DSS-induced inflammation in the colon drives a proinflammatory signature in the brain that is ameliorated by prophylactic treatment with the S100A9 inhibitor paquinimod

Nearly 1.5 million North Americans and 2.2 million Europeans suffer from inflammatory bowel disease (IBD), and the annual incidence and prevalence continue to increase [1, 2]. IBD is characterized by chronic inflammation in the intestinal tract, consisting of ulcerative colitis (UC) and Crohn’s disease (CD). In UC, inflammation typically involves the rectum and extends proximally, resulting in lesions of variable severity that remain localized to the colon [3]. In CD, inflammation is transmural, producing segmented lesions that can occur in any part of the gastrointestinal tract [3]. Both UC and CD are life-long diseases that cycle between periods of active flare and remission. While pathology primarily affects the intestine, extraintestinal manifestations are prevalent and can involve virtually any organ site [3‐5]. In this regard, IBD is increasingly linked to CNS inflammation and behavioral alterations in patients. A relatively large number of IBD patients report depressive symptoms and cognitive dysfunction compared to the general population, ranging from symptoms that seem to be transient and coincide with flare onset to cognitive issues that a recent study reveals as a long-term comorbidity in IBD patients [6‐11]. IBD patients are also at increased risk for neurodegenerative diseases, such as Parkinson’s disease and dementia [12‐17]. Understanding the mechanistic basis of this connection between colonic inflammatory disease and neurological manifestations will be necessary to develop therapies designed to alleviate or prevent colitis-induced CNS dysfunction.

We have recently developed a transgenic mouse expressing a caspase-1 biosensor that allows the measurement of inflammatory responses in living animals and ex vivo [18]. Initial characterization of this mouse demonstrated that the biosensor can effectively identify regions of segmented inflammation in the colon of mice treated with DSS [18]. In this study, we employed our reporter system to monitor inflammation in mice with colitis. Our findings demonstrate that local inflammation in the colon triggers systemic and neuroinflammation. We further characterized this proinflammatory signature in the brains of mice with colitis and found that prophylactic administration of paquinimod can mitigate colonic and CNS inflammation.

To understand the inflammatory response in mice with ongoing colitis, we employed the DSS model of colitis, which recapitulates human UC in mice. Similar to our previous results [18], caspase biosensor mice given 2% DSS for 7 days exhibited significant weight loss, colonic shortening, increased histopathological hallmarks of UC and blood in the stool, as well as biosensor activation and increased inflammatory cytokine/chemokine levels in the inflamed colon (Fig. 1). Female mice developed less severe disease, as measured by reduced weight loss, less colonic shortening, reduced inflammatory cytokine and chemokine levels in the colon, and less female mice had a positive Hemoccult test (Fig. 1, Additional file 1).

We next tested whether the induction of colitis could drive inflammasome activation in other tissues, specifically the CNS. We observed a significant increase in biosensor activation in the brains of mice treated with DSS compared to control mice (Fig. 2A). We were concerned that a disruption of the blood–brain barrier (BBB) could lead to increased luciferase substrate diffusing into the brains of mice treated with DSS, and therefore may explain the apparent increase in biosensor activation observed in the brains of these mice. To address this concern, we generated organotypic slice cultures from brains isolated from DSS-treated and control mice immediately following ex vivo IVIS imaging. After 7 days of standard culture conditions, bioluminescence was quantified in vitro, and we again observed increased biosensor activation in slice cultures generated from DSS-treated mice, relative to control mice (Fig. 2B).

