1. Development of the strains; propagation of stocks

Broadhurst (Reference Broadhurst and Eysenck1960) developed the Maudsley strains at the Animal Psychology Laboratory of the University of London’s Institute of Psychiatry in Beckenham, Kent, following an earlier successful genetic selection experiment by Calvin Hall, initiated in the 1930s (described in Hall, Reference Hall and Stevens1951). Broadhurst selected for high and low defecation in a brightly illuminated arena accompanied by white noise as an acoustic stimulus by breeding male and female rats with the highest OFD scores (MR strain) and, conversely, those with the lowest scores (MNR strain) for 15 generations. After discontinuing selection at Generation 15, he retested the strains at Generation 20 and found the differences in OFD had been maintained (Broadhurst, Reference Broadhurst1962). The strains have been inbred by brother/sister mating to fix key loci in the homozygous state. It was noteworthy that the primary deviation from the foundation population mean OFD score was in the direction of decreased defecation and that it was achieved within a few generations. Analyses of data for the first five generations of MNR rats could not reject the hypothesis of involvement of a major gene in the OFD response to selection (Liang, Reference Liang1978, personal communication). The strains were exported to the National Institutes of Health in Bethesda, Maryland, and to the University of Northern Iowa (Harrington, Reference Harrington1981). Responding to concern that OFD scores of MNR rats of the North American stocks had increased (Harrington, Reference Harrington1972), strain differences in OFD were found to resemble those typical of English stocks when they were maintained under similar conditions to those in use at the Animal Psychology Laboratory in England (Harrington & Blizard Reference Harrington and Blizard1983), a result consistent with the idea that the genetic variants underlying OFD in the strains were fixed. Cryopreserved embryos from sublines of the strains (MR/Har/Bliz and MNRA/Har/Bliz and from the MR/N strain from the former NIH colony) are kept at the University of Missouri Rat Resource Center in Columbia, Missouri (https://www.rrrc.us), and can be rederived as necessary.

2. Validation of selection criterion

Broadhurst relied on Calvin Hall’s earlier attempts (Hall, Reference Hall1934, Reference Hall and Stevens1951) to validate OFD as a measure of emotionality and provided additional support of his own in unselected Wistar rats (Broadhurst, Reference Broadhurst1957a, Reference Broadhurst1957b, Reference Broadhurst1958). In humans, gastrointestinal (GI) discomfort (both diarrhea and constipation) accompanies anxiety and stress. Broadhurst also chose to cite the excretory reactions of humans in warfare (Stouffer et al., Reference Stouffer, Lumsdaine, Lumsdaine, Williams, Smith, Janis and Cottrell1949) to support the relevance of this physiological system to human stress response (Broadhurst, Reference Broadhurst and Eysenck1960). Thus, OFD in rodents as a measure of mammalian reactivity to stress is plausible both from the perspective of experimental approaches in laboratory rats and from observations in humans during everyday life and when under extreme stress.

Cross-species relevance of elements of the open-field test should be considered in evaluating its appropriateness for the development of a model system. The very bright illumination in the open-field test is a plausible stressor when considering the photophobic nature of the rat but response to bright lights is not at the forefront of discussions of human N. More generally, Archer (Reference Archer1973, Reference Archer1975) argued that, rather than representing a unitary dimension (as conceptualized in humans), rat emotionality is better conceived as a range of expressive behavioral patterns displayed in a situation-specific manner. From this perspective, selection on a single criterion such as OFD might not capture the full range of emotionality in rats.

3. Behavioral and neurochemical differences in strains

Two reviews summarized the many behavioral comparisons of the strains that were conducted in England and suggested that the results were consistent with the idea that MR rats have greater stress reactivity (see later discussion of this characterization) than MNR (see tabular presentation of these results in Broadhurst, Reference Broadhurst1975; Eysenck & Broadhurst, Reference Eysenck, Broadhurst and Eysenck1964). Later behavioral research (reviewed in Blizard & Adams, Reference Blizard and Adams2002) tested the strains in social contexts including agonistic behavior. Resident MR males were observed attacking intruders introduced into a home cage or a small colony and attacking familiar colony mates more often than MNR males across these settings. Differences in social behavior between the strains may emerge in early life: when weanlings were tested for emergence from their nest box while either a strange male or the rat’s mother was confined at the center of their home cage, MRs of both sexes were quicker to emerge and were more active near the strange male compared to MNRs, whereas both strains uniformly emerged and made contact when their mother was the stimulus animal. This latter test was unique in presenting a social test stimulus in a familiar setting.

