Commensal gut microbiota-based strategies for oral delivery of therapeutic proteins

Abstract

Therapeutic proteins are rarely available in oral dosage form because the hostile environment of the human gastrointestinal tract and their large size make this delivery method difficult to design. The commensal bacteria in the gut face the same situation; however, not only do they survive but low levels of their structural fragments and proteins such as LPS, peptidoglycan and flagellin are consistently detectable in the circulatory systems of healthy individuals. This opinion article discusses how gut bacteria survive in the gut, how their components penetrate the body from the perspective of the bacteria’s and host’s proactivity, and how orally administered therapeutic proteins may be developed to enter the body through similar mechanisms.

Commensal microbes as part of the oral delivery strategy for therapeutic proteins

The oral delivery of therapeutic proteins is often thought to be impossible because of their sizes and the hostile environment in the human gastrointestinal (GI) tract. Stomach acid and the digestive enzymes in the intestines can degrade these drugs, and the GI tract is impermeable to large molecules because of the intercellular tight junctions at the epithelium (Figure 1A). Even if these proteins successfully pass the brush border (see Glossary), they can elicit mucosal immunity, resulting in the production of anti-drug antibodies (ADAs) to inhibit them (Figure 1A). Nanoparticle drug delivery is the most-investigated method to enhance the oral absorption of therapeutic proteins, but it is limited by the size of the encapsulated material [1]. Although pharmaceutical companies make great effort to develop the oral dosage form for therapeutic proteins, the commercial availability is limited because of the aforementioned difficulties. Many reviews list the different types of intestinal barriers and suggest that breaking through these barriers is the key to successful oral delivery [2,3]. However, the interactions between commensal bacteria and the gut suggest that the intestinal barriers are not insurmountable. Although the penetration of the bacterial materials is mostly studied in disease states where the intestinal barrier is disrupted [4], detectable levels of bacterial structural units, including DNA; flagellins, the building blocks of flagella; peptidoglycans, cell wall components; and lipopolysaccharides (LPS), which are cell membrane components of Gram(−) bacteria, have been reported in the circulation of healthy control individuals in various studies (Table 1) [5–9]. The use of bacteria as drug vehicles to deliver chemo- and radiation therapies have been explored in various preclinical studies; for example, Escherichia coli MG1655 can be loaded with doxorubicin and Listeria monocytogenes with Rhenium-188 before they are administered intravenously to target tumors [10]. Commensal bacteria not only survive in the hostile gut environment but also avoid various host surveillance mechanisms in sophisticated ways. In this opinion article, we will discuss this bacterial evasion and the host’s intentional admittance and suggest that mimicking the entry of these bacterial materials under healthy conditions presents a new way to design the oral delivery of therapeutic proteins.

Figure 1. Proposed oral delivery strategies for therapeutic proteins.

(A) The extreme pH in stomach, the tight junctions at the intestinal epithelium and the lymphoid follicles, e.g. Peyer’s patch, are the major barriers for oral delivery of therapeutic proteins. (B) Biofilm can protect therapeutic (Tx) proteins from denaturation by stomach acid but must be degraded by anti-microbial peptides (AMP) in the intestine. Tagging bacterial peptide on Tx proteins can avoid the detection by antigen-presenting cells (APCs) and mucosal antibodies. Partial components of commensal bacteria can be used as conjugates of prodrugs to escape phagocytotic clearance but must be cleaved by the enzymes in the circulation or action sites. The Tx proteins that evade the mucosal immunity then enter the lymphatic circulation and subsequently the central circulation, and lastly are eliminated by the liver. (C) To use commensal bacteria in drug delivery, we must consider their strains and prevalence and implement quantitative control. To tag on bacterial peptides or encapsulate using partial bacterial structures to avoid host’s immunosurveillance, we must prevent functional interference of Tx proteins and ensure selective delivery and their eliminations.

Table 1.

