Growth differentiation factor 15 alters intestinal barrier and increases permeability: A new molecular target in inflammatory bowel disease

Abstract

BACKGROUND

Inflammatory bowel disease (IBD) is a group of chronic, inflammatory disorders that include Crohn’s disease and ulcerative colitis. IBD arises from the interaction of various environmental and genetic factors. Altered gut permeability and mitochondrial stress in the colonic mucosa are two mechanisms previously implicated in IBD pathogenesis. We have previously demonstrated activation of the mitochondrial unfolded protein response (UPR^mt) in the colonic mucosa of IBD patients and linked this activation to pro-inflammatory signaling. Growth differentiation factor 15 (GDF15) is an important downstream mediator of the UPR^mt.

AIM

To investigate whether GDF15 has a role in IBD and how GDF15 impacts colonic epithelium.

METHODS

Circulating levels of GDF15 were assessed in plasma samples from IBD patients and healthy controls using an enzyme-linked immunosorbent assay. To study the effects of GDF15 on the colonic mucosa, we employed two different in vitro culture models: Colonic organoids and T84 cells.

RESULTS

We found that circulating GDF15 Levels were elevated in IBD patients and correlated with markers of inflammation (C-reactive protein) and intestinal permeability [haptoglobin and lipopolysaccharide-binding protein (LBP)]. Additionally, we demonstrated that GDF15 alters the intestinal barrier and increases permeability by decreasing the levels of zonula occludens 1 and claudin 1, critical components of tight junctions. Thus, our findings confirm previous reports of increased circulating GDF15 levels in IBD patients and the activation of UPR^mt.

CONCLUSION

In the present study, we describe a novel mechanism in IBD pathophysiology, linking mitochondrial stress to the disruption of the intestinal barrier and increased intestinal permeability.

Key Words: Growth differentiation factor 15; Inflammatory bowel disease; Crohn's disease; Ulcerative colitis; Mitochondrial stress; Intestinal permeability; Tight junctions

Core Tip: Mitochondrial dysfunction in intestinal mucosa is emerging as a novel mechanism in inflammatory bowel disease pathogenesis. In the present study, we show that growth differentiation factor 15 (GDF15), a marker of mitochondrial stress, is elevated in plasma from both ulcerative colitis and Crohn’s disease patients. Furthermore, levels of GDF15 correlate with markers of inflammation and intestinal permeability. In vitro assays using colonic organoids further showed that GDF15 alters the intestinal barrier, reducing the levels of zonula occludens 1 and claudin 1, and subsequently increases intestinal permeability. Targeting GDF15 may offer a promising strategy to prevent disruption of the intestinal barrier, potentially reducing immune overactivation.

INTRODUCTION

Inflammatory bowel disease (IBD) is a group of chronic, inflammatory disorders that affect the gastrointestinal tract. IBD primarily includes Crohn’s disease (CD) and ulcerative colitis (UC). The development of IBD is thought to result from the interaction of multiple environmental factors in genetically predisposed individuals. Alteration of gut permeability is a key mechanism in IBD pathophysiology, facilitating the contact between epithelial cells and gut microbiota and toxins, which further activate the immune system and trigger proinflammatory pathways[1,2]. Another, yet poorly understood, mechanism in IBD is the disruption of mitochondrial homeostasis[3]. A better understanding of the specific pathogenetic mechanisms involving mitochondrial dysfunction is needed to uncover potential therapeutic targets within this pathway. Our group has previously reported activation of the mitochondrial unfolded protein response (UPR^mt) linked to pro-inflammatory signaling in IBD[4]. To further investigate the role of UPR^mt in IBD, we aimed to study growth differentiation factor 15 (GDF15), a member of the transforming growth factor (TGF)-β superfamily and a downstream mediator of UPR^mt.

