Spatio-temporal patterning of different connexins in developing and postnatal human kidneys and in nephrotic syndrome of the Finnish type (CNF)

Connexins (Cxs) are membrane-spanning proteins which enable flow of information important for kidney homeostasis. Changes in their spatiotemporal patterning characterize blood vessel abnormalities and chronic kidney diseases (CKD). We analysed spatiotemporal expression of Cx37, Cx40, Cx43 and Cx45 in nephron and glomerular cells of developing, postnatal kidneys, and nephrotic syndrome of the Finnish type (CNF) by using immunohistochemistry, statistical methods and electron microscopy. During kidney development, strong Cx45 expression in proximal tubules and decreasing expression in glomeruli was observed. In developing distal nephron, Cx37 and Cx40 showed moderate-to-strong expression, while weak Cx43 expression gradually increased. Cx45/Cx40 co-localized in mesangial and granular cells. Cx43 /Cx45 co-localized in podocytes, mesangial and parietal epithelial cells, and with podocyte markers (synaptopodin, nephrin). Different Cxs co-expressed with endothelial (CD31) and VSMC (α –SMA) markers in vascular walls. Peak signalling of Cx37, Cx43 and Cx40 accompanied kidney nephrogenesis, while strongest Cx45 signalling paralleled nephron maturation. Spatiotemporal Cxs patterning indicate participation of Cx45 in differentiation of proximal tubules, and Cx43, Cx37 and Cx40 in distal tubules differentiation. CNF characterized disorganized Cx45 expression in proximal tubules, increased Cx43 expression in distal tubules and overall elevation of Cx40 and Cx37, while Cx40 co-localized with increased number of interstitial myofibroblasts.


Materials and methods
Human tissue processing. In our study, 11 human embryonic and foetal tissues were acquired from the Department of Gynaecology and Obstetrics after spontaneous abortions, or after tubal pregnancies from the Department of Pathology. Tissues were processed with permission of the Ethical and Drug Committee of the University Hospital in Split in accordance with Helsinki Declaration. The age of conceptuses was evaluated between 8 th and 38 th developmental week from menstrual data and corresponding to the external measurements (crown-rump length) and the Carnegie staging system. All tissues were without sings of macerations and morphologically regular. In our research we also included postnatal tissue taken during autopsy of healthy 1,5-year old boy and nephrotic kidney tissue from 3 nephrectomised CNF patients (homozygous missense mutation c.1096 A > C; pSer366Arg in NPHS1 gene was detected in all three patients). Informed parental consent was obtained for the research purposes 2 . Tissues were furtherly processed and glass slides were prepared as previously described 39 . Tissue was fixed in 4% paraformaldehyde in phosphate buffer saline (PBS), dehydrated in ethanol dilutions, paraffin-embedded and serially cut as 5 µm thick sections. Appropriate tissue preservation was confirmed by Haematoxylin and Eosin staining of every 10 th section. Tissue was analyzed using Olympus BX51 light microscope (Olympus, Tokyo, Japan).
Immunohistochemistry and immunofluorescence staining. Tissue sections were deparaffinized and then rehydrated in ethanol following standard protocol 40 and treated with sodium citrate buffer. After washing with PBS, blocking buffer (Protein Block, Abcam, UK) was administered on the tissue covered area. Primary antibodies were applied (Table 1) overnight and rinsed in PBS. Suitable secondary antibodies were applied (Table 1) and incubated in humidity chamber for one hour. Considering the double staining with lectins, after the mentioned procedure of administering primary antibodies, sections were incubated with FITC-conjugated lectins (Dolichos Biforus Agglutinin -DBA or Lotus Tetragonolobus Lectin -LTL, Table 1) in dark. They were washed in PBS following the 2 hr incubation period at room temperature 41 . Slides were afterwards washed in PBS and treated with DAPI nuclear staining. Control for specificity was excluding primary antibody from the staining procedure. For imaging, Olympus fluorescence microscope (BX61; Tokyo, Japan) with a digital camera (DP71) was used. Images were captured by using the Olympus CellA software and assembled in the plates by Adobe Photoshop. For Fig. 5 exclusively, we used DS-Ri2 digital camera for imaging, and NIS-Elements F software for image capture.
