Metformin strengthens uroepithelial immunity against E. coli infection

Urinary tract infection frequently caused by E. coli is one of the most common bacterial infections. Increasing antibiotic resistance jeopardizes successful treatment and alternative treatment strategies are therefore mandatory. Metformin, an oral antidiabetic drug, has been shown to activate macrophages in the protection against certain infecting microorganisms. Since epithelial cells often form the first line of defense, we here investigated the effect on uroepithelial cells during E. coli infection. Metformin upregulated the human antimicrobial peptides cathelicidin LL-37 and RNase7 via modulation of the TRPA1 channel and AMPK pathway. Interestingly, metformin stimulation enriched both LL-37 and TRPA1 in lysosomes. In addition, metformin specifically increased nitric oxide and mitochondrial, but not cytosolic ROS. Moreover, metformin also triggered mRNA expression of the proinflammatory cytokines IL1B, CXCL8 and growth factor GDF15 in human uroepithelial cells. The GDF15 peptide stimulated macrophages increased LL-37 expression, with increased bacterial killing. In conclusion, metformin stimulation strengthened the innate immunity of uroepithelial cells inducing enhanced extracellular and intracellular bacterial killing suggesting a favorable role of metformin in the host defense.


Results
Metformin enhances the expression of the antimicrobial peptides LL-37 and RNase7. Metformin treatment upregulated the expression of the AMPs CAMP and RNASE7 mRNA and corresponding intracellular proteins LL-37 and RNase7, compared to untreated uroepithelial cells, TERT-NHUC ( Fig. 1A-F;  Fig. S1A-D) and 5637 ( Fig. S1E-J). 4 mM metformin treatment was found to be optimal based on the dose response profiles of CAMP and RNASE7 expression in both TERT-NHUC and 5637 (Fig. S1A,B,E,H). The intracellular activity of AMPs, especially within the lysosomes is vital in the uroepithelial cells, as E. coli can neutralize the lysosomal pH, thus escaping the autophagy machinery 16 . Interestingly, metformin treatment resulted in increased localization of LL-37, but not RNase7 in the lysosomes of TERT-NHUC (Fig. 1G) and 5637 cells (Fig. S1K), as shown by confocal microscopy. Furthermore, expression of LL-37 and RNase7 was confirmed by flow cytometry (Fig. S1C,D). The differential expression and localization of these AMPs in uroepithelial cells could be regulated by different signaling pathways triggered by metformin.

Metformin upregulates CAMP and RNASE7 expression via distinct mechanisms.
Recently metformin has been suggested to act on TRPA1 channels 17 . TRPA1 is known to be present in bladder epithelial cells 18 and is activated upon binding to the E. coli lipopolysaccharide 19 . Similar to uroepithelial cells, TRPA1 is  green) and RNase7 (lower panel, green) with lysosomes (red) (n = 3). All experiments were performed in human uroepithelial cells, TERT-NHUC and treated with metformin (Met, 4 mM) for either 24 h or 36 h for mRNA and protein respectively, compared to only vehicle treated control. Data are presented as mean with SEM. Significance levels **P < 0.01; ***P < 0.001. www.nature.com/scientificreports/ also active at the lysosomes of the dorsal root ganglion neurons 20,21 . We therefore investigated if uroepithelial cells also respond to metformin stimulation via TRPA1. Metformin treatment did not increase the intracellular calcium levels in TERT-NHUC, whereas TRPA1 activation by Allyl Isothiocyanate (AITC) increased intracellular calcium (Fig. S2A,B; Videos 1, 2). Pretreatment of TERT-NHUC with either metformin or TRPA1 inhibitor A-967079 (A96) prevented the AITC induced calcium influx (Fig. S2C,D; Videos 3,4). This indicated that metformin acts as a TRPA1 inhibitor in uroepithelial cells, consistent with a previous report in neurons 17 . Metformin upregulated TRPA1 mRNA and protein expression in TERT-NHUC, with prominent enrichment of TRPA1 specifically in the lysosomes but not in other organelles such as mitochondria ( Fig. 2A-C). In addition, the TERT-NHUC showed significant upregulation of CAMP upon treatment with metformin or TRPA1 inhibitor A96, and additive effect on treating A96 pretreated cells with metformin (Fig. 2D). TRPA1 activation by AITC itself had no impact on CAMP expression (Fig. 2D). Furthermore, AITC was not capable of blocking CAMP expression induced by pretreatment with TRPA1 inhibitors metformin or A96, as the ion channels were not activated in the presence of their inhibitors. In contrast, AITC treated TERT-NHUC showed a trend of increased RNASE7 expression and an additive increase of RNASE7 was observed upon treatment with both metformin and AITC (Fig. 2E). A96 itself was able to decrease RNASE7 expression, while A96 pretreatment blocked the metformin and AITC induced increase in RNASE7 expression (Fig. 2E). Thus, CAMP expression was regulated by TRPA1 channel inhibition, while RNASE7 expression was likely regulated by a mechanism that is triggered by metformin and TRPA1 activation, probably by AMPK activation. These findings are in line with our calcium imaging data (Fig. S2) and previous reports showing that metformin acts as an inhibitor of TRPA1 17 .