GDF15 is considered a biomarker of mitochondrial dysfunction, and its circulating levels are elevated in primary mitochondrial disorders[5,6]. However, GDF15 is also gaining much attention beyond the field of mitochondrial disorders, as its levels have been found to be increased in many other diseases and conditions. Previous studies have shown that GDF15 contributes to regulating metabolism and appetite in diabetes mellitus and obesity, and increased circulating GDF15 levels are also associated with cardiovascular diseases and chronic lung diseases[7-11]. A 2022 study reported increased GDF15 levels in IBD patients with sarcopenia in a Japanese cohort of CD patients[12]. More recently, it was reported that circulating GDF15 levels are increased in IBD patients compared to healthy controls[13]. However, this finding requires validation in an independent cohort, and the role of GDF15 in IBD pathophysiology remains to be fully elucidated. We previously observed that, in colonic organoids, GDF15 and other UPR^mt components are elevated following treatment with non-steroidal anti-inflammatory drugs, which are known to exacerbate disease in IBD[4]. In the present study, our goal was to confirm elevated circulating levels of GDF15 in IBD patients compared to healthy controls and to investigate the role of GDF15 in the molecular and cellular mechanisms underlying intestinal inflammation.

MATERIALS AND METHODS

Subjects

Healthy controls (HC) (volunteers from the colon cancer screening program with non-pathological findings) and patients with a confirmed IBD diagnosis attending for a follow-up colonoscopy were recruited at the Gastroenterology Department of the Virgen de la Victoria University Hospital (Málaga, Spain). Myocardial infarction (MI) patients were used as positive controls for high plasma levels of GDF15. MI patients were recruited at the Cardiology Department of the Virgen de la Victoria University Hospital (Málaga, Spain). The demographic and clinical data of subjects are described in detail in Tables 1 and 2. All participants gave their written informed consent, the study protocol was carried out in accordance with the ethical guidelines of the Declaration of Helsinki, and the study was approved by the Málaga Provincial Research Ethics Committee, Málaga, Spain (No. PI0244-2021, No. ORG-Digest-2021, and No. PI-131-2020). All subjects were older than 18 years.

Table 1 Demographic characteristics of subjects and biochemical data.

	HC (n = 23)	IBD (n = 39)		MI (n = 19)	HC vs IBD
		CD (n = 21)	UC (n = 18)
Sex ratio, female/male (%)	11/12, (47.8/52.2)	11/10, (52.4/47.6)	11/7, (61.1/38.9)	10/9, (52.6/47.4)	NS
Age (years)	45.4 ± 13.6	40.48 ± 12.6	46.3 ± 15.0	78.4 ± 5.8	NS
BMI	24.0 ± 3.3	24.3 ± 5.1	24.2 ± 3.5	28.8 ± 3.9	NS
Current smoker, yes/no (%)	4/19, (17.4/82.6)	8/13, (38.1/61.9)	3/15, (16.7/83.3)	0/19, (0.0/100.0)	NS
Fecal calprotectin	N/A	785.9 ± 1128.2	663.6 ± 1213.7	N/A	N/A
CRP	4.97 ± 5.66	11.2 ± 17.5	8.8 ± 11.3	22.1 ± 23.1	P = 0.006
Glucose	83.5 ± 10.1	85.4 ± 19.9	103.7 ± 81.3	147.8 ± 43.4	NS
Creatinine	0.81 ± 0.16	0.75 ± 0.12	0.76 ± 0.20	1.3 ± 0.4	NS
Urea	26.0 ± 6.3	23.3 ± 11.4	26.2 ± 7.2	52.3 ± 16.2	NS
LDL	123.7 ± 26.2	95.5 ± 32.5	102.1 ± 27.9	85.1 ± 26.5	NS
Haptoglobin	130.4 ± 56.6	147.4 ± 32.9	158.2 ± 97.2	N/A	NS

For age, body mass index, and biochemical parameters from blood test, mean ± SD is represented. HC: Healthy controls; IBD: Inflammatory bowel disease; CD: Crohn’s disease; UC: Ulcerative colitis; MI: Myocardial infarction; CRP: C-reactive protein; BMI: Body mass index; LDL: Low-density lipoprotein; NS: No significance; N/A: Not applicable.

Table 2 Clinical characteristics of patients, n (%).

	Crohn’s disease	Ulcerative colitis
n	21	18
Time of disease (years)	11.3 ± 9.0	13.0 ± 8.5
Treatment
Corticoids	6 (28.6)	2 (11.1)
5-aminosalicylic acid	3 (14.3)	14 (77.8)
Sulfasalazine	3 (14.3)	1 (5.6)
Azathioprine	4 (19.0)	4 (22.2)
anti-TNF	3 (14.3)	1 (5.6)
Disease phenotype	Inflammatory 8 (38.1)	Left-sided colitis 8 (44.4)
	Stricturing 4 (19.0)	Extensive colitis 9 (50.0)
	Fistulizing 7 (33.3)

For the time of disease, mean ± SD is represented. TNF: Tumor necrosis factor.