Semi thin and ultra-thin sections and electron microscopy. In preparation for electron microscopy we used tissue samples of 10 th and 22 nd week-old human kidneys, 1.5 years healthy and 3-year CNF kidney tissue. After the 24-hour fixation in 4% paraformaldehyde, specimens were post-fixed in 1% osmium tetroxide for an hour, dehydrated in courses of ethanol and embedded in LX 112 resin. Toluidine blue was used for staining of previously cut semi thin sections (1 µm). Semi thin sections were additionally cut into ultrathin sections (0.05 µm thick) which were examined by transmission electron microscope (Zeiss 902 A, Germany) after being treated with uranyl acetate and lead citrate staining 2 .
Semi-quantification. The staining intensity of chosen antibodies was semi-quantitatively assessed by four categories: with (−) indicating the absence of any reactivity; (+) as a mild reactivity; (++) as moderate reactivity; and (+++) as strong reactivity (Table 2). Three researchers semi-quantitatively analyzed the staining intensity separately by image analysis software ImageJ (National Institutes of Health, Bethesda, MD, USA) 1 .

Statistical analysis.
Florescence intensity histograms were acquired for red and green fluorescence channels in ImageJ software (NIH, Bethesda, MD, USA). Pixels that had fluorescence intensity between 0 and 15 were considered background. Expression of different connexins was quantified as the area under the curve (AUC) of florescence intensity histograms, and we refer to this as florescence intensity unit (FIU) in the rest of the article. AUCs and their interval estimates were calculated by using AUC analysis routine in GraphPad Prism 8.0 software (Graph Pad, La Jolla, CA, USA). Statistical significance and effect sizes as well as respective 95%CI were calculated by ANOVA followed by Tukey post hoc test in GraphPad Prism 8.0 software (Graph Pad, La Jolla, CA, USA). Level of significance was set at p = 0.05. Expression of connexins during embryonic development was analyzed by finding peaks or nadirs in time series using the single sample t test.

Results
During early human development, the ureteric bud induces metanephric cup mesenchymal cells to gradually differentiate into renal vesicle epithelial cells, with centrally positioned lumen. The described process is known as mesenchymal-to epithelial transition 1 . During further development renal vesicle transforms into S-shaped body, which gives rise to immature glomeruli and different parts of nephron tubules.
Co-expression of Cx43 and Cx45 in the developing and postnatal kidney tissue. Double immunofluorescence method reveals co-expression of Cx45/Cx43 in the metanephric cup and collecting tubules at the earliest stages of kidney development ( Fig. 1a-d). In immature glomeruli, Cx43 characterizes endothelial cells of vascular wall, while Cx45 is positioned in more peripheral part of the vascular wall (VSMC), thus Cx43/Cx45 expression mostly does not overlap. Cx45 is strongly expressed in some glomerular cells (probably corresponding to mesangial cells) ( Fig. 1e-h).
In the second half of intrauterine life, strong Cx45 expression characterizes all kidney structures, while weak Cx43 expression is observed in glomeruli, proximal and distal tubules. Cx43 expression is observed in the afferent and efferent arterioles and parietal epithelial cells. Again, Cx43/Cx45 co-expression is observed in in different cellular compartments of the same structures ( Table 2, Fig. 1i-l).
In postnatal kidneys, Cx43/ Cx45 co-express in different cellular or membranous compartments of the same kidney structures including podocytes, walls of blood vessels and parietal epithelial cells, Cx43 is expressed moderately, while Cx45 strongly ( Fig. 1m-p, inset Fig. 1p).
