Indole-3-propionic acid links gut dysfunction to diabetic retinopathy: a biomarker and novel therapeutic approach

Abstract

Background:

Both host and microbe metabolism of tryptophan (Trp) is altered in diabetes; however, the molecular mechanisms are incompletely understood.

Objective:

We utilized strategies to increase either angiotensin converting enzyme-2 (ACE-2) dependent or independent Trp absorption in a model of type 2 diabetes, db/db mice and tested whether the strategies could prevent development of diabetic retinopathy (DR), the most common microvascular complication of diabetes. Additionally, we investigated levels of Trp metabolites in humans with and without DR.

Design:

Enhanced ACE-2 dependent Trp absorption was achieved with gavage of genetically modified bacteria that preserved intestinal ACE2:B⁰AT1 expression. ACE-2 independent Trp absorption was achieved by gavage of the Trp dipeptide (Isoleucine-Trp; IW) absorbed via SLC15A1. Both strategies were used as either as a prevention (6 months treatment) or intervention (3 months treatment) and at the conclusion, intestinal, metabolic and retinal studies were performed including spatial mass spectroscopy (MS). Plasma Trp metabolites and gut permeability markers were measured in individuals with T2D with (n=30) and without (n=40) DR and compared to healthy controls (n=35).

Results:

LP-ACE2 or IW treatment prevented DR, corrected dysbiosis, enriched Trp-metabolizing bacteria, improved gut barrier integrity, boosted incretin secretion, and restored glucose homeostasis in db/db mice. Spatial MS identified indole propionic acid (IPA) as a metabolite in the retinal pigment epithelial layer protecting the posterior blood retinal barrier. T2D individuals with DR demonstrated elevated serum markers of endotoxemia and intestinal barrier disruption while showing reduced levels of the beneficial metabolite IPA and elevated levels of the toxic metabolite indole sulfate.

Conclusion:

Nutraceutical strategies that restore Trp metabolism or IPA serve as both a biomarker and a treatment for DR.

Graphical Abstract

Visual abstract illustrating the gut-retina axis and the protective role of microbial-derived indole metabolites. Trp is absorbed by the host through epithelial transporters (ACE2:B⁰AT1 and PEPT1) but is also metabolized in the gut by microbes that convert indole into beneficial metabolites such as IPA. IPA promotes tight junction stability, while reducing inflammation via downregulation of TLR4, MyD88, and NLRP3 signaling. IPA activates the PXR, leading to suppression of intestinal IL-1β and increases secretion of incretins, contributing to both gut and systemic anti-inflammatory effects. Improved gut barrier function reduces leakage of microbial peptides into circulation and endotoxemia. Circulating IPA reaches the retina, where it activates PXR in both the inner (vascular endothelium) and outer (RPE) blood-retinal barriers preventing the progression of DR. The schematic of experimental design was created using BioRender, an online scientific illustration software. Green arrows indicate activation, whereas red blocked lines represent inhibition.

Introduction:

The human host is inextricably linked to its microbiome and has coevolved with it resulting in the outsourcing of approximately 60% of human metabolism to microbes^{1, 2}. The gut provides the dominant microbiome in the body, representing approximately 1000 microbial species³. Dysregulation of the gut microbiome is associated with a range of pathologies such as cerebrovascular disease⁴, parkinson’s disease⁵, cancer⁶, obesity⁷ and diabetic retinopathy (DR)^8–10.

Trp is an indispensable dietary amino acid, playing a pivotal role in regulating the intestinal immune system, suppressing inflammation, and maintaining microbial homeostasis^11–14 but Trp absorption is reduced in diabetes^{15, 16}. The metabolism of Trp is orchestrated by multiple signaling pathways, primarily the kynurenine, serotonin, and indole pathways. Kynurenine and serotonin related metabolites are generated by the host cells whereas indole metabolism occurs by the gut microbes and the metabolites subsequently act on host cells. Approximately 85 microbial species metabolize dietary Trp to indole via the enzyme tryptophanase¹⁷.

Previously, we discovered that epithelial ACE2 serves as a chaperone for the intestinal epithelial cell trafficking of the Trp transporter, sodium coupled neutral amino acid transporter B⁰AT1¹⁸ and forms stable dimer of heterodimers ([ACE2:B⁰AT1]2) in the apical membrane of the gut epithelium in vivo and in vitro^18–20. We showed in the diabetic small intestine that expression of ACE2 parallels that of B⁰AT1 expression. Using genetically modified Lactobacillus paracasei expressing soluble ACE2 (LP-ACE2), we showed generation of Angiotensin 1–7 (Ang 1–7) from microbe derived Ang II and by activation of the MAS receptor preserved endogenous ACE2:B0AT1 expression. Thus, fecal Ang II served as the substrate for bacterial derived soluble ACE2²¹ preserving host epithelial function and preservation of host ACE2 expression.

In this study, we demonstrate that strategies that improve Trp absorption via either ACE2 dependent or independent mechanisms of Trp or direct supplementation of beneficial Trp metabolites such as indole propionic acid (IPA) prevented development of DR, mitigated diabetes induced dysbiosis and intestinal epithelial barrier damage and improved metabolic homeostasis by increasing incretin release. To our knowledge, we show for the first time that the beneficial metabolite IPA is present in the retinal pigment epithelium (RPE), the cells that generate the posterior blood retinal barrier and confirm using in vitro studies that IPA directly increases expression of key nuclear receptors, aryl hydrocarbon receptor (AhR) and Pregnane X receptor (PXR) that modulate barrier function in RPE cells. Using a cohort of diabetic individuals that only suffered from the microvascular complication of DR and no other vascular complications, we found low levels of the IPA and elevated levels of the detrimental metabolite Indole sulfate (IS) compared to age-gender matched diabetic and nondiabetic controls. Together, these data identify dysregulated Trp metabolism as highly impactful on development of DR by showing that IPA may serve as a biomarker of DR and that its supplementation may be effective in the therapeutic management of DR.

Materials and Methods:

Detailed material and methods are available in the supplementary section.

Results:

DR pathology is directly linked to reduced intestinal Trp absorption

Db/db mice recapitulate the full spectrum of diabetic vascular complications, including DR. We evaluated two independent strategies to test the role of Trp absorption in DR pathogenesis. One approach targeting the ACE2-dependent mechanism and the other ACE2-independent pathway (Figure 1A). For the ACE2-dependent approach, we administered LP-ACE2, previously shown to preserve intestinal epithelial ACE2 expression in a type 1 diabetes model, the Akita mouse²¹. LP-ACE2 was delivered via oral gavage three times weekly for six months, beginning at the onset of hyperglycemia (Figure 1A). As expected this treatment resulted in preservation of ACE2:B⁰AT1 dimer expression in the diabetic intestinal epithelium and also increased ACE2 levels in the plasma (Figure S1A–F). To assess ACE2-independent Trp absorption, we administered IW, which is absorbed via the peptide transporter 1 (PepT1, also known as SLC15A1). After six months of treatment, IW treated db/db mice showed preserved luminal ACE2 and B⁰AT1 expression but no increase in plasma ACE2 levels (Figure S1A–F). To further investigate the regulatory effects of LP-ACE2 and IW on the renin angiotensin system (RAS), we quantified Ang 1–7 levels using ELISA. Ang 1–7, a key component of the protective arm of the RAS, was significantly reduced in untreated db/db mice, indicating the expected RAS imbalance typically seen in diabetes. Remarkably, both LP-ACE2 and IW treatments significantly restored intestinal Ang 1–7 levels (p < 0.0001) (Figure S1G), supporting their role in enhancing endogenous intestinal ACE2 activity. In contrast, components of the deleterious arm of the RAS, including angiotensin-converting enzyme (ACE), Ang II, and the angiotensin II type 1 receptor (AT1R), were markedly elevated in untreated db/db mice, consistent with RAS overactivation in diabetes. Pro-inflammatory and pro-fibrotic RAS mediators were significantly reduced following LP-ACE2 and IW treatments, approaching levels observed in normal controls (Figure S1H–J). These findings indicate that the activation of either the ACE2 dependent or independent mechanism of Trp absorption restored the intestinal RAS homeostasis.

Figure 1: Oral administration of LP-ACE2 and IW prevents the development of DR in db/db mice.

