Paper I – Nerves can sense UPEC alpha-hemolysin in the kidney and signal for

Figure 8. Intact splenic nerve signaling is needed for splenic IFNγ responses to UPEC kidney infection.

(A) IFNγ levels in spleens of infected (LT004), compared to sham-infected (PBS) animals measured by ELISA. n = 3-5. (B) mRNA expression of Ifng in splenic tissue of infected (LT004, black symbols) or sham-infected (PBS, unfilled symbols) animals, with (squares) or without (circles) nerve block prior to infection.

Relative Ifng expression (delta CT) is given in relation to Gapdh. n = 3. Individual data points and median values (red bars) are shown in both graphs. * = p<0.05 determined by Mann-Whitney in (A) and Kruskal-Wallis analysis with Dunn’s correction in (B) (Steiner et. al. 2021, Paper I).

5.1.2 Delineating important players at the infection site and responses of the renal epithelium during kidney infection

To get a more detailed picture of what happens at the infection site in the first 4 h of infection and delineate important players that might contribute to neural sensing of the kidney infection, we performed intravital multiphoton imaging during the time course of the infection and at endpoint we microdissected the infection foci to perform tissue analyses.

Initial bacterial colonization of the renal tubule and subsequent multiplication of bacteria could be seen through intravital imaging, similar to what had been observed in earlier studies^6,7,90,96. This confirms that the initial host-pathogen interaction occurs at the renal epithelium in this model. As infection progress, however, renal epithelial cells slough off and enable paracellular movement of bacteria down to the basement membrane⁷. Using ex vivo immunofluorescent analysis this epithelial breakdown could be seen already at 4 h of infection (Figure 9A). Bladder epithelial cells have also been found to slough off during UPEC infection, and this has been associated with apoptosis dependent on FimH/Type 1 pili⁴⁶. The sloughing response of both renal epithelial cells and bladder epithelial cells to UPEC infection may have two effects: it can facilitate increased clearance of adhering bacteria (benefiting the host), but it may also provide access to underlying cells and tissues (benefiting the bacteria). However, in this study, the bacteria were still confined to the renal tubule, as no bacteria were found in blood cultures, suggesting that the basement membrane hinders systemic spread of bacteria at this timepoint.

Ditting et al. have found that the kidneys have sensory nerves both in the renal pelvis and in the cortex along renal vasculature, glomeruli, and tubules¹⁷⁶. As our main objective of this work was to investigate if nerves could sense an infection in the kidney and signal to the spleen, we investigated if sensory nerves might be present in areas of the kidney where they may encounter products released during infection in our model. We were able to confirm that

sensory nerves are present in the renal cortex using ex vivo immunofluorescent analysis, and found that sensory nerve fibers are suitably present in the tubular basement membrane where they could come in contact with signals from infected renal epithelial cells, as well as bacterial compounds (Figure 9B).

Figure 9. Important players during the first 4 h of a localized kidney infection. (A) Ex vivo analysis of an infected kidney tubule 4 h after infection shows UPEC (green) in the renal tubule lumen (L), and some areas of disrupted epithelial lining (arrow) where bacteria can reach the basement membrane (collagen IV, white). Scale bar = 25 μm. (B) Ex vivo analysis of uninfected rat renal cortex shows nerve fibers (β3-tubulin in top panel and PGP9.5 in bottom panel, red) in the basement membrane (collagen IV, white) of glomeruli (G, white arrow head), arterioles (A, cyan arrow head) and proximal tubules (PT, white arrow). Some of the fibers surrounding renal tubules were found to be sensory nerve fibers (TrkA, cyan, bottom panel), indicated by arrows. Actin (yellow) is stained for histological reference. Scale bar = 15 μm in top panel and 20 μm in bottom panel (Steiner et. al. 2021, Paper I).

Utilizing the biomimetic flow chamber to study the renal epithelial cell response in more detail in vitro, we again observed sloughing of renal epithelial cells upon infection with CFT073. Infection with CFT073 also resulted in the expected release of cytokines (IL-6 and IL-8) and extracellular ATP (eATP). As CFT073 is known to express HlyA, we hypothesized that eATP release might be a sign of cell lysis. However, further probing revealed that the eATP release occurred while HlyA was at sub-hemolytic concentrations, and prior to sloughing of renal epithelial cells and cell death. This is in coherence with reports that ATP can be released by different cell types, during both cell lysis and via non-lytic mechanisms^173,224. Curious about the role of HlyA in triggering the eATP response, we stimulated renal epithelial cells with an isogenic mutant of CFT073 that lacks the expression of HlyA (LT002). This stimulation did not result in any eATP release from the renal epithelial cells, suggesting that the response is HlyA dependent.