Bacterial components are detected in healthy individuals

Detected component	Molecular size	Health condition	Concentration	Reference
DNA	/	Recruited healthy control	Detectable	[8]
Flagellins	26–140 kDa	Overweight healthy participants	0.56 ng/mL (mean)*	[6]
		Lean healthy control before treatment	Detectable	[9]
Lipopolysaccharides (LPS)	50–100 kDa (after treatment with SDS and heat)	No depression and anxiety control	<50 pg/mL	[75]
		Recruited healthy control	44 pg/mL (mean)	[76]
		Recruited healthy control	0–250 pg/mL; 50 pg/mL (mean)	[77]
		Recruited healthy control	<5.1 pg/mL (mean)	[78]
		Lean healthy control before treatment	99 pg/mL (mean)	[9]
Peptidoglycans	50 kDa	Recruited healthy participants	0.6–5 μg/mL	[5]
		Non-obese	Detectable	[7]

^*Unpublished data

Survival of Commensal bacteria in harsh host environment may inform delivery strategy for therapeutic proteins

In extreme environments, bacteria can form a biofilm to enhance their survival. For example, after culturing in solid minimal salts glycerol glutamate plates, Bacillus subtilis, a probiotic strain, form a layer of biofilm that improves survival for up to 48 hours after being orally delivered to mouse intestines [11]. Such protection of biofilm remains the same when the bacteria are delivered to other species, such as pigs [11]. The bacterial biofilm may serve as a bio-shield for therapeutic proteins against stomach acid. However, the biofilm itself can harm the host, as it may protect opportunistic pathogenic bacteria as well, e.g., Enterococcus faecalis. The anti-microbial peptides produced by Paneth cells in intestine have long been investigated as an anti-biofilm agent [12]. These host peptides can stop the bacterial formation of biofilm or disperse in it to disrupt the structure. The biofilm must be degraded at the site of absorption to release the drugs, so the ideal type of biofilm for encapsulating therapeutic proteins should be resistant to the acidic pH in the stomach but susceptible to anti-microbial peptides in the small intestine (Figure 1B). Furthermore, since recombinant proteins are often produced by lab-engineered Escherichia coli strains, commensal bacteria in the gut can also be engineered to produce therapeutic proteins. Oral administration of Lactococcus lactis engineered to produce human proinsulin is under clinical trials for treating Type I diabetes (NCT03751007). Several commensal strains are known to produce biofilm, such as Lactobacillus and Bifidobacterium. Hence, biofilm-forming bacteria can be engineered to produce and secrete therapeutic proteins into the biofilm matrix which provides protection against the degradation by stomach acid and enhance absorption. However, quantitative regulation of the production, as well as the nature of the commensal strains, e.g., opportunistic or truly beneficial, must be considered.

Entry mechanism of large bacterial materials in the intestine may inform delivery strategy for therapeutic proteins

Large bacterial materials encounter multiple barriers to enter host systems after surviving the extreme pH of the stomach. Recent studies suggest that barriers imposed by secretions of the intestinal cells, and tight intercellular junctions in normal health conditions may allow bacterial materials to enter host’s systems. The studies also advance our understanding of the variations in bacterial evasion of immune surveillance mechanisms in the host. Moreover, gut microfold cells (M-cells) that allow food-borne antigens to enter the host through transcytosis to modulate immune response may facilitate absorption of large intact proteins.

Intestinal cell defense

In a non-infectious situation, the absorption of live bacteria in the intestine is unlikely. In the context of this discussion on gut microbiota, gut permeability means the entry of bacterial components into the host’s system. The permeation of the gut is thought to occur primarily in the colon of the large intestine because of its many resident bacteria. Approximately 10¹² cfu of bacteria per mL of luminal fluid were found in the large intestine, versus about 10³ to 10⁷ cfu/mL in the small intestine [13,14]. The small intestine – especially the ileum, which is adjacent to the colon – may be an additional access point for bacterial materials to enter the host as the absorptive rather than secretory nature of the small intestine may compensate for the relatively lower bacterial load [15,16].

The small intestine has distinctive structures which include villi and microvilli, and cellular components which consist of enterocytes, Paneth cells, goblet cells, stem cells, tuft cells and enteroendocrine cells. The Paneth and goblet cells secret anti-microbial peptides and mucins respectively, providing the first line of defense against bacteria in the small intestine. Nonetheless, a recent study shows that certain commensal strains are resistant to anti-microbial peptides, and these strains maintain and stabilize the composition of the gut microbiota [17]. The large intestine has no villi, microvilli and Paneth cells. Colonocytes are the major cell types present [18]. In the absence of Paneth cells for anti-microbial defense, it relies on many goblet cells to produce mucins, which form a dense mucus layer to prevent bacterial penetration [19]. However, strains like Akkermansia muciniphila, which are known to be beneficial for promoting the epithelial development from intestinal stem cells and maintaining gut integrity, express mucin-degrading enzymes to break down the mucus [20]. These suggest that commensal gut bacteria have various mechanisms to fight against the host’s defense but maintain their mutualistic benefits to the host. To select the optimal strains of bacteria to be a carrier or manufacturer of therapeutic proteins, we must take into the account their sensitivity and resistance toward the host’s defense mechanisms.