Scientific
Metformin is a well-established activator of the AMPK pathway 22 , however it is not known to stimulate CAMP or RNASE7 expression. In the uroepithelial cells, TERT-NHUC treated with metformin and AMPK modulators, CAMP expression remained unaffected upon AMPK activation by AICAR (5-aminoimidazole-4-carboxamide ribonucleotide) or inhibition by compound C (Fig. 2F). On the other hand, RNASE7 was significantly upregulated upon AMPK activation by both metformin and AICAR (Fig. 2G), confirming the direct involvement of the AMPK pathway. Interestingly, pretreatment with the AMPK inhibitor compound C prevented metformin induced upregulation of both CAMP and RNASE7. These results suggest that metformin induced CAMP expression is regulated by inhibition of TRPA1 channels while RNASE7 expression is mediated by the activation of the AMPK pathway. In addition to these antimicrobial peptides, other innate immune mechanisms might also be triggered by metformin.  23 , metformin is believed to act primarily on mitochondrial respiration 24 . Mitochondria are a source of free radical species including reactive oxygen species (ROS) and reactive nitrogen species (RNS) 25 . Besides AMPs, the uroepithelial cells also produce ROS and RNS which help to control invading pathogens 26 . In uroepithelial cells, 5637, metformin treatment increased nitric oxide synthase (NOS2) mRNA and protein expression ( Fig. 3A-C), along with higher release of free nitric oxide (Fig. 3D), without altering the expression of the human antioxidant genes catalase (CAT ) and superoxide dismutase (SOD2) (Fig S3A, B). Notably, metformin treatment showed specific increase in mitochondrial, but not cytosolic ROS ( Fig. 3E-G), which was decreased by pretreatment with the ROS inhibitor N-acetyl-l-cysteine (NAC). This is in agreement with previous reports attributing the antimicrobial activity of metformin to increased mitochondrial ROS production in macrophages 13,14 . Metformin promotes bacterial clearance. Interestingly, metformin was recently shown to upregulate the peptide hormone, growth/differentiation factor 15 (GDF15) 27,28 which was demonstrated to support host survival during bacterial infection and sepsis 29 . Notably, metformin treated uroepithelial cells upregulated GDF15 mRNA expression in both 5637 ( Fig. 4A) and TERT-NHUC (Fig. 4B). GDF15 expression was upregulated by metformin, TRPA1 inhibitor A96 (Fig. S4A) and AMPK inhibitor compound C (Fig. S4B). Neither AITC nor AICAR showed upregulation in GDF15 expression. However, we did not observe any direct antibacterial effect of GDF15 peptide on uropathogenic E. coli (data not shown). Treatment of uroepithelial cells with GDF15 peptide did not increase CAMP or RNASE7 expression in 5637 (Fig. S4C,D) or TERT-NHUC cells (Fig. S4E,F). The induction of these AMPs in uroepithelial cells therefore seems to be GDF15 independent. This observation raised the possibility that GDF15 produced by uroepithelial cells stimulates other immune cells like macrophages. Interestingly, GDF15 peptide treatment upregulated both mRNA and protein expression of LL-37 in human THP1 monocyte derived macrophages ( Fig. 4C-E). In line with this finding, increased bacterial killing was observed in the conditioned media and cell lysate of GDF15 peptide stimulated macrophages (Fig. 4F,G).
Since uropathogenic E. coli are known to invade uroepithelial cells and survive intracellularly 4,30 , we explored whether metformin pretreated cells were better prepared to fend off invading bacteria. Upon infection with uropathogenic E. coli, metformin treated 5637 and TERT-NHUC cells showed upregulation in LL-37 at the mRNA and protein levels ( AMPs are also known to stimulate proinflammatory cytokines which help in the recruitment of immune cells such as macrophages, neutrophils and NK cells that promote resolution of bladder infection 31,32 . Metformin treated 5637 cells increased mRNA expression of the proinflammatory cytokines IL1B and CXCL-8 upon infection (Fig. 5G,H).