Blizard (Reference Blizard1981, Reference Blizard and Palermo1989) and Blizard and Adams (Reference Blizard and Adams2002) reviewed later research of primarily North American stocks, focusing more on the physiological and neurochemical basis of emotionality as represented in the central and peripheral sympathetic nervous system. In part, these findings cohere with the idea that genetic selection for OFD in the Maudsley model has affected both the central noradrenergic system where MNRs exhibit greater sustained cerulear response to chronic stress (Blizard, Freedman, & Liang, Reference Blizard, Freedman and Liang1983) and the peripheral noradrenergic system where MNRs possess increased sympathetic tone in many organs (Blizard, Altman, & Freedman, Reference Blizard, Altman and Freedman1982; Liang & Blizard, Reference Liang and Blizard1978; Slater, Blizard, & Pohorecky, Reference Slater, Blizard and Pohorecky1977).

4. Interpretive issues underlying behavioral comparisons

The large number of behavioral comparisons of the strains referred to above consisted, for the most part, of screening the two strains on standard behavioral tests without exploration of the specific mechanisms as to why strain differences occurred. For example, MNR rats were found to perform escape avoidance conditioning more efficiently than MR rats, and this was interpreted to imply that the putatively greater fear/emotionality of MR rats impaired their acquisition of the task. Nevertheless, protocols used for avoidance conditioning vary widely: for example, shock intensity imposed, nature of the conditioned stimulus, including its duration, physical parameters inside the conditioning chamber, etc., can interact with characteristics of the tested subjects to enhance or obscure group differences. If the Maudsley strains differed in pain thresholds, or in activity levels in the chambers before conditioning trials commenced, would such differences affect interpretation of any strain differences that emerged during conditioning? This concern is especially cogent given the lack of replication of the strain difference in active avoidance conditioning by Harrington (Reference Harrington1979) and others (for discussion, see Blizard and Adams, Reference Blizard and Adams2002).

Pertinent questions can be asked of many of the other tests in which MR/MNR differences were found. More generally, the use of electric shock in several of the relevant test situations introduced a stimulus that is not ecologically relevant to the rat. To a large extent, this concern is met by introducing social stressors as a complement to or even as a substitute for some of the existing traditional stress stimuli (Adams & Blizard, Reference Adams and Blizard1987; Blanchard et al., Reference Blanchard, Spencer, Weiss, Blanchard, McEwen and Sakai1994; Martinez, Calvo-Torrent, & Pico-Alfonso, Reference Martinez, Calvo-Torrent and Pico-Alfonso1998). Clearly, there is face validity for evolutionarily meaningful social contexts where defeat or low social status may correspond to lowering of reproductive success (Adams & Boice, Reference Adams and Boice1983; Blanchard, McKittrick, & Blanchard, Reference Blanchard, McKittrick and Blanchard2001).

Differences in stress reactivity were invoked earlier to account for differences between the strains. Activation of the hypothalamic-pitutitary-adrenal (HPA) axis is commonly used as an index of stress, and the Maudsley strains have not been found to differ in adrenocorticotropic hormone (ACTH)/corticosterone response to painful stimuli (Blizard, Eldridge, & Jones, Reference Blizard, Eldridge and Jones2015). It would be helpful to characterize the stress dimension more specifically when considering behavioral outcomes so as to be able to predict the direction of differences between groups before a test is administered rather than in a post hoc manner.

Aside from the results of standardized behavioral testing, a notable difference between the strains was observed during routine handling; MNR rats were more tractable, had lower muscle tone, and were flaccid, often hanging limply when held gently by the shoulders and neck. In contrast, MRs tended to struggle and resist handling (Blizard, personal communication, Reference Blizard1990).