Co-expression of different Cxs with CD31 and α -SMA in developing and postnatal kidney tissue. Between 8 th and 10 th developmental week, human kidneys contain all described stages of nephron formation, with immature glomeruli positioned closest to the future medulla. Metanephric cup cells strongly express Cx45, moderately Cx40 and Cx37, and weakly Cx43. In immature glomeruli and collecting ducts, Cx43 is expressed weakly, Cx45 strongly, while Cx40 is expressed moderately-to-strongly in the form of dense granules. Cx37 is moderately expressed in some metanephric cup cells and immature glomeruli, but strongly in collecting tubules in the form of coarse grains ( Table 2,  During further development (10 th to 38 th week), glomeruli became mature, while proximal convoluted tubules and distal convoluted tubules become morphologically distinguishable. During that period, Cx45 is moderately expressed in the glomeruli and distal tubules, while strongly in proximal tubules (Fig. 2m,q). Cx43 slightly increases to moderate in the distal tubules, while Cx40 display moderate-to-strong expression in all three structures. Cx37 expression is moderate in the whole kidney, while weakly in the proximal tubules (Table 2).
In the 22 nd and 38 th developmental week, moderate-to-strong co-expression of Cx45 co-expresses with strong expression of CD31 and α -SMA in the walls of blood vessels (Table 1, Fig. 2m-p,q-t).
In postnatal healthy kidneys, expression of Cx45 increases only in glomeruli, while Cx43 increases from weak to moderate in all kidney structures. While Cx37 shows no changes in its expression pattern, expression of Cx40 is postnatally mild in glomeruli, moderate in proximal tubules, but increases to strong in distal tubules, (Table 2., see Fig. 2u). Double staining of Cx40 and α -SMA shows their co-localization in intraglomerular and extraglomerular blood vessels (Fig. 2u-x) Co-expression of different Cxs with CD31 and α -SMA in CNF. In CNF kidneys, glomeruli display different degrees of pathological changes, while some proximal tubules appear dilated thus forming small cysts. Compared to healthy postnatal kidneys, in CNF kidneys expression and distribution of different Cxs changes.: Cx45 and Cx43 expression slightly decreases in glomeruli, while Cx40 and Cx37expression increases in some distal tubules (Table 2). Cx43/Cx45 co-express in different cellular-compartments of the same kidney structures ( Table 2, Fig. 3a-d). Cx37 shows irregular expression in the form of coarse granules within glomeruli, some parts of proximal and distal tubules. Double staining of Cx37 and CD31 shows their co-expression in the walls of blood vessels (Fig. 3e-h). Cx40 displays reduced expression in affected glomeruli, increased expression in distal tubules and moderate expression in interstitium, Double staining of Cx40 with α -SMA shows their co-localization in interstitial myofibroblasts (Fig. 3i-l).  www.nature.com/scientificreports www.nature.com/scientificreports/ Co-expression of nephrin and synaptopodin with different Cxs in developing and CNF kidneys. Up to the 10 th developmental week, synaptopodin expression is absent in the kidney structures (Table 2.). In the 22 nd foetal week, Cx43 is mildly expressed in the glomeruli, proximal and distal tubules, and strongly in the juxtaglomerular apparatus (jga), while synaptopodin expression is moderate in the glomeruli, and very mild in proximal and distal tubules ( Table 2). Nephrin expression has been previously shown to increases during human development 2 Synaptopodin only partly co-expresses with Cx43 in the glomerular cell population probably corresponding to podocytes (Fig. 4a-d), as well as does Cx45 with nephrin ( Fig. 4e-h).
CNF kidneys show decreased and only partial expression of synaptopodin in the glomeruli, and very weak expression in proximal and distal tubules ( Table 2; Fig. 4i-l). Cx43 and synaptopodin do not co-express in the podocytes (Fig. 4i-l), while nephrin expression in CNF glomeruli is very mild (Fig. 4m-p).

Co-expression of different Cxs with markers of proximal tubules (DBA) and distal tubules (LTL)
in developing, postnatal and CNF kidneys. In 8 th to 10 th developmental week, only small number of kidney tubules showed strong signal to LTL, while there was no signal to DBA at that developmental stage. Double staining of Cx37 with LDL shows absence of their co-expression in distal tubule, while positive signal to Cx37 was seen in collecting ducts ( Table 2, Fig. 5a-d). In contrast to developing kidneys, postnatal kidneys showed strong reactivity to both DBA in proximal tubules and LDL in distal tubules. Double staining of Cx40 with DBA showed their co-expression in proximal tubules (Fig. 5e-h), while Cx43 co-expressed with LDL in distal tubules (Fig. 5i-l).