Samples

Colon biopsy samples were obtained from normal-appearing mucosa from HC and CD patients, and were processed immediately after their reception at the Virgen de la Victoria University Hospital Biobank (Andalusian Public Health System Biobank) either for RNA isolation or for the development of colonic organoids. Blood was collected, and complete biochemistry tests were run. Plasma was isolated immediately and aliquoted for enzyme-linked immunosorbent assay assays. Fecal calprotectin values within ± 1 month were obtained for IBD patients. Blood samples from MI patients were obtained within 24 hours after stabilization, and plasma was isolated and aliquoted immediately.

Assessment of biomarker levels in plasma

Plasma GDF15 levels and plasma lipopolysaccharide-binding protein (LBP) were determined by enzyme-linked immunosorbent assay (No. EHGDF15X10 for GDF15 and No. EH297RB for LBP, Invitrogen, CA, United States) following the manufacturer’s instructions. One aliquot of plasma from each participating subject was thawed. Each sample was run in duplicate, and the average of the duplicates was calculated. A standard curve was used to determine the concentration of each sample in ng/mL.

Messenger RNA isolation and assessment of gene expression

RNA was isolated using QIAzol (QIAGEN Science, Hilden, Germany) following the manufacturer’s procedure and quantified by spectrophotometry (Nanodrop 2000, Thermo Scientific, Waltham, MA, United States). Complementary DNA was synthesized using M-MLV reverse transcriptase (Promega, Madison, WI, United States) from 500 ng of RNA. Gene expression in colonic biopsies was assessed by quantitative polymerase chain reaction in a LightCycler 480 (Roche Diagnostics S.L., Barcelona, Spain) using 1 μL from a 1/5 dilution of the complementary DNA, SensiFAST™ SYBR^® Hi-ROX (Bioline, London, United Kingdom). The forward (F) and reverse (R) primers used have been previously described[4]. The ΔCt method was used to analyze the data. S18 was used as a housekeeping gene. Within each assay, each sample was run in triplicate (technical triplicates), and the average of the triplicates was used for statistical analysis. Gene expression in T84 cells was assessed by quantitative polymerase chain reaction in a QuantStudio 12K Flex (384) (Applied Biosystems, Barcelona, Spain) using specific TaqMan™ assays: Zonula occludens (ZO)-1 (No. Hs01551871), occludin (No. Hs05465837), claudin 1 (No. Hs00221623), and claudin 2 (No. Hs00252666). RNase P (Cat. No. 4316844, Applied Biosystems, Barcelona, Spain) was used as a housekeeping gene. Each sample was run in duplicates, and the ΔCt method was used to analyze the data.

Cell cultures and in vitro assays

Colonic organoids were developed from colonic biopsies using the medium IntestiCult™ Organoid Growth Medium for human (catalog number: No. 100-0190, Stemcell Technologies, Saint Égrève, France), following the manufacturer’s protocol. Colonic organoids (passage 3-5) were seeded in a 10 μL dome of 1:1 dilution of DMEM: F12 [15 mmol/L N-2-hydroxyethylpiperazine-N’-2-ethanesulphonic acid, 1% bovine serum albumin (BSA)] and Matrigel on an 18-well μ-slide (Cat. No: 81816, IBIDI GmbH, Gräfelfing, Germany). After 10 minutes of incubation at 37 °C, 100 μL of intesticult organoid growth medium (1 × penicillin/streptomycin) was added. The day after seeding, the medium was replaced with 50 μL of either fresh intesticult medium or fresh intesticult medium containing 100 or 400 ng/mL of GDF15 (catalog number: No. 32-1249, Abeomics, San Diego, CA, United States). The following day, 50 μL of intesticult was added, either alone or containing GDF15 at 100 or 400 ng/mL, or interferon gamma (IFNg) (catalog number: No. 32-6390, Abeomics, San Diego, CA, United States) at 100 ng/mL for a final IFNg concentration of 50 ng/mL. Concentrations of GDF15 and IFNg were based on previous publications[14,15]. For assessment of the intestinal barrier function, 11 μL of 1mg/mL of fluorescein isothiocyanate (FITC)-Dextran 4K (No. FD4-100MG, Sigma-Aldrich, Burlington, MA, United States) was added on the following day, and organoids were visualized on a SP5 confocal microscope (Leica Microsystems, Wetzlar, Germany) 24 hours later. Three representative areas from inside and outside of each analyzed organoid were measured (mean gray value), and mean values were calculated. To correct for the background signal, we calculated the ratio between the mean gray value inside and the sum of the mean gray values inside and outside the colon organoid. For each organoid line, at least 10 organoids in three different wells (a total of at least 30 organoids) were analyzed for each condition.

