Human-Gut Bacterial Protein-Protein Interactions: Understudied but Impactful to Human Health

Abstract

The human gut microbiome is associated with a wide range of diseases, yet the mechanisms these microbes use to influence human health are not fully understood. Protein-protein interactions (PPIs) are increasingly identified as a potential mechanism by which gut microbiota influence their human hosts. Similar to some PPIs observed in pathogens, many disease-relevant human-gut bacterial PPIs function by interacting with components of the immune system or the gut barrier. Here, we highlight recent advances in these two areas. It is our opinion that there is a vastly unexplored network of human-gut bacterial PPIs that contribute to the prevention or pathogenesis of various diseases, and that future research is warranted to expand PPI discovery.

A paradigm shift for studying human-bacterial interactions

The gut microbiome is associated with a wide range of human diseases, from intestinal diseases like colorectal cancer (CRC) [1] and inflammatory bowel disease (IBD) [2] to extra-intestinal diseases like diabetes mellitus [3] and Alzheimer’s disease [4]. Despite the profound modulatory effects that the microbiome has on human health [5], our understanding of the biomolecular mechanisms that microbes use to influence host physiology is incomplete. Much of what is currently known about the gut microbiome’s influence on host physiology relates to small molecules like short-chain fatty acids and secondary bile acids [6–8], and there has been relatively less focus on the potential of modulatory proteins. Each year, a growing number of bacterial proteins are found to directly interact with human proteins and contribute to disease (Figure 1). However, the attention on bacterial proteins, and more specifically, human-bacterial PPIs, is rooted in pathogenicity studies of several well-studied bacteria, such as Mycobacterium tuberculosis [9] and Salmonella enterica serovar Typhimurium [10]. In contrast, only a few PPIs have thus far been identified between gut bacteria and their human hosts.

Figure 1:

Human-bacteria protein-protein interaction (PPI) publications and discovered PPIs increased substantially over the past two decades.

Information on PPIs between humans and bacteria was collected from three publicly available databases: A few additional PPIs were manually collected. The publications each PPI originated from were determined using NCBI Entrez, and the year was noted for each PPI and relevant publication. PPI data was collected in August 2022. Abbreviations: PPI, protein-protein interaction.

Some host-microbe PPIs involve bacterial proteins known as microbe-associated molecular patterns (MAMPs) containing highly conserved molecular motifs recognized by pattern recognition receptors (PRRs) that are part of the host’s innate immune system. The history of how the term MAMP arose is indicative of the growing appreciation for human-bacterial interactions. Coined by a pioneering immunologist, Charles Janeway, in 1989, the concept of immune-detectable motifs in bacteria was originally termed pathogen-associated molecular pattern (PAMP) [11]. However, over a decade later, the field shifted to the term MAMP to encompass the many commensals that also harbor detectable protein motifs like flagellin and non-protein motifs like lipopolysaccharide (LPS) [12]. Mirroring the historical shift of PAMP to MAMP, human-bacterial PPIs, once primarily considered a pathogen-specific phenomenon, are now including human-associated microbiota that are implicated in various diseases, playing both protective and pathogenic roles. Their effects are not limited to PRR-MAMP signaling and include other surface receptor engagements, as well as secreted proteins with internal targets. Moreover, gut bacteria and their products may travel beyond the gut into more distal tissues, like the pancreas and joints [13,14], increasing their potential for influencing various aspects of host physiology. Here, we discuss the growing recognition of human-gut bacterial PPIs, focusing on two pathways in which they have been more deeply studied: immune signaling and gut barrier regulation.

Gut bacterial proteins directly interact with immune system components

One way many microbes, both pathogenic and commensal, interact with their human hosts is through modulation of the immune system (Figure 2). Unsurprisingly, the most well-known example of an immune-modulating human-bacterial PPI involves bacterial flagellin, a MAMP, and human Toll-like receptor 5 (TLR5), a PRR. When flagellin is recognized by TLR5, a signaling cascade is triggered which activates NF-κB—a pro-inflammatory transcription factor that controls the expression of various inflammatory cytokine genes—ultimately leading to an inflammatory anti-bacterial response [15]. Although this finding dates back over two decades, recent work on flagellin variants encoded by different gut bacterial species shows that some flagellins have evolved to avoid TLR5 detection altogether—termed “evaders” [16], or to strongly bind to TLR5 without eliciting an inflammatory response. The latter phenomenon, termed “silent recognition”, may help to explain how the immune system tolerates commensal flagellin while still remaining responsive to various pathogenic flagellins in the gut [17]. However, among the host-microbiome PPIs that have been identified to date, no other involves a protein as taxonomically wide-spread as flagellin.

Figure 2:

Gut bacterial proteins are immune-modulatory and target various host proteins.