Further, to explore the effect of metformin in inhibiting bacterial growth in uroepithelial cells, survival assays were performed in conditioned media, cell lysates and intact cells. Significantly increased bacterial killing was observed in the conditioned media from metformin treated 5637 (Fig. 5I) and TERT-NHUC cells (Fig. S5G). Interestingly, the cell lysates from metformin treated uroepithelial cells retained higher antimicrobial activity www.nature.com/scientificreports/ (Fig. 5J, Fig. S5H). Finally, metformin also increased intracellular E. coli killing in uroepithelial cells, 5637 (  inhibitor, compound C (CompC, 20 µM), Compound C pretreatment followed by metformin or Compound C pretreatment followed by AICAR (n = 3, performed in duplicates). All experiments were performed in human uroepithelial cells, TERT-NHUC, treated with respective drugs for 24 h for mRNA and 36 h for protein expression, compared to only vehicle treated control. Data are presented as mean with SEM. Significance levels *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Discussion
We demonstrate that metformin exerts its effect through different host protective mechanisms, prominent being upregulation of the AMPs, LL-37 and RNase7. The AMPs are particularly interesting antibacterial agents as they can act intracellularly as well as extracellularly. Moreover, their intracellular localization, especially within the lysosomes are vital. LL-37 exerts its activity against intracellular pathogens upon enrichment in lysosomes of epithelial cells 33 as well as in macrophages 34 . In vitro metformin treatment induced lysosomal enrichment of LL-37 could therefore better prepare uroepithelial cells against bacterial invasion. Interestingly, RNase7 was not localized in the lysosomes upon metformin treatment. This is in line with a previous report demonstrating that RNase7 does not accumulate in lysosomes, though its related antimicrobial peptide RNase6 can localize to lysosomes 35 . It is worthy of note that metformin also increased TRPA1 expression in the lysosomes of uroepithelial cells. Our calcium imaging, together with previous reports demonstrates that TRPA1 is active in the lysosomes 20,21 .
Lysosomes also act as a hub of AMPK activity 36 , which is crucial for lysosomal biogenesis and function 37 . Hence, we explored the role of metformin mediated AMPK activation in regulating CAMP and RNASE7 expression. While RNASE7 expression was AMPK-dependent, CAMP expression was largely AMPK-independent. Despite the fact that compound C can inhibit several other kinases 38 , we observed that compound C itself had no effect on RNASE7 expression. Compound C significantly blocked metformin and AICAR induced RNASE7 expression, suggesting that metformin induced RNASE7 expression is dependent on AMPK activation. On the other hand, CAMP expression is dependent on TRPA1 inhibition by metformin or TRPA1 specific inhibitor A96. These evidences point towards activation of at least two distinct pathways governing CAMP and RNASE7 induction by metformin.
Metformin can increase the antimicrobial effect of macrophages further by direct stimulation and via immunomodulatory agents produced by uroepithelial cells. GDF15 is one such host protective molecule, which is upregulated by metformin and is protective against bacteria induced sepsis 29 . Metformin and A96 upregulated GDF15 mRNA expression, indicating that TRPA1 inhibition can increase GDF15 expression. Notably, this mechanism is similar to the induction of CAMP by TRPA1 inhibition. However, treatment of uroepithelial cells with GDF15 peptide did not impact CAMP or RNASE7 expression. Hence, the expression of these antimicrobial peptides is independent of GDF15 in uroepithelial cells. Metformin induced GDF15 might influence multiple host protective mechanisms. It is likely that metformin induced GDF15 in uroepithelial cells improves the antibacterial efficacy of other innate immune cells such as macrophages by stimulating antimicrobial peptide production.
Most importantly, both the uroepithelial cells tested, TERT-NHUC and 5637, showed significantly improved killing of intracellular and extracellular bacteria upon metformin treatment, indicating that metformin prepares www.nature.com/scientificreports/ these uroepithelial cells to kill pathogenic bacteria more efficiently. The increase in CAMP and RNASE7 mRNA levels during infection of metformin treated uroepithelial cells could be attributed to rapid transcriptional activation 2 . The significant upregulation of LL-37 peptide, which was not observed for RNase7 in metformin treated and infected TERT-NHUC uroepithelial cells, suggest differential translational regulation of these AMPs. Most of the cytosolic pool of AMPs are released to the extracellular medium when the bacteria contact the plasma membrane of uroepithelial cells. Hence, higher transcriptional induction of AMPs in metformin pretreated cells upon infection, could be advantageous for host defense. Taken together, our data demonstrate that metformin induces multiple antimicrobial mechanisms in uroepithelial cells. The enhanced bacterial clearance by metformin treated human uroepithelial cells is most likely influenced by the combined action of increased innate immune response including AMPs, ROS, NOS and cytokines. Therefore, in the future reorienting drugs that also strengthen host mediated antibacterial effects, like metformin, can be of interest to complement traditional treatment strategies.

