Ulcerative colitis: Recent advances in the understanding of disease pathogenesis

Genetics

Genetic studies (including genome-wide association [GWA], whole genome sequencing [WGS], and fine mapping studies) have been particularly successful in identifying 260 susceptibility loci (both common and rare genetic variants) associated with IBD^10–14. There are several key findings. Firstly, most genetic factors are shared between UC and CD. In an initial analysis of 15 GWA datasets, Jostins et al. showed that 110 out of 163 (67%) susceptibility loci were associated with both UC and CD¹¹. These shared genes encode both innate and adaptive immune pathways, cytokine signalling, and immune sensing (e.g. IL23-R, IL-12, JAK2, CARD9, TNFSF18, and IL-10). Many of these genes (70%) are also shared with other autoimmune diseases such as ankylosing spondylitis and psoriasis. Secondly, the strongest genetic signals within UC-specific loci are associated with the human leukocyte antigen (HLA) region in chromosome 6. Sixteen HLA allelic associations (mostly class II) are described for UC, including HLA DRB1*01*03 for IBD colonic involvement on deeper fine mapping genetic analysis¹⁵. Further analyses show that these are associated with colonic involvement for UC and CD¹⁶. It is of interest to note that HLA allelic associations with extensive and aggressive UC have been noted even prior to GWA studies¹⁷. Recent WGS of nearly 2,000 UC patients identified a new but rare missense variant (present in 0.6% of cases) in the adenylate cyclase 7 gene (ADCY7) that doubles the risk of UC¹². Outwith the HLA region, the ADCY7 gene has the strongest genetic association observed with UC. ADCY7 is one of a family of 10 enzymes that convert ATP to the ubiquitous second messenger cAMP. In addition to this, many UC-specific genes are involved in the regulation of epithelial barrier function (further discussed below). Thirdly, despite the identification of many susceptibility loci, genetics explain only 19% of disease heritability in UC¹⁸. The concordance rate amongst monozygotic twins for UC is only 6.3% (compared to nearly 60% in CD). Collectively, genetic factors confer a small but definite increase in susceptibility for UC. Many individuals, however, have no genetic predisposition when assessed by a polygenic risk score that accounts for all of the susceptibility loci¹⁹. This suggests a key role for aberrant adaptive immune responses and epithelial barrier dysfunction in UC disease pathogenesis. Non-genetic factors (notably epigenetics^20,21) may also play an important role.

Environmental factors

The rapid rise of UC incidence in newly industrialised countries suggests that environmental factors are important¹. This parallels the patterns observed in the Western world during the early 20^th century. Specifically, UC appears first in urban areas, its incidence rising rapidly then slowing; subsequently, CD incidence rises and eventually approaches that of UC²². Westernisation is accompanied by new urban lifestyle, exposure to pollution, change in diet, access to antibiotics, better hygiene, and fewer infections, all considered as general contributory factors²³. Notwithstanding this, more specific environmental factors associated with UC have been known for some time. The strongest example is seen in the protective effect of cigarette smoking and the notable observation of new-onset UC in individuals who stop smoking. The global patterns of smoking and IBD are changing; an increasingly large former smoker population with UC in China is suggestive of a rapid expansion of the at-risk population²⁴. The anti-inflammatory effect conferred by cigarette smoking in UC is intriguing and may be mediated by carbon monoxide²⁵. Further examples include the protective effect of appendicitis against future development of UC^26,27, the bimodal incidence with a second peak associated with older age in men²⁸, and, more recently, the curious association with Parkinson’s disease (another condition associated with non-smoking and old age)^29,30. These all provide more specific aetiological insights into the development of UC. Epidemiologic data have shown a potential protective effect of high dietary n-3 polyunsaturated fatty acids (PUFAs), present in oily fish³¹, and a diet high in red meat in the development of UC^32–34.