In CNF kidneys, dilated proximal tubules strongly reacted with DBA. Double staining of Cx37 with DBA shows their co-expression in proximal tubules (Fig. 5m-p).

Quantification and semi-quantification of Cxs expression during development and postnatal period. Semi-quantification of Cxs expression in different parts of nephron is shown in
Analysis of the mean fluorescence signal of different Cxs revealed significant difference of maximal signal (peak signalling) between Cxs: Cx37, Cx43 and Cx40 displayed very similar course of fluorescent intensity signalling, being strongest in the 21 st developmental week, but overall intensity of Cx40 was stronger (p = 0.0003, p = 0.02, p < 0.0001, respectively). The strongest fluorescence signalling characterized Cx45, with peak in the 38 th developmental week (p = 0.03). In postnatal period, intensity of florescence signal dropped, except for Cx40 which remained high (Fig. 2y). All Cxs signals are much stronger in CNF kidneys compared to healthy postnatal kidneys (Fig. 3m).

Discussion
Connexins are transmembrane proteins that form gap junctions, which in the kidney tissue primarily contribute to renal haemodynamics, as they mostly localize in the walls of blood vessels. However, they also seem to have essential role in tubular epithelial and glomerular function 42 , in prenatal development, morphogenesis and tissue differentiation. By now, numerous experimental studies resulted in controversial data about spatio-temporal expression of specific Cx sub-types in the kidney tissue.
In the present study, strong expression of Cx45 displayed decreasing course in glomeruli of developing kidneys, while its strong expression in proximal tubules continued into postnatal period. In contrast, during www.nature.com/scientificreports www.nature.com/scientificreports/ development of mouse kidneys, Cx45 expression paralleled appearance of renin producing cells, but postnatally disappeared 38 . Compared to healthy postnatal kidneys, CNF kidneys showed disorganized distribution of Cx45 in proximal tubules, what might be associated with faulty signalling. CNF has been previously shown to be histologically characterized by tubular primary cilia dysregulation and appearance of proximal tubules cysts 43 . We also found both Cx45 and Cx40 expression in glomerular mesangial and granular cells, thus confirming possibility of appearance of heterotopic channels between different Cxs, which has been shown in experimental animals 19,38,44 . However, use of different techniques in studies analysing Cxs expression led to conflicting conclusions. Thus, Cx45 was detected in mouse distal tubules by immunohistochemistry 23 but its mRNA failed to be confirmed 32 .
In our study, different kidney structures showed weak Cx43 expression, which during development gradually increased to mild only in distal tubules. While Cx43 expression increased in glomeruli and proximal tubules of postnatal healthy kidneys, it became irregular and strong in distal tubules of CNF kidneys. In contrast to our results, previous study on human foetal kidneys missed to show early Cx43 expression in the metanephric cup cells 34 , what might be explained by higher sensitivity of immunofluorescence antibodies used in the present study. Discrepancies in the reported expression of CX43 and Cx45 might also reflect interspecies differences in Cxs expression 20 . In comparison to Cx45, Cx43 expression was confined only to apical cytoplasmic compartment of tubular nephron cells, while Cx43 / Cx45 co-expression characterized different cellular compartments in mesangial cells, podocytes, and parietal epithelial cells. Our study also showed that both Cx43 and Cx45 co-localized with markers for podocyte pedicles, thus pointing to importance of those two Cxs subtypes in podocyte signalling. Dedifferentiation of human podocytes in CNF kidneys has been previously shown to be associated with changes in cytoskeletal arrangement, presence of primary cilia and re-expression of nestin 1,2 . Additionally, interactions between connexins were suggested to be crucial in regulation of EMT of developing kidney cells 45 , which in CNF seems to be disturbed.
