Kidney Intercalated Cells Are Phagocytic And Acidify Internalized Uropathogenic Escherichia Coli
Feb 28, 2022
Kidney intercalated cells are involved in acid-base homeostasis via vacuolar ATPase expression. Here we report six human intercalated cell subtypes, including hybrid principal intercalated cells identified from single cell transcriptomics. Phagosome maturation is a biological process that increases in biological pathway analysis rank following exposure to uropathogenic Escherichia coli in two of the intercalated cell subtypes. Real time confocal microscopy visualization of murine renal tubules perfused with green fluorescent protein expressing Escherichia coli or pHrodo Green E. coli BioParticles demonstrates that intercalated cells actively phagocytose bacteria then acidify phagolysosomes. Additionally, intercalated cells have increased vacuolar ATPase expression following in vivo experimental UTI. Taken together, intercalated cells exhibit a transcriptional response conducive to the kidney’s defense, engulf bacteria and acidify the internalized bacteria. Intercalated cells represent an epithelial cell with characteristics of professional phagocytes like macrophages.
Keywords: kidney disease；kidney tissue；kidney cells；kidney; renal
The kidney-collecting tubule is the terminal location for acid–base regulation 1 . The collecting tubule consists of principal cells and intercalated cells (ICs). Principal cells (PCs) mediate salt and water resorption and are characterized by cytoplasmic aquaporin 2 (AQP2) 2 . ICs help maintain acid–base homeostasis and have three described subtypes, type A (A-IC), type B (B-IC), and nonA–nonB 2,3 . A-ICs secrete protons via apical vacuolar H-ATPase (V-ATPase) and regenerate bicarbonate via the basilar chloride/bicarbonate (Cl − /HCO 3 ) transporter, anion exchange 1 (AE1/band 3/SLC4A1) 1,4 . B-ICs secrete HCO 3 − via an apical Cl − /HCO 3 transporter, pendrin (SLC26A4), and express basolateral V-ATPase 2,3 . Rodent nonA–nonB ICs have apical pendrin and V-ATPase, but no AE1, and have been identified in the collecting duct and connecting tubule (CNT) 3 . Increasing evidence supports an innate immune role for ICs along with their traditional function of pH regulation. First, the majority of ten unrelated patients with distal renal tubular acidosis, a condition characterized by IC dysfunction, were reported to have a UTI history 5 . Second, uropathogenic Escherichia coli (UPEC), the organism isolated in 70% of acute pyelo-nephritis in males and 80% in females, selectively localizes to the cytoplasm of ICs 6,7 . Third, we identified that ICs express antimicrobial peptides (AMPs) such as ribonuclease 7 (RNASE7) 8–11 . These AMPs have direct activity against a range of pathogens, including bacteria, at baseline and/or in response to infection 8–11 . Fourth, we and others have independently reported that mice with abnormal IC development have increased susceptibility to urinary tract infections (UTIs) and/or pyelonephritis 12,13 . Fifth, ICs both sense and mediate inflammation 14 . Last, we have demonstrated innate immune gene expression in isolated murine ICs 15,16
CISTANCHE WILL IMPROVE KIDNEY/RENAL FUCTION
Understanding how ICs defend the urinary tract from infection is important because UTIs are frequently encountered. Over 50% of women experience a UTI during their lifetime, and UTIs account for 1–6% of all medical visits 17 . When bacteria ascend to the kidneys resulting in pyelonephritis, potential complications include hypertension, chronic kidney disease, and urosepsis 18,19 . An estimated 250,000 cases of pyelonephritis occur annually in the United States 17 . Multidrug resistance is emerging as a critical challenge regarding treatment of UTIs and other infections 20 . Therefore, the mechanisms ICs use to defend against ascending uropathogens might be therapeutic targets that could be exploited to develop kidney-specific pyelonephritis therapy. Because most IC studies involve animal models, a better understanding of human ICs will be needed to guide basic science studies into improved clinical care.