Human urothelial cells have previously been shown to rapidly release eATP upon stimulation with UPEC^173,174. Interestingly, infection-induced eATP release is believed to have an autocrine effect and stimulate IL-8 release from the urothelial cells^173,174. It has also been shown to promote gut inflammation in an in vivo model of Shigella infection¹⁷², suggesting that eATP responses contribute to pro-inflammatory signaling. Further, there is evidence that ATP signaling via purinergic receptors (e.g. P2X and P2Y) can regulate host immunity by modulating neutrophil phagocytosis, chemotaxis, and cytokine production²²⁵. Thus, HlyA-mediated ATP release during UPEC kidney infection may contribute to pro-inflammatory

signaling. This could explain why children with febrile UTI caused by UPEC expressing HlyA have been reported to have higher concentrations of IL-8 and IL-6 in urine²²⁶.

5.1.3 Indirect and direct nerve sensing of UPEC toxin HlyA

Having established that intact splenic nerve signaling is needed for the observed splenic IFNγ responses and shown that both renal epithelial cells and sensory nerves come in contact with bacteria at different timepoints during the infection, we next investigated how the nervous system might sense the presence of infection. ATP release from urothelium has been proposed to trigger the sensation of bladder pain by activation of P2X receptors on sensory nerves²²⁷, and ATP has also been reported to activate TRPV1²²⁸. Considering our observation that renal epithelial cells release eATP upon stimulation with UPEC expressing HlyA, we stimulated primary mouse DRG cells with ATP and could confirm that the sensory nerves were activated and released CGRP (Figure 10A). This suggests that UPEC expressing HlyA indirectly, by stimulating renal epithelial cells to release eATP, can activate sensory neurons.

In vivo such eATP stimulation of sensory nerves may occur as soon as bacteria interact with the renal epithelium, and without bacteria coming in direct contact with the sensory nerve endings in the basement membrane.

Signaling by epithelial eATP is, however, unlikely to persist throughout the entire duration of the infection, as infected renal epithelial cells eventually slough off and die. At later timepoints, when the basement membrane is denuded, live bacteria may instead come into close proximity to the sensory nerves and directly activate sensory signals. Considering the accumulating evidence that sensory nerves can directly sense bacterial products¹⁰¹, we investigated if our UPEC strains could activate the DRG cells. As HlyA had been found necessary for the renal epithelial response, we stimulated primary mouse DRG cells with both LT004 (HlyA+), as well as the isogenic mutant strain lacking HlyA expression (LT005, HlyA-). For other virulence factors (e.g. LPS and flagella) these strains are identical. While we found that infection with live LT004 (HlyA+) indeed resulted in activation of sensory nerves and subsequent CGRP release, no such activation was found when sensory nerves were infected with LT005 (HlyA-) (Figure 10B). Collectively our results indicate that UPEC expressing HlyA indirectly and directly can activate sensory neurons.

To investigate if HlyA is important for the activation of the nerve-mediated inter-organ communication between the kidney and the spleen, we microinfused LT004 (HlyA+) or LT005 (HlyA-) into renal tubules of rats and analyzed splenic tissue after 4 h of infection.

We found upregulation of splenic Ifng in LT004 (HlyA+) infected animals, but not in LT005 (HlyA-) infected animals (Figure 10C). Thus, we could confirm a role for HlyA in causing the inter-organ communication between the infected kidney and the spleen in vivo.

Figure 10. HlyA directly and indirectly activates sensory neurons, and facilitates inter-organ communication during UPEC kidney infection. CGRP release from primary DRG cells stimulated with (A) ATP or (B) live bacteria LT004 (HlyA+) and LT005 (HlyA-). Capsaicin and unstimulated/uninfected served as positive and negative controls respectively. n = 3-10. (C) mRNA expression of Ifng in splenic tissue of animals infected with LT004 (HlyA+) or LT005 (HlyA-) compared to sham-infected (PBS) animals.

Relative Ifng expression (delta CT) is given in relation to Gapdh. n = 5. Individual data points and mean values (red bars in A-B) or median values (red bars in C) are shown in both graphs. * = p<0.05 determined by one-way ANOVA with Turkey’s correction (A-B) and Kruskal-Wallis analysis with Dunn’s correction in (C) (Steiner et. al. 2021, Paper I).

At sublytic levels, insertion of pore-forming toxins leads to an influx of extracellular Ca²⁺, and efflux of intracellular K⁺ in infected cells⁶⁷. The pore-forming toxin of S. aureus, Hla, has been shown to activate sensory nerves through pore formation¹⁴⁹, resulting in such influx of Ca²⁺ and action potential generation¹⁰¹. While this might be an explanation for how HlyA expressing UPEC directly can activate sensory nerves, the pore forming ability of E. coli HlyA remains to be shown.