For assessment of tight junction proteins, the media was removed 24 hours after adding IFNg, and organoids were fixed with 10% formalin, permeabilized with Triton X 0.2%, and blocked with phosphate buffered saline (PBS) 1 × BSA 2%. Primary antibody against ZO-1 (No. ZO1-1A12, Invitrogen, CA, United States) was added at a 1:1000 dilution in PBS 1 × 0.1% BSA, and incubated 24 hours at 4 °C. Afterwards, primary antibody was washed and secondary antibody against mouse immunoglobulin G (IgG) AlexaFluor^®488 (No. ab150113, Abcam, Cambridge, United Kingdom) was added at a 1:2000 dilution in PBS 1 × and incubated 1 hour at room temperature. Secondary antibodies were washed, and 4’-6-diamidino-2-phenylindole (DAPI) (No. 62247, Thermo Scientific, Waltham, MA, United States) was added. Cells were incubated for 6 minutes, and DAPI was washed to remove excess. Wells were covered with ProLong^® diamond antifade mountant (catalog number: No. 39695, Life Technologies, Carlsbad, CA, United States), and organoids were visualized 24 hours later on a SP5 confocal microscope. For each organoid line, at least three organoids in three different wells were analyzed. Fluorescence was measured as integrated density, and values from ZO-1 were normalized to fluorescence from DAPI.

T84 cells (ATCC^® CCL-248™) were obtained from the European Collection of Authenticated Cell Cultures. All reagents were from Sigma-Aldrich (Sigma-Aldrich, St. Louis, MO, United States) unless otherwise specified. T84 cells were cultured in DMEM: F12, 15 mmol/L, N-2-hydroxyethylpiperazine-N’-2-ethanesulphonic acid media supplemented with 10% fetal bovine serum (FBS), 1 mmol/L L-glutamine, 100 IU/mL penicillin, and 100 μg/mL streptomycin. For gene expression assessment of tight junction markers, T84 cells were seeded on 48-well plates. When cells were semi-confluent (90%), the media were replaced with 0.2 mL of either fresh medium (untreated and IFNg wells) or fresh medium containing 100 or 400 ng/mL of GDF15 (catalog number: No. 32-1249, Abeomics, San Diego, CA, United States). The following day, 0.2 mL of fresh medium was added to untreated wells, 0.2 mL of fresh medium containing 100 ng/mL of IFNg (catalog number: No. 32-6390, Abeomics, San Diego, CA, United States) was addedd to IFNg wells for a final concentration of 50 ng/mL, and 0.2 mL of fresh medium containing 100 or 400 ng/mL of GDF15 was added to GDF15 100 ng/mL and GDF15 400 ng/mL wells, respectively. Cells were collected in QIAzol for RNA isolation. A total of three experiments were run. Within each experiment, conditions were run in triplicate.

For immunofluorescence analysis of tight junction proteins in the T84 cell line, cells were seeded on an 18-well μ-slide (Cat.No: 81816, IBIDI GmbH, Gräfelfing, Germany). When cells were semi-confluent (90%), the media was replaced following the same protocol as for gene expression assays in T84 cells, but using 0.05 mL instead of 0.2 mL, for a final volume of 0.1 mL in wells. Media was removed 24 hours after adding IFNg, and cells were fixed with 10% formalin, permeabilized with Triton X 0.2%, and blocked with PBS 1 × BSA 2%. Primary antibody against ZO-1 and claudin 1 (No. ab211737, Abcam, Cambridge, United Kingdom) were added at a 1:1000 dilution in PBS 1 × BSA 0.1%, and incubated 24 hours at 4 °C. Afterwards, primary antibody was washed and secondary antibody against mouse IgG AlexaFluor^®488 (No. ab150113, Abcam, Cambridge, United Kingdom) and rabbit IgG Alexa Fluor^®647 (No. ab150075, Abcam, Cambridge, United Kingdom) were added at a 1:2000 dilution in PBS 1 × and incubated 1 hour at room temperature. Secondary antibodies were washed, and DAPI was added. Cells were incubated for 6 minutes, and DAPI was washed to remove excess. Wells were covered with ProLong^® diamond antifade mountant (catalog number: No. 39695, Life Technologies, Carlsbad, CA, United States), and cells were visualized 24 hours later on a SP5 confocal microscope. A Total of three experiments were run, and 5 pictures per condition were analyzed. Fluorescence was measured as integrated density, and values from ZO-1 and claudin 1 were normalized to fluorescence from DAPI.