Because the kidney is a complex organ, mRNA profiles can be difficult to interpret. For example, the nephron has regional specialization with distinct tubular segments and cell subtypes. The kidney contains endothelial, epithelial, interstitial, and recruited inflammatory cells, all with distinct biological functions and gene expression patterns. Single-cell RNA sequencing (scRNA-seq) represents a rapidly advancing methodology to identify and profile distinct cell types at baseline or in response to stressors 21 . Here, we show using scRNA-seq technology that human kidney ICs consist of six different subsets that include three A-ICs, one hybrid IC–PC, one nonA–nonB IC, and one B-IC. RNA velocity analysis revealed transcriptional shift from PCs toward A-IC subsets upon UPEC exposure. Ingenuity Pathway Analysis revealed phagosome maturation as the key biological pathway in an A-IC subset upon UPEC exposure. Within this pathway, V-ATPases are the key involved components. We used intravital microscopy to visualize in vivo UPEC uptake by ICs and confirmed acidification of UPEC internalized in these ICs. Finally, a murine UTI model demonstrated increase in V-ATPase (Atp6v1b1) mRNA expression. This study will provide the foundation to explore and therapeutically manipulate V-ATPase expressing epithelial cells as bacterial phagocytes.
Results
Human ICs can be enriched via magnetic-activated cell sorting (MACS) with mast/stem cell growth factor receptor (CD117/c- KIT) coated magnetic beads. Human kidney tissue, primarily from normal margins of kidney mass resections, was dissected into 2–4-mm pieces then overnighted to our lab in Dulbecco’s Modified Eagle Medium (DMEM) on ice by the Cooperative Human Tissue Network [https://www.chtn.org]. We evaluated c- KIT as cell surface marker to sort human ICs because c-KIT has been successfully used for this purpose in mice 22 . c-KIT expression localized to human ICs based on the finding that c-KIT and the human IC marker V-ATPase, B1 subunit (gene ATP6V1B1), co-labeled the same cells in the human kidney section (Supple-mentary Fig. 1) 22,23 . Following magnetic sorting which included removal of CD45 + immune cells and dead cells; the enrichment of c-KIT-positive presumed ICs was demonstrated by ATP6V1B1 mRNA expression that increased by 8–12-fold in c-Kit-positive compared to the c-KIT-negative cells. Both SLC4A1 and SLC26A4 were variably enriched in ICs demonstrating that both A-IC and/ or B-IC enrichment is possible using this methodology (Supplementary Fig. 2). The background of the kidney tissue used in this study is presented in Supplementary Table 1.
Integrated cluster analysis identified six IC subtypes. ScRNA-seq was completed on 1861 dissociated kidney cells that were enriched for ICs as described above. The enriched ICs were obtained from the normal margins of a single kidney mass resection. Quality control (QC) filtering was passed by 859 cells exposed to UPEC in vitro for 1h and 1002 cells exposed to saline in vitro for 1h. Seurat function analysis identified 12 clusters in the sorted cell preparation (Fig. 1a). The nine most conserved markers within each cluster are presented in Supplementary Figs. 3–14. Overall, IC subtypes accounted for 1066 (57%) of the 1861 cells. Six of the clusters represented IC subtypes. Specifically, we found three A-IC, one B-IC, one nonA–nonB-IC, and one hybrid PC–IC subtype(s). Kidney epithelial cell nomenclature recommended by Chen and colleagues, conserved makers that have been previously reported to be cell-type markers or con- served markers that differentiated one of the 12 clusters from others were used to assign a cell type to each cluster (Supplementary Table 2, Fig. 1b, c, Fig. 2) 24–28 .
A-IC subtype and hybrid PC–IC protein expression correlates with scRNA-seq results. We validated key scRNA-seq findings, notably that SLC8A1 (also referred to as Na + /Ca + exchanger 1 (NCX1)) is a marker for hybrid PC–ICs and that heat shock protein family A (Hsp70) member 1A (HSPA1A) along with early growth response 1 (EGR1) are markers that can differentiate A-IC subtypes (Fig. 3). Because we used the normal margins of kidney cancer resections, we validated that our SLC8A1, HSPA1A, and EGR1 kidney expression patterns were consistent to that seen in normal kidney tissue from the Human Protein Atlas available from http://www.proteinatlas.org (Supplementary Fig. 15) 29
Collecting duct cell subtype marker profiles previously demonstrated in murine kidney cells remain largely intact in human kidney cells. Because it will be important to contextualize future murine model findings to human pathophysiology, we evaluated human IC expression of murine-collecting duct cell markers identified previously in scRNA-seq studies performed by Chen et al. 22 and Ransick et al. 30 . A dot plot of the results is presented in Fig. 4. Murine-collecting duct cell-type markers are mostly conserved in human collecting duct cells.