Because the bioluminescent reporter can be turned on downstream of either inflammatory and apoptotic caspases [18], we next wanted to determine whether brain tissue isolated from DSS-treated mice exhibited an inflammatory phenotype. Brains were isolated from control or DSS mice 7 days following DSS exposure and caspase-1 expression was measured. We detected an increase in caspase-1 p45 and p20 expression in some, but not all, brains isolated from mice with ongoing colitis (Fig. 2C). As a positive control for neuroinflammation induced by a peripheral stimulus, mice were injected intraperitoneally with lipopolysaccharide (LPS) to trigger inflammation in the CNS. We detected significant bioluminescence in brain tissue 24 h following LPS injection (Fig. 2A). As with the DSS-treated mice, we also measured increased caspase-1 expression in brains isolated from some, but not all, LPS-injected mice (Fig. 2C). Notably, in the same DSS and LPS brains where we detected cleaved caspase-1 (p20), we did not detect cleaved caspase-3 (p17/p19), although we did measure some cleaved caspase-8 (Additional file 2), which could suggest that other caspases may be contributing to biosensor activation in this system. To further elucidate if inflammatory pathways were upregulated in the CNS in mice with colitis, brain tissue isolated from DSS-treated mice was separated into forebrain, midbrain, and hindbrain regions and expression of inflammatory transcripts was quantified. DSS treatment induced an upregulation of TNF-α, IL-1β and KC (Fig. 2D). We also detected reduced expression of brain-derived neurotrophic factor (BDNF), which is well established to be reduced in response to inflammation in a number of pathological conditions [21] (Fig. 2D). Next, we measured inflammatory cytokine and chemokine expression using a proteome profiler array to simultaneously measure the relative expression levels of a panel of 40 inflammatory markers in the brains of control, DSS- and LPS-treated mice. Relative to control mice, DSS-treated mice exhibited increased expression of a number of inflammatory cytokines/chemokines, most notably and consistently, KC (Fig. 2E, F, Additional file 3). KC upregulation in the brain was confirmed by CBA (Fig. 2G).

In order to determine if this increased inflammatory response was the result of increased immune cell infiltration into the brain, immune cell populations in brain tissue from control and DSS-treated mice were assessed. We detected a significant increase in inflammatory monocytes, neutrophils and lymphocytes in the brains of LPS-treated mice, while in DSS-treated mice, we measured minimal increases in monocyte and neutrophil infiltration and no lymphocyte recruitment into the brain (Fig. 2H, Additional file 4). To determine if resident cells in the CNS were driving this inflammatory response in the brain in mice with colitis, we isolated brains from control, DSS- or LPS-treated mice, clarified and stained the intact organs with an anti-Iba1 antibody and visualized microglia using light sheet microscopy. We detected increased Iba1 immunoreactivity and a change in signal distribution in DSS brains compared to control, and this signal was even more pronounced in brains isolated from mice injected with LPS (Fig. 2I). Collectively, these data suggest that colonic inflammation activates proinflammatory pathways.

We next performed RNA sequencing analysis to better understand the underlying changes in the brains of mice with colitis. Mice were treated with 2% DSS for 5 days and then given normal drinking water for 2 days. On day 7, brain tissues were harvested from DSS and control mice and RNA was prepared for sequencing. Pathway analysis of the differentially expressed genes identified several significantly enriched terms related to inflammation, including genes involved in the IL-17 signaling pathway, regulation of inflammatory responses, and antimicrobial peptides pathway (Fig. 3A, B). A number of the genes identified are well established to play a role in neuroinflammation, such as Ptgs2 (also known as COX-2), the receptor for IL-1α/β, IL1R1, and S100 calcium-binding proteins S100A8/A9, endogenous alarmins/DAMPs. We focused on Lcn2, S100A8 and S100A9, which were upregulated in 3 of the 4 pathways identified. Lcn2 is an antimicrobial peptide that is secreted by various cell types and has become a biomarker for inflammation. Notably, it is increased in inflamed intestinal tissues of patients with active ulcerative colitis and active Crohn’s disease [22, 23] and our data demonstrate that it is also highly upregulated in the brain of mice with ongoing colitis. Likewise, S100A8 and S100A9, also known as myeloid-related protein 8 and 14 (MRP8 and MRP14), respectively, are emerging preclinical biomarkers upregulated in various inflammatory disease models including IBD [24‐28]. To this end, we quantified S100A9 and Lcn2 transcript and protein levels in brain tissue isolated from control and DSS-treated mice, using the original experimental paradigm (2% DSS, 7 days). We confirmed upregulation of Lcn2 and S100A9 transcripts and measured increased Lcn2 and S100A8/A9 protein expression in brains isolated from mice with colitis, albeit expression is markedly less than the levels detected in mice injected with LPS (Fig. 3C, D). Together, these data reveal transcriptomic changes in the brain consistent with an inflammatory response in the CNS following colitis induction. Importantly, this inflammatory phenotype appears to be transient, as most of the differentially expressed genes detected in the brain on day 7 returned to baseline by day 14, 9 days following DSS cessation (Fig. 3B).

In order to determine if there was a systemic inflammatory response in mice with colitis, we measured inflammatory cytokines in the serum. We detected a notable increase in serum IL-6 and KC in mice treated with DSS for 5–7 days (Fig. 4A, B, Additional file 5), supporting the idea that acute colitis promotes a systemic inflammatory response [29, 30]. We also measured increased S100A8/A9 in the serum of DSS-treated mice (Fig. 4C). However, consistent with previous reports [29], we were unable to detect a significant increase in endotoxin levels in the serum of mice with colitis, despite measuring high levels of serum endotoxin in LPS-treated mice (Fig. 4D).

