Self-Injurious Behaviour (SIB), which has been defined as intentional, direct injuring of body tissue without suicidal intent (Baguelin-Pinaud et al., 2009), is a serious global health problem occurring in several high risk populations with neuropsychiatric diseases / disorders (including borderline personality disorder, PTSD, depression, autism spectrum disorders) and in other neurodevelopmental disorders (Zlotnick et al., 1997, Mork et al., 2013, Al-Sharqi et al., 2012). Of particular concern is its lack of effective treatment (Navines et al., 2013, Novak et al., 2014) and growth in children and adolescence, most commonly in cases of depression or neurodevelopment (Jacobson and Gould, 2007). Macaques have been noted to spontaneously develop SIB (Novak, 2003), an important distinguishing factor from other animal models (namely rodents) which tend to be induced (with some exceptions (Bechard et al., 2017), making the macaque model a closer match with the human analog (Novak et al., 2014, Tiefenbacher et al., 2005, Davenport et al., 2008). This fact is especially important when considering the prevalence of SIB in populations with neurodevelopmental disorders. Such similarities with human SIB phenotypes suggest that a molecular investigation of SIB macaques could provide beneficial insight into the human pathology of SIB.

The use of macaques in research appears to represent the best compromise between phylogenetic and physiological relatedness to humans, cost efficiency, life-span, resources, expertise in animal husbandry practices, and adaptability for translation of results to humans. Aspects of research studies that utilise primarily rhesus macaques include neurobiology, anatomy, cognition and behavior (Urbanski and Sorwell, 2012, Voytko and Tinkler, 2004, Peters, 2002, Bailey et al., 2011, Schultz et al., 2001), reproductive senescence (Atsalis and Margulis, 2008), and immune senescence (Asquith et al., 2012, Messaoudi et al., 2011, Vaccari and Franchini, 2010). The rhesus macaque has been considered the “gold standard” model for human research, has been in use since the 1960’s, and has proven to have very high translational validity with respect to neurologic and behavioral assessments of infants (Nelson and Winslow, 2009). Equally important is the striking similarity between the innate immune systems of rhesus monkeys and humans (Evans and Silvestri, 2013). Due to this similarity, rhesus monkeys are susceptible to infection by many pathogens that result in disease states almost identical to those found in humans (a noteworthy example being Simian Immunodeficiency Virus, which induces an AIDS-like disease, including CNS compliactions, in rhesus monkeys (Lackner and Veazey, 2007). Importantly, the immune system of rhesus monkeys is more similar to humans than the immune systems of rodents, making monkeys the best choice for evaluating and correlating immunologic responses and CNS disease. Indeed, recent studies have linked severe stress to specific immune activation (Beurel and Lowell, 2017) and glial activation (Nijs et al., 2017).

While once controversial, the concept of glia regulating complex behaviours is rapidly gaining ground. While much work focusses on schizophrenia (see review by Xia and colleagues (Xia et al., 2016)), the basic physiology of abnormal astrocyte activation would be conserved. The very recent study by the Goldman group definitively showed astrocytes are responsible for behavioral traits (Windrem et al., 2017, Han et al., 2013). We have shown that astrocytes of macaques with SIB have a distinct morphological atrophy, and that the glial activation may be a component of the behaviors (Lee et al., 2013b, Lee et al., 2015a).

Support for such potential findings is grounded in the neuroinflammatory response of Spontaneous SIB Macaques. Such subjects demonstrated astrocyte activation, which has been demonstrated by the upregulation of intermediate filaments (IF), most specifically glial fibrillary acidic protein (GFAP), vimentin, and nestin. Furthermore, such responses have been linked to the dysregulation of downstream gene expression, suggesting neuroinflammation in SIB macaques can also be studied at the molecular level. To this end, we examined brain tissues from macaques with SIB, and compared them with matched tissues from animals without reported abnormal behaviors.

In control macaques, vimentin expression was limited to endothelial cells (Figure 1A). In macaques with SIB, we noted that astrocytes had increased levels of vimentin (Figure 1B), that reached significance in white matter astroctyes only (1C). We were very surprised to note that there was no increase in nestin expression, at least as measured as the proportion of GFAP / nestin double positive cells (1D).

As a first step to determine the underlying mechanisms of cell activation in animals with SIB, we performed RNA Seq analysis. We filtered data to determine genes that were either up-regulated or down-regulated to the highest degree. The top ten genes that were up- or down-regulated are presented in Table 2. Seven of the top ten genes that were either up- or down-regulated have antimicrobial functions (noted by *), confirming our previous studies of immune activation in these animals (Lee et al., 2013b, Lee et al., 2015b). Further, two genes were directly associated with astrocyte activation or maturation (§), and eight were linked to behavioral abnormalities and/or intellectual disability (#). Of the top 20 genes up or down regulated, eight were previously associated with behavioral or developmental disabilities, and seven with inflammation.