Statistical analysis

The Kolmogorov-Smirnov test was used to assess whether data followed a normal distribution. The difference in sex ratio between HC and IBD patients was assessed using the χ² test. Age differences, body mass index, and biochemical parameters from blood test between HC and IBD patients were assessed using a t-test. The difference in time of disease between CD and UC patients was assessed using a t-test. Data grouped in columns are expressed as either median ± interquartile range or mean ± SD, according to the distribution of the data. Mann-Whitney test (one-tailed) was used to compare plasma levels of GDF15 between different groups, and mRNA levels of intestinal barrier markers between untreated T84 cells and GDF15-treated cells. Paired t-test (one-tailed) was used to compare different conditions within the same organoid/cell line in confocal microscopy assays. Pearson’s or Spearman’s correlation tests were used to assess the relationship between different variables, according to the distribution of the data. For all tests, a P value lower than 0.05 was statistically significant. All authors had access to the study data and had reviewed and approved the final manuscript.

RESULTS

Demographic and clinical characteristics of subjects

A total of 23 HC, 39 IBD patients, and 19 MI patients were recruited for this study. The demographic characteristics of subjects participating in the study are described in Table 1. We found no statistical differences in sex ratio, age, or body mass index between HC and IBD patients. As expected, MI patients, who were used as positive control for elevated plasma GDF15, were significantly older (P = 0.003) compared to HCs. Results from the blood biochemical test are also shown in Table 1. As expected, IBD patients showed significantly higher levels of C-reactive protein (CRP), as well as albumin. Haptoglobin, another acute-phase protein, was higher in IBD patients, although not reaching statistical significance. Clinical characteristics of IBD patients (time of disease and treatment) are described in Table 2. No significant differences were found in the time of disease between CD and UC patients.

Circulating levels of GDF15 are increased in IBD patients and correlate with GDF15 gene expression in the colon

Levels of GDF15 in plasma from IBD patients were significantly increased compared to HC (P = 0.0119) (Figure 1A). When stratified (Figure 1B), both CD and UC patients maintained significantly higher levels (P = 0.0498 and P = 0.0132). We next inquired about the source of plasma levels of GDF15 found in IBD patients. To that end, we assess the correlation between gene expression from colonic biopsies of CD patients with plasma levels from the same subjects (both samples collected on the same day). We found a strong positive correlation between gene expression and plasma levels (Spearman r = 0.6002, P = 0.0020) (Figure 1C), suggesting that the colon may be an important source of circulating GDF15.

Figure 1 Circulating levels of growth differentiation factor 15. A: Plasma growth differentiation factor 15 (GDF15) levels in healthy controls (healthy controls, n = 23), inflammatory bowel disease patients (inflammatory bowel disease, n = 39), and myocardial infarction patients (myocardial infarction, n = 19). Data are presented using Tukey’s boxplots; B: Plasma GDF15 levels in healthy controls (healthy controls, n = 23), Crohn’s disease patients (n = 21), and ulcerative colitis patients (n = 18). Data are presented using Tukey’s boxplots; C: Correlation between plasma GDF15 levels and colonic GDF15 gene expression (n = 21 pairs). ^aP < 0.05 and ^cP < 0.001. HC: Healthy controls; IBD: Inflammatory bowel disease; MI: Myocardial infarction; CD: Crohn’s disease; UC: Ulcerative colitis.