Studies on animals showed that Cx43 might be primarily involved in vasomotion, but also in gene transcription, ATP and vesicle release, cytoskeletal dynamics and cell stress 11 . In CKD, deletion of Cx43 caused anti-inflammatory effect and reduced interstitial fibrosis, while similar to our study -its disruption led to depletion of podocyte pedicles 46 . In contrast, Cx43 upregulation promoted pro-inflammatory environment 21 thus suggesting that alterations in expression of Cxs might be associated with development of CKD 37 . www.nature.com/scientificreports www.nature.com/scientificreports/ (m-p) Postnatal CNF kidneys (3,5 years). Cx37, Cx40 and Cx45 expression (arrows), expression of markers for proximal (DBA) and distal (LTL) tubules (arrows), proximal tubules (pt), distal tubules (dt), collecting tubules (ct), interstitium (i), glomeruli (g), blood vessels (bv). Co-expression of Cx37 with LTL is missing in 10 th developmental week kidneys (a-d), while co-expression of Cx40/DBA (arrowhead) (e-h) and Cx43/LTL (arrowhead) (i-l) is seen in proximal and distal tubules of postnatal kidneys. In CNF, Cx37 co-expresses with DBA (arrowhead) in dilated proximal tubules. Double immunofluorescence staining of DAPI with Cx37/LTL, Cx40/DBA, Cx43/LTL and Cx37/DBA x40. www.nature.com/scientificreports www.nature.com/scientificreports/ During development, moderate-to-strong expression of Cx37 in human kidney structures continued into postnatal period. Similar to our study, previous animal studies have also shown expression of Cx37 primarily in distal nephron 20 . Importance of early expression of Cx37 and Cx43 was implicated in kidney development, while their elevation was referred in neonatal unilateral ureteral obstruction 32 . We also noticed overall elevation of Cx40 and Cx37 expression in CNF kidneys, particularly in distal tubules.
Moderate-to strong expression of Cx40 during human kidney development decreased in glomeruli and increased in distal tubules in postnatal period. Opposite to our study, Cx40 expression in rats was observed in intra and extraglomerular mesangial cells and non-glomerular endothelial cells 29 .
While use of different techniques in analysing Cxs expression in kidney tissue often led to conflicting results, application of immunohistochemical method in kidney tissues of different animal species and humans revealed a certain degree of conservation in their spatio-temporal expression.
In conclusion, in developing human kidneys co-expression of different Cxs with markers for endothelial cells (CD31) and VSMC (α -SMA) indicated importance of communications between those two cell populations for vascular wall integrity. Overall signalling intensity of Cx37, Cx43 and Cx40 had its peak in the 22 nd developmental week and accompanied a process of nephrogenesis, while peak of Cx45 expression in the 38 th week corresponded to period of functional nephron maturation 47,48 . Based on the Cxs spatiotemporal expression in glomerular and tubular human kidney cells, we suggested importance of Cx45 in differentiation of proximal tubules, and Cx43, Cx37 and Cx40 in differentiation of distal tubules. Additionally, fine cytoplasmic balance between Cx43 and Cx45 co-expression might be crucial for differentiation of glomerular cell population, including podocytes, mesangial and parietal epithelial cells. The observed changes in Cxs expression characterized all structures in developing and CNF kidneys, thus pointing to Cxs as possible target in treatment of kidney diseases and repair processes 29 . Alterations in gap junction activity have been shown to cause structural and functional damage in several kidney diseases 21,29,37 . We also showed increased number of myfibroblasts co-localizing with Cx40 in the interstitium of CNF kidneys, thus implying possible involvement of Cx40 in regeneration processes 49 . Namely, expansion of myofibroblast population, which acquired some functional and structural characteristics of smooth muscle cells 50 lead to extensive deposition of interstitial extracellular matrix, which is the key characteristic of renal fibrosis in CKD. In addition, some dedifferentiated renal fibroblasts re-expressed mesenchymal markers, what might be considered as a sign of EMT 6 . Therefore, treatment that would re-establish mature fibroblast phenotype could be another option for treating renal fibrosis 6,51 .