S100A8 and S100A9 are upregulated in numerous models of neuroinflammation, cognitive dysfunction and neurodegeneration [31‐34]. Recently, a class of orally active chemical compounds called quinolone-3-carboxamides (Q-compounds) were found to be protective in several inflammatory disease mouse models [32, 35‐43] and have been tested in clinical trials [44‐46]. One quinolone-3-carboxamide derivative, called paquinimod (ABR-215757), binds S100A9 and inhibits its interaction with TLR4/MD2 and RAGE [47]. To determine if pharmacological inhibition of S100A9 signaling attenuated DSS-induced systemic and neuroinflammation, paquinimod was provided in the drinking water one week prior to and during the onset of colitis. Mice prophylactically treated with paquinimod exhibited detectable but mitigated signs of colitis, as determined by a reduced trend in weight loss, reduced colonic shortening and inflammation in the colons compared to DSS-only treated mice, although most mice still had detectable blood in the stool (Fig. 5A–D). Consistent with the reduced colonic inflammatory cytokines and chemokines, we also measured reduced levels of S100A8/A9 in the colons of DSS-treated mice (Fig. 5E). Paquinimod treatment also prevented the upregulation of Lcn2 and S100A8/A9 in the brain in DSS-treated mice (Fig. 2F–I). These data indicate that prophylactic administration of paquinimod ameliorated colonic and neuroinflammation in mice treated with DSS.

Inflammatory manifestations outside of the intestines are common in patients with IBD, and typically coincide with colonic inflammation [5]. To identify extraintestinal sites of inflammation during ongoing colitis, we took advantage of a reporter mouse that becomes bioluminescent at sites of inflammasome/caspase activation. As expected, we detected colonic inflammation and clinical signs of colitis in mice treated with DSS for 7 days (Fig. 1). We also detected significant biosensor activation in the brains of mice with colitis (Fig. 2). It is important to note that while we have demonstrated that this reporter becomes activated in sites of inflammation, it may also be activated by other inflammatory and apoptotic caspases in addition to caspase-1 [18, 48]. Consistent with increased biosensor activation, we detected increased expression of caspase-1 in brains isolated from DSS and LPS-treated mice (Fig. 2C). The increased expression of p45 is likely indicative of increased upregulation of procaspase-1, triggered by the activation of inflammatory signaling cascades, while the increased expression of p20 is a downstream indicator of inflammasome activation. The presence of cleaved caspase-1 together with biosensor activation in the brains of mice treated with DSS is a strong indicator that an inflammasome(s) is active in this tissue during colitis. Identifying the inflammasome-forming receptor(s) that is activated in response to DSS treatment and the cell type or brain region where the inflammasome is activated requires further investigation.

We also detected cleaved caspase-8 in the brains of some DSS- and LPS-treated mice (Additional file 2). Caspase-8 activation in microglia has been reported previously in vitro in response to challenge with LPS and other inflammatory stimuli [49], as well as in microglia in brain tissue from patients with multiple sclerosis (MS) [50]. In addition to its canonical role as a driver of apoptosis, caspase-8 has recently emerged as a critical player in inflammatory responses. Like caspase-1, active caspase-8 cleaves inflammatory cytokines IL-1β, IL-18, and can be recruited into inflammasome complexes, particularly in cells where caspase-1 is absent or deficient, driving cell death [51‐57]. Microglia specifically have been shown to process IL-1β by a NLRP3–ASC–caspase-8-dependent mechanism [50]. While we were unable to detect cleavage of caspase-3 in brains from either group at this time point, we cannot rule out the possibility that this caspase is activated at low levels below the limit of detection in our assay or that caspase-3 may ultimately become activated at later time points. In support of this, one group found increased caspase-3 in the hippocampus of mice treated with 5% DSS for 7 days [29]. Further experiments should be performed to delineate the temporal kinetics of caspase activation in the CNS in mice with colitis.