5. Peripheral sympathetic nervous system and emotionality

Research conducted on North American stocks of the Maudsley rats showed that, under resting conditions, MNR rats had substantially higher concentrations of norepinephrine (NE) in peripheral organs than MR rats (e.g., Blizard, et al., Reference Blizard, Altman and Freedman1982; Liang & Blizard, Reference Liang and Blizard1978; Slater et al., Reference Slater, Blizard and Pohorecky1977), and one interpretation of these differences is that MNR rats’ organs are under tonic peripheral sympathetic stimulation. A corollary of this finding is that, when stressed, there is the potential for greater efflux of NE onto organs of MNR rats. In the colon, such an event would have the potential to relax smooth muscle and inhibit colonic motility and is consistent with MNR rats’ behavior when placed in the open-field test (defecation scores are effectively zero). Additional strain differences in GI processing have been reported, which must be considered in developing an appropriate understanding of how genetic selection has impacted this system in the strains. For example, following mild food deprivation, after being fed a small meal in their home cage, MNR rats excrete more fecal boli than MR rats, the opposite of the strain difference that occurs in the open-field test (Blizard, Reference Blizard1989, personal communication). Thus, neurological or neurochemical systems favoring increased GI and/or colonic motility under basal conditions may have also been selected in MNR rats. Furthermore, the cited changes in the peripheral sympathetic nervous system, brought about by genetic selection, have the potential to account for strain differences in emotionality without positing primary variations in the CNS.

Another incidental observation was that the eyes of the MNR rats were much more prominent than those of MR rats, a larger amount of the eyeball could be seen (Blizard, Reference Blizard1987, personal communication). If this was the result of sympathetic overstimulation of upper eyelid retractor muscles, as is typical in patients suffering from Graves disease, it would be consistent with the increased sympathetic tone in MNR rats noted above.

6. Central noradrenergic system and emotionality

MNR rats showed greater sustained tyrosine hydroxylase elevation in the locus ceruleus (LC), following chronic stress than MR rats (Blizard et al., Reference Blizard, Freedman and Liang1983), and opened the possibility that an important central noradrenergic nucleus might have been the focus of genetic selection (Blizard, Reference Blizard1988). Intriguingly, an earlier attempt to explore the neurological basis of anxiety (Gray & McNaughton, Reference Gray and McNaughton2000) emphasized the potential role of the dorsal noradrenergic bundle (DANB) in the medial septal-hippocampal inhibition system in anxiety. Stimulation of hippocampal theta rhythm from electrodes implanted in the medial septum has a frequency-specific relationship to current driving intensity, with a minimum at 7.7 Hz in outbred male rats, which is lacking in putatively less emotional female rats and MNR rats. MRs, on the other hand, resemble unselected outbred rats with a minimum at 7.7 Hz. Administration of antianxiety drugs and depletion of DANB noradrenaline content by neurochemically specific toxins also abolished the frequency-specific minimum at 7.7 Hz in unselected rats. The DANB originates in the LC, so the discovery of differences in the magnitude of strain-specific changes in response to chronic stress in this locus in the Maudsley model is an intriguing correlation that draws attention to the need for further examination of the central noradrenergic system in these strains. The LC projects to many areas of the forebrain and group differences in its biosynthetic capacity may have important functional implications in its terminal regions. Understanding of the LC and its role in neurophysiological and behavioral response has changed considerably since the findings described above were reported. At that time, an idea promoted by Amaral and Sinnamon (Reference Amaral and Sinnamon1977), that activation of the LC improved signal-to-noise ratio in its terminal regions, was influential. On the other hand, this was a hypothesis relevant to the neurophysiological dimension and was not necessarily easy to translate into specific behavioral outcomes. More recent approaches (Poe et al., Reference Poe, Foote, Eschenko, Johansen, Bouret, Aston-Jones and Sara2020) emphasize the discrete role of ascending projections from the LC, and none of these have been related to the manifold behavioral differences between the two strains. Later, we discuss recent work on the role of descending projections of the LC on colonic function.