Flagellin from various gut bacteria (in blue) binds to TLR5 on gut epithelial cells, triggering the activation of pro-inflammatory NF-κB and the subsequent increase in inflammation. SlpA from L. acidophilus NCFM (in yellow), targets CD209 on dendritic cells, inducing anti-inflammatory IL-10 in the gut. MAM from F. prausnitzii (in green) enters epithelial cells and reduces inflammation by blocking NF-κB activity, though the specific human protein interactor is unidentified. Fap2 from F. nucleatum (in purple) specifically targets colon tumor tissue and binds to TIGIT on T cells and NK cells, inhibiting their anti-tumor cytotoxicity. Abbreviations: TLR5, Toll-like receptor 5; TIGIT, inhibitory receptor T cell immunoglobulin and ITIM domain.

Several species-specific effector proteins have been shown to bind to and manipulate members of the NF-κB pathway, downstream of various PRRs. As expected, some pathogens reduce NF-κB activity, impairing the hosts’ inflammatory defenses. Shigella flexneri’s protein OspJ, for example, prevents the ubiquitylation and degradation of the NF-κB inhibitory protein IkB, leading to prolonged NF-κB inhibition and reduced inflammation [18]. Conversely, Listeria monocytogenes induces inflammation; its Listeria adhesion protein (LAP) activates NF-κB and the subsequent inflammation increases the permeability of the gut barrier, allowing it to translocate into other host tissues [19]. Similar to these pathogens, some gut bacteria also modulate the NF-κB inflammatory response. Microbial anti-inflammatory molecule (MAM) is a secreted NF-κB protein inhibitor made by Faecalibacterium prauznitzii, a gut commensal inversely associated with multiple diseases including two types of IBD (Crohn’s disease and ulcerative colitis), type-II diabetes mellitus, and CRC [20]. What is most striking is that this bacterial protein has a cytoplasmic target and simply exposing Caco2 intestinal cells grown in culture to MAM is sufficient to reduce NF-κB activation, suggesting that secreted MAM enters epithelial cells and interrupts NF-κB signaling directly. Though the specific human protein interactor has not been confirmed, MAM colocalizes with IκκB, a regulatory kinase in the NF-κB pathway, suggesting protein interaction [21]. More research is needed to determine whether reducing inflammation via PPIs with components of the NF-κB pathway is common for gut bacteria.

Probiotics, a subset of ‘generally regarded as safe’ to consume organisms commonly found in fermented foods, have also been studied for their anti-inflammatory effects [22], a phenomenon that may be due in part to PPIs. For example, S layer protein A (SlpA) from Lactobacillus acidophilus NCFM, a probiotic often used in fermented foods and dietary supplements [23], helps promote gut immune cell homeostasis by binding directly to dendritic cell-specific ICAM-3-grabbing nonintegrin (DC-SIGN) (also known as CD209), increasing the expression of anti-inflammatory cytokine IL-10 [24]. Similar to the NF-κB pathway, pathogens also have proteins that induce IL-10 production and promote the development of chronic infections. The protein PPE18 from the pathogen M. tuberculosis, for example, interacts with Toll-like receptor 2 (TLR2) on macrophages and activates p38 MAPK, a kinase critical for IL-10 induction [25]. Aside from proteins, many gut bacteria also produce small molecules like short-chain fatty acids (SCFAs) that induce IL-10 through G-protein coupled receptor 43 (GPR43), exemplifying how gut bacteria are already known to regulate IL-10 and protect against excessive intestinal inflammation [26]. Since defects in IL-10 production and signaling are associated with inflammatory diseases like IBD and some autoimmune diseases, and gut bacteria are already known to play important roles in maintaining gut homeostasis through other means, research is warranted to further explore whether additional gut bacterial proteins modulate IL-10 levels in the gut [27].

Bacterial proteins can also impact the immune system’s response to tumors. Fusobacterium nucleatum, a common oral bacterium that is frequently detected in the gut microbiomes of CRC patients, particularly those with adenocarcinomas, inhibits tumor cytotoxicity via its protein Fap2. Fap2 binds to and activates inhibitory receptor T cell immunoglobulin and ITIM domain (TIGIT) on natural killer (NK) cells and T cells, inhibiting their anti-tumor cytotoxicity and promoting tumor immune evasion [28]. Interestingly, since Fap2 is also a carbohydrate-binding protein, or lectin, that specifically binds to the glycan Gal-GalNAc, it is likely that the interaction between Fap2 and TIGIT is mediated through these bound glycans, providing an additional mechanism for host-bacterial protein interaction. A second F. nucleatum protein, FadA, promotes tumorigenesis in an orthogonal way. FadA mediates binding to E-cadherin on CRC cells, mediating F. nucleatum attachment and internalization, which consequently stimulates CRC cell proliferation through the activation of the Wnt/β-catenin pathway, a signaling pathway central in tumorigenesis [29,30]. An important note here is that tumor-microbe interactions are not restricted to the gut. Breast and lung tumors similarly have tumor specific microbiota, suggesting that these microbes may also play some role in tumorigenesis and pins the tumor environment as an interesting place to search for additional PPIs that mediate tumor-bacterial interactions [31].