Tight intercellular junctions

Other than the specialized cells like Paneth and goblet cells, both the small and large intestines depend on tight intercellular junctions of the villus to limit the entry of harmful materials into the body [21]. Small-molecule drugs, usually defined as the size <1 kDa, can enter the intestinal epithelium through the gaps between the epithelial cells, referred as the paracellular pathway. It is generally believed that these tight junctions at the intestinal epithelium are the barrier for the absorption of larger molecules like therapeutic proteins. The tight junctions are composed of different types of proteins such as zonula occludens and claudins, and the various tight junction compositions among different cell types allow different degrees of penetration (Figure 1A) [22] [23]. A study using mouse small-intestine organoids demonstrates that areas enriched with enterocytes or goblet cells are more permeable to 4-kD-10-kDa dextran than those enriched with stem cells and Paneth cells [23]. Another study reports that small-intestine organoids are impermeable to 40-kDa dextran, but the maximum molecule size that can penetrate tight junctions remains unknown [24]. Conversely, because of the secretory nature of large intestine, the tight junctions in the colon are stiffer and human colon organoids are impermeable to even 4-kDa dextran [25]. Although whether intact peptides or proteins can be absorbed under normal conditions due to the presence of digestive enzymes remains controversial [26], molecules smaller than 10 kDa seem not to be limited for the absorption in small intestine. Moreover, doubts about the absorption of intact proteins mostly originate from early studies that focused on the nutritional perspective [26]. A recent study shows that LPS, which is larger than 50 kDa, is detectable in the portal vein at a level five times higher than that in the hepatic vein [27]. In another study, LPS is found complexed with LPS-binding protein (LBP) and high-density lipoprotein-3 (HDL₃) in the portal vein [28]. These studies suggest that the low levels of such bacterial components in the circulation do not result from size sequestration but the effective removal by the liver. Using the advanced proteomics approach to evaluate bacterial proteins in mesenteric blood at pre- and post-absorption will provide definitive answers about the types of proteins that can infiltrate the intestines. We know that the tight junctions allow molecules up to 10 kDa to permeate them [23] [29], suggesting that oral delivery of drugs with even larger molecular sizes is possible.

Host gut M-cells may facilitate absorption of large proteins.

M-cells, a specialized epithelial cell type found in the small intestine, are a gateway for the entry of large molecules derived from gut microbiota. Through M-cell transcytosis, gut-derived antigens or bacteria are transported to antigen-presenting cells (APCs) in lymphoid follicles, e.g., Peyer’s patches, and then drained to mesenteric lymph nodes [30]. These APCs including macrophages and mucosal dendritic cells process and present the ingested foreign objects such as food-borne antigens or pathogens to the lymphocytes in the intestinal lymphoid follicles. The unique function of intestinal APCs is to orchestrate the oral tolerance of food antigens and the protection against pathogens [31]. The close contact between M-cells and APCs ensures tight intercellular transport; however, materials can remain in the extracellular space without being taken up by APCs [32] (Figure 1B). Approximately 10% of epithelial cells in lymphoid follicles are estimated to be M-cells [33]. Thus, the transcytosis of M-cell offers a potential gateway for the entry of therapeutic proteins.

M-cell has long been explored in the development of oral vaccines [34]. For example, probiotics such as Escherichia coli Nissle 1917 that are camouflaged with a yeast membrane are readily taken up by M-cells and then stimulate IgA production [35]. The delivery of non-vaccine types of therapeutic peptides via M-cells has also been investigated recently. Insulin encapsulated in aminoclay and subsequently coated with Ulex europaeus agglutinin-1, a type of lectin that binds to glycoproteins and glycolipids, for M-cell targeting has been shown to be taken up in the epithelium and Peyer’s patches to decrease the blood glucose level in diabetic mice [36]. β-glucans, the cell wall unit of yeast, are considered minimally immunogenic [37], and microcapsules derived from bakery yeast, are “generally recognized as safe” (GRAS) by the FDA, have been investigated as vehicles to deliver nanodrugs via M-cell uptake [38]. However, therapeutic proteins must avoid triggering adaptive immunity through prolonged exposure, resulting in the production of ADA and inactivation. Further investigation of the mechanisms by which bacterial proteins escape the host’s immune surveillance will illuminate alternate designs for the oral delivery of therapeutic proteins.