7. Human colon function and rodent models

As noted, the differences in OFD between the Maudsley strains are associated with variation in colonic function under resting conditions and correlated with alterations in the peripheral sympathetic nervous system. IBS in humans represents a variety of functional GI disorders in which stress and personality factors have often been implicated, and exploration of mechanisms underlying these associations has revealed new insights into a biological mechanism via which genetic selection for OFD may have been achieved. Recently, Kurahashi’s laboratory (Kurahashi et al., Reference Kurahashi, Kito, Baker, Jennings, Dowers, Koh and Sanders2020a, Reference Kurahashi, Kito, Hara, Takeyama, Sanders and Hashitani2020b) has provided evidence of the existence of two α₁ adrenergic receptors (α_1A and α_1D) in mouse and human colon, which have opposing effects on colonic muscle. Specifically, α_1D receptors located on smooth muscle cells (SMC) of mouse and human colon resulted in contraction of colonic muscle when exposed to 1 µm NE, while α_1A receptors, located on PDGFRα+ cells, inhibited SMC cells via the AR-SK signaling pathway when exposed to 10 µm NE. Alpha₁ receptors in rat colon could represent the substrate upon which the altered levels of NE existing in colonic tissues (Blizard et al., Reference Blizard, Altman and Freedman1982) of the two strains could exert their effect. Thus, the higher levels of NE in tissues of MNR rats released onto α_1A receptors in the colon during stress could diminish the amplitude of colonic contractions and reduce or prevent OFD. Conversely, in nonstress situations, such as the home cage, release of lower or basal levels of NE could stimulate colonic contractility via α_1D receptors and result in the higher fecal output following a meal in MNR rats (described earlier). The suggested role of α₁ receptors in human and mouse colon is a novel finding discovered by investigators chiefly interested in IBS. We hypothesize that alterations in sympathetic tone, such as we have suggested to exist in the Maudsley model, could interact with these receptors to produce these functional disorders. Aside from presynaptic influences, the possibility exists that genetic selection for OFD may have also assorted different densities of colonic α₁ receptors or different colonic receptor types in the two strains.

Earlier, differences in the function of the LC between the Maudsley strains were discussed in relation to their potential impact on the CNS and their role in strain differences in behavior. New insight into the role of descending projections of the LC on autonomic function has recently been reported. Kong et al. (Reference Kong, Zhang, Gao, Pan, Deng, Xu and Wen2023), studying C57BL/6J mice, found that stimulation of a projection of the LC to the rostral ventromedial medulla has an hyperalgesic effect on nociceptive input from the descending colon and rectum. Previous research had shown that stimulation of direct projections from the LC to the dorsal horn had an opposite analgesic effect in response to colorectal distention in rats (Liu, Tsuruoka, Maeda, Hayashi, & Inoue, Reference Liu, Tsuruoka, Maeda, Hayashi and Inoue2007). These results show that the LC plays an important role in influencing afferent information from a key organ (colon) in GI processing. More generally, the discovery of important relationships between the LC and the colon provides an important focus for exploration of the relationship between the central and peripheral nervous systems in the Maudsley model.

8. IBS, personality, and genetics

In human genetic studies, there is a developing literature on the relationship of IBS to personality in which the contribution of genetics has been examined: Eijsbouts et al. (Reference Eijsbouts, Zheng, Kennedy, Bonfiglio, Anderson, Moutsianas and Parkes2021) conducted genome-wide analyses of 40 548 people with IBS from the UK Biobank and 12 852 from the Bellygenes initiative (a worldwide study aiming to identify genes linked to IBS) and compared them to 433 201 controls from the UK Biobank. They identified six susceptibility loci, which were replicated in a 23andMe panel of approximately four times as many cases and controls. Four of the loci were also associated with mood and anxiety disorders or were expressed in the nervous system. Specifically, there were significant genome-wide correlations between risk of IBS, anxiety, N, and depression of 0.5 or higher. In contrast, genetic correlations with other psychiatric disorders were substantially lower. The predominant conception of the relationship between IBS and anxiety is that the latter “causes” abdominal symptoms. However, Eijsbouts et al. conducted additional analyses to explore the role of shared genetic risk versus other conceptual models. They removed participants with anxiety from the IBS GWAS and removed participants with IBS from the GWAS for anxiety. The genetic correlation between IBS and anxiety remained and was estimated at 0.31 (SE = 0.06). Bidirectional Mendelian randomization and other analyses also showed that anxiety or depression and IBS are the results of shared etiologic pathways rather than one causing the other. Applying the same logic to the Maudsley model, it is possible that any influence of genetic selection on brain and behavior reflects the effect of the same genes that altered the peripheral sympathetic nervous system, not because one caused the other, as implied by the choice of the selection criterion, but because they are separate outcomes of the same neural pathways acting in the brain and the periphery.