A-IC subtype A increases in relative percentage with 1 h of UPEC exposure. The percentages of cells within clusters were consistent between exposure type with the exception of A-IC subtype A/cluster 0, which increased from 16.2 to 20.1%, and A-IC subtype C/cluster 5, which decreased from 9.4 to 6.6%, with saline vs. UPEC exposures, P =0.03 (Table 1).
Hybrid PC–IC cells change their RNA velocity away from PCs and toward A-ICs in response to 1 h of UPEC exposure. The relative abundance of recently transcribed unspliced pre-mRNAs vs. mature spliced mRNA can be used to calculate the change in mRNA abundance, termed mRNA velocity 31 . A positive mRNA velocity in single-cell studies indicates that genes are being upregulated where a negative mRNA velocity means that genes are being downregulated 31 . The RNA velocity direction infers that a cell has a mRNA expression trajectory toward another cell type 31,32 . The mRNA velocity in response to UPEC exposure is presented in Fig. 5. Of importance, hybrid PC–ICs change their RNA velocity away from PCs and toward ICs in response to UPEC. In addition, ICs maintain their transcription activity, whereas transcriptional activity becomes minimal in other kidney cell types.
The kidney innate immune profile demonstrates some early single-cell expression changes following 1 h of UPEC exposure. scRNA-seq expression patterns of select innate immune genes following UPEC vs. saline exposure for 1h are presented in Sup- plementary Figs. 16 and 17. Key findings include high expression of the AMP adrenomedullin (ADM) in A-IC subtypes A and B, hybrid PC–ICs, and B-ICs. PCs do not express adrenomedullin at baseline but have significantly increased expression in response to UPEC exposure. Cytokine-inducible SH2-containing protein (CISH) and barrier to autointegration factor 1 (BANF1) expression is significantly induced in nonA–nonB ICs following UPEC exposure. Interleukin 18 (IL18), galectin 3 (LGALS3), beta defensin 1 (DEFB1), and signal transducer CD24 lipocalin 2/neutrophil gelatinase-associated lipocalin (LCN2/NGAL) was only identified
in hybrid PC–ICs and only minimal toll-like receptor 4 mRNA expression was present. We did not identify expression of ribo-nuclease 7 (RNASE7) on scRNA-seq. We evaluated RNASE7 mRNA expression in magnetically sorted pooled IC and non-IC kidney cells. RNASE7 mRNA expression was identified in ICs from one out of four kidneys (Supplementary Fig. 18).
The biological process of phagosome maturation is associated with ICs exposed to UPEC for 1h. The ten leading biological pathways associated with each cluster were ranked by p value as determined by Ingenuity Pathway Analysis. Three collecting duct-assigned clusters A-IC subtype A/cluster 0, A-IC subtype B/ cluster 3, and A-IC subtype C/cluster 5 had differential biological pathway processes following UPEC exposure compared to saline (Fig. 6a–c). Phagosome maturation was the top ranked function in cells exposed to UPEC and saline in A-IC subtype C/cluster 5, moved from second to first position in A-IC subtype A/cluster 0 and from fourth to third highest position in A-IC subtype B/ cluster 3. None of the remaining collecting duct cell subtypes (Supplementary Fig. 19) had top ten associated pathways that were different following UPEC exposure compared to saline. The gene expression profile that resulted in the phagosome maturation ranking for A-IC subtype A/cluster 0 is presented in Sup- plementary Table 3.
