Perspectives

The experimental studies of Part II in this dissertation investigated the functions of the aging canine brain mainly from an ethologist’s perspective, but several investigations are rooted in neuroscience and genetics. The Senior Family Dog Project at the Department of Ethology Eötvös Loránd University (Fgure 37) addresses economical, societal, mental health, animal welfare, and educational aspects of the canine aging.

The recently finished, ongoing and future projects are summarized briefly below.

20.1 Cross-sectional and longitudinal behavioural investigations

Behavioural studies mainly looked at the effect of age-related changes in general, by comparing young adults to healthy aged dogs. These studies provided contradictory results in connection with activity and spatial orientation (Mongillo, Pitteri, et al., 2013;

Rosado et al., 2012; Vas et al., 2007). However, the performance of an old individual in a cognitive test depends on the previous maximum level of the skill and age-related change too, which means that only longitudinal studies can separate the two effects. In

the framework of the Senior Family Dog Project a specifically chosen dog population is followed over a 5 year period since 2016. We perform tests to measure learning and memory, inter-specific communication, activity levels, motor skills and impulsivity, attachment, and sensory performance (to check dogs general vision, hearing and olfactory abilities, labeled as “Cognitive Battery” Piotti et al, in prep, Figure 35). We use a psychometric approach to measure cognitive phenotypes and look for associations with age and other demographic variables using multivariate statistical methods. We re-test the dogs three months after their first test for short-term comparisons, then further test sessions are conducted one year apart for long-term analyses.

With the Cognitve Battery we test whether a three month-long physical and/or cognitive intervention is effective when applied to family dogs aged above 8 years. We are interested to observe whether transfer effects (i.e. improved performance on related but untrained tasks or improvement in tests measuring abilities which were not trained directly) took place in the sample. The candidates are pre-screened before enrollment to flag individuals which suffer from sensory or physical impairments. Baseline and post treatment tests have been completed in three cohorts. (Szabó et al. in prep).

We also seek age-related changes in vocalisation as a potential biomarker for cognitive decline (Marx et al, in prep).

In collaboration with the Clever Hans Lab in Vienna we analyse personality trait-age associations in a personality test battery test, and investigate the re-testing of dogs 4 years later. We confirmed the long-term consistency of the personality traits measured, found both linear and quadratic associations between these traits and age across the age range of 0.5-15 years, and also identified three individuals who showed behavioural markers of cognitive decline (Turcsán et al, in prep). Behaviours related to sociability and frustration were stable over ca. 4 years in Border collie dogs. Activity decreased over repeated testing, mostly due to systematic age effect. Novelty seeking also decreased, while Problem orientation increased from Test1 to 2, both partly due to age effect and partly due to random effect (e.g. experience with the task which led to lower interest in the ‘novel’ object, but higher success in solving a familiar problem).

This is one of the first studies that shows behavioural consistency over such a long time in dogs.

Previous studies have documented the benefits of physical activity and cognitive enrichment on the performance of dogs in cognitive tasks. We described in detail a touchscreen apparatus, software and training methods that we have used to facilitate dog computer interaction (DCI). We compared dogs of different ages in their performance of learning different tasks on the touchscreen, and found that young dogs learn the tasks twice as fast as old dogs. However, feedback from the owners suggested that the learning experience is enjoyable for both the owner and the dog. DCI has the potential to improve the welfare of older dogs in particular through cognitive enrichment (Wallis et al, 2017).

We have assigned unfamiliar, nonaggressive dogs randomly to three types of dyads defined by sex, and observed their unrestrained, spontaneous behaviours in an unfamiliar dog park. Sex, dyad composition, neuter status, and age influence different aspects of the interactions in dyads. The time companion dogs spent in proximity to

each other and number of approaches decreased with age. Indicating decreased intraspecific sociality and activity in older dogs (Iotchev et al., 2019).

Figure 35. Collage about the subtests of the Cognitive Battery.