Schematic of experimental design shows the preventive treatment regimen in db/db mice. Starting at two months of age, immediately after the establishment of diabetes, mice were gavaged with LP-ACE2 (three times per week) or IW (daily) for six months. At study completion (eight months of age), tissues were collected for analysis. Saline-treated db/db mice and non-diabetic WT mice served as controls (A). Retinal function was assessed using ERGs. Photopic a- wave (n=3–5/group) (B), Photopic b-wave (n=4–12/group) (C), Scotopic a-wave (n=4–9/group) (D), and Scotopic b-wave (n=4–8/group) (E) Amplitudes were significantly improved in LP-ACE2- and IW-treated mice compared to saline-treated db/db controls. Visual acuity, measured by OKN, showed increased spatial frequency thresholds in treated groups (n=10–11/group) (F). DR was evaluated by quantifying acellular capillaries in retinal flat mounts. Representative brightfield images and quantification revealed a significant reduction in acellular capillaries in LP-ACE2- and IW-treated mice, indicating reduced retinal vascular pathology (n=4–6/group) (G, H). Each black data point represents an individual experimental animal. The schematic of experimental design was created using BioRender, an online scientific illustration software. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons. Scale bar = 20 μm.

To assess retinal function, electroretinography (ERG) was performed after six months of diabetes. Photopic a-wave amplitudes were significantly reduced in db/db-saline mice compared to wild-type (WT) controls (41.95±8.23 vs.115.3±7.41; p=0.0005). Both LP-ACE2 and IW treatments significantly restored a-wave amplitudes (94.49±15.16, p=0.0118; and 106.0±9.30, p=0.0052, respectively; Figure 1B). Photopic b-wave amplitudes were also reduced in db/db-saline mice compared to controls (75.48±1.26 vs. 228.1±23.89; p=0.0023), with significant improvement observed only in the LP-ACE2 group (200.0±21.39; p=0.0378; Figure 1C). This supports that LP-ACE2 beneficially impacts the ON-bipolar cells and Muller glial cells. Under scotopic conditions, both a- and b-wave amplitudes were significantly diminished in db/db-saline mice (140.1±12.10 vs. 229.0±15.10, p=0.0043 for a-waves; 175.8±12.23 vs. 424.9±23.97, p<0.0001 for b-waves; Figure 1D, 1E). IW treatment significantly improved both amplitudes (219.6±10.77, p=0.0382 for a-waves; 317.4±22.76, p=0.0208 for b-waves), whereas LP-ACE2 did not. This supports that IW beneficially impacts rod photoreceptors and rod bipolar cells. Visual acuity, assessed by spatial frequency (Figure 1F), was significantly reduced in db/db-saline mice compared to WT (0.30±0.015 vs. 0.38±0.008; p<0.0001). Both treatments significantly improved spatial frequency (0.38±0.006, p<0.0001 for LP-ACE2; 0.35±0.007, p=0.0061 for IW).

Retinal vasodegeneration, a hallmark of DR, was quantified by counting acellular capillaries. As expected, db/db-saline mice exhibited significantly more acellular capillaries than WT controls (11.0±1.15 vs. 4.80±0.66; p=0.0031; Figure 1G, 1H). Both LP-ACE2 and IW treatments significantly reduced acellular capillary number (6.5±1.19, p=0.0451; and 6.25±1.25, p=0.0331, respectively). Together, these findings demonstrate that enhancing Trp absorption through either ACE2-dependent or independent mechanisms improves DR and that the transporters used for Trp absorption appears to influence specific retinal cell functions as supported by the ERG results, but visual acuity and vascular degeneration were beneficially impacted by either LP-ACE2 or IW administration.

Enhancing Trp absorption improves glucose homeostasis, gut dysbiosis, and intestinal structure and function in db/db mice.

Metabolic endpoints were evaluated to assess the systemic effects of LP-ACE2 and IW treatments. Only IW treatment resulted in a reduction in weight (Figure S2A) and blood glucose (Figure S2B). There was no reduction in hemoglobin A1C (Figure S2C) but plasma insulin levels (Figure S2D) were reduced and insulin sensitivity and β-cell function improved, as indicated by HOMA-IR (Figure S2E) and HOMA-%β indices (Figure S2F). Enhanced serum incretins, GLP-1 (Figure S2G) and GIP (Figure S2H) were also seen with LP-ACE2 and IW treatment. These findings suggest that both LP-ACE2 and IW treatment enhance glucose homeostasis and insulin responsiveness in diabetic mice.

To explore the impact of these treatments on the gut microbiome, shotgun metagenomic sequencing of cecal samples was performed. As shown in Figure 2A, db/db mice exhibited significantly reduced alpha diversity (109.4±10.52) compared to WT controls (152.3±10.96; p=0.03). Unexpectedly, alpha diversity was further reduced in LP-ACE2 treated mice, whereas IW dipeptide treatment significantly increased alpha diversity (198.4±11.3; p=0.0001 vs. untreated db/db). Beta diversity, assessed via partial least squares-discriminant analysis (PLS-DA), revealed distinct clustering of db/db mice from WT controls (Figure 2B). Neither treatment significantly altered beta diversity, though both showed a modest shift toward the WT microbial profile. These results highlight the distinct microbial signatures in diabetes and suggest that IW exerts a more favorable effect on microbial richness.

Figure 2: LP-ACE2 and IW treatment ameliorates gut dysbiosis, structural alterations, and goblet cell dysfunction in db/db mice.

To evaluate the impact of LP-ACE2 and IW on diabetes-induced gut dysbiosis, 16S rRNA sequencing and metatranscriptomic analyses were performed. Microbial diversity was assessed by measuring α-diversity using observed species counts (n=4–9/group) (A), β-diversity via partial least squares discriminant analysis (PLS-DA) (n=4–9/group) (B), and phylogenetic differences (n=4–9/group) (C). To assess structural changes in the gut, immunohistochemistry was performed on small intestinal sections. Representative brightfield images show H&E staining, Ki67-positive proliferating cells, and Alcian blue staining for mucin-producing goblet cells (D). Quantitative analysis included villus-to-crypt length ratio (n=6–10/group) (E), Ki67 proliferative index (n=8/group) (F), and goblet cell counts (n=13/group) (G). Both LP-ACE2 and IW treatments significantly improved gut architecture, enhanced epithelial proliferation, and increased mucin production, indicating restoration of gut barrier integrity. Each black data point represents an individual experimental animal. All statistically significant changes were normally distributed and determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons. Scale bar = 50 μm.

Microbial composition at the phylum level revealed notable shifts in db/db mice compared to WT, including reductions in Actinobacteria, Tenericutes, and Proteobacteria, and increases in Firmicutes, Verrucomicrobia, and Bacteroidetes (Figure 2C). However, neither LP-ACE2 nor IW significantly altered phylum-level distributions.

At the genus level, we focused on bacteria involved in Trp metabolism, gut barrier integrity, glucose homeostasis, and inflammatory cascade. Bacteroides and Bifidobacterium, key producers of beneficial indole metabolites, were reduced in db/db mice (by 87% and 64%, respectively), while Barnesiella, a producer of the toxic metabolite indoxyl sulfate, was elevated by 113.5%. Both treatments restore Bacteroides while reducing Barnesiella abundance, supporting a shift toward a more favorable Trp-metabolizing microbiome (Figure S3A–C). We also examined genera associated with gut barrier integrity. Muribaculaceae, Lactobacillus, and Enterorhabdus were significantly depleted in db/db mice (by 87%, 80%, and 69%, respectively). LP-ACE2 and IW restored Muribaculaceae and Lactobacillus levels, though Enterorhabdus remained unchanged. Parvibacter, reduced by 44% in diabetes, was partially restored by IW treatment. Conversely, Akkermansia and Duncaniella were markedly elevated in db/db mice (1494% and 136%, respectively) and normalized following IW treatment (Figure S3D–I). Notably, IW treatment resulted in a more pronounced restoration of beneficial genera involved in gut barrier maintenance compared to LP-ACE2. Together, these findings suggest that IW exerts broader and more beneficial effects on microbial diversity and gut barrier associated taxa than LP-ACE2.

Next, we also assessed the abundance of microbial genera involved in glucose metabolism and inflammation. The microbial genera, associated with glucose metabolism, Eggerthellaceae was completely depleted in db/db mice and restored only with LP-ACE2 treatment. Oscillibacter abundance was reduced by 21% in diabetes and trended toward normalization with IW. In contrast, Lachnospiraceae was significantly elevated in db/db mice (180%) and shifted toward WT levels following either treatment. These microbial shifts align with the observed improvements in glucose homeostasis and insulin sensitivity (Figure S3J–L).