IC Atp6v1b1 mRNA expression increases in response to experimental pyelonephritis in vivo. The Ingenuity phagosome maturation pathway highlighted V-ATPase as involved in ICs (Supplementary Table 3). To determine if V-ATPase is activated upon UPEC exposure, we used a murine UTI model. One hour post transurethral UPEC vs saline inoculation “IC reporter” mice, which have tdTomato (tdT, a red fluorescent protein variant) expression in ICs, were euthanized and ICs were enriched from dissociated kidney cells. Enriched ICs had higher Atp6v1b1 mRNA expression in mice with UPEC inoculation compared to saline control (Fig. 6d).
ICs phagocytize UPEC in vivo. We developed methodology to perfuse a single kidney tubule with UPEC in a live mouse with an intact kidney. Green fluorescent protein (GFP)-expressing UPEC aggregates microperfused into a single kidney tubule lumen of “IC reporter” mice were visualized flowing through tubules, selectively adhering to the luminal surface and being internalized by ICs (Fig. 7) during intravital microscopy.
ICs acidify UPEC containing phagolysosomes in vivo and in vitro. We also perfused tubules with pHrodo Green E. coli BioParticles that only fluoresce when acidified. Uptake and fluorescence of these bioparticles was only visualized during\ intravital microscopy in ICs (Fig. 8). Imaris segmentation analysis demonstrated that green fluorescence significantly increased over time in tdT + ICs within 2 tubules perfused with pHrodo Green E. coli BioParticles compared to a control tdT + tubule (Fig. 9a, b). When fluorescence was evaluated on a cellular basis, 12/16 (75%)tdT + cells demonstrated an increase in green fluorescence over time following tubular perfusion with pHrodo Green E. coli BioParticles (Fig. 9b). The linear regression results from each cell are presented in Supplementary Table 4. The live in vivo intravital imaging findings of phagocytosis and acidification of PHrodo Green E. coli BioParticles by ICs were validated using an in vitro flow cytometry assay. Murine kidney cell suspension from “IC reporter” mice in which ICs (CD45 − tdT + ) endogenously express tdT was exposed to pHrodo Green E. coli BioParticles versus cells with media alone. CD45 + resident immune cells in cell suspension were gated as a positive control and CD45 − tdT − (non-ICs) cells were gated as a negative control. Comparing media alone to
bioparticle exposure, green fluorescence indicative of bioparticle acquisition and acidification increased from 0.7 to 62.8% of CD45 + cells. This finding demonstrates the expected bacterial uptake by professional phagocytes such as neutrophils. Interestingly, 22.2% of the ICs had uptake of green bioparticles compared to 5.6% of non-ICs (Supplementary Fig. 20), media control had uptake in 0.7% of ICs and 0% of non-ICs.
Discussion
ICs are difficult to study, particularly in humans, because they comprise only 7% of kidney cells, consist of a range of subtypes, and do not retain their phenotype in culture 33 . Chen et al. 22 reported single-cell analysis on murine kidney cells and Lake et al. reported a single-nucleus RNA sequencing pipeline on human kidney cells 34,35 . These studies identified two subtypes of A-ICs. Here, we have developed methodology to perform single-cell sequencing on viable enriched human ICs. We enriched for viable ICs rather than sequencing all kidney cells to allow increased focus on this cell type. Our primary findings were (a) identifi- cation that hybrid PCs–ICs can switch their RNA velocity direction away from PCs and toward A-ICs in response to UPEC exposure, (b) at least three subtypes of A-ICs exist and their relative frequencies can shift in response to UPEC exposure, (c) phagosome maturation is a leading biological pathway in multiple A-IC subtypes and can increase in relative significance with UPEC exposure, and (d) ICs phagocytose and acidify UPEC in vivo during imaging of live mice.