20.2 Neural correlates of non-pathological aging: (f)MRI- EEG studies, and brain atlas

In humans, disturbances in memory and executive function during senescence (Beth Adams, Chan, Callahan, & Milgram, 2000; Cotman & Head, 2008) are accompanied by specific neural changes, like ventricle volume growth due to cerebral atrophy, the reduction in neuronal cell formation and the accumulation of beta-amyloid plaques (Head, 2011). Several studies have been carried out to reveal more specific alterations at the neural level that could be related to mental decline in dogs. A detailed descriptive study (Borras, Ferrer, and Pumarola 1999) found various age related changes in the dogs’ brain, involving a wide variety of tissue types in the central nervous system (e.g.

neurons, glia cells, vascular endothelial cells), such as retraction of the cerebral gyri, cerebral haemorrhages and lipofuscin accumulation, but they found no age effect with

regard to cerebrovascular amyloidosis. Only laboratory dogs (beagle breed) were involved in these analyses so far (eg. Head, 2013). However, the laboratory beagle populations exhibit significant behavioural differences affecting socio-cognitive functions compared to the family beagle dogs (Turcsán et al. in prep.), so the results have only limited applicability in dogs and humans living in various conditions.

We aim to identify neural markers that are associated with cognitive aging using non-invasive methodologies (fMRI, EEG) on family dogs. First, functional magnetic resonance imaging (fMRI) studies are applied to dogs that have been trained to remain lying motionless for 6-8-minute-long intervals (a globally unique sample of dogs, Figure 36). Second, the polysomnography method provides spectral data (EEG) from the dogs, which is directly comparable with that of humans. In parallel, we also collect behavioural data, including cognitive and memory performance, personality ratings, as well as sensory data for each dog. We expect to find that brain-related changes correspond to cognitive performance.

Figure 36. Dogs participating in fMRI tests.

We successfully set up the methodological background to study rs-fMRI networks of aging dogs, describing multiple human analogous network. In the first study, we collected resting state fMRI data from 22 dogs (Szabó et al., 2019). Based on the results of the first study, and the experience gained from it, we continued data collection for the second study with an improved protocol. Data collection for the second study is nearly finished. We will analyse the age-related changes in connectivity, to describe in which regions the reduction of connectivity is most pronounced.

We used fMRI in order to measure brain activity while processing auditory and visual stimuli. We analysed multilevel fMRI adaptation which revealed a human-analogue lexical processing hierarchy in the dog brain (Gábor et al., 2019).

Voxel-based morphometry of the aging to the changes in the size of the lateral cerebral ventricles. The volumetry we performed revealed an interesting fact, namely that despite the sometimes remarkable (and statistically significant) ventricular size increase in aged dogs, no negative effect was shown in association with this on the performance in a highly specialised fMR task (sustained attention and self-inhibition, Gunde et al, in prep).

We have established a novel method for obtaining high-resolution macro-anatomical images of an in-situ dog brain, which, together with the structural images (CT and MRI) gained from the same study, could form the basis of a comparative canine brain atlas (Czeibert, Baksa, et al., 2019). We also summarized the existing terms of the canine gyriation nomenclature to show the inconsistencies in their usage regarding sulci of the brain. Based on this summary, we clarified their definition to provide a common base for future terminology (Czeibert, Piotti, Petneházy, & Kubinyi, 2019) and made an MR-normalization protocol for the fMRI studies to provide the means for a more reliable statistical analysis, and also included the creation of an individual label-based MRI-atlas for research purposes (Czeibert, Andics, Petneházy, & Kubinyi, 2019).

Figure 37. Overlaying the labels on the individual template with 50% opacity (for reference of the numbers, see the original paper). (A) Transverse plane. (B) Sagittal plane. (C) Dorsal plane. (D) Three-dimensional surface view of the labels from the right lateral aspect. Picture from Czeibert, Andics, Petneházy, & Kubinyi, 2019.

20.3 Canine Brain and Tissue Bank

We expected that some old dogs participating in our different studies would die due to natural causes during the 5 year long course of the project. This offered the possibility

of comparing their behavioural performance to those dogs that live/lived longer. We also expected that some of our subjects’ owners will offer the brain of their deceased dog for post-mortem analysis. Therefore, we have established the Canine Brain and Tissue Bank in Budapest at the Eötvös Loránd University in 2016 (https://familydogproject.elte.hu/canine-tissue-bank/). We developed a unique pet dog body donation system for those owners, who, in agreement with their veterinarian, voluntarily offer their dog’s body after a medically reasoned euthanasia (Sándor et al, submitted). This donation system is in harmony with the international and national guidelines and laws. As an integral part of this, we established a state-of-the-art communication, transportation, and sampling system. The samples, together with thorough documentation of the dogs’ previous, cognitive performance and medical history allow us to correlate the post-mortem physiological data with the behavioural measurements. So far we have obtained more than 100 dog brains and other tissues (e.g.

liquor cerebrospinalis, lymphoid tissue, ganglia, muscle and skin; Sándor, Tátrai, et al.