Host immune surveillance evasion by commensal bacteria

Phagocytic clearance and the development of ADA are the major mechanisms underlying drug tolerance for therapeutic proteins [39] (Figure 1A). Nanoparticles are currently one of the focal designs to escape recognition by APCs [40]. For example, nanomicelles, including a modified “self-peptide” based on human CD47, were shown to avoid macrophage recognition as foreign particles [41]. However, a design based on host peptides may prolong the half-life of nanoparticles increasing the chance of undesirable distribution to nontarget tissues. To gain insight into how to avoid immune surveillance but remain sensitive to elimination, we can refer to the bacterial protein flagellin. Flagellin is a monomeric unit that forms the filament of flagellum in bacteria. It is a ligand of a member of the pattern recognition receptor (PRR) family— toll-like receptor 5 (TLR5). Interaction between flagellin and host cell TLR5 results in various types of immune responses [42]. Each unit of flagellin has a hypervariable region that differs among species and a conserved region that TLR5 recognizes. Flagellins from certain commensal bacteria can evade recognition by TLR5 through alternations in either region. For example, the flagellin from Helicobacter pylori is less potent to stimulate TLR5, and those of α and ε Proteobacteria cannot be detected by TLR5 [43,44], allowing them to bypass detection by epithelial cells at the brush border and mucosal dendritic cells at Peyer’s patches. After passing Peyer’s patches, the flagellin then reaches the mesenteric lymph nodes. A proteomics analysis of pre- and post-nodal materials in the lymph indicates that the protein size is not a relevant factor for the lymph clearance capacity [45]. Rather, lymphatic clearance is limited by the types and amounts of materials passing through [45]. If a foreign protein cannot be recognized as non-self by nodal phagocytotic cells, they are likely to bypass the lymph nodes and enter the central circulation. Because macrophages and conventional dendritic cells are hyporesponsive to flagellins, these escaped flagellins can remain in circulation [46]. The content of the intestinal lymphatic trunk drains into the thoracic duct and eventually enters the central circulation through the subclavian vein, and the highly perfused liver is the final location to eliminate these bacterial products [47] (Figure 1B). It is reported that the TLR5 expressed in hepatocytes rather than dendritic cells is the primary driver of the clearance of flagellins [48]. Unlike mucosal or oral vaccines where flagellin has been investigated as an adjuvant to trigger immunity through recognition by APCs in the mucosa [49], therapeutic proteins may avoid APC detection in intestinal follicles or lymph nodes through a tagging sequence derived from immunity-evading flagellins (Figure 1B).

Prodrug design based on bacterial components

One way to improve oral delivery of drugs is to generate chemically stable prodrugs that are resistant to degradations before reaching the target sites. The biotransformation of prodrugs to active drugs relies on specific enzymes between the site of administration and the site of action. Various prodrug designs take advantage of the function of gut microbiota. Sulfasalazine, a drug used for Crohn’s disease and rheumatoid arthritis, relies on the azoreductase in gut bacteria to cleave it into the two active ingredients: sulfapyridine and 5-aminosalicylic acid. A prodrug system can be based on the host enzymes that inactivate bacterial components as well. For example, peptidoglycans detected in the circulation of healthy individuals (Table 1) can be broken down by a soluble form of peptidoglycan recognition protein (PGLYRP, Box 1). PGLYRPs are PRRs that are broadly distributed in various tissues. One of the isoforms, PGLYRP2, a plasma protein produced by the liver, is a N-acetylmuramoyl-L-alanine amidase that hydrolyzes peptidoglycans into N-acetylmuramoyl and L-amino acid residues in the circulation [50,51]. The identification of potential substrates of PGLYRP2 such as glycosylated proteins, will inform the structural design of prodrugs. Conversely, commensal bacteria can alter the molecular structure of their components to resist degradation. Paneth cells produce lysozymes to hydrolyze peptidoglycans, but Lactococcus lactis controls the degree of O-acetylation of its peptidoglycans to resist hydrolytic degradation by lysozymes [52]. Moreover, bacteria can modify the glycan or peptide portions of their peptidoglycans to avoid recognition by host PRRs and the resulting immune response [53]. Such modified units can also serve as templates for constructing prodrug conjugates to avoid mucosal degradation and immune detection. The ideal conjugates for therapeutic proteins will be structures that can both avoid immune recognition at the GI mucosa and be cleaved by a specific circulating enzyme or an enzyme at the site of action (Figure 1B).