ICs differentiate in response to acidosis by reversing their polarity 36,37 . ICs have been reported to arise from AQP2- expressing cells 38 . In 2017, Chen and colleagues identified double-positive cells that expressed Aqp2 along with Slc4a1 or Slc26a4 by murine single-cell profiling. These findings were validated by Park and colleagues the following year 22,35 . In 2019, we demonstrated by protein and mRNA immunolabeling that hybrid PC–ICs exist 15 . Specifically, 9% of murine ICs defined by V-ATPase B1 fluorescent immunolabeling also expressed Aqp2 mRNA defined by fluorescent in situ hybridization 15 . The present study defined a population of human kidney-collecting duct cells that express AQP2, ATP6V1G3, and SLC4A1 consistent with hybrid PC–ICs. On tSNE and UMAP plots, these hybrid cells clustered closest to PCs, but were distinct from traditional PCs based on IC marker expression. SLC8A1 expression was distinct to this cell type. Interestingly SLC8A1 is highly involved with epithelial-to-mesenchymal transition. Specifically, the absence of SLC8A1 transforms epithelial cell phenotypes via β-catenin- mediated destabilization of E-cadherin 39 . Here, we demonstrate that ICs differentiate in response to uropathogen exposure. Hybrid PC–IC cells change their RNA velocity, the time derivative of mRNA expression, away from PCs and toward A-ICs in response to UPEC. In mice, Ransick et al. 30 identified Slc8a1 expression primarily in the distal convoluted tubule and cells resembling PCs in the connecting tubule. On our confocal microcopy images evaluating kidney SLC8A1 expression, we found isolated cells that expressed SLC8A1 in AQP2-positive tubules (e.g., collecting ducts or connecting tubules). We also identified AQP2-negative tubules in which most cells expressed SLC8A1, potentially consistent with the aforementioned distal convoluted tubule Slc8a1 expression in mice. Some A-IC subtypes have differential leading biological pathways following UPEC exposure and A-IC subtype A increased in relative frequency in response to UPEC. These findings indicate that A-IC subtype A may represent an innate immune variant. The role of SLC8A1 in IC differentiation warrants investigation in future studies.
CISTANCHE WILL IMPROVE KIDNEY/RENAL FAILURE
We performed scRNA-seq on 1861 human kidney cells of which 1066 were ICs. To contextualize murine experimental models to human pathophysiology, it will be important to compare human and murine ICs. Chen and colleagues enriched for murine ICs with c-KIT and performed scRNA-seq 22 . They classified 74 cells as PCs, 87 as A-ICs, and 23 cells as B-ICs 22 . Ransick et al. 30 performed scRNA-seq on 688 ICs. They did not enrich for ICs but rather divided the kidney into the cortex, inner medulla, and outer medulla to evaluate zonal differences 30 . We demon strated that murine-collecting duct cell-type marker expression appeared to be largely conserved in human collecting duct cells. Whether mice have comparable A-IC subtypes to humans will likely require a targeted single-cell evaluation of a larger number of murine ICs.
We identified scRNA-seq IC expression of innate immune proteins such as the immune and inflammatory mediator galectin 3 (LGALS3) and the ADM that we have previously reported in rodent ICs 15,40–42 . However, some key innate immune proteins previously reported in ICs, such as Lipocalin 2 (LCN2/NGAL), RNASE7, and toll-like receptor 4 (TLR4) had minimal expression or were not seen in human ICs on the single-cell level in our analysis 6,9 . The paucity of LCN2/NGAL and TLR4 expression in ICs is consistent to what has been reported in murine ICs by our research group along with Ransick et al. 30 kidney cellexplorer/] 15 . Chen et al. 22 did report some TLR4 and NGAL mRNA expression in ICs, but several fold less than in PCs. We did evaluate pooled human ICs for RNASE7 expression and identified it in ICs from 1 out of 4 kidneys (Supplementary Fig. 18), indicating that its expression may be intermittent depending on region, time point, and physiological conditions. CISH expression, which is induced in nonA nonB ICs in response to UPEC, has been demonstrated to regulate the innate immune response to Mycobacterium tuberculosis in the lung and spleen 43 . ICs phagocytosed bacteria over several minutes, and our cells for scRNA-seq were exposed to UPEC for a relatively short 1 hour time point. Future studies are needed to determine if other IC pathways, such as AMP expression or regulation of cell death, become more prominent in later time points following UPEC exposure.