2019). We evaluated the histological quality of the first canine brains donated to the Canine Brain and Tissue Bank by hematoxylin-eosin (HEMA) staining. Together with HEMA staining we tested the applicability of these tissue samples in immunohistochemistry analysis using three different antibodies against autophagy related proteins (SQSTM1, MAP1LC3B and a microtubule associated protein). These studies were only intended as a methodological set-up and yielded positive results.

Figure 38. The Canine Brain and Tissue Bank collects samples from euthanised pet dogs whose bodies have been formerly donated by their owners. Credit : Kálmán Czeibert

We also tested several available methods for the proteomic analysis of dog cerebrospinal fluid (CSF) that was collected post mortem and we have evaluated the levels of autophagy-related genes (SQSTM1, MAP1LC3B) in dog cortical samples by western blot and Real-Time qPCR analysis. In the case of MTMR14, which has been shown to down-regulate autophagy, our results were in accordance with the findings of our collaborators utilising Drosophila melanogaster as a model species, regarding the age-related changes in the expression of this gene (Kovács et al., 2019).

An analysis of the associations between gene expression levels (set of all RNA molecules, transcriptome) and cognitive performance is also possible using suitable brain tissues. In this project, we aim to quantify the genes that are differentially expressed in the brains of old and young dogs, which will provide information about the genetics of cognitive aging in dogs. The total RNA set will be isolated from the brains of old and young dogs (brain tissue samples are available from the Dog Brain-bank) and the RNA samples will be sequenced. Old-vs-young sample comparisons will be performed according to the literature (Jónás et al, in prep.).

20.4 Genetics and molecular biology

Age-related cognitive decline, either being normal or pathological, is a fairly unique attribute of humans. Even in the case of neurodegenerative diseases, which can usually be easily characterized by certain behavioural or physiological changes, the underlying genetic factors often remain barely understood (Sándor and Kubinyi, 2019). In addition, the role of these factors in the development of disease is also strongly influenced by environmental context. For example, the average heritability of late-onset AD was estimated to be 0.58 (Gatz et al., 2006), although one genetic variant, the ε4 allele of the APOE gene were shown to outmatch any other variants as the main risk factor for developing late-onset AD (Brousseau et al., 1994; Corder et al., 1993; Lambert et al., 2013; Strittmatter et al., 1993). Interestingly, the APOE e4 was also associated with cognitive function both in subjects suffering from Alzheimer’s disease and in non-impaired old individuals (Wisdom, Callahan, & Hawkins, 2011). A GWAS study confirmed that the APOE region is significantly associated with non-pathological cognitive aging due to a functional, regulatory non-protein-coding effect (Davies et al., 2014). This finding also pinpoints to the necessity of considering genetic factors that can modify the course of non-pathological cognitive aging, when we want to get a better picture of human aging. Our goal is to identify age-related genetic markers.

Some genes which have been implicated in gene-behaviour associations in dogs show associations with aging in humans, which makes them good prospective candidates for further research. In humans, the major cognitive domains, memory, in particular, show high heritability (Harris & Deary, 2011; Reynolds & Finkel, 2015).

Several candidate genes were tested for association with cognitive ability and cognitive decline during aging in older people. The brain-derived neurotrophic factor (BDNF) gene, which is implicated in hippocampal-dependent memory (Egan et al., 2003), depression (Martinowich, Manji, & Lu, 2007) and susceptibility to psychiatric disorders (Hall, Dhilla, Charalambous, Gogos, & Karayiorgou, 2003; Notaras, Hill, &

van den Buuse, 2015), showed that it was associated with reduced cognitive function in elderly people (Payton, 2009; Tapia-arancibia, Aliaga, Silhol, & Arancibia, 2008).