To determine the molecular mechanisms underlying this differential activation of glial cells, we have begun analyses of the RNA Seq data. Comparing total genes up and down regulated, there were 437 genes down regulated, and 526 up regulated. At the molecular level, we have noted there were multiple pathways activated linked to tissue remodeling and inflammation (Figure 2A). When we imported genes either up- or down-regulated into String software, to explore how these gene changes were inter-related.

We examined vimentin upregulation by tracking other molecular components associated with IF upregulation, which may be part of a mechanism of altered glial cell function in SIB. While vimentin protein was clearly increased (Figure 1A), there was no increase in vimentin at the mRNA level (from RNA seq data). However, several other genes important for vimentin expression were differentially regulated (Figure 2B). Importantly, within this pathway, TPM3 (1.6 fold), TNNT2 (3.3-fold), CASP7 (4.5-fold), NTRK2 (3.1-fold) and MYH14 (5.3-fold) were upregulated, whereas MYL2 (9.1-fold) was downregulated. In summary, one could reasonably explain increased vimentin within astrocytes.

Separately, we noted that multiple genes associated with Notch and Wnt signaling pathways were also upregulated (Figure 2C). Of note, HES4 (4.3-fold), DLL1 (2.7-fold), PSEN2 (3.3-fold), WNT2B (3.2-fold), TCF (2.4-fold) and CTBP2 (3.6-fold) were all upregulated. These genes have roles in neurogenesis and neuronal differentiation (HES, PSEN2, TCF4 and CTBP2)(Kageyama et al., 2005), further strengthening the potential for altered neuronal plasticity in macaques with SIB.

We have previously reported that animals with SIB exhibited shorter and less complex astrocytes that were more likely to be activated with regards to TLR2 (Lee et al., 2013b) expression. In this study, we aimed to better understand the molecular basis for spontaneously occurring SIB in rhesus macaques, and by inference, self-injury in humans. As we had previously showed innate immune activation and altered expression patterns of GFAP in astrocytes, we were interested in examining other glial activation markers and underlying molecular changes in frontal cortex of macaques with SIB. That vimentin, GFAP and TLR2 were altered, but not nestin, indicates at least a degree of specificity to the astrocyte activation (Duan et al., 2015). Molecular analyses of differentially-regulated genes were linked to cellular differentiation, innate immune activation and vascular remodeling.

Both Biocarta and the NCI-Nature curated pathway analysis tools indicate that the Rho pathway was most upregulated in the SIB macaques. RhoGTPases have been identified as potential targets in rodent models of Rett syndrome (De Filippis et al., 2012). While astrocytic degeneration in animals and humans with depressed phenotypes could point to Rho pathway activation, this study provides further evidence in primates. Upregulation of HES4 could lead to extensive remodeling in key brain areas (Bai et al., 2015) through activation of the rho and rac (Hall and Lalli, 2010), as well as the wnt pathways. Further, activation of the wnt pathway can increase expression of vimentin (Knoll et al., 2014), leading to further remodeling.

Astrocytes are glial cells important in maintaining homeostasis and the blood brain barrier, regulating neurogenesis and supporting metabolism at synapses (Lee and MacLean, 2015). There are several reports of GFAP-low/negative astrocytes, including in depressive disorders (Fatemi et al., 2004, Torres-Platas et al., 2015, Tynan et al., 2013, Miguel-Hidalgo et al., 2010). Altered expression of GFAP, vimentin (Al-Ahmad et al., 2011), and nestin (Strong et al., 2004, Pekny et al., 1998) is linked to altered vesicle motility in astrocytes (Potokar et al., 2007). Therefore IF changes could logically be linked with altered secretion of pro- or anti-inflammatory molecules as well as with morphological changes. This hypothesis is supported by several lines of evidence: astrocytes show altered morphology concomitant with inflammation (Xing et al., 2008), associated with altered expression of IFs.

Self-injurious behaviour in macaques is a spontaneously occurring phenomenon, affecting approximately 5% of the captive population. The underlying molecular mechanisms have remained somewhat a mystery. As such, these studies expand upon those of induced depressive-like behaviors in rodents that have shown glial activation and neuroinflammation as probable mechanisms (Tynan et al., 2013, Hinwood et al., 2013). We acknowledge that ideally tissues would be collected prospectively, allowing simultaneous collection of CSF and plasma for cytokine studies, and deeper studies of links between mRNA and protein expression. Notable in this study was increased vimentin expression at the protein, but not the mRNA levels. In vitro studies could dissect this and other disconnects between protein and mRNA expression. Further, as indicated in Figure 2C, several genes associated with vim expression were differentially expressed, potentially facilitating increased protein expression. Future studies will validate the top hits by PCR, and provide pathway targets for potential behavior modifying medicaments.

An important caveat in these studies is the retrospective nature of the studies. It would be very interesting to repeat these studies using an induced depressive behaviour model (Li et al., 2013) with age and social hierarchy matched animals with no history of self-injury. Further investigation is needed to determine precisely which cells are activated, and if treatment with antidepressants reverse all the components of the activation observed (Lee et al., 2015a).