To confirm this inflammatory phenotype, we measured the upregulation of specific inflammatory transcripts by qRT-PCR and found increased IL-1β, KC and TNF-α particularly in the hindbrain region (Fig. 2D), which is the region of the brain that is reproducibly bioluminescent in DSS-treated mice. Proteome mouse cytokine array and CBA also revealed increased expression of a number of inflammatory cytokines and chemokines, most notably KC (Fig. 2E–G), which was also significantly elevated in the colon (Fig. 1D) and serum (Fig. 4B). We next took an unbiased approach to identify differentially expressed genes in the brains of mice with colitis. Pathway analysis of the RNA sequencing data revealed that the most significantly upregulated pathway was the regulation of inflammatory response pathway, followed by IL-17 signaling, arginine biosynthesis and antimicrobial peptides pathways (Fig. 3). Inflammatory and antimicrobial pathways are intimately linked, so it is not surprising that both of these pathways were upregulated in our model. We confirmed upregulation of three proteins identified in three of the pathways, S100A8, S100A9 and Lcn2, all of which are established biomarkers for inflammation. Other groups have previously demonstrated that S100A9 [58, 59] and Lcn2 [22, 60, 61] are upregulated in the colon of mice with colitis. However, the finding that they are also increased in the brain in mice with colitis is novel.

Inflammatory responses can occur in diverse cell types in the CNS, though neuroinflammation has mainly been attributed to microglia. LPS-induced microglia activation in the CNS is well supported (reviewed in [62]), but whether colitis drives microglia activation in the rodent brain remains debated [29, 63‐65]. Our data revealed that brains isolated from DSS-treated mice show an increase in Iba1 immunoreactivity and signal distribution (Fig. 2I). Since Iba1 is a marker of both resting and activated microglia, the increased signal most likely represents a change in microglia morphology and localization rather than an increase in cell number. The Iba1 signal was even more evident in brains isolated from mice 24 h after LPS injection and in some cases, appeared to localize around the vasculature (Fig. 2I). While light sheet microscopy is not the best system for quantification, the increased Iba1 signal in both DSS and LPS brains is visually apparent and consistent with increased microglial reactivity [66]. Neuroinflammation can also result from disruption of the BBB and infiltration of immune cells into the nervous system, which drive pathological sequelae in a number of disease models. There is some evidence for BBB permeability in trinitrobenzene sulfonic acid (TNBS)-treated rats/rabbits [67, 68] and mild BBB permeability in DSS-treated mice [29, 69]. We therefore wanted to investigate whether peripheral immune cells infiltrated the CNS in mice with colitis. We measured a modest but significant increase in inflammatory monocytes and neutrophils in the brains from mice treated with DSS (Fig. 2H). These data suggest that in addition to resident glia cells, innate immune cells may be contributing to neuroinflammation in mice with colitis.

It is worth noting that although our data demonstrate that proinflammatory pathways are upregulated in the CNS during colitis, this proinflammatory signature appears to be transient and moderate. Our RNA sequencing data that identified upregulation of inflammatory pathways 7 days after DSS exposure also revealed that these genes returned to baseline by 9 days following DSS cessation, indicating that these inflammatory responses resolve upon the resolution of colitis. We also included mice injected systemically with LPS as a positive control for neuroinflammation induced by a peripheral stimulus in our experiments to provide scale. While the proinflammatory phenotype is significantly and reproducibly detected by multiple measurements of inflammation in the brains of DSS-treated mice, it is a mild-to-moderate inflammatory response relative to that detected in the brains of LPS-treated mice (Figs. 2, 3, 4, Additional file 3, Additional file 4).

Mechanistically, how colonic inflammation can trigger neuroinflammation has yet to be elucidated. Colonic inflammation likely drives neuroinflammation either by a directed gut–brain signaling axis, potentially mediated by the gut microbiota or afferent neurons, or inflammation in the colon could trigger systemic inflammation that subsequently activates inflammatory pathways in the CNS. These two pathways may not be mutually exclusive. Our data demonstrate a significant increase in the inflammatory chemokine, KC, as well as inflammatory biomarkers, S100A8 and S100A9, in the colons, serum and brains of mice with colitis (Figs. 1, 2, 3, 4, 5). As mentioned above, colitis-induced extraintestinal inflammation is well documented in mice and in humans. This, taken together with the increase in inflammatory markers in both the serum and the brain argue in favor of a model, whereby colitis drives a global inflammatory response that culminates in inflammation in the CNS. More mechanistic insight into how gut-driven inflammation contributes to neuroinflammation is needed.