Future research on IBS focused on specific mechanisms, regardless of their location in the body, will have to be conducted in a manner that recognizes the strongly held view (Mayer, Ryu, & Bhatt, Reference Mayer, Ryu and Bhatt2023) that this functional disorder must be investigated and treated using a multidimensional approach that encompasses multiple biological mechanisms as well as attention to environmental factors.

9. Heritability of neuroticism trait in humans

Large-scale studies of the N dimension have provided more precise estimates of its heritability. N is a heritable trait in humans with a broad-sense heritability of 48% based on a meta-analysis of six twin cohorts (total N 29 496 twin pairs; van den Berg et al., Reference van den Berg, de Moor, McGue, Pettersson, Terracciano, Verweij and Boomsma2014) and confirmed by pedigree analyses (Boomsma et al., Reference Boomsma, Helmer, Nieuwboer, Hottenga, de Moor, van den Berg and de Geus2018) as well as literature reviews (Sanchez-Roige, Gray, MacKillop, Chen, & Palmer, Reference Sanchez-Roige, Gray, MacKillop, Chen and Palmer2017). Vukasović and Bratko (Reference Vukasovic and Bratko2015) noted the higher heritability estimates in twin studies (47%) compared to family and adoption studies (22%) and attributed these differences to nonadditive genetic effects, which contribute to resemblances of twins and full siblings, but not to resemblance of nearly all other relatives.

Large-scale studies have also enhanced our understanding of the N dimension by providing analyses of its relationship to other personality dimensions, based on GWAS results. Exemplary of this trend are the findings of a comprehensive analysis which examined the relationship of the Big Five personality dimensions to loneliness (Abdellaoui et al., Reference Abdellaoui, Chen, Willemsen, Ehli, Davies, Verweij and Cacioppo2019a) in more than 29 000 twins and their family members. All personality traits were correlated with loneliness, but only N showed a significant relationship to loneliness (r = 0.50) after correcting for the remaining four personality traits. Loneliness has an estimated heritability of 42% (Distel et al., Reference Distel, Rebollo-Mesa, Abdellaoui, Derom, Willemsen, Cacioppo and Boomsma2010). In the study of Abdellaoui et al. (Reference Abdellaoui, Chen, Willemsen, Ehli, Davies, Verweij and Cacioppo2019a), single-nucleotide polymorphisms (SNPs) data were available in ∼4000 subjects. From the molecular data, it was estimated that the SNP (i.e., narrow-sense) heritability was 22% for N and 14% for loneliness. Narrow-sense heritability, of course, is particularly important when considering selection experiments because it is the magnitude of the narrow-sense heritability that gauges the additive genetic variance which selection can assort. A genetic correlation between these traits was estimated at .71, and second larger study (Abdellaoui et al., Reference Abdellaoui, Sanchezj-Roige, Sealock, Treur, Dennis, Fontanillas and Boomsma2019b) confirmed this estimate (r _g = 0.69). N’s correlation with loneliness draws attention to the social milieu for the expression of this dimension and underlines the potential significance of studies of social behavior in the Maudsley strains referred to earlier.

