PolySia is expressed during embryonic and postnatal murine kidney development

To explore possible implications of polySia in the glomerular vasculature development, we examined the spatiotemporal polySia expression during postnatal kidney development in wildtype mice. We performed immunohistochemical stainings with the monoclonal antibody (mAb) 735, detecting polySia with DP ≥ 10 [39, 40] (Fig. 1C). Specific binding was confirmed routinely by the treatment of consecutive tissue sections with endosialidase F (Endo), specifically degrading polySia [41] (Supplemental Fig. S1). At birth, the murine kidney exhibited robust polysialylation with prominent staining in the cortical area including the nephrogenic zone (NZ), where nephrogenesis occurs, and the medulla (Fig. 1C). At post natal day (P)0.5, the formation of new nephrons, the functional units of the kidney, is still in progress and polySia is found on interstitial cells and different stages of nephron development, like the cap mesenchyme (CM), comma-shaped bodies (CB) and S-shaped bodies (SSB). The renal corpuscle (RC) which forms the blood-filtering unit of the mature nephron, tubular structures and ureteric bud tips (UBTs) that give rise to the collecting duct system were devoid of polySia (Fig. 1D). A similar pattern was observed in renal sections at embryonic day (E)14.5 (Fig. 1E), with polySia expression in interstitial cells as well as CM and SSBs, but not in RCs and UBTs. As renal development progresses for at least one week after birth, we monitored polySia expression at P7.5 and P15.5. Although decreasing with age, polySia expression continued (Fig. 1F, G). By the time of kidney maturation (P15.5), its expression became confined to regions adjacent to the Bowman capsule, interstitial cells adjoining blood vessels and tubules, and medullary interstitial cells. Western blot analysis of wildtype kidney homogenates confirmed the decrease of polySia expression during postnatal development (Fig. 1H). We neither observed polySia in kidney homogenates nor at cellular level in renal sections from St8sia2−/− St8sia4−/− mice (Fig. 1I-K). Collectively, these findings suggest a pivotal role for this glycan with regulated spatiotemporal expression pattern in kidney morphogenesis.

PolySia-deficient mice display reduced numbers of glomerular endothelial cells

Next, we comparatively analysed kidneys isolated from wildtype and St8sia2−/− St8sia4−/− mice. Mutant kidneys were similar in gross morphology to wildtype kidneys, and the overall kidney architecture appeared unaltered, as evidenced by in situ hybridization (Supplemental Fig. S2). In H&E stained kidney sections of mature kidneys (P15.5), the glomeruli of polySia-negative mice appeared smaller compared to control animals (Fig. 2A), with morphometric analysis confirming significantly smaller glomerular tuft areas in St8sia2−/− St8sia4−/− mice (Fig. 2B). Notably, a similar histological phenotype was seen in mice that are deficient in NCAM, the major polySia carrier protein [42], but not in single knockout mice lacking either St8sia2−/− or St8sia4−/− (Fig. 2B). Next, we analysed the functionality of blood filtration by measuring protein/creatinine ratios in urine samples from control and polySia-negative mice from P0.5—P15.5 (Fig. 2C). Consistent with