Gut bacterial proteins regulate the gut barrier

The gut barrier protects the host against various external factors, including the microbiota, and a compromised barrier permits the translocation of microbial components that can trigger systemic inflammation in the host. An increase in gut permeability, controversially termed “leaky gut”, is observed in individuals with IBD, obesity, type-II diabetes mellitus, and some autoimmune diseases, among others, and is thought to contribute to the pathogenesis of these diseases [32]. Thus far, a few disease-relevant human-gut bacterial PPIs have been identified that regulate the gut barrier (Figure 3). One such protein, Amuc_1100 from Akkermansia muciniphila, a gut commensal deficient in individuals with obesity and diabetes mellitus, has been noted for its beneficial effects on metabolic disease markers [33]. Amuc_1100, via its interaction with Toll-like receptor 2 (TLR2), can improve gut barrier integrity in obese and diabetic mice while simultaneously reducing high-fat diet-induced hypercholesteremia and fat mass gain [33]. Although TLR2 is an innate immune receptor, it also functions to regulate the expression of tight junction genes like claudin 3 and occludin that play central roles in maintaining gut barrier integrity [34]. The previously discussed protein MAM from F. prausnitzii also modifies the gut barrier by interacting with proteins in the tight junction pathway, such as Zona occludens-1 (ZO-1), increasing gut barrier integrity in a diabetes mellitus mouse model [35]. Since proteins like Amuc_1100 and MAM can ameliorate “leaky gut” in a diabetes mellitus mouse model and both A. muciniphila and F. prausnitzii abundances are reduced in individuals with diabetes mellitus versus healthy individuals [35], it is possible that reduced levels of these proteins help prevent increased intestinal permeability and associated diseases.

Figure 3:

Gut bacterial proteins impact gut barrier integrity through direct interactions with host proteins.

Amuc_1100 from A. muciniphila (in red) binds to TLR2 on epithelial cells, inducing higher expression of tight junction genes like claudin 3 and occludin that increase barrier integrity. F. prausnitzii’s MAM (in green) enters epithelial cells and targets tight junction genes like ZO-1, also increasing barrier integrity. The metalloprotease GelE from E. faecalis (in pink) binds to and degrades the adherens junction protein E-cadherin, decreasing strength of the intercellular barrier. Proteases from various gut bacteria (in grey) target and degrade components of the ECM, reducing overall gut barrier integrity. Abbreviations: TLR2, Toll-like receptor 2; ZO-1, Zona occludens-1; ECM, extracellular matrix.

Unlike MAM and Amuc_1100, various microbiota-derived proteases instead increase the permeability of the gut barrier, most of which are associated with IBD. The secreted metalloprotease gelE from Enterococcus faecalis decreases gut barrier integrity by degrading the adherens junction protein E-cadherin on epithelial cells [36], similar to the stomach pathogen Helicobacter pylori’s protease HtrA [37]. For H. pylori, degrading E-cadherin facilitates its access to the intercellular space of epithelial cells and, consequently, the basal side of the gut barrier. Interestingly, as an opportunistic pathogen, E. faecalis has also been observed to translocate across the gut barrier, possibly via proteases like gelE [38]. The extracellular matrix (ECM), an extracellular network of proteins that provides structural support and maintains the barrier between cells and the external environment, is another gut barrier PPI target. Various gut bacterial proteases, such as those from Bacteroides fragilis, degrade components of the ECM and may contribute to extracellular matrix remodeling, an integral part IBD progression [39]. Lastly, ulcerative colitis-associated proteases from the gut microbe Bacteroides vulgatus induce barrier dysfunction and worsen colitis in mice, though their specific host targets have not been identified [40].

Concluding remarks and future perspectives

Gut bacterial proteins interact with several immune and gut barrier components, and their involvement in human disease is just now being appreciated. Since gut bacteria are in constant close contact with their human hosts, primarily with the epithelium and underlying immune cells, it is logical that some of their proteins interact with these surrounding tissues. Human-gut bacterial PPIs potentially occur due to random chance and accidental affinity, existing without any selective pressures on either gut bacteria or their human hosts. However, there may be fitness benefits that contribute to the evolution of host-gut bacterial PPIs. For example, the protein Bxa, a phage-encoded bacterial ADP-ribosyltransferase from Bacteroides stercoris, binds to human non-muscle myosin II proteins and induces both actin cytoskeleton changes and inosine secretion, which thereby increases the ability of B. stercoris to colonize the gut [41]. An alternative hypothesis was recently proposed—that host-microbiome interactions developed due to a dependency on bacteria for normal human physiological functions, termed evolutionary addiction [42]. This hypothesis may help explain the presence of human-gut bacterial PPIs, specifically those involved in metabolic interdependence and intestinal and immune tissue maturation.