Phagocytosis involves the cellular uptake of particulates >0.5 μm by the plasma membrane envelope 44 . “Professional phagocytes”, myeloid-derived immune cells, including macrophages, neutrophils, and osteoclasts differentiate invading microbes from the microbiota and healthy cells, engulf the target into a phagosome, generate reactive oxygen species (ROS), secrete AMPs, and present antigens to other cells 45–49 . Of importance, robust acidification by V-ATPase of phagolysosomes creating an acidic microenvironment sufficient for killing most microbes is a hallmark characteristic of professional phagocytes 48,50 . For example, inhibiting V-ATPase with bafilomycin A1 suppresses macrophage bactericidal activity 51 . Certain epithelial cells phagocytose microbes but are less efficient at bacterial killing than professional phagocytes and have been described as nonprofessional phagocytes 52 . ICs represent an epithelial lineage that has phagocytosis and antigen-presentation capabilities (Figs. 6–9) along with robust AMP expression and generous mitochondria to generate ROS; thus, ICs appear to have more comparable bacterial-killing potential to macrophages than the aforementioned nonprofessional phagocytes 8,53–55 . A-ICs have previously been demonstrated to be capable of high rates of apical endocytosis of dextran into cytoplasmic vesicles 56 . We speculate that IC phagocytosis of UPEC is an extension of characteristics shared with macrophages, including V-ATPase expression, redox potential, AMP secretion, and endocytosis capabilities 8,16,49,57,58 . Whether ICs can phagocytose cellular debris and bacteria other than UPEC in a similar manner to professional phagocytes remains to be determined.
Confocal imaging of a single-kidney tubule perfused in vivo with bacteria expands on prior in vitro techniques like cell culture and perfusion of dissected tubules. Here, we developed metho- dology to directly test in live mice whether phagocytosis is an IC function. Intravital microscopy allows for intricate study of dynamic cellular processes within functioning murine kidneys 59 . Past strategies to utilize intravital microscopy have involved systemic injection of intravital dyes or markers into mice to study proximal tubule endocytosis, kidney blood flow dynamics, and vascular or glomerular permeability 60 . When intravital micro- scopy is combined with sophisticated surgical techniques, difficult-to-evaluate early time points of infective processes can be studied 61 . Previously, bacteria were microperfused in proximal tubules of rat to evaluate blood flow, neutrophil recruitment, and urinary obstruction 62,63 . The vast majority of UTIs are due to ascending pathogens and would initially encounter the collecting duct 64,65 . Thus, we propose that the collecting duct is a critical tubular region to evaluate host-pathogen interactions. Our intravital model system effectively recreates a classical
CISTANCHE WILL IMPROVE KIDNEY/RENAL DISEASE
single collecting tubule perfusion study, but in a live mouse with an intact kidney, blood, lymphatic, and neuronal supply along with retained ability to interact with immune cells. Pyelonephritis pathophysiology has male versus female distinctions. For example, in humans, females are five times more likely to develop acute pyelonephritis but male mice have androgen-mediated increased severity of pyelonephritis 66,67 . Ransick and colleagues demonstrated differential male and female differences in proximal tubule cells, particularly in organic anion transporters 30 . Male ICs were evaluated in both our single-cell sequencing and intravital studies. Because of the role of ICs in the bacterial defense of the kidney and electrolyte balance, sex-related diversity in IC functions, such as phagocytosis and single-cell gene expression, will be key areas for future studies. This study does have limitations. The UPEC exposure to human ICs in our scRNA-seq was in vitro to a single kidney and may have distinctions compared to in vivo UPEC exposure. In addition, our findings represented the anatomical region of the human kidney sampled and future studies with different locations may have regionalized diversity as demonstrated by Ransick et al. 30 . ICs comprised 57% of the cells in our scRNA-seq, enriched from the ~7% IC composition in kidneys at baseline 33 . Our sample did contain some other epithelial and endothelial cell types, but was negative for CD45 + immune cells that were sorted off prior to enrichment of ICs (Fig. 1, Supplementary Fig. 1). With scRNA-seq, the target ICs could be identified and evaluated separately. Hematopoietic stem cells express c-KIT, but they do not likely represent the kidney resident myeloid cell populations 68 . We did not identify other c-KIT-expressing cells present in healthy human kidneys, such as stromal cells/telocytes 69 . Some isolated loop of Henle and proximal tubule c-KIT positivity has been previously reported on kidney immunohistochemistry consistent with our scRNA-seq results 70