5.1.4 Neural control of inflammation through IFNγ during kidney infection There are several neuro-immune reflexes that culminate in the modulation of host responses to maintain immune homeostasis¹⁰². Boekel et al. found that IFNγ regulated genes are upregulated after 5 and 8 h of infection, suggesting that the inter-organ communication indeed modulates the host responses in the infected kidney⁹⁶. IFNγ is known to have both anti-inflammatory^229-232, and pro-inflammatory effects^232-236. We investigated how the splenic IFNγ release might modulate local immune responses during kidney infection by returning to our in vitro biomimetic flow model of UPEC kidney infection. Renal epithelial cells were pre-incubated with IFNγ prior to infection, whereafter IL-6 and IL-8 levels in the flow through were measured for up to 5 h. We found that co-incubation with IFNγ results in dampening of IL-8 responses of the renal epithelial cells (Figure 11A). Significant effect on the IL-6 secretion was not seen within this timeframe. Dampening of the IL-8 response was still observable when IFNγ was added after 1 h of infection, indicating that IFNγ exerts this effect on an ongoing infection. Wanting to see if a similar phenomenon could be seen in vivo, and to investigate the role of the splenic activation, we compared C-X-C Motif Chemokine Ligand 1 (CXCL1, rat homologue of human IL-8) responses in rats that had been splenectomised or sham-splenectomised prior to infection. We found a trend of lower expression of Cxcl1 in renal tissue of sham-splenectomised and infected animals (with intact splenic IFNγ responses), compared to splenectomised and infected animals (with disrupted

IFNγ signaling) (Figure 11B). Together, our results suggest that the inter-organ communication between the infected kidney and the spleen results in an anti-inflammatory neuro-immune reflex, with down regulation of renal IL-8 responses via IFNγ.

Figure 11. Splenic IFNγ has immune-modulatory effects at the infection site. (A) IL-8 release from A498 cells infected with LT004 without (black circles) or in the presence of 250 μg/ml IFNγ (gray squares), compared to uninfected cells (unfilled circles). Means ± SD are shown, n = 4. (B) Cxcl1 mRNA expression in kidneys of LT004 infected (black symbols) and sham-infected (unfilled symbols) subjected to splenectomy (squares) or sham-splenectomy (circles) prior to infection. Relative Cxcl1 expression (delta CT) is given in relation to Gapdh. Individual data points and median values are shown, n = 3-5. * = p<0.05 determined by two-way ANOVA with Turkey’s correction in (A) and Kruskal-Wallis analysis with Dunn’s correction in (B). # = p<0.05 with significance for comparison between uninfected and both LT004 infected and LT004 infected cells co-incubated with 250 μg/ml IFNγ in (A) (Steiner et. al. 2021, Paper I).

IL-8 is a proinflammatory cytokine that is involved in recruitment and activation of neutrophils^237-239. In experimental models of UTIs, mice lacking the expression of C-X-C Motif Chemokine Receptor 2 (CXCR2, homologue to the human IL-8 receptor) have been found to exhibit unrestrained progression of UTI with increased acute mortality and renal tissue destruction^94,240. Further probing has shown that pyelonephritis-prone children have a reduced expression of IL-8 receptors compared to healthy controls²⁴⁰, and mutations in the IL-8 gene are associated with increased risk of UTI and worse outcomes of UTIs (e.g.

urosepsis)^241-243. However, while IL-8 responses might be important in early stages of infection, a persisting IL-8 response may lead to excess recruitment of neutrophils and collateral tissue damage. In fact, higher urine levels of IL-8 in children with UTI has been associated with higher risk of renal scarring²⁴⁴. We do not yet know if the observed inter-organ communication between the infected kidney and the spleen is of harm or benefit for the host in the long run. Based on our findings, however, we suggest an immune-modulatory role for nerve-driven inter-organ communication during kidney infection. Such neural modulation of the immune response could contribute to a fine-tuned immune response to infection and prevent perpetuated inflammation.

While we report an immune-modulating effect of IFNγ at the infection site in the kidney, the effects of IFNγ might be manyfold. Different cytokines, including IFNγ, have also been reported to sensitize nociceptors^157-161. IFNγ could thus potentiate neural sensing of the bacterial infection by sensitizing the sensory nerves. Further, as neuro-immune signaling to

the spleen is known to activate neutrophils²⁴⁵, the observed splenic activation may prime cells in the spleen that later move to the site of infection. Studies examining this would, however, need to go beyond the 4 h infection window, as the effects we have observed on Cxcl1 expression in the kidney occurs before massive neutrophil infiltration in the kidney.

Thus, in Paper I we show that both renal epithelial cells and sensory nerves can come in contact with UPEC during a kidney infection and respond to the infection. With regards to sensory nerves, we show that they have neural responses to both direct contact with UPEC expressing HlyA, and indirectly through eATP released from renal epithelial cells infected with HlyA-expressing UPEC. We found that this direct and indirect sensing of HlyA triggers inter-organ communication between the infected kidney and the spleen in vivo in an anti-inflammatory neuro-immune reflex.

5.2 PAPER II – NERVES CAN SENSE UPEC LPS AND SIGNAL FOR LOCAL