Many studies reported that the catechol-O-methyl-transferase (COMT) gene, which is involved in the dopaminergic pathways of neurotransmitters, is associated with cognition, but a meta-analysis have not found strong evidence for this link (Barnett, Scoriels, & Munafò, 2008). Variation in the disrupted in schizophrenia 1 (DISC1) gene was linked to cognitive aging specifically in women (Thomson et al., 2005). However, the findings did not replicate in a follow-up study on a different sample (Palo et al.,

2007). On the other hand, for example, the dystrobrevin binding protein (DTNBP1) gene has been associated with cognitive function in multiple studies (Zhang, Burdick, Lencz, & Malhotra, 2010). Polymorphisms in the dopamine receptor D4 (DRD4) gene were found to be linked to activity-impulsivity trait in dogs (e.g. Hejjas et al., 2007), and alleles of the human homolog are associated with longevity (Grady et al., 2013;

Szekely et al., 2016). Furthermore, BDNF was shown to be differently expressed in dogs fed with an antioxidant cocktail or exposed to environmental enrichment (Fahnestock et al., 2012; Sechi et al., 2015), thus it may have a similar function in neuroprotection as its human counterpart. Oxytocin receptor gene (OXTR) polymorphisms in dogs were linked to human directed social behaviour (Kis, Bence, et al., 2014), while in humans, together with BDNF, it was linked to old age depression (Chagnon, Potvin, Hudon, & Préville, 2015). COMT, which has already been mentioned in connection with human cognitive aging above, is a suggestive marker of human contact seeking in laboratory beagles (Persson, Wright, Roth, Batakis, & Jensen, 2016). In summary, it is also possible that age-related canine personality changes (Fratkin et al., 2013) are also affected by these genes, however, this question requires further research efforts to get answered.

We found associations between specific gene (oxytocin receptor, OXTR; opioid receptor, OR) polymorphisms and the greeting behaviour of dogs, but we did not find an age effect in the test population (Kubinyi et al., 2017).

We designed experiments to detect the activity of mobile genetic elements in the dog genome. Specifically, we aimed at measuring the activity of the LINE-1 retrotransposon in the brain and its relation with aging. We have developed a western blot method to measure the amount of the ORF2 protein in the brain and muscle samples. Currently, we are collecting samples from euthanized dogs to compare age groups. We have also developed a quantitative real time PCR (RT-qPCR) method to measure the LINE-1 mRNA level in brain and muscle samples. For this purpose, we have designed a primer pair on the most conservative 200 bp range in the LINE-1 ORF2 sequence. Currently, we are looking for the best control genes and are focusing on collecting more samples to compare age groups (Tátrai et al, in prep).

We aimed to investigate how cognitive aging in dogs is associated with the composition of their intestinal microbiota. We collected faecal samples from old and young dogs (N=50) and analysed the compositions of their intestinal microbiota by 16S RNA sequencing using an Ion Torrent platform from Life Technologies. We have already performed PCR for 36 samples and sent them for sequencing to the University of Michigan (USA). Currently, we are collecting more faecal samples and are continuing with the PCR and sequencing analyses.

20.5 Closing remarks

The dog has grown to be one of the most important animals for researchers who aim to understand the biological background of complex traits. The significance of our best friend as a model animal originates in its unmatched morphological and behavioral variability and the unique population structure of purebred dogs being selected for various purposes through centuries. Dog breeds show a huge phenotypic, genetic and

longevity variability and have more shared ancestral DNA sequences with humans than rodents do. Dogs have six-twelve times shorter lifespan than humans, making aging-related investigations feasible. Companion dogs live closely with humans thus share the same, varied environment, unlike laboratory animals. Importantly they also have multiple, naturally developing age-related diseases as humans, for example, the Alzheimer-like canine cognitive dysfunction. Therefore, it is expected that research on dogs is more applicable to humans than those conducted with laboratory animals. The results of the present thesis showed age-related differences in brain activity, cognition, personality, and social status in several aspects similar to human aging, supporting the dog’s potential as a model of cognitive aging and providing knowledge for increasing the quality of life of dogs and owners alike.

Figure 39. Members of the Senior Family Dog Project in 2017 (above) and 2019 (below).