Some of these human-gut bacterial PPIs likely involve multifunctional proteins known as moonlighting proteins. Moonlighting often occurs within a bacterial cell, but notably, one-fourth of known moonlighting proteins are virulence factors, involving interactions with human proteins [43]. Elongation factor thermo unstable (EF-Tu), for example, is a bacterial moonlighting protein that not only transports aminoacyl-tRNAs to the ribosome to facilitate translation in the cytoplasm, but can be surface-localized or secreted where it interacts with various human proteins like fibronectin and Factor H, promoting intestinal adhesion and immune evasion [44]. However, not all bacterial species’ EF-Tu proteins are cell surface-localized or secreted. Even though many bacterial proteins may have natural affinity for certain human proteins, a fitness benefit may only be realized if the proteins are utilized in a specific environmental context.

So far, discovering new PPIs has remained low-throughput, largely directed by disease-association studies that identify bacteria enriched or depleted in disease. Disease-relevant proteins are subsequently found by performing proteomics on fractions of the bacterial supernatants or membranes that produce an effect (decreased inflammation, increased barrier integrity, etc.). In general, human-gut bacterial interactions have been difficult to study, not only due to the vast number of coexisting species found in the microbiome, but also due to the difficulty in culturing and genetically engineering many of these bacteria.

Recent technological advances are key in expanding our understanding of host-gut bacterial PPIs. High-throughput experimental approaches have emerged in recent years, though have so far only been aimed at identifying human-pathogen PPI networks. A human-SARS-CoV-2 PPI network, formed by 739 PPIs, was generated by systematically screening all pairs of SARS-CoV-2 proteins and human proteins using a high-throughput yeast two-hybrid assay [45]; however, this experimental method is only feasible for one organism with a limited number of genes. Some improvements have been made to increase the throughput of yeast two-hybrids and have been applied to human protein interactions with Bacillus anthracis, Francisella tularensis, and Yersinia pestis [46]. A human-Acinetobacter baumanii PPI network, conversely, was generated using mass spectrometry of crosslinked proteins in A. baumanii-infected lung epithelial cells [47]. While mass-spectrometry based methods are similarly promising for developing a human-gut bacteria PPI network, this method needs modification to eliminate the need for human cell infection, which is inapplicable to many gut microbes.

Computational approaches hold promise for predicting PPIs between human proteins and gut bacterial proteins which can later be experimentally verified. A recent computational study, for example, analyzes metagenomic data from different disease cohorts and, using sequence homology to known PPIs, predicts over 1,000 gut bacterial protein clusters to be involved in human-gut bacterial PPIs associated with CRC, IBD, obesity, and Type 2 diabetes mellitus [48]. Artificial intelligence (AI) was also recently adapted to predict protein structure and interaction and is a major step in PPI discovery [49]. AlphaFold, a revolutionary AI tool used to predict a protein’s 3D structure from their amino acid sequence, maintains a high level of accuracy and can be trained to predict PPIs [50,51]. By leveraging the potential of these technologies to explore human-gut bacterial PPIs at a larger scale, we anticipate the rate of human-gut bacterial PPI discovery to increase exponentially, enabling us to address some of the outstanding questions that remain (see Outstanding Questions).

Outstanding Questions Box.

• How common is it for gut bacterial proteins to influence host physiology, as opposed to small molecules like short chain fatty acids?

• What other host pathways are modulated by human-gut bacterial PPIs besides immune and gut barrier pathways?

• How did these PPIs evolve between gut bacteria and humans? Do any of these PPIs confer fitness advantages for the bacteria? For the host?

• How do gut bacterial proteins access the cells they interact with? Can they travel through the mucus layer and other protective layers of the gut to reach the epithelium? Or can they only access the epithelium if the gut barrier is already damaged?

• How do secreted bacterial proteins like MAM enter cells to interact with their target proteins? Is this a mechanism that extends to other gut bacterial proteins?

• Can the disease amelioration noted in mouse studies by human-gut bacterial PPIs like MAM and Amuc_1100 translate to humans? Can these proteins be used as therapeutics to reduce inflammation and help repair the integrity of the gut barrier?

Highlights.

The human gut microbiome is associated both positively and negatively with human disease.

The mechanisms for these associations are not completely understood, yet a few disease-relevant interactions between gut bacterial proteins and human proteins have recently been identified.

Many of the human-gut bacterial PPIs identified to date impact human health and disease by targeting the immune system and the gut barrier.

With newer technologies like artificial intelligence and high-throughput interaction screens, we expect the discovery rate of disease-relevant human-gut bacterial PPIs to increase substantially, as we see for many pathogens.