The microbial genera, Adlercreutzia, Muribaculum, Eubacteriales, Eubacterium, Turicibacter, and Ralstonia contribute to inflammatory signaling and their abundance is impacted by diabetes^22–30. Beneficial genera such as Eubacterium, Adlercreutzia, and Muribaculum were reduced by 86–90% in db/db mice and were restored toward WT levels following either LP-ACE2 or IW treatment. Ralstonia was undetectable in diabetic mice but was present in both treatment groups. In contrast, Eubacteriales and Eubacterium were markedly elevated in diabetes (1313% and 3475%, respectively) and trended toward normalization with treatment (Figure S3M–R). These results suggest that both LP-ACE2 and IW modulate the gut microbiome to reduce inflammation and support metabolic health.

These microbial changes were accompanied by improvements in intestinal morphology. Villus length, which was reduced in db/db mice, was restored to near-normal levels in both treatment groups (Figure 2D, top panels). The villus-to-crypt ratio, significantly lower in db/db mice (3.39±0.12 vs. 4.99±0.30; p=0.03), was normalized following treatment (Figure 2E). Ki67 immunostaining revealed reduced crypt cell proliferation in untreated db/db mice, which was increased in both LP-ACE2 and IW treated groups (Figure 2D, middle panels; Figure 2F). These findings suggest enhanced epithelial renewal and absorptive capacity. Additionally, Alcian blue staining demonstrated a significant increase in goblet cell numbers³¹ in both treatment groups compared to untreated db/db mice (Figure 2D, bottom panels; Figure 2G), indicating improved mucosal barrier function. Collectively, these results demonstrate that LP-ACE2 and IW treatment not only restore eubiosis but also improve intestinal structure and function, with IW showing a broader impact on beneficial microbial populations.

LP-ACE2 and IW treatment protects diabetes induced gut barrier defects and mitigates the inflammatory response in db/db mice.

To further evaluate the impact of LP-ACE2 and IW on diabetes-induced gut barrier dysfunction, we assessed the expression of junctional proteins in intestinal tissue sections and measured plasma levels of key intestinal permeability markers. Immunofluorescence staining of the small intestine revealed decreased expression of the epithelial tight junction proteins ZO-1 (green) and the adherens junction protein p120-catenin (red) in db/db mice compared to WT controls. Treatment with LP-ACE2 or IW maintenance of both proteins to near-normal levels (Figure S4A–C). Similarly, analysis of the gut-vascular barrier showed elevated expression of plasmalemma vesicle-associated protein 1 (PV1; green) and reduced expression of VE-cadherin (red) in db/db mice, indicating increased endothelial permeability and compromised barrier integrity³². Both treatments normalized PV1 and VE-cadherin expression (Figure S4D–F). The plasma levels of key gut permeability markers, zonulin and FABP2 were significantly elevated in db/db mice compared to wild-type (WT) controls (p< 0.0001; Figure S4G–H), and levels were markedly reduced following treatment with either LP-ACE2 or IW. These structural improvements were accompanied by reductions in circulating markers of endotoxemia or microbial translocation, including PGN, liposaccharides (LPS), and LPS-binding protein (LBP). The circulating levels of these microbial peptides significantly increased in untreated db/db mice compared with WT mice (Figure S4I–K). Further, LP-ACE2 and IW treatment reduced the levels of these gut microbial antigen (GMA) in the plasma of the diabetic mice, suggesting improved integrity of the gut barriers and support that both LP-ACE2 and IW enhance gut barrier integrity.

Therefore, next we explored the levels of inflammatory cytokines in the intestine. IL-1β, IL-2, IL-6, IL-17, IFN-γ, and TNF-α were significantly elevated in untreated db/db mice compared to WT controls. Notably, IL-1β levels were 447.4±30.28 pg/mL in db/db mice vs. 69.09±25.14 pg/mL in WT (p<0.0001), and IL-6 levels were 528.9±62.87 pg/mL vs. 98.67±39.69 pg/mL (p=0.0003). Both LP-ACE2 and IW treatments significantly reduced IL-1β (230.3±18.25 and 215.3±31.61 pg/mL, respectively) and IFN-γ levels (57.76±5.25 and 62.96±2.62 pg/mL). LP-ACE2 significantly reduced IL-2 and TNF-α, while IW did not. However, IW significantly reduced IL-6 (p<0.01), whereas LP-ACE2 did not (Table S1).

Anti-inflammatory cytokines, including TGF-β, IL-12, and IL-10, were suppressed in saline-treated db/db mice. TGF-β levels were significantly increased with LP-ACE2 (p< 0.001) but not IW treatment. IL-12 levels increased modestly with both treatments, while IL-10 was significantly elevated only in the IW group (p=0.0027) (Table S1). Western blot analysis of intestinal lysate from the experimental cohorts revealed that both LP-ACE2 and IW treatment reduced TLR4 expression; however, only LP-ACE2 significantly suppressed downstream effectors including MyD88, Caspase-1 and NLRP3 (Figure S5A–D). In summary, LP-ACE2 and IW treatments enhance gut barrier integrity, reduce microbial translocation, and attenuate intestinal inflammation in diabetic mice, through modulation of TLR4 signaling and restoration of Trp-metabolizing microbial populations.

IPA treatment corrected gut barrier defects and blocks incretin reduction in db/db mice.

Bacterial Trp metabolism occurs via transamination, decarboxylation, and tryptophanase pathways, producing metabolites that influence host physiology. Indole is generated via the tryptophanase pathway, while IPA is produced through transamination. Diabetic mice gavaged with saline demonstrated small intestine indole levels that were significantly reduced compared to WT mice (52.31±6.79 vs. 146.9±16.81; p<0.0001). Treatment with either LP-ACE2 or IW restored indole levels toward those observed in WT mice (Figure 3A), suggesting that both treatments enhanced Trp absorption and microbial indole production. As expected, IPA levels were reduced in db/db mice compared with the WT cohort (Figure 3B). Although LP-ACE2 treatment did not increase intestinal IPA levels, a significant increase was observed in the IW treated db/db mice.

Figure 3: IPA treatment mitigates gut barrier defects and improves glucose metabolism in db/db mice.

LP-ACE2 and IW treatment restores diabetes-induced depletion of indole levels measured by ELISA (n=4–8/group) and IPA levels presented as fold change and measured by LC/MS (n=4–5/group) (A-B). The presence of IPA in the WT retina was detected by MALDI imaging (n=5) (C). Schematic of the experimental design for IPA treatment regimen in db/db mice. Starting at two months of age, immediately after diabetes onset, mice were gavaged with IPA (daily) for six months. At study completion (eight months of age), tissues were collected for analysis. Saline-treated db/db mice and non-diabetic WT mice served as controls (D). The effect of IPA on gut epithelial and endothelial barrier dysfunction were evaluated using immunofluorescence staining. Representative images of intestinal sections showing the expression of epithelial junctional proteins ZO-1 (green) and p120-catenin (red) and endothelial barrier markers PV1 (green) and VE-cadherin (red) (E). The quantification of immunostaining of ZO-1(n=8/group) (F), p120-catenin (n=8/group) (G), PV-1 (n=6/group) (H), and VE-cadherin (n=6/group) (I). To assess the effect of IPA on glucose metabolism, the levels of incretins, GIP (n=6–7/group) (J) and GLP-1 (n=6–7/group) (K) were measured in plasma samples. Each black data point represents an individual experimental animal. The schematic of experimental design was created using BioRender, an online scientific illustration software. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with HSD correction for multiple comparisons. Scale bar = 50 μm (20X magnification).

IPA is a microbiota-derived Trp metabolite known for its potent antioxidant and anti-inflammatory properties. While IPA is produced in the gut by commensal bacteria, its biological influence extends systemically, implicating roles in distant organs beyond the intestine. To explore whether bacterial metabolites reached the retina, we utilized spatial mass spectrometry (MS). Spatial MS analysis confirmed the presence of IPA in the RPE layer which constitutes the posterior blood retinal barrier (Figures 3C & Figure S6). To understand IPA’s therapeutic potential, we investigated its effects on both intestinal and retinal endpoints, as outlined in Figure 3D, employing a combination of in vitro and in vivo approaches.