Materials and methods
Cell lines and culture conditions. Telomerase-immortalized human uroepithelial cells, TERT-NHUC (provided by M. A. Knowles, Leeds, UK) and human bladder epithelial cell line, 5637 (HTB-9, American Type Culture Collection) were cultured as previously described 4 . In brief, TERT-NHUC cells were grown in Epilife medium (MEPI500CA, Life Technologies) supplemented with 1% of human keratinocyte growth supplement (S0015, Life technologies) and 5637 cells, human monocytes THP1 (TIB-202, ATCC) were grown in RPMI 1640 medium (Life Technologies) supplemented with 10% fetal bovine serum (Life Technologies), and cultured at 37 °C and 5% CO 2 . THP1 monocytes were differentiated to macrophages by treatment with 150 ng/ml of phorbol myristate acetate (PMA) for 24 h followed by replacement with PMA-free RPMI 1640 medium for another 24 h. Cell treatment. In vitro cell experiments were carried out in Primaria plates (BD Falcon) for uroepithelial cells, TERT-NHUC, and polystyrene (Costar) for 5637 and THP1 differentiated cells. Nearly confluent cells, 70-80%, were treated with freshly prepared metformin (4 mM, Sigma-Aldrich) for 24 h, deionized water served as vehicle control. 4 mM metformin concentration was found to be optimal for both the cell lines based on the dose response expression profiles of CAMP and RNASE7 expression (Fig. S1). 5' adenosine monophosphate-activated protein kinase (AMPK) activator 5-Aminoimidazole-4-carboxamide ribonucleotide (AICAR, 1 mM, Sigma-Aldrich) or AMPK inhibitor, Compound C (20 µM, Sigma-Aldrich) were used and cells were treated for 24 h. In blocking experiments, Compound C was added 1 h prior to adding AICAR or metformin. Chinese hamster ovary cell line expressed growth/differentiation factor 15 peptide (GDF15, 50 ng/ ml, R&D systems) was treated to the cells for 24 h. For TRPA1 RT-PCR or imaging experiments, cells were stimulated with the activator, allyl isothiocyanate (AITC, 10 µM, Sigma-Aldrich) and inhibitor A-967079 (A96, 20 µM, Sigma-Aldrich), for 24 h for mRNA analysis. Equal amount of DMSO served as vehicle in control cells.
Total RNA extraction and quantitative real-time PCR. Total RNA extraction was performed as previously described 7 and up to 500 ng of RNA was transcribed to complementary DNA with the Superscript cDNA Synthesis Kit (Life Technologies) according to the manufacturers' instructions. Gene expression was analyzed by SYBR Green-based assay (Life Technologies) in a Rotor-Gene PCR cycler (Corbett Life Science) using gene spe- Immunofluorescence analysis. Uroepithelial cells were grown on glass cover-slips (VWR) and treated as described above. Cells were processed for immuno-staining as previously described 39  Calcium imaging. TERT-NHUC cells were grown on glass bottom dishes (ibidi), treated with 2 µM of the membrane permeant Ca 2+ indicator dye Fluo-4 AM (Invitrogen) for 30 min or 1 µM of LysoTracker red (Invitrogen) for 30 min. After indicated time, cells were washed with HBSS, freshly supplemented with medium and time series imaging was performed in green or red channel using 63 × oil immersion objective of Zeiss LSM700 confocal microscope for 300 frames at 1.56 s/frame interval. Drugs were administered to the imaging dish at the time points indicated in the graphs.
Reactive Oxygen Species measurements. Uroepithelial cell, 5637 were treated with metformin (4 mM) with or without 1 h pretreatment with the ROS chelator N-acetyl cysteine (NAC, 3 mM Sigma Aldrich) for 24 h. Cells were stained with MitoSox Red (5 µM, Invitrogen) for 30 min and images were captured in LSM700 confocal microscope. For cytosolic ROS measurements, uroepithelial cells, 5637 were treated with metformin (4 mM) or NAC for 24 h. Cytosolic ROS was quantified by staining the cells with 2' ,7'-dichlorodihydrofluorescein diacetate (H2DCFDA, 10 µM, Sigma Aldrich) and fluorescence signal quantified using Fluorimeter (Omega) at excitation of 485 nm and emission 530 nm. Data were represented as relative fold change with respect to untreated control.
Nitric oxide measurement. Nitric oxide in the cell culture supernatants from untreated 5637 cells and cells treated with metformin (4 mM) for 24 h were incubated with Griess reagent for 10 min at room temperature, followed by absorbance recording at 540 nm.
Conditioned media and antimicrobial killing assay. Cell free supernatants were collected from metformin (4 mM) treated uroepithelial cells after 36 h incubation. Cell extracts were prepared by lysing the cells in 1% Triton-X-100, centrifuged at 8000 g for 5 min to remove cell debris. 50 µl from 10 4 CFU/ml of E. coli were added to 150 µl of conditioned media or cell extract and incubated at 37 °C CO 2 at 150 rpm. After 30 min incubation, 100 µl of cell extract was plated, while 50 µl of conditioned media was cultured after 6 h of incubation for enumerating the CFU.