Text Box 1. Peptidoglycan recognition protein family.

Peptidoglycan recognition protein (PGLYRP) family is a type of pattern recognition receptors that is less commonly studied in human compared with toll-like receptors (TLR) and nucleotide-binding oligomerization domain-like receptors (NLR). They are secreted soluble proteins. Four isoforms have been identified in humans, namely PGLYRP1 to PGLYRP4. PGLYRP1 (also called PGRP-S where S denotes short) is expressed in various types of immune cells including neutrophils and eosinophils. However, unlike other PGLYRPs, PGLYRP1 does not have amidase activity to cleave the amide bond between the glycan and peptide moieties in peptidoglycan. PGLYRP2 (also called PGRP-L where L denotes long) belongs to N-acetylmuramoyl-L-alanine amidase-2 family. It is synthesized in the liver and subsequently secreted into the circulation. PGLYRP3 (also called PGRP-Iα where I denotes intermediate) and PGLYRP4 (also called PGRP-Iβ) also possesses N-acetylmuramoyl-L-alanine amidase activity, and the mRNA expressions of these two isoforms are detected in the salivary epithelial cells. There are very few studies on these two isoforms. Although this family is named as peptidoglycan recognition protein, PGLYRPs can also bind to LPS, but the binding site is different from that for peptidoglycans. The additional functions of PGLYRP are pending for further investigations.

Benefits of using commensal bacteria or bacterial components as conjugates

The human body is not proof against infiltration by commensal bacterial materials. The well-known purpose of allowing penetration of these materials at low levels is to develop immunotolerance [54–57]. Additionally, a moderate degree of permeation appears to be beneficial. Pre-exposure to low level of LPS protects hepatocytes from apoptosis [58]. Because the liver is essential for bacterial clearance, this resistance to apoptosis allows it to function properly during persistent assaults. Moreover, LPS helps to maintain the stemness of hepatocytes in vitro, and a decrease in the stemness of hepatocytes was observed in wild-type mice treated with oral antibiotics, suggesting the involvement of the gut microbiota [59]. As the portal vein is the first site of contact after the intestinal barrier, the surrounding hepatocytes are constantly handling enriched portal venous blood, and the maintenance of juvenescence ensures their functional efficacy. Alternatively, flagellin derived from commensal flagellated bacteria can stimulate apolipoprotein-A1 (ApoA1) production in hepatocytes and increase the high-density lipoprotein (HDL) level [6]. ApoA1 is the key lipoprotein present in HDL particle which provides cardioprotective effects through reverse cholesterol transport. This action of flagellin may protect against potential damage from other bacterial materials as HDL can neutralize LPS. Similar to the action of low level of LPS, treating flagellin without the presence of bacteria activates the antiapoptotic factors in epithelial cells, which protects the cells from subsequent pathogenic bacteria-induced apoptosis [60]. Based on the antiapoptotic effect of flagellin, KMRC011, a recombinant TLR5 agonist, is now undergoing clinical trials as a potential treatment for acute radiation syndrome [61]. Tagging a flagellin-mimetic on a therapeutic protein may serve as an adjuvant to increase HDL level and protect the intestinal mucosa while shielding a therapeutic protein from host’s immunological detection (Figure 1B).

Selective delivery based on differential expressions of PRRs

Bacterial components bind to various types of plasma-membrane and cytoplasmic PRRs in the body [62]. Cell membrane receptors or proteins are good candidates for targeted delivery [63–65]. Different cell types and tissues, including both immune and non-immune, have different sensitivities toward bacterial components based on the presence and absence of PRRs and the degree of their expressions. Hence, using bacterial components that interact with the plasma-membrane PRRs as conjugates can also help to achieve selective delivery. For instance, TLR5 is expressed in hepatocytes but not macrophages, and in the crypt but not the villi of the intestine [66]. TLR4, a PRR for LPS, is found in the colon but not the small intestine [66]. All of the cell types in the liver are nonresponsive to 100 ng/mL of LPS in terms of cytokine secretion except for Kupffer cells, and none responds to a concentration of 1 ng/mL [67]. The risk of inducing inflammatory responses by tagging or conjugation with partial bacterial components can therefore be avoided by adjusting the dose. Moreover, allosteric binding sites have been identified in some of these receptors [68], and these additional binding sites broaden the screening options for potential conjugation candidates to achieve selective delivery.