10. GWAS studies of neuroticism

Research in humans has led to multiple discoveries of regions in the genome that are implicated in N. Inspired initially by the seminal paper by Flint et al. (Reference Flint, Corley, DeFries, Fulker, Gray, Miller and Collins1995) that demonstrated the feasibility of identifying quantitative trait loci (QTLs; regions in the genome influencing quantitative phenotypes or traits) in mice for complex traits such as anxiety, linkage studies of N were undertaken (e.g., Wray, Kemper, Hayes, Goddard, & Visscher, Reference Wray, Middeldorp, Birley, Gordon, Sullivan, Visscher and Boomsma2008), as well as candidate gene studies that, for example, exploited the known synteny between mouse and human loci to screen the human genome for loci identified in mice (Fullerton et al., Reference Fullerton, Willis-Owen, Yalcin, Shifman, Copley, Miller and Flint2008). The initial enthusiasm for linkage studies in humans diminished rapidly when power simulations for complex traits were carried out based on realistic effect sizes. Likewise, candidate gene studies did not prove a fruitful approach. For example, QTLs for major depressive disorder discovered in genome-wide studies do not generally confirm the significance of candidate genes (e.g., Bosker et al., Reference Bosker, Hartman, Nolte, Prins, Terpstra, Posthuma and Nolen2011). Breakthroughs in genotyping technology enabled screening large numbers of participants for SNPs and genome-wide association studies became feasible, leading to several projects on N. Van der Walt, Campbell, Stein, and Dalvie (Reference van der Walt, Campbell, Stein and Dalvie2023) identified 32 GWASs of anxiety disorders, nondiagnostic anxiety traits, and N that reported 563 independently significant variants, with 29 replicated nominally in independent samples and three replicated significantly. In considering the low replication rate, van der Walt et al. reached the sobering conclusion that future GWAS investigations would need to increase sample sizes into the millions. We took their supplementary report for N based on eight meta-analyses of GWAS studies, the first one published in 2015 and the most recent one in 2019 (Baselmans et al., Reference Baselmans, Jansen, Ip, van Dongen, Abdellaoui, van de Weijer and Bartels2019; de Moor et al., Reference de Moor, van den Berg, Verweij, Krueger, Luciano, Arias Vasquez and Boomsma2015; Kim et al., Reference Kim, Kim, Yun, Heo, Cho, Kwon and Kim2017; Lo et al., Reference Lo, Hinds, Tung, Franz, Fan, Wang and Chen2017; Luciano et al., Reference Luciano, Hagenaars, Davies, Hill, Clarke, Shirali and Deary2018; Nagel et al., Reference Nagel, Jansen, Stringer, Watanabe, de Leeuw, Bryois and Posthuma2018; Okbay et al., Reference Okbay, Baselmans, De Neve, Turley, Nivard, Fontana and Cesarini2016; Smith et al., Reference Smith, Escott-Price, Davies, Bailey, ColodroConde, Ward and O’Donovan2016). Figure 2 summarizes these genome-wide significant results for N and plotted the total number of hits for each chromosome (Fig. 2A) and the proportion of hits per chromosome (Fig. 2B). Generally, the larger chromosomes (chromosome length is included under the X-axis) tend to have a larger number of significant hits, with some notable exceptions for chromosomes 8 and 18. The genes coding for the receptors relevant to contraction and relaxation of colonic muscle mentioned earlier do not align with any of the GWAS regions in Figure 2B.

Figure 2. Based on Supplement Table 3 in van der Walt et al. (Reference van der Walt, Campbell, Stein and Dalvie2023). X-axis: chromosome number and length of chromosome (number of base pairs, corresponding to https://www.ncbi.nlm.nih.gov/grc/human/data). Y-axis: 2A: number of hits on each chromosome is a sum of hits on each chr from Sup Table 3. 2B: proportion of hits per chromosome.

11. Developments in rat behavioral genetics

When Broadhurst conducted his selection experiment, the laboratory rat was the animal of choice in most psychology departments. During the course of the molecular revolution, most studies focused on the mouse because it was the traditional organism of choice for mammalian genetic studies, and investigators interested in behavior genetics accordingly switched their focus to that species. Nevertheless, the enormous literature on rat behavior remains a huge resource and has stimulated developments in rat genetics. In one mapping study, investigators studied F2 offspring of a cross between two rat lines selected for differences in sensation-seeking behavior. Relevant to the present interest in defecation response, they identified a statistically significant locus on rat chromosome 18 for frequency of defecation in the open field (Chitre et al., Reference Chitre, Hebda-Bauer, Blandino, Bimschleger, Nguyen, Maras and Palmer2022). Another study (Munro et al., Reference Munro, Wang, Chitre, Polesskaya, Ehsan, Gao and Mohammad2022) has mapped regulatory genes affecting expression of genes in five brain regions using genetically heterogeneous rats derived from a cross of eight strains originating in the former NIH colony. These results from the Palmer laboratory are available online at RatGTEx.org where expression data for individual genes as well as expression QTLs can be accessed. These developments in genetics will be an important resource for future investigations of rat models of human personality dimensions.