Gut microbiota

The IBD gut microbiome is significantly less diverse and stable over time, as recently extensively characterised in the Integrative Human Microbiome Project ([iHMP], where 132 IBD and healthy individuals were followed up longitudinally for 1 year)³⁵ and demonstrated in a case-control study involving 1,800 IBD and irritable bowel syndrome patients³⁶. A depletion of protective bacteria such as short-chain fatty acid (SCFA)-producing Ruminococcaceae and Lachnospiraceae that coincides with an expansion of pro-inflammatory microbes such as Enterobacteriaceae, including Escherichia coli, and Fusobacteriaceae has been noted^37,38. These changes, however, are less obvious in UC compared to CD³⁹. It is not known if dysbiosis is a consequence of, or plays a causal role in, gut inflammation in UC. In this regard, the virome and mycobiome are also less diverse in UC^40–43. In the longitudinal iHMP, microbiome patterns did not predict the likelihood of a disease flare. To add to the complexity, a further study in UC showed that microbial abundance did not necessarily correlate with transcriptional activity⁴⁴. Therapeutically, however, faecal microbial transplantation (FMT) from healthy donors can treat UC. There are four controlled positive FMT clinical studies^45–49. The restoration of microbial diversity, including bacterial species responsible for SCFA production in donor stool, has been suggested as an important contributor^46,50. Hence, one of the main effects of dysbiosis in UC is likely to be a reduction in epithelial health or a state of epithelial dysfunction that further primes innate susceptibility to UC. In support of this, faecal diversion away from the rectum worsens inflammation, giving rise to “diversion colitis” in UC; the opposite is true for CD, where faecal diversion improves inflammation⁵¹.

Epithelial dysfunction

With the histologic observation of subepithelial inflammation, many studies implicate an impaired epithelial barrier as a pathogenic factor for UC. This is through either altered or impaired secretion (e.g. of antimicrobial peptides, damage-associated molecular patterns, or mucus) or physical defects (e.g. from disruption of epithelial tight junctions or defective regeneration or detoxification) (Text box 1)^52,53. GWA studies show UC-specific susceptibility genes that regulate epithelial morphogenesis (hepatocyte nuclear factor 4 α, Hnf4α⁵⁴), adherens junction stability via E-cadherin (CDH-1), basement membrane anchoring and stability (via laminins, LAMB-1, and extracellular matrix, ECM1), tight junction assembly (guanine nucleotide binding protein alpha 12, GNA12), ion transport (solute carrier family-26, SLC26A3)⁵⁵, and epithelial health via endoplasmic reticulum stress (orsomucoid-1-like gene 3, ORMDL3)⁵⁶. Of interest, a protein truncating genetic variant in RNF186, a single-exon ring finger E3 ligase with strong colonic epithelial expression, protects against UC; however, the underlying mechanism is not yet clear^14,57. Hence, there is a potentially innate “at risk” phenotype where exposure to additional injurious stimuli such as non-steroidal anti-inflammatories⁵⁸ (that reduce the synthesis of protective prostaglandins) and dietary components such as emulsifiers (that reduce the thickness of the mucus layer)⁵⁹ may be the second trigger that precipitates colitis. As discussed earlier, dysbiosis results in loss of SCFA production³⁵, which is essential for epithelial energy provision, mucus production, and proliferation in the colon. Hence, clinical trials involving butyrate⁶⁰, propionic acid⁶¹, prebiotics^62–66, and L-carnitine⁶¹, which facilitate SCFA transport, have demonstrated some efficacy in treating UC⁶⁷. During active UC, key pro-inflammatory cytokines such as tumour necrosis factor-alpha (TNF-α), interferon (IFN)-γ, and interleukin (IL)-13 have direct deleterious effects on epithelial barrier integrity^68,69. Drugs that maintain remission in UC, such as mesalazine, may exert some of their therapeutic effect by maintaining epithelial health⁶⁷. Hence, protecting the “at risk” or restoring colonic epithelial health may be a viable strategy to maintain long-term remission in UC.