Concerns and solutions to using bacterial components

We have proposed three strategies using commensal bacteria in oral delivery of therapeutic protein: biofilm protection, tagging with bacterial peptides and encapsulation using bacterial conjugates (Figure 1B). However, the use of commensal bacteria is not free of problems, and several strategies may help to eliminate some of the drawbacks. Using biofilm-forming bacteria as a manufacturer to produce therapeutic proteins may face the problem of resistance development, which may then result in overpopulation that causes infection-like harm. Hence, we should avoid using opportunistic strains and select the truly or universally beneficial bacteria, for example, the Lactobacillus genus offers various benefits such as immunomodulation and cholesterol-lowering effect and [69,70]. The use of biofilm also raises concerns about resistance development and the release of biofilm-associated toxins [71,72]. Selecting biofilm types that can be degraded by anti-microbial peptides in the small intestine or modifying the original biofilm structure will be necessary. Conversely, the avoidance of immune detection by tagging drugs with bacterial peptides may result in non-specific delivery as the tagged drugs can enter the circulation before reaching the target site (Figure 1B). To solve this, co-tagging with peptides targeting specific organs may be needed [73]. For example, cetuximab, a therapeutic antibody for treating advanced colorectal cancer, tagged with octreotide, a synthetic peptide analogue of somatostatin, targets the somatostatin receptor on cancer cells to achieve tissue-specific delivery [74]. However, tagging additional peptides may alter the structure of therapeutic proteins resulting in functional interference. The understanding of chemical biology of such modifications will help to optimize the design. Moreover, knowing the differential expressions of plasma-membrane PRRs in different types of cells may also help to achieve selective delivery. No method is completely free of concern, and Table 2 lists the advantages and disadvantages of using commensal bacterial components compared with the current approaches. In regard of regulatory issues, safety of using bacterial material may raise a concern on infection-related adverse effects. As probiotics are generally considered safe as food supplement, the choice of commensal strains that have been used in probiotics is a good starting point to explore the abovementioned strategies. The FDA has implemented improved methodology to examine purity of probiotics because they have been increasingly investigated as disease treatment. We envision the extended regulations will be applied for using commensal bacteria in drug development.

Table 2.

Advantages and disadvantages of using commensal bacterial components for oral delivery

Barriers	Current approach	Proposed strategy	Advantage	Disadvantage
Extreme pH in stomach	Nanoparticle, microsphere, etc. as carrier	Use of biofilm	Able to produce therapeutic protein simultaneously Chemically stable	Difficult for quantitative control Variations in bacterial survival due to ethnic, gender, geographic differences
ADA production	Host peptides to ensure being recognized as self	Commensal bacterial peptides to avoid host immune detection	Avoid detection by multiple types of APCs	Unsuitable for immunocompromised patients
Elimination by APCs or mucosal degradations	Prodrug design subjected to bacterial activation in gut	Prodrug design based on the clearance of bacterial components	Rather consistent in half-life due to the reliance on host’s enzymes Predictable efficacy	Limited by the locations of the activating enzymes May be interfered by infections

Concluding remarks and future perspectives

The extreme pH in stomach, the tight junctions at the intestinal epithelium and the immunological defense at the intestinal mucosa are the major barriers to orally deliver therapeutic proteins. To use commensal bacterial components to facilitate drug delivery, we must consider the prevalence of bacterial strains among different populations and the behavior of the selected strains. Patients may response very differently because of the inherited absence of certain bacterial strains in the gut due to ethnic, dietary and geographic differences (Figure 1C). If we engineer the commensal strains to produce the therapeutic proteins or use opportunistic strains, we must also control their proliferation. To use bacterial peptides tagged on therapeutic proteins, we must identify the precise sequences attributing to the evasion of host’s immunity. The potential interference of therapeutic proteins due to structural change upon tagging should be evaluated, and additional mechanisms or guiding peptides may be required for selective delivery (Figure 1C). To use partial bacterial structure as prodrug conjugates, we must consider the location of the activating enzymes. For the uses of bacterial peptides and structures, we must ensure the elimination of such components to avoid the potential development of resistance arising from chronic exposure.

