2021) to accelerate the detection of mutation effects. In these studies, however, (i) the genetic background (until now C57) is never mentioned in the method section of articles, suggesting that it is per se irrelevant if the mutant mouse is viable; (ii) the focus is predominantly placed on the pathogenic inflammatory/metabolic cell alterations engendered by multiple Aβ species and on the possibility to rescue them by rectifying the mutated genomic sequences; (iii) mutation effects are investigated separately at the neural (Jun et al., 2020) or behavioral (Sakakibara et al., 2019; Sutoko et al., 2021) levels and therefore do not inform on the status of neural circuits when mice face cognitive challenges, that is, when these circuits actually come into play.