Circulating levels of GDF15 correlate with markers of inflammation and intestinal barrier permeability

We also assessed the correlation between circulating levels of GDF15 and different markers from the biochemical blood test. We did not find a correlation of plasma levels of GDF15 with calprotectin, a specific marker of intestinal inflammation (data not shown). However, GDF15 showed a positive correlation with CRP (Spearman r = 0.3290, P = 0.0066), a hallmark of inflammation (Figure 2A). These results suggest that GDF15 is not constantly elevated in IBD patients, but rather shows a dynamic behavior, paralleling inflammatory processes. Circulating levels of GDF15 also correlated with plasma levels of haptoglobin (Pearson r = 0.4066, P = 0.0017), another acute-phase protein (Figure 2B). Because haptoglobin is related to intestinal permeability, and leaky gut or dysfunctional intestinal barrier function is frequently observed in IBD patients, we aimed to assess levels of LBP, a validated biomarker of intestinal permeability. As expected, LBP plasma levels were increased in IBD patients compared to HC (P = 0.0127, Figure 2C), and further increased in MI patients (P = 0.0037), who also presented with higher levels of GDF15 in plasma. In fact, we found a correlation between plasma levels of GDF15 and LBP (Spearman r = 0.2032, P = 0.0423, Figure 2D), suggesting a role of GDF15 in intestinal barrier function. We also found a weak correlation between GDF15 plasma levels and glucose (Pearson r = 0.2576, P = 0.0469) and urea (Pearson r = 0.2788, P = 0.0295).

Figure 2 Correlation between circulating levels of growth differentiation factor 15 and blood markers. A: Correlation between plasma levels of growth differentiation factor 15 (GDF15) and plasma levels of C-reactive protein. Sample size: Healthy controls (HC) = 23 pairs, Crohn’s disease patients (CD) = 21 pairs, ulcerative colitis patients (UC) = 15 pairs, myocardial infarction patients (MI) = 18 pairs; B: Correlation between plasma levels of GDF15 and plasma levels of haptoglobin. Sample size: HC = 18 pairs, CD = 14 pairs, UC = 18 pairs; C: Lipopolysaccharide-binding protein levels across different subject groups. Sample size: HC = 24, CD = 21, UC = 17, MI = 11; D: Correlation between plasma levels of GDF15 and lipopolysaccharide-binding protein. Sample size: HC = 22 pairs, CD = 20 pairs, UC = 17 pairs, MI = 13. ^aP < 0.05 in a Mann-Whitney test combining CD and UC patients. ^bP < 0.01 in a Mann-Whitney test. CRP: C-reactive protein; GDF: Growth differentiation factor; LBP: Lipopolysaccharide-binding protein.

GDF15 alters the intestinal barrier and increases permeability

To inquire about the impact of higher levels of GDF15 on the intestinal barrier, we treated colon organoids with GDF15 and performed a permeability assay (Figure 3A and B). Lumen of colon organoids treated with GDF15 showed higher signal from FITC-Dextran 4K compared to untreated colon organoids (Figure 3C), suggesting that GDF15 induces alteration of intestinal barrier (P = 0.0053 for 100 ng/mL dose and P = 0.0096 for 400 ng/mL dose). Colon organoids from both IBD patients and HC showed similar patterns in response to GDF15, with mean increases of dextran 4k-FITC signal in response to the lower dose of 15.7% and 11.1%, respectively. To delve into the mechanism by which GDF15 alters the intestinal barrier, we studied ZO-1, which is a key component of the tight junction. Levels of ZO-1 were assessed in two colon organoid lines from HC and one line from an IBD patient (Figure 4A-D). In all organoid lines assessed, treatment with 100 ng/mL of GDF15 induced a significant down-regulation of ZO-1 (P = 0.0065), while treatment with 400 ng/mL of GDF15 did not induce significant changes (Figure 4E). To confirm the effects of GDF15 on the intestinal barrier, we investigated the levels of ZO-1 and claudin 1 in another model of colonic epithelium, T84 cells. We observed a down-regulation of both ZO-1 and claudin 1 after treatment with GDF15 at either 100 or 400 ng/mL (Figure 5). To further inquire about the mechanisms by which GDF15 induces a down-regulation of tight junction proteins, we assessed the gene expression of different intestinal barrier markers. Opposite to our results on protein levels, we found no differences in the expression of ZO-1 and claudin 1 after treatment with GDF15 (Figure 6). However, we observed a trend towards down-regulation of occludin and claudin 2 in T84 cells treated with either 100 or 400 ng/mL of GDF15.