Many studies focus on how to avoid increasing gut permeability to prevent the penetration of harmful bacterial materials into the body. This is useful to prevent or treat disease conditions. However, leakage and limited entry are two sides of the same coin. Understanding how commensal bacterial materials escape host surveillance and how the body handles these materials for beneficial effects will inspire alternative methods to enhance the oral delivery of therapeutic agents. Nonetheless, some questions remain to be answered before we can fully take advantage of this entry point (see Outstanding Questions Box). In particular, we should understand how the large molecular units derived from commensal bacteria travel in the body and interact with both immune and non-immune cells under normal or healthy conditions.

Outstanding Questions Box.

• How large of bacterial materials can pass through intestinal tight junctions?

• Where do the components of commensal microbiota circulate and how are they eliminated after evading the host’s immunosurveillance?

• What are the differences in the interactions between host cells and commensal microbial materials among immune and non-immune tissues?

• What are the differential protein expressions of plasma-membrane PRRs in different cell types and tissues?

Highlights.

• Structural fragments of gut commensal bacteria enter systemic circulation in healthy condition.

• The entry of these bacterial materials is mediated by both transcellular and paracellular pathways in the intestine.

• Commensal bacteria can be engineered to produce therapeutic proteins and at the same time provide biofilm for shielding against stomach acid.

• Prodrug design can be based on the breakdown of bacterial components in the body.

• Conjugation of therapeutic proteins with partial components of commensal bacteria can possibly avoid undesirable immune responses.

Glossary

Adaptive immunity

a type of immune response that takes days or weeks to develop. It is activated by exposure to pathogens and relies on immunological memory to recognize the threat and elicit defense mechanisms in future

Aminoclay

a type of layered silicates referred as amino organophyllosilicates, which is synthesized from metal salt and organotrialkoxylane

Anti-drug antibodies (ADAs)

antibodies elicited upon the exposure of biological drugs such as proteins, peptides and antibodies that can decrease the efficacy of these drugs by inactivation and enhanced clearance. They may also induce adverse effect such as allergic response and anaphylactic shock

Anti-microbial peptides

a group of peptides produced by Paneth cells in intestines that exhibit broad-spectrum microbicidal activities

Apolipoprotein-A1 (ApoA1)

the major type of apolipoprotein in high-density lipoprotein particle

Biofilm

a consortium of single or multiple types of microorganisms embedded in self-produced matrices of extracellular polymeric substances that keep the cells together and allow the growth on surfaces

Brush border

the microvilli-covered layer of epithelium at the luminal side of small intestine

cfu

colony-forming unit, a unit of measurement to estimate the concentration of microorganism in a sample

Dextran

a polymer of glucose produced by bacteria that is commonly used as a tracer of gut permeability in pre-clinical studies

Enteroendocrine cell

a type of epithelial cells that secrete a variety of hormones or signaling molecules to activate nerve fiber or communicate with other organs

Flagellin

a protein unit composing the filament of flagellum which is the bacterial structure providing motility

Goblet cell

a type of intestinal epithelial cells that synthesize mucins, which are the macromolecules responsible for the biochemical and biophysical properties of mucus

High-density lipoprotein-3 (HDL₃)

HDL is categorized into two subtypes, HDL₂ and HDL₃, depending on the size. HDL₃ is the smaller and denser type. It is enriched with apolipoprotein-A2 (ApoA2) but contains fewer ApoC2, ApoC3, and ApoE compared with HDL₂

Lectin

a family of proteins found in many plants that highly bind to carbohydrates

Lymphoid follicle

an aggregate of lymphocytes including dendritic cells and B-lymphocytes in the intestinal lamina propria. They exist in the form of isolated lymphoid follicles or in groups along the mucous membrane known as Peyer’s patches

Paneth cell

a type of intestinal epithelial cells that contain granules of anti-microbial peptides such as defensins for secretion upon exposure to microbial invasion

Pattern recognition receptors (PPRs)

important players in innate immunity that interact with microbe-associated molecular patterns resulting in induction of inflammatory signaling pathway. Toll-like receptors, NOD-like receptors and RIG1-like receptors are the most studied PRRs

Reverse cholesterol transport

a process mediated by HDL particle to transport cholesterol from cells in peripheral tissues to the liver. The removal of cholesterol in vascular cells at atherosclerotic lesions provides cardioprotective effect

Tuft cell

a type of secretory epithelial cells that regulate intestinal type 2 innate lymphoid cell (ILC2)–epithelial response circuit