Abnormal immune response: innate

In active UC, there is a complex inflammatory milieu of innate and adaptive immune cells infiltrating the lamina propria. Neutrophils, the short-lived “first responder” cells, are recruited in abundance with characteristic histology of “crypt abscesses” in UC, where neutrophils transmigrate across the colonic epithelium and die within the colonic crypts⁷⁰. The UC inflammatory environment promotes neutrophil survival (potentially via HIF-1 and hypoxia)^71,72. This increased survival escalates its inflammatory action and tissue damage (via many means, including the release of serine and matrix metalloproteases, reactive oxygen species, and pro-inflammatory cytokines)⁷³. The high number of neutrophils undergo uncontrolled pro-inflammatory cell death (necrosis, necroptosis, and NETosis), which potentiates and amplifies the pro-inflammatory environment^74,75. This is supported by high levels of s100a8/9 proteins (or calprotectin), usually found in neutrophils, that are released in blood and stool^76–78 and a prominent serological response to self perinuclear anti p-neutrophil cytoplasmic antibodies (pANCA) in UC, both likely indirect indicators of uncontrolled neutrophil cell death⁷⁹. Neutrophil extracellular traps (NETs) can act as a sump for immunogenic molecules that sustain the inflammatory response⁷⁵. Hence, there is a rational paradigm that, following disease initiation, the preceding wave of innate inflammatory neutrophils and monocytes (with their pro-inflammatory cytokine repertoire, e.g. IL-1 family, IL-6, and TNF-α) creates an inflammatory milieu (nutritional, metabolic, and cytokine) that promotes a pathologic adaptive (likely T-cell) immune response⁸⁰. Such a milieu can also shape newly arriving inflammatory monocytes, monocyte–macrophage function, their survival, and their phenotype, and further factors that influence the host’s ability to resolve inflammation, restore homeostasis, and repair the UC mucosa^81,82.

Abnormal immune response: adaptive

UC’s strong genetic associations with HLA (mostly class II) suggest that abnormal antigen(s) driving the aberrant T-cell response, which then further shape the pathologic cytokine milieu, are likely to be a crucial causative factor. How HLA influences commensal and/or self antigen presentation (and the identities of these) to T cells and thereafter downstream pathogenic T-cell response remains unclear and challenging. Approaches to study, screen, and define T-cell epitopes have improved considerably and progress is likely⁸³. Traditionally, UC is associated with a Th2 response with high IL-4, IL-5, and IL-13, whereas CD has a more dominant Th1/Th17 response⁸⁴. Earlier studies that show less IL-4 in UC, with CD1d-restricted natural killer T-cells producing IL-13, point to a non-classical Th2 response⁸⁵. Some recent developments have overtaken this area. These include the identification of IL-23 as a key driver of Th17 responses⁸⁶, genetic associations with IL-23 and its related genes^11,87, and the presence of Th17⁸⁸ (and Th9⁸⁹) cells in UC. The Th2 angle becomes less clear where anrukinzumab (a drug that blocks IL-13 by binding with IL-4Ra, a shared subunit for IL-13 and IL-4 receptors)⁹⁰ and tralokinumab (a drug that blocks binding to both IL-13Ra and IL-13Ra2) are not effective in UC⁹¹. Blocking IL-23, however, is effective in UC, e.g. mirikizumab (anti-p19 subunit of IL-23)⁹² and ustekinumab (anti-p40 subunit of IL-23)⁹³. The example of anti- TNF treatment first used in CD and then shown to be equally effective in UC⁹⁴ demonstrates that basing a translational approach on pure Th-cytokine profile may be oversimplified. Furthermore, although CD4 T cells are considered to be more important in IBD pathogenesis, it is CD8 T cell transcriptomic signatures that have been found to influence whether UC and CD adopt a more aggressive disease course (in this study, CD4 T signatures were not useful)⁹⁵. New data characterising the adaptive immune populations at a transcriptomic (and at a single cell) level⁹⁶ will yield many more new insights. The recent discovery of innate lymphoid cells (ILCs)^97,98 as a further mediator of IL-23-driven inflammatory response in the colon⁹⁹ is a further new dimension in UC.