Figure 3 In vitro assessment of the effects of growth differentiation factor 15 on intestinal barrier permeability. Colonic organoids were treated with fluorescein isothiocyanate-dextran 4k (FD4), and the relative FD4 intensity (ratio between the mean gray value inside the colonic organoid and the sum of the mean gray values inside and outside) was calculated for each organoid. A: Untreated colonic organoid (healthy control), relative FD4 intensity = 0.092; B: Growth differentiation factor 15-treated organoid (100 ng/mL), relative FD4 intensity = 0.163. Scale bar: 100 μm; C: Relative FD4 intensity in organoid lines under different conditions, presented using a Tukey’s box plot. Images were obtained with a Leica SP5 confocal microscope using a 10 × objective and analyzed in ImageJ. For each organoid line, ≥ 10 organoids from three different wells (≥ 30 total organoids) were analyzed per condition. ^aP < 0.05 and ^bP < 0.01, one-tailed paired t-test. GDF: Growth differentiation factor; FD4: Fluorescein isothiocyanate-dextran 4k; IFNg: Interferon gamma.

Figure 4 In vitro assessment of the effects of growth differentiation factor 15 on intestinal barrier tight junctions. A and B: Representative images of an untreated organoid stained with 4’-6-diamidino-2-phenylindole (DAPI) (A) and zonula occludens (ZO)-1 (B); C and D: Representative images of an organoid from the same line, treated with 100 ng/mL of GDF15, stained with DAPI (C) and ZO-1 (D). Scale bar: 100 μm; E: Relative ZO-1 fluorescence values normalized to DAPI. Data are presented as mean ± SD relative to the untreated organoid. Images were obtained using a Leica SP5 confocal microscope with a 20 × objective and analyzed in ImageJ. For each organoid line, ≥ three organoids from three different wells were analyzed. ^bP < 0.001, one-tailed paired t-test. GDF: Growth differentiation factor; ZO-1: Zonula occludens-1; IFNg: Interferon gamma.

Figure 5 Zonula occludens-1 and claudin 1 expression in T84 cells. A-C: Representative field of T84 cells stained with 4’-6-diamidino-2-phenylindole (DAPI) (A), zonula occludens (ZO)-1 (B), and Claudin 1 (C); D-F: Representative field of T84 cells treated with 100 ng/mL of growth differentiation factor 15, stained with DAPI (D), ZO-1 (E), and claudin 1 (F). Scale bar: 100 μm; G and H: Relative ZO-1 and claudin 1 fluorescence values, respectively, normalized to DAPI. Data are presented as mean ± SD of relative fluorescence. Images were obtained using a Leica SP5 confocal microscope with a 40 × objective and 1.5 zoom, and analyzed in ImageJ. ^aP < 0.05 and ^bP < 0.01, one-tailed paired t-test. GDF15: Growth differentiation factor 15; ZO-1: Zonula occludens-1; IFNg: Interferon gamma.

Figure 6 Messenger RNA levels of intestinal barrier markers. Data are presented as mean ± SD of the ΔCt values from three experiments. Within each experiment, conditions were run in triplicate, and each sample was run in duplicate. ZO-1: Zonula occludens-1; IFNg: Interferon gamma; GDF15: Growth differentiation factor 15.

DISCUSSION

GDF15 is a mitokine expressed by various tissues in response to stress, particularly mitochondrial stress[16]. The role of GDF15 in human health remains controversial, with studies suggesting both protective and detrimental effects. For example, GDF15 prevents hepatic steatosis by inhibiting oxidative stress, mitochondrial dysfunction, and inflammasome activation[17], and also prevents obesity and insulin resistance by increasing lipolysis in adipocytes and hepatocytes[18]. In contrast, GDF15 has been shown to promote the proliferation of cervical cancer cells[19] and to upregulate mucin expression in alveolar cells exposed to cigarette smoke, contributing to the pathogenesis of chronic obstructive pulmonary disease[20].