Immunohistochemical analysis revealed that IPA treatment significantly increased the expression of the intestinal epithelial markers, ZO-1 and p120-catenin (Figure 3E–G). Similarly, endothelial integrity of the intestinal vasculature was also improved, as evidenced by increased VE-cadherin expression and reduced PV1 expression (Figure 3E, 3H–I). Indole derivatives, including IPA, have been reported to improve glucose metabolism and insulin sensitivity, though the mechanisms remain unclear³³. In db/db-saline mice, GLP-1 levels were significantly reduced compared to WT controls (93.04±6.83 vs. 252.8±15.18; p<0.0001; Figure 3J), as were GIP levels (140.0±2.68 vs. 239.0±7.83; p<0.0001; Figure 3K). Treatment with IPA, LP-ACE2, or IW significantly increased GLP-1 and GIP plasma concentrations toward WT levels (Figures 3J, 3K, S2G, and S2H). Together, these findings suggest that IPA, similar to LP-ACE2 and IW, improves gut barrier integrity and enhances incretin secretion, potentially contributing to improved glucose homeostasis in diabetes.

IPA modulates TLR-4 mediated canonical signaling pathway and attenuates diabetes-induced intestinal and ocular inflammation.

Activation of TLR4 by metabolic stimuli triggers the recruitment of the adaptor protein MyD88, which initiates a downstream signaling cascade to promote transcription of pro-inflammatory genes, including IL-1β, and components of the NLRP3 inflammasome³⁴. Concurrently, this pathway primes and activates the NLRP3 inflammasome, leading to cleavage and activation of Caspase-1, which in turn processes IL-1β into its mature and active form. The release of these cytokines amplifies local and systemic inflammation, contributing to tissue damage in both intestinal and retinal environments under diabetic conditions^{35, 36}.

As shown in Figure 4A–E, db/db mice exhibited significantly elevated expressions of TLR4, MyD88, Caspase-1, and NLRP3 in lysate of the intestine as shown by western blots, indicating robust activation of this pro-inflammatory pathway. IPA treatment significantly mitigated diabetes induced activation of this canonical pathway.

Figure 4: IPA treatment preserves canonical signaling and mitigates diabetes induced inflammation.

To assess the impact of IPA on canonical signaling, the expression of TLR-4 and its downstream effectors MyD88, Caspase-1, and NLRP3 were detected in the small intestine tissue lysates by Western Blots and quantified (n=3/group) (A-E). The effect of IPA on TLR-4, MyD88, and IL-1β expression in RPE cells was detected using immunostaining and flow cytometry (F-K). Representative images of TLR4 (red) expression and its quantification (n=7–8/group) (F,G), the number of MyD88 positive cells measured by flow cytometry (n=3–4/group) (H, I), and IL-1β (green) staining and its quantification (n=7/group) (J,K). Each black data point represents an independent biological replicate. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons. In Western Blot images, each lane represents a sample from an individual experimental mouse. Scale bar = 50 μm (20X magnification).

Similar to its protective role in the intestine, IPA in the retina may help preserve cellular homeostasis, lessening inflammatory signaling and supporting vascular integrity. To evaluate the anti-inflammatory potential of IPA, we employed an in vitro approach using RPE cells. Human RPE cells were treated with LPS (1μg/mL) for 24 hours, in the presence or absence of IPA (10μM). IPA treatment significantly diminished LPS-induced TLR4 and MyD88 activation, as well as IL-1β expression (Figure 4F–K). Specifically, TLR4 expression was markedly elevated in LPS-treated cells but was significantly reduced following IPA treatment (p<0.0001; Figure 4F–G). Similarly, the number of MyD88-positive cells increased upon LPS exposure but returned to baseline levels with IPA treatment (p=0.0016; Figure 4H–I). A comparable reduction was observed in IL-1β expression (p<0.0001; Figure 4J–K), further supporting the anti-inflammatory role of IPA.

IPA regulates intestinal Th17 cell activity and mitigates inflammatory signaling between the gut and retina.

Given that the gut plays a vital role in immune regulation, we next examined IL-17, a cytokine produced by Th17 cells that orchestrates both innate and adaptive immune responses. The levels of IL-17 cytokine were significantly elevated (216.9±11.82 vs. 22.58±5.06; p<0.001) in the intestines of db/db mice compared with WT cohort. Significantly reduced levels of IL-17 were observed following treatment with either LP-ACE2 (144.8±22.87; p<0.008) or IW (106.1±13.42; p<0.0001) (Table S1).

To explore whether intestinal Th17 cells contribute to retinal inflammation in DR, we utilized CAG::KikGR transgenic mice, which express the photoconvertible fluorescent protein Kikume Green-Red. Upon targeted photoactivation in the intestine, RORγ⁺ Th17 cells convert from green to red fluorescence, enabling tracking of their migration from the intestine (site of photoactivation) (Figure 5A). Diabetic CAG::KikGR mice underwent photoactivation of the small intestine. Red-labeled intestinal RORγ⁺ Th17 cells were detected in the retina (Figure 5B) providing direct evidence of gut-to-retina trafficking of pro-inflammatory Th17 cells under diabetic conditions. Eyes were not exposed to photoactivation and resident RORγ⁺ Th17 cells in the retina remained green allowing for clear distinction between migratory and local Th17 populations. Flow cytometry revealed a significant elevation of RORγT⁺ Th17 cells in both the gut and retina of db/db mice (Figure 5C, D). IPA treatment reduced RORγT⁺ cell numbers in the small intestine (0.31±0.06 vs. 0.15±0.03) and in the retina (6.69±0.80 vs. 4.00±1.07) in db/db mice, supporting its role in restoring immune homeostasis.

Figure 5: IPA treatment modulates intestinal immune cell populations and reduces retinal immune cell infiltration in db/db mice.

A schematic illustrates the experimental approach involving photoconversion of intestinal immune cells and their migration to the retina (A). Immunostaining of retinas from KIK::GR mice revealed a higher number of pathogenic Th17 cells in diabetic mice (lower panel) compared to controls (upper panel) (B). Red Th17 cells represent cells that migrated from the photoactivated gut region to the retina. The effect of IPA treatment on hematopoiesis in diabetic mice was assessed by detecting long-term hematopoietic stem cells (LT-HSC⁺) (n=3–4/group) and granulocyte-monocyte progenitors (GMP⁺) (n=3–4/group) in the bone marrow (C,D), peripheral monocytes(n=3–4/group) and classical monocytes(n=3–4/group) (E,F) and Th17 cells in the gut (n=3–4/group) and retina (n=2–3/group) (G, H) using flow cytometry. Each black data point represents an independent biological replicate. The schematic of experimental design was created using BioRender, an online scientific illustration software. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons. Scale bar = 20 μm (20X magnification).

To further investigate the disruption of immune homeostasis in db/db, we assessed stem/progenitor populations in the bone marrow and peripheral blood monocytes (Figure 5E–H). db/db mice displayed a significant increase in short term repopulating-hematopoietic stem cells (HSC) cells, accompanied by an expansion of granulocyte myeloid progenitor (GMP⁺) populations (p=0.017) in the bone marrow, and elevated levels of CD11b⁺ monocytes and Ly6C⁺ classical monocytes in the circulation (Figure 5E–H) suggesting diabetes induced skewing of hematopoiesis toward the myeloid linage. IPA treatment significantly mitigated diabetes-induced alterations in bone marrow stem/progenitors and circulating monocytes.

LP-ACE2 and IW enhance AhR and PXR signaling to promote intestinal immune balance in db/db mice.

Activation of aryl hydrocarbon receptor (AhR) and pregnane X receptor (PXR) nuclear receptors is essential for maintaining intestinal immune homeostasis and epithelial barrier integrity³⁷. AhR regulates mucosal immune responses and epithelial barrier maintenance via IL-22 signaling, while PXR plays a significant role in detoxifying microbial metabolites, maintaining epithelial integrity, and suppressing NF-κB-mediated inflammatory pathways. Importantly, PXR is also known to transcriptionally regulate tight junction proteins and xenobiotic transporters, making it a pivotal mediator of host-microbe interactions and intestinal homeostasis. Western blot analysis revealed a significant upregulation of protein levels for AhR and PXR in LP-ACE2 and IW treated db/db mice (Figure 6A–C). These findings indicate that both treatments restore Trp-sensing nuclear receptor expression in the diabetic gut. In line with this, IL-22 levels were significantly reduced in untreated db/db mice compared to wild-type controls (64.38±7.08 vs. 157.9±12.96; p=0.0255; Table S1). However, both LP-ACE2 and IW treatments restored IL-22 expression to near-normal levels (153.6±44.52, p< 0.04 for LP-ACE2; and 154.4±34.05, p< 0.02 for IW), further supporting their role in promoting gut epithelial repair and immune regulation through AhR/PXR signaling. Collectively, these results suggest that restoration of AhR and PXR activity by LP-ACE2 and IW contributes to improved intestinal immunity and barrier function, representing a promising strategy to combat gut inflammation and systemic immune dysregulation in diabetes.