The mitochondria and UC

Recent progress has been driven by a strong focus on direct studies on the inflamed mucosa specifically in newly diagnosed or drug-naïve individuals^38,100,101. Of interest, using a bulk RNAseq approach in 206 newly diagnosed paediatric UC individuals (PROTECT study), Haberman et al. showed a significant reduction in the expression of mitochondrial genes that encode the oxidative phosphorylation chain (responsible for energy production) and nuclear encoded genes such as PPARGC1A (responsible for mitochondrial biogenesis), implicating mitochondriopathy as a pathogenic process in UC¹⁰⁰. Mitochondria are intracellular double-membrane-bound organelles with many essential physiological roles such as in energy production and the regulation of cell death and immune responses¹⁰². In the last 10 years, many seminal studies have highlighted the mitochondria as the major previously unknown “jigsaw piece” in inflammation¹⁰³. Mitochondrial dysfunction has long been implicated in UC, as far back as 1980^104,105 (reviewed by Novak et al.¹⁰⁶), but new data from the last 3 years have re-focused this concept^100,107,108. Such dysregulation of genes that control mitochondrial function have been shown in earlier colonic microarray studies in UC¹⁰⁹.

Functional studies show that mitochondria are sited in a uniquely damaging environment (in the colon, more so than other tissue sites)^107,110. Loss of mitochondrial homeostasis (including mitophagy and the autophagic removal of damaged mitochondria—IBD GWAS susceptibility genes PARK7 and LRRK2) can lead to defective energy production¹¹¹, increased mitochondrial oxidative stress¹⁰⁷, and even the release of mitochondrial products (mitochondrial DNA) as pro-inflammatory DAMPs^108,112. These lines of evidence contribute to key UC themes such as epithelial dysfunction, the pro-inflammatory mucosal milieu, and direct triggers of the inflammatory response. Such convergence of data has culminated in new approaches in targeting the pro-inflammatory mitochondria, for example mitochondrial anti-oxidant therapy in active UC.

Single cell profiling of the inflamed UC mucosa

Single cell RNA sequencing (scRNA) technology was developed in 2009 before becoming more widely available in 2014. It provides a comprehensive analysis and census of the cell populations (“who is all there?”) in a complex inflamed UC mucosal milieu¹¹³. In UC, three recent scRNA studies (Parikh et al.¹¹⁴, Smillie et al.⁹⁶, and Kinchen et al.¹¹⁵—scRNA analyses on colonic epithelium, whole layer, and mesenchyme, respectively) have provided some compelling insights^96,114. These studies have identified new and rare cell types, unique cell-type-specific expression, and deep cell–cell interaction and cell lineage relationships. Secondly, mucosal compartments that have previously received less attention—notably, the colonic mesenchyme—are now implicated as key mediators of inflammation¹¹⁶. Thirdly, they show entirely new disease angles and have unexpectedly reinvigorated some older mechanistic theories. We highlight the key findings below.

Colonic epithelium: novel cell population and cell-specific changes. A main question is whether there are specific subsets of colonic epithelial cells that display intrinsic molecular pathology that can be pathogenic drivers in UC. Both scRNA studies identified a previously unknown epithelial cell population characterised by distinct expression of the calcium-sensitive chloride channel bestrophin-4 (BEST4), the protease cathepsin E, and the OTOP2 gene. Intriguingly, this colonocyte likely has the ability to sense pH in the luminal environment and to set the epithelial cGMP tone in response. Smillie et al. showed that BEST4⁺ enterocytes are distinct from epithelial cells and they are also enriched in genes including otopetrins 2 and 3 (OTOP2/3), proton channels that detect pH and underlie sour taste perception, and carbonic anhydrase VII (CA7). In another novel finding, Parikh et al. demonstrated a positional remodelling of goblet cells that coincides with downregulation of WFDC2, an anti-protease molecule that is expressed by goblet cells and inhibits bacterial growth. In vivo, WFDC2 preserves the integrity of tight junctions between epithelial cells and prevents invasion by commensal bacteria and mucosal inflammation. WFCD2 has been proposed as a regulator of innate immunity through inhibition of serine and cysteine proteases¹¹⁷.