In the present study, we observed elevated plasma levels of GDF15 in patients with IBD, consistent with our previous findings showing activation of the UPR^mt in colon tissues from IBD patients[4]. This finding supports mitochondrial stress in the colon as a pathophysiological mechanism in IBD. Additionally, we confirmed previous reports of elevated circulating GDF15 levels in IBD patients[13]. Importantly, we observed significantly higher levels in both CD and UC patients. We next observed a correlation between GDF15 and CRP, a hallmark of inflammation, and haptoglobin. Haptoglobin is well known for its role in binding hemoglobin from lysed erythrocytes, preventing oxidative damage[21]. However, it has also been reported to bind lipopolysaccharide, thus reducing immune overactivation[22]. Recent studies have identified associations between haptoglobin and plasma levels of lipopolysaccharide, as well as other markers of increased intestinal permeability, such as LBP[23-25]. To confirm an association between GDF15 and increased intestinal permeability, we evaluated its correlation with plasma levels of LBP, a validated biomarker of increased intestinal permeability. We found a positive correlation between GDF15 and LBP, which led us to investigate the effects of GDF15 on the intestinal barrier in vitro. Using colonic organoids, we demonstrated for the first time that GDF15 increases intestinal permeability, likely through the downregulation of key components of tight junctions, such as ZO-1 and claudin 1. The lack of further permeability increases at higher doses of GDF15 suggests that the mechanism is saturated at 100 ng/mL. Inconsistent results from our gene expression experiments prevent us from clearly identifying downregulation of tight junction players at the mRNA level as the main mechanism of GDF15 action.

This effect of GDF15 on the gastrointestinal tract has not been previously described and raises new questions about its mode of action. The only currently known receptor for GDF15, glial-derived neurotrophic factor receptor alpha-like, is expressed exclusively in the brain[26]. It has been hypothesized that GDF15 might interact with the Transforming growth factor (TGF)-β receptors TGFBR1 and TGFBR2[27]. However, studies suggesting such interactions should be interpreted with caution, as commercially purified GDF15 has been found to be contaminated with TGF-β[14]. The source of this contamination is likely the constitutive overexpression of TGF-β in mammalian cell cultures used for the overexpression of GDF15. To avoid contamination in our study, we used GDF15 purified from Escherichia coli, ensuring that the observed effects on intestinal tight junctions were induced by GDF15. Further studies are needed to identify the specific receptor mediating the effects of GDF15 on intestinal tight junctions.

The source of GDF15 Likely varies across different conditions, with the placenta exhibiting the highest expression levels[16]. In patients with cardiovascular diseases, it might seem logical to assume that the heart is the primary source of elevated plasma GDF15. However, the literature indicates that, while GDF15 gene expression in the heart correlates with circulating levels, extra-cardiac expression (particularly in the liver) contributes more significantly to plasma GDF15[28]. The gastrointestinal tract may also serve as a major source of circulating GDF15 under certain conditions. For example, the gastrointestinal tract has been implicated in the increase in plasma GDF15 levels induced by metformin treatment[29]. In our study, we found a positive correlation between plasma GDF15 levels and GDF15 gene expression in colonic biopsies, suggesting that the colon may be a major source of circulating GDF15. Comparative studies of GDF15 gene expression across different tissues in a mouse model of colitis are necessary to better clarify the source of circulating GDF15. Nonetheless, the parallelism between circulating GDF15 levels and gene expression in the colon supports a potential impact of GDF15 on the gastrointestinal tract.

Intestinal barrier dysfunction has been consistently reported in IBD as a key factor contributing to immune system dysregulation[1]. In our study, we report a novel mechanism by which GDF15 affects the intestinal tract, altering the intestinal barrier and increasing permeability. Increased intestinal permeability facilitates contact between epithelial cells, gut microbiota, and toxins, potentially leading to immune system overactivation. Furthermore, the correlation between circulating GDF15 levels and CRP may reflect immune overactivation resulting from increased intestinal permeability. This finding is relevant not only for IBD patients, but also for patients with cardiovascular diseases. Elevated plasma GDF15 levels have been found in several cardiovascular conditions[10]. At the same time, there is growing interest in studying the impact of intestinal microbiota on cardiovascular health as an important factor in the development of cardiovascular disorders[30-32]. In fact, our cohort of patients with MI, which served as a positive control for elevated plasma GDF15 levels, also showed elevated plasma LBP levels, highlighting the effect of GDF15 on the intestinal barrier in disorders other than IBD.

CONCLUSION

Our study confirms increased circulating levels of GDF15 in IBD patients and reveals a novel molecular mechanism by which this mitokine disrupts the intestinal barrier. Targeting GDF15 may offer a promising strategy to prevent disruption of the intestinal barrier, potentially reducing immune overactivation.

ACKNOWLEDGEMENTS

Experiments on confocal imaging were performed in the ICTS “NANBIOSIS”, Unit U28 at IBIMA Plataforma-BIONAND.