Figure 6: LP-ACE2, IW and IPA treatments attenuates diabetes-induced reduction of AhR and PXR expression in intestine and HIEC under diabetic mimicking conditions.

The effect of LP-ACE2 or IW treatment on AhR and PXR protein expression was determined by Western Blots and quantified (n=3/group) (A-C). After 24 hrs of DIZE, IW and IPA treatment, AhR and PXR expression in HIEC was determined under diabetic mimicking conditions. HIECs were exposed to LPS and high glucose (HG) for 24 hours to mimic diabetes in vitro. Representative images of AhR (green) (n=10–19/group) and PXR (red) (n=8–18/group) expression in all the cohorts and their quantification (D-F). Each black data point represents an independent biological replicate. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons. In Western Blot images, each lane represents a sample from an individual experimental mouse. Scale bar = 50 μm (20X magnification).

LP-ACE2, IW and IPA restore AhR and PXR expression in human intestinal epithelial cell cultures under diabetes mimicking conditions.

To validate our murine findings in a human system, we conducted in vitro studies using human intestinal epithelial cells (HIEC). HIECs were exposed to LPS and high glucose (HG) for 24 hours to emulate diabetic conditions following treatment with the ACE2 activator diminazene aceturate (DIZE), IW, and IPA. HIECs were subsequently stained to assess the expression of AhR and PXR. As shown in Figure 6D–F, LPS and HG exposure markedly downregulated both AhR and PXR expression, confirming impairment in Trp-sensing nuclear receptor signaling under diabetes mimicking conditions. These findings not only support the therapeutic potential of Trp-derived metabolites and ACE2-based interventions but also highlight the importance of Ahr and PXR as targetable nuclear receptors in restoring epithelial immune balance in diabetes induced intestinal inflammation.

Diabetes-induced retinal PXR downregulation and its restoration by IW in db/db mice.

Retinoid X receptors (RXRs) are nuclear receptors and critically regulate gene expression involved in vascular development, angiogenesis, and maintenance of the BRB. RXR expression is downregulated in both human and murine diabetic retinas, which contributes to retinal vascular pathology³⁸. PXR forms a functional heterodimer with RXR which regulates genes involved in xenobiotic metabolism, anti-inflammatory responses, and barrier protection, primarily in the liver and gut. However, emerging evidence suggests its relevance in the other tissues as well, where detoxification and regulation of inflammation are critical for tissue homeostasis.

Given this, we next investigated PXR expression in both human and mouse diabetic retinas. As shown in Figure 7A–D, PXR was detectable in normal human and mouse retinal tissues but was significantly downregulated in diabetes. In humans with DR, PXR expression was markedly reduced (15.06±2.06 vs. 23.83±2.71; p = 0.02), with a similar trend observed in db/db mice (33.06±2.34 vs. 56.03±4.61; p = 0.002) compared to controls. While LP-ACE2 treatment did not restore PXR expression in the diabetic retina, IW treatment enhanced PXR expression (64.28±1.10; p = 0.0006), suggesting a key role of Trp metabolites in maintaining PXR-mediated signaling in the diabetic eye. These results further support the role of PXR as a targetable regulator of retinal function and underscore the potential of Trp-derived interventions in protecting against DR.

Figure 7: IW restores diabetes-induced retinal PXR downregulation in db/db mice and DIZE, IW, and IPA enhance PXR expression in RPE and HRECs.

Representative images of human retina from healthy controls or diabetics showing PXR expression (red) and its quantification (n=5–6/group) (A,B). Representative images of PXR expression in the retinal tissue sections of all three experimental cohorts, WT, db/db treated with saline and db/db treated with IW and its quantification (n=3–4/group) (C,D). After 24 hours of exposure to DIZE, IW, or IPA, PXR expression was determined in RPE (green; E, G; n=7–8/group) and HRECs (red; F, H; n=6–8/group) under diabetic mimicking conditions. Each black data point represents an independent biological replicate. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons. In Western Blot images, each lane represents a sample from an individual experimental mouse. Scale bar = 50 μm (20X magnification).

Under diabetic mimicking conditions, Trp metabolites restore PXR expression in cells of the BRB.

RPE cells and retinal endothelial cells (RECs) are key cellular components of the outer and inner BRBs, respectively. RPE cells while maintaining photoreceptor homeostasis through phagocytosis of photoreceptor outer segments these cells also maintain the posterior BRB from the choroid³⁹. RECs form tight junctions that restrict vascular leakage and preserving retinal immune privilege. These tight junctions are essential for regulating the selective passage of molecules from the circulation into the retina or choroid⁴⁰.

Given that gut epithelial cells are responsive to Trp metabolites via nuclear receptors like PXR, we investigated whether RPE and human REC (HREC) exhibit similar responsiveness under diabetic mimicking conditions. RPE and HRECs were exposed to HG and LPS. Parallel groups were treated with DIZE, IW, and IPA for 24 hours and PXR expressions were assessed. As shown in Figure 7E–H, both RPE and HRECs expressed basal levels of PXR. However, HG+LPS treatment significantly reduced PXR expression, mimicking the diabetic inflammatory milieu. Remarkably, all three interventions, DIZE, IW, and IPA effectively restored PXR expression in both cell types, counteracting the suppressive effects of HG and LPS. Collectively, this shows that targeting Trp metabolism and PXR signaling may exert protective effects on retinal barrier cells, offering a promising strategy for managing DR and related complications.

Restoration of Metabolic, Retinal, and Immune Homeostasis in db/db mice following 3-month treatment with LP-ACE2, IW, and IPA

To evaluate the therapeutic potential of these compounds in established diabetes, 5-month-old db/db mice, representing approximately three months of chronic hyperglycemia, were treated with LP-ACE2, IW, or IPA for an additional three months, following the same protocol used in the prevention cohort (Figure 8A). As previously observed in the prevention arm, all three treatments demonstrated efficacy across multiple gut and retinal endpoints. Although none of the interventions resulted in significant weight loss (Figure S7A), they led to comparable improvements in glycemic control, as evidenced by reductions in fasting blood glucose (Figure 8B) and hemoglobin A1c levels (Figure 8C). Insulin sensitivity was enhanced, indicated by decreased serum insulin (Figure S7B), improved HOMA-IR scores (Figure S7C), and increased HOMA-IR % β-cell mass (Figure S7D). All three treatments elevated incretin expression, which promoted insulin secretion and suppressed glucagon levels, suggesting a shared mechanism of action (Figure S7E–F).

Figure 8: Intervention with LP-ACE2, IPA and IW in db/db mice significantly enhanced gut barrier integrity, preserved retinal structure and function, and modulated immune responses.

Schematic of experimental design shows the intervention regimen in db/db mice. Starting at five months of age, three months after the establishment of diabetes, db/db mice were gavaged with LP-ACE2 (three times per week), IW and IPA (daily) for another 3 months. At study completion (6 months of diabetes), tissues were collected for analysis. Saline-treated db/db mice and non-diabetic WT mice served as controls (A). The levels of blood glucose (n=3–7/group) (B) and HgA1c (n=3–7/group) (C) were measured in all five cohorts. The effects of LP-ACE2, IW and IPA on diabetes-induced gut epithelial and endothelial barrier dysfunction were evaluated using immunofluorescence staining. Representative images show expression of epithelial junctional proteins p120-catenin (red), ZO-1 (green) and endothelial barrier integrity markers PV1 (green) and VE-cadherin (red) (D). The quantification of epithelial markers (n=8–10/group) (E-F) and endothelial markers (n=6–10/group) (G-H) confirmed changes in these markers in saline-treated db/db mice with restoration in all three treatment cohorts. Plasma levels of gut permeability markers were measured by FABP2 ELISA (n=3–6/group) (I) and LPS (n=3–6/group) (J). Each black data point represents an individual experimental replicate. The schematic of experimental design was created using BioRender, an online scientific illustration software. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons. Scale bar = 50 μm (20X magnification).