Colonic epithelium: intrinsic changes associated with UC inflamed and non-inflamed mucosa. The sharp demarcation between inflamed and non-inflamed UC mucosa in the colon provides the unique opportunity for scRNA approaches to find distinct changes that may explain this transition from normal to affected mucosa. Interestingly, both areas exhibit many similar dysregulated gene expressions. This suggests a role for mucosal epigenetics: the transcriptional signature of UC precedes inflammation, arises as the result of a dominance of regenerative over damage cues or even as a protective mechanism in anticipation of damage, and persists after resolution. All epithelial subtypes in the inflamed UC mucosa showed upregulation of several inflammatory pathways, notably IFN-γ signalling and cytokine production. Epithelial cells downregulated metabolic processes and induced genes that are needed to produce reactive oxygen species and for microbial killing. Absorptive and secretory progenitor cells upregulated differentiation and cell migration pathways, which suggests an active attempt to repair colitis-induced damage.

Colonic immune cell population: dominant functional cellular hubs. In Smillie et al.’s study that explored the overall colonic immune cell population, cell–cell interaction analyses in the inflamed UC mucosa showed strong compartmentalisation around inflammatory monocytes and novel network hubs around the poorly characterised CD8⁺IL17⁺ T cells and microfold-like (M) cells that are usually rarely found in the healthy colon. CD8⁺IL17⁺ T cells induce IL17A/F, IL23R, and cytotoxic, co-stimulatory, and co-inhibitory programs in UC. M cells are typically associated with lymphoid tissue in the human small intestine, where they are important for recognition of the gut microbiota but are rarely found in the healthy colon¹¹⁸. A further striking cell–cell interaction hub is centred on a mesenchymal subset of inflammation-associated fibroblasts (IAFs)⁹⁶. In some UC patients, IAFs are expanded nearly 190-fold and enriched with many genes associated with colitis, fibrosis, and cancer (including IL13RA2).

Colonic mesenchyme: a newly identified inflammatory component contributing to an anti-tumour necrosis factor response. In the mesenchyme-focused scRNA study, Kinchen et al. identified a distinct activated mesenchymal cell population that expressed TNF superfamily member 14 (TNFSF14), fibroblastic reticular cell-associated genes, IL-33, CCL19, and lysyl oxidases¹¹⁵. One of the most enriched genes in IAFs is oncostatin M (OSM), a putative risk gene, and its receptor OSMR¹⁰. In an earlier study^38,119, West and colleagues identified significant overexpression of OSM in inflamed IBD mucosa¹¹⁶. OSM is part of the IL-6 cytokine family that can induce JAK-STAT, phosphatidylinositol-3-kinase (PI3K), and mitogen-activated protein kinase (MAPK) downstream signalling pathways. Further characterisation showed that OSMR is highly expressed in the mesenchyme (as later also shown to be the case). Using UC clinical trial datasets on anti-TNF treatment (infliximab and golimumab), high mucosal OSM expression is associated with poor response to anti-TNF^120,121.

Future insights from scRNA studies. These recent studies provide a vast “library reference” level amount of data that the IBD research community is only beginning to assimilate and understand. Tantalising new discoveries such as epithelial pH sensing, the roles of new enterocytes marked by BEST4⁺, and colonic anti-bacterial responses mediated by WFDC2 and CD8⁺IL17⁺ T cells will require more detailed studies. These are early days of moving from census to understanding function and biology. Other leads such as OSMR blockade and CCL9 inhibition are nearer to translation as potential therapeutic targets. The International Human Gut Atlas Project (https://helmsleytrust.org/rfa/gut-cell-atlas) will generate an even larger compendium of scRNA data in the next 5 years. Rationalising these enormous data (with other -omics datasets, e.g. genetics and microbiome), or, in lay-terms, how we combine the knowledge of “what and where are the cells?” with “what genes?” and “what bacteria?”, will be both challenging and exciting¹²².