There has been recent interest in S100A8 and S100A9 because they are found at such high concentrations at sites of inflammation. S100A8 and S100A9 make up nearly half of the total cytoplasmic proteins in neutrophils and are significantly upregulated by macrophages in response to inflammatory stimuli. Once released into the extracellular environment, they activate immune cells and vascular endothelial cells, perpetuating inflammation in tissues. Intriguingly, upregulation of these proteins does not appear to be specific for a particular inflammatory disease, as S100A8 and S100A9 have been implicated in a number of seemingly unrelated inflammatory conditions, including rheumatoid arthritis [70‐74], CD [27, 28], systemic lupus erythematosus [75‐78], cardiovascular diseases [79‐81], psoriasis [82, 83], MS [84, 85], and others. More recently, S100A8/A9 was found to be upregulated in patients with COVID-19 and expression correlated with disease severity [86, 87].

Paquinimod is an orally active immunomodulatory quinoline-3-carboxamide derivative that binds to S100A9 (and S100A12 [88]), and prevents its interaction with TLR4 and RAGE receptors to prevent inflammatory signaling [47]. Paquinimod and related Q-compounds have been tested in clinical trials for a number of human autoimmune and inflammatory diseases, including CD [44‐46, 89], and paquinimod specifically has been shown to be protective in mouse models of infection and autoimmune-induced neuroinflammation [32, 35] as well as depression [31]. Because we detected S100A8/A9 upregulation in the serum and brains in mice with colitis (Figs. 3, 4), we hypothesized that inhibition of S100A9-signaling would reduce the neuroinflammatory signature in mice with colitis. To determine the effect of S100A9 inhibition on colitis-induced inflammation, mice were prophylactically treated with paquinimod 1 week prior to and during DSS treatment. Paquinimod-treated mice exhibited reduced inflammatory cytokines and chemokines in the colon and reduced inflammatory biomarkers in the brain (Fig. 5). Paquinimod and DSS-treated mice also exhibited detectable but improved clinical signs of colitis (Fig. 5). These data are supported by the previous finding that treating mice with a neutralizing anti-S100A9 antibody prevented DSS-induced colitis and colitis-associated cancer in mice [58].

The consequence of colitis-induced neuroinflammation on behavior and cognition was not addressed in this work. Previous studies have demonstrated memory impairment, anxiety-like behavior, and altered responses to stress in mouse models of DSS-induced colitis [30, 90‐93], which return to normal as acute colitis resolves [92, 93]. Systemic LPS administration has been shown to drive hypothalamic, hippocampal and cortical proinflammatory cytokine expression, and elevation of proinflammatory cytokines in the brain have been linked to memory impairment and neurodegeneration [94‐99]. The cognitive deficits and increased stress/anxiety measured in DSS-treated mice and in humans may be, in part, due to this inflammatory milieu in the CNS that persists during colonic inflammation. Stress-induced upregulation of S100A8 and S100A9 has also been reported previously in mouse hippocampus [100], though it is unclear if S100A8/A9 expression in the CNS is beneficial or detrimental to stress and anxiety. In two separate mouse models, one group found that that upregulation of S100A8/A9 may be protective against anxiety-like behavior following sepsis [101], while another group found that pharmacological inhibition of S100A9 with paquinimod effectively ameliorated depressive behaviors in a model of chronic unpredictable mild stress (CUMS) [31]. It will therefore be important to assess the effect of S100A9 inhibition on DSS-induced stress and anxiety.

One caveat with this study is that broadly targeting inflammatory pathways will not only prevent neuroinflammation, but also colonic inflammation, the driver of neuroinflammation in our model. In other words, although we detected reduced systemic and neuroinflammation in paquinimod-treated mice, this could be because most mice never developed significant colitis. It will be important to determine if therapeutic administration of paquinimod mitigates systemic and neuroinflammation in mice that have ongoing colitis. However, we were unable to test this, because mice with colitis often stop drinking their water, meaning that drugs given ad libitium may not be taken at the proper dose, if at all. An oral gavage route of administration could be tested in the future but was avoided in this study since this approach can induce stress and weight loss, which would likely affect measurements of neuroinflammation and colitis clinical scores. In humans, however, drugs that target S100A9 could be given after colitis onset as soon as patients become symptomatic. It would be interesting to test such therapeutics to determine if they alleviate colitis and other inflammatory sequelae in patients with IBD.

In conclusion, this study found that colitis drives mild to moderate inflammation in the brain and increased inflammatory cytokines in the serum, and that colonic and neuroinflammation were mitigated by prophylactic treatment with the S100A9 inhibitor paquinimod.