Gut epithelial integrity was improved, as shown by increased ZO-1 and p120 catenin staining (Figure 8D–F), while endothelial integrity was supported by elevated VE-cadherin and reduced PV-1 expression (Figure 8D, G–H). Circulating markers of gut damage and endotoxemia, including FABP-2 (Figure 8I), LPS (Figure 8J), and PGN (Figure S7J), were significantly reduced in all treatment groups. MUC1, a key component of the mucosal barrier, was upregulated following treatment (Figure S7G, H), further supporting improved epithelial defense. IPA and other tryptophan metabolites markedly increased intestinal PXR expression (Figure S7G, I), while LP-ACE2 treatment enhanced epithelial ACE2 and B⁰AT1 expression, including ACE2:B⁰AT1 dimer formation (Figure S8A–C). As expected, LP-ACE2 elevated serum ACE2 levels (Figure S8E), whereas IW and IPA did not.

Diabetic mice exhibited the expected reductions in photopic a- and b-waves (Figure S9A–B) and scotopic a- and b-waves (Figure S9C–D). IW and IPA restored photopic a-wave amplitudes (Figure S9A), while photopic b-waves remained unchanged. All three treatments significantly improved scotopic a- and b-wave amplitudes (Figure S9C–D). Spatial frequency, a measure of visual acuity, was improved exclusively in IW-treated mice (Figure S9E). Spatial MS analysis of retinal sections revealed reduced L-Tryptophan and IPA levels and increased accumulation of toxic metabolites, including IS and IAAD, in the diabetic retina (Figure S9F). Treatment with LP-ACE2 and IW attenuated the elevation of IS and IAAD and restore L-Try abundance. Notably, neither LP-ACE2 nor IW improved the expression of IPA, whereas direct IPA supplementation enhanced IPA abundance in the retina. These findings suggest that while LP-ACE2 and IW provide partial protection against toxic tryptophan derivatives, IPA treatment uniquely restores beneficial metabolites and suppresses harmful ones in the diabetic retina. VE-cadherin immunostaining revealed increased expression in the retina across all treatment groups (Figure S9G), indicating enhanced blood-retinal barrier integrity.

In T2D, immune cells contribute to chronic-low grade inflammation, vascular damage, and insulin resistance. Using the gating strategies shown in Figure S10A–D, immune cell phenotyping was performed across cohorts in blood and small intestine. B cells (Figure S11A) in the blood of diabetic mice return to wild-type (WT) levels following IPA treatment. Similarly, CD8⁺ which are elevated in diabetic blood are reduced with LP-ACE2 and IPA (Figure S11B). CD4⁺ T cells, which are diminished in diabetes, are restored to WT levels with IW and IPA (Figure S11C). Th17 cells, which are elevated in the blood of db/db mice are unaffected by the treatments (Figure S11D). Classical monocytes (CD11b⁺F4/80⁺Ly6C⁺), elevated in diabetic blood, are significantly reduced by LP-ACE2 and IPA treatment (Figure S11E), whereas non-classical monocytes remain unchanged (Figure S11F). In the small intestine, B cells and CD8⁺T are unaffected by the treatments. CD4⁺ T cells are decreased in diabetes and are increased by LP-ACE2 or IW treatment. Th17 cells, which are elevated in the diabetic gut, are reduced only by IPA treatment. In the small intestine, proinflammatory M1 macrophages, upregulated in diabetes, are reduced by all three treatments. Conversely, Anti-inflammatory proangiogenic M2 macrophages are reduced in diabetes, but restored by IW or IPA treatment (Figure S11G–L). Collectively, these findings demonstrate that LP-ACE2, IW, and IPA are effective as interventional therapies in chronic diabetes, improving glucose homeostasis, reducing gut permeability, and ameliorating retinal dysfunction mirroring the benefits observed in the prevention cohort.

Retinopathy severity linked to L-tryptophan (L-Trp) metabolism in human plasma.

To determine the clinical relevance of our findings, we measured circulating levels of L-tryptophan (Trp) and its metabolites in diabetic individuals with and without DR and age matched controls. Targeted metabolomics profiling of 38 compounds was performed in plasma samples from the three cohorts. Demographic characteristics of all study subjects are given in Table S2. As shown in Figure 9A, plasma Trp levels were significantly reduced in T2D individuals compared to healthy controls (19.86±0.71 vs. 24.28±1.25; p=0.0012) with a further reduction in individuals with DR (16.5±0.53; p=0.0063).

Figure 9: Impaired tryptophan metabolism and barrier function in T2D subjects with DR.

Targeted metabolomic analysis was performed to assess L-Trp and its downstream metabolites in the plasma of T2D individuals and age- and sex-matched healthy controls. Plasma levels of L-Trp (n=18–29/group) (A), indole (n=15–24/group) (B), IPA (n=16–25/group) (C), IA (n=8–12/group) (D), ICAld (n=11–20/group) (E), ILA (n=12–25/group) (F), IGA (n=15–19/group) (G), and IAA (n=14–17/group) (H) in T2D subjects with and without DR. Plasma levels of deleterious indole metabolite IS in T2D subjects with and without DR (n=17–23/group) (I). Gut barrier integrity was evaluated by quantifying plasma levels of gut permeability markers: FABP2 (n=18–28/group) (J), PGN (n=18–29/group) (K), and LBP (n=16–28/group) (L) using ELISA. All markers were significantly altered in T2D subjects compared to controls with further changes correlating with DR severity. Each black dot represents an individual human subject. Data are presented as mean ± SEM. Statistical significance was determined using one-way ANOVA with Tukey’s HSD correction for multiple comparisons.

Unabsorbed dietary Trp is metabolized by gut microbiota into a range of bioactive metabolites, including indole. Indole levels were significantly elevated in both T2D (0.74±0.07 vs. 0.38±0.03; p=0.0055) and T2D with DR (0.79±0.08; p=0.0010) groups compared to healthy controls (Figure 9B). Indole can be metabolized into either beneficial metabolites such as IPA, indole acrylic acid (IA), indole-3-carboxaldehyde (ICAld), indole-3-lactic acid (ILA), indole-3-glyoxylic acid (IGA), and indole-3-acetic acid (IAA) or harmful derivatives such as indole sulfate (IS). IPA levels were significantly reduced in T2D (140.5±16.29;p<0.0001) and further decreased in T2D with DR (80.04±8.32) compared to healthy individuals (272±32.25; p<0.0001) (Figure 9C). Similar to IPA, the levels of IA and ICAld were also reduced significantly in T2D subjects with further reduction in T2D with DR (Figure 9D, 9E). Unexpectedly, the levels of ILA, IGA, and IAA were significantly elevated in both T2D and T2D with DR subjects (Figure 9F–H). We reasoned this may represent a compensatory increase. The level of detrimental metabolite IS was significantly elevated in T2D patients with and without DR (Figure 9I). In addition to the eight indole metabolites, the levels of eleven other Trp metabolites, 2-Aminobenzoic acid (2-AA), Indole-3-acetamide (IAM), Serotonin (SER), Quinolinic acid (QA), L-kynurenine (L-KYN), N-Formylanthranilic acid (NFAA), N-formylkynurenine (NFK), 5-Hydroxyindoleacetic acid (5-HIAA), Kynurenic acid (KYNA), Xanthurenic acid (XA), and Picolinic acid (PA) were altered across the three cohorts (Table S3). The levels of 2-AA (20.42±4.55; p=0.04 and 14.05±2.0), SER (62.45±6.96; p=0.01 and 52.40±6.76; p=0.003), and 5-HIAA (15.18±1.21; p=0.002 and 17.30±1.71; p<0.0001) were significantly altered in T2D and T2D with DR, compared with healthy subjects (9.56±1.04, 117.7±23.09, and 8.22±0.55), respectively, while the levels of others were unchanged. Among the 38 compounds, 19 metabolites were undetectable in all groups: 2-aminophenol (2AF), 3-hydroxyanthranilic acid (3-HAA), 2-amino-3-methoxybenzoic acid (3-MeOAA), 3-hydroxykynurenine (3HKYN), N-methyltryptamine (MTRM), melatonin (MLT), 6-hydroxymelatonin (6-HMLT), 5-methoxyindoleacetic acid (5-Me-IAA), hydroxytryptophan (5-HTP), 5-hydroxytryptophol (5-HTOL), 3-indoxyl-β-D-glucopyranoside (indican), 3-indoleacetonitrile (IACN), tryptophol (IET), tryptamine (TRM), N-acetylserotonin (NAS), 5-methoxytryptamine (5-MeOT), 2-ketoadipic acid (2-KA), Cinnabarinic acid (CA), and nicotinic acid (N-acid). These findings underscore the complex interplay between beneficial and detrimental Trp metabolites in the pathogenesis of vascular complications in diabetes, particularly diabetic retinopathy.

Intestinal barrier function is essential for maintaining host health.

Previously, we demonstrated in T1D individuals with DR that plasma levels of intestine-specific fatty acid binding protein 2 (FABP2), an indicator of damaged gut epithelium were elevated. In the present study, we observed significantly higher FABP2 levels in individuals with T2D compared to healthy controls (20.61±2.98 vs. 6.86±0.93; p=0.0424), with a further increase in those with DR (36.47±4.99; p=0.0134), suggesting a link between disrupted L-tryptophan metabolism and gut barrier dysfunction in DR (Figure 9J).

Gut barrier impairment contributes to endotoxemia, which we assessed by measuring plasma levels of the GMA, peptidoglycan (PGN). PGN levels were significantly elevated in both T2D and T2D with DR cohorts (45.10±2.35 and 52.15±2.68, respectively) compared to healthy individuals (25.25±1.22) (Figure 9K). Additionally, levels of lipopolysaccharide-binding protein (LBP), a key mediator of immune recognition of LPS, were significantly increased in both diabetic groups (Figure 9L). Collectively, these findings highlight a strong association between altered Trp metabolism and gut barrier dysfunction in diabetes, with IPA and FABP2 emerging as potential predictive biomarkers for the presence of DR.

Discussion:

This study collectively supports a mechanistic link between Trp metabolism, gut barrier integrity, and the development of DR. Trp absorption occurs via two distinct intestinal pathways: an ACE2-dependent mechanism (ACE2:B⁰AT1) and an ACE2-independent mechanism (SLC15A1/PEPT1), both of which are impaired in T2D rodent models. Notably, the Ile-Trp dipeptide, absorbed via SLC15A1, bypasses the impaired ACE2:B⁰AT1 pathway in diabetes and exerts protective effects on the intestine comparable to those observed with use of LP-ACE2, a genetically engineered probiotic that enhances the activity of the protective arm of intestinal RAS system²¹. Both interventions enhance incretin secretion, reduce hyperglycemia, and restore gut epithelial and endothelial barrier integrity and underscore the importance of the gut-retina axis in DR pathogenesis. Elevated circulating gut leakage markers indicate endotoxemia and result in increased production of immune cells from both the adaptive and innate immune systems. Using multiple functional and structural endpoints including acellular capillary counts (a marker of vasodegeneration), and ERG, we demonstrate that both LP-ACE2 and IW significantly improve retinal outcomes confirming disease modification through gut-targeted interventions. Enhancing gut Trp levels or administering IPA prevents the development of DR, in part by reducing Th17 cell recruitment to the retina and limiting both innate and adaptive immune cell infiltration into the retina. Th17cell homing to the retina is primarily mediated by CCR6/CCL20 interaction, as well as upregulation of VCAM-1 and ICAM-1 on retinal endothelial cells^{41, 42}. In diabetes, these adhesion molecules are upregulated, facilitating T-cell infiltration. Our data shows that IPA can reduce the recruitment of Th17 cells to the retina.

Trp and IPA levels are significantly higher in healthy individuals compared to those with diabetes, with the lowest levels observed in individuals with DR. Plasma levels of Trp and IPA inversely correlate with gut barrier dysfunction and endotoxemia, suggesting a systemic link between microbial metabolism and vascular health. Moreover, a stepwise reduction in Trp and IPA levels across experimental cohorts suggests a threshold-dependent influence on disease progression. IPA further promotes intestinal homeostasis by upregulating PXR and AhR, which suppress NLRP3 and TLR-4-mediated inflammatory signaling in db/db intestine and RPE cells. IPA reduces the recruitment of Th17cells to the retina, highlighting its anti-inflammatory potential.

While LP-ACE2 treatment improved gut barrier integrity and reduced inflammation, its effect on microbial alpha diversity was somewhat paradoxical. Alpha diversity within a single sample reflects both species’ richness (the number of species) and evenness (the relative abundance of each species). We reasoned that LP-ACE2 selectively promotes the growth of beneficial commensals, which dominate the microbial community and drive the observed physiological improvements, namely, enhanced barrier function and reduced inflammation while suppressing deleterious species. This may occur without significantly altering the total number of species.

We observed distinct differences in the effects of LP-ACE2 and IW on specific ERG waveforms. Although we did not investigate the underlying mechanisms responsible for these differences, LP-ACE2 is known to increase Ang 1–7 levels, which can exert vasodilatory, anti-inflammatory, anti-fibrotic, and anti-thrombotic effects across various tissues, including the retina. Notably, LP-ACE2 had a more pronounced impact on the b-wave, which reflects inner retinal activity. Given that LP-ACE2 elevates serum ACE2 levels, it may enhance retinal blood flow, particularly through the central retinal artery, which supplies the inner retina. IW also enhanced b-wave amplitudes, likely by improving the function of ON bipolar cells, the primary generators of the b-wave. Their activity reflects the integrity of the inner nuclear layer and retinal synaptic transmission. Additionally, the b-wave incorporates contributions from Müller cells, further underscoring its role as a composite measure of inner retinal health. While we currently lack direct evidence of specific effects of Trp metabolites on these cell types, this remains an active area of investigation.

We are at a watershed moment in human biology where an auxiliary microbial genome has been identified that affects human cell function and can be leveraged to tackle complex diseases such as DR^43–46. Trp-derived bacterial metabolites exhibit a biosignature specific to DR. We found that levels of IPA are reduced in individuals with T2D, and even further diminished in those with T2D and DR. Similarly, IPA decreases in db/db mice, a well-established model of T2D. However, dietary supplementation with IW restores IPA levels, likely by enriching populations of Trp-metabolizing bacteria that convert unabsorbed Trp into bioactive compounds such as IPA⁴⁷. IPA exhibits anti-inflammatory, antioxidant, and neuroprotective effects⁴⁸, while stimulating the expression of tight junction proteins in both the intestinal epithelium and the RPE⁴⁹ thus inhibiting the penetration of toxic factors from the systemic circulation. IPA improves glucose metabolism⁴⁹, potentially reducing the risk of T2D and beneficially influencing liver lipid synthesis³³. IPA has antimycobacterial activity,⁵⁰ enhances mitochondrial function⁵¹ and has cardioprotective effects^{52, 53}. Importantly exogenous IPA administration even at high doses has minimal toxicity⁵⁴. However, not all Trp metabolites are beneficial. The IS is a metabolic toxin in diabetes, contributing to oxidative stress, renal injury, and vascular complications⁵⁵. Its levels reflect both gut microbial activity and renal clearance capacity⁵⁶, making it a potential biomarker and therapeutic target. In diabetes, elevated IS levels have been linked to worsening vascular and renal outcomes⁵⁷.

Trp is metabolized to indole by the enzyme tryptophanase, which is expressed by over 85 species of Gram-positive and Gram-negative enteric bacteria^{37, 58}. IPA showed beneficial effect on bone marrow resulting in increased retention of long-term repopulating HSC suggesting its role in improving the microenvironment necessary for HSC maintenance and self-renewal⁵⁹. Trp metabolites, including IPA, activate AhR, a transcription factor that regulates immune function⁵⁸. Upon activation, AhR influences the differentiation of CD4⁺T cells into regulatory T cells (Tregs) and Th17 cells⁵⁸. While Tregs suppress inflammation, excessive activity of Th17 cells is implicated in autoimmune and inflammatory diseases. AhR activation also promotes the production of anti-inflammatory cytokines IL-22 and IL-10, which are essential for maintaining epithelial homeostasis both in the gut and retina^{37, 60, 61}. In line with these mechanisms, WT mice exhibited higher intestinal levels of indole, AhR expression, IL-22, and IL-10 compared to db/db mice. Treatment of db/db mice with LP-ACE2 or IW increased intestinal indole levels and restored AhR, IL-10, and IL-22 expression, demonstrating the anti-inflammatory potential of Trp metabolites in diabetes. These findings were further validated in human epithelial cells, confirming translational relevance.

In addition to AhR, Trp metabolites such as IPA also activate PXR, which downregulates TNF-α and enhances the expression of tight junction proteins like ZO-1 and p120-catenin, thereby strengthening epithelial barrier function⁵⁸. Both ACE2-dependent and ACE2-independent strategies that we utilized to enhance Trp absorption increased PXR expression and tight junction protein levels in gut and RPE cells, while reducing TNF-α expression. Gut leakage markers including PGN, LBP, and LPS were significantly elevated in db/db mice and were reduced following treatment with LP-ACE2, IW or IPA. These interventions also increased the villus/crypt ratio and proliferation index in the small intestine, enhancing absorptive surface area and function. Interestingly, IW treatment unexpectedly increased ACE2:B⁰AT1 dimerization. We hypothesize that this may be due to the ACE-inhibitory properties of this dipeptide, which could lead to a compensatory increase in endogenous ACE2 expression within the gut epithelium which we have previously shown results in concomitant B⁰AT1 expression⁶².

While all three treatments showed benefit to the gut and retina, their mechanisms are different. LP-ACE2 treatment prominently enhanced intestinal and systemic ACE2 expression which we previously showed also improved recruitment of myeloid angiogenic cells to the intestinal vasculature²¹. IW absorbed via PEPT1, bypasses ACE2-dependent Trp uptake and serves as a precursor for microbial IPA synthesis. Although IW is partially hydrolyzed by aminopeptidase N and DPP-IV, it retains sufficient stability for intestinal transport and bioactivity⁶³. IW has antioxidant and ACE-inhibitory properties, along with its ability to activate AhR/PXR signaling, position IW as a promising candidate for therapeutic development as a functional food additive or nutraceutical. IPA modulated immune responses via AhR and PXR signaling to enhance retinal and gut barrier function. Gut derived IPA can reach the retina and is localized to the RPE layer supporting its role as a mediator of posterior blood retinal barrier integrity. All three interventions LP-ACE2, IW and IPA stimulated GLP-1 and GIP secretion from enteroendocrine cells, contributing to both local anti-inflammatory effects and systemic metabolic improvements. IPA treatment alone also significantly lowered fasting glucose levels compared to controls. These findings suggest that incretin activation represents an additional mechanism for Trp-based interventions and provides further support to the gut-retina axis by mitigating vascular pathology in T2D.

DR has been shown to be associated with changes in the gut microbiome^64–66. Microbiome analysis revealed expected reductions in alpha diversity in db/db mice, with IW treatment significantly restoring diversity toward WT levels. Interestingly, beta diversity was increased in db/db mice, but this was driven by expansion of deleterious rather than commensal species. While neither LP-ACE2 nor IW altered microbiome composition at the phylum level, important shifts were observed at the genus level, particularly in Akkermansia, Bacteroides, Lachnospiraceae, Eubacterium, and Lactobacillus. Akkermansia, known for its anti-inflammatory properties and ability to enhance gut barrier integrity, is also a producer of IPA^{48, 67}. IPA is primarily produced by commensal Clostridium spp., particularly from Clostridium cluster IV and XIVa. The commensal Clostridium spp. requires higher levels of Trp in the intestine to flourish and the bacteria favor a specific anaerobic niche and metabolic environment^{51, 68, 69}.

This study has several limitations. First, we used only male mice, which limits generalization of findings to females. Second, the diabetic model employed primarily captures early-stage pathological features of DR, such as retinal inflammation, vascular leakage, and an increased number of acellular capillaries. While this model is well-suited for investigating the initial molecular and cellular events associated with DR onset, it does not fully replicate the advanced retinal pathology observed in later stages of human DR. Another limitation is that while we show a reduction in IPA levels and an increase in IS in DR subjects, this does not demonstrate causality and additional studies need to be performed to determine whether levels of IPA and IS represent biomarkers that can predict disease progression or the response to systemic therapies that target the gut-retinal axis. In addition, other Trp-derived metabolites that we did not study, such as indole-3-lactic acid (ILA), indole-3-acetaldehyde (IAAld), and kynurenine pathway intermediates, may also influence gut and retinal homeostasis. Future studies are needed to explore the impact of these metabolites on the gut-retina axis, which may uncover additional biomarker candidates. Finally, we have not yet identified the specific transporters responsible for IPA uptake in RPE and HREC cells. Although IPA is a small, lipophilic molecule likely capable of passive diffusion into the retina to activate AhR and PXR, it may also be transported by organic anion transporters (OATs), such as members of the SLC22 family, which are expressed in retinal tissues⁷⁰. Investigating the mechanisms of IPA entry into retinal cells will be a focus of future research. Finally, our intestinal immunofluorescence analyses are semi-quantitative and were not validated using western blotting or ELISA, which limits the robustness of some of the findings.

Conclusion

This study demonstrates that gut-targeted interventions, including genetically modified food-grade probiotics (LP-ACE2), Trp-containing dipeptides (IW), and direct administration of beneficial Trp metabolites like IPA, represent viable nutraceutical strategies to prevent vascular complications such as DR. Importantly, the systemic nutraceutical strategies described can be combined with the current standard of care direct ocular treatments. Our findings highlight that bacterial metabolites reach the retina where they function to enhance retinal barrier integrity and reduce inflammatory signals.

What is already known on this topic:

1. Diabetic retinopathy (DR) is the most common diabetic microvascular complication, but the mechanisms causing this condition remains incompletely understood.

2. DR is associated with disruption of the gut-retinal axis.

What this study adds:

1. Activation of the ACE2 dependent pathway for Trp absorption corrected diabetes induced gut barrier dysfunction, increasing incretin secretion, and reducing intestinal inflammation and prevented development of DR in a T2D murine model.

2. Nutraceutical supplementation with Trp containing dipeptides prevented development of DR and corrected diabetes induced gut dysfunction similar to what was observed through the ACE2 dependent pathway of Trp absorption.

3. Indole-3-propionic acid (IPA) levels are reduced in individuals with DR compared to diabetics without retinopathy and age-sex matched controls.

4. Spatial mass spectroscopy of healthy murine retina demonstrated IPA in the retinal pigment epithelial cells supporting that this bacterial metabolite serves to protect the posterior blood retina barrier (BRB).

5. In a mouse model of T2D, IPA supplementation either as a prevention or intervention attenuates hyperglycemia, gut inflammation, intestinal and retinal epithelial and endothelial barrier damage and prevented development of DR in a mouse model of T2D.

6. IPA increases PXR expression and activation resulting in reduced NLRP3 activation and IL-1 expression in gut and retinal epithelial cells.

How this study might affect research, practice, or policy:

This study reveals that impaired tryptophan metabolism and reduced levels of IPA contribute to the vision threatening complication, DR. The study supports the use of nutraceuticals and engineered probiotics to restore gut barrier function and modulation of inflammation, offering a novel, safe, inexpensive therapeutic strategy and also the potential use of IPA as biomarker for diabetic eye disease thus fostering early disease intervention.

Non-standard Abbreviations and Acronyms:

T1D

Type 1 Diabetes

T2D

Type 2 Diabetes

DR

Diabetic Retinopathy

OKN

Optokinetic nystagmus

ERG

Electroretinography

RAS

Renin-angiotensin system

ACE

Angiotensin converting enzyme

Ang II

Angiotensin

AT1R

Angiotensin type 1 receptor

ACE2

Angiotensin converting enzyme 2

hACE2

Human ACE2

B0AT1

Sodium-coupled neutral amino acid transporter

SLC6A19

Solute carrier family 6 member 19

LP

Lactobacillus paracasei

GMA

Gut microbial antigens

FABP2

Fatty acid binding protein 2

LBP

Lipopolysaccharide binding protein

LPS

Lipopolysaccharide

PGN

Peptidoglycan

IW

Isoleucine-tryptophan dipeptide

PepT1

Peptide transporter 1

PXR

Pregnane X receptor

SLC15A1

Solute carrier family 15 member 1

TRP

Tryptophan

IPA

Indole-3-Propinoic Acid

ILA

Indole-3-Lactic Acid

IA

Indole Acrylic Acid

IAA

Indole-3-Acetic Acid

IGA

Indole-3-Glyoxylic Acid

IS

Indole Sulfate

ICAld

Indole-3-Carboxaldehyde

Data availability statement:

Data are available upon reasonable request from the senior corresponding author, presented in the main text or supplementary materials. Correspondence and requests for materials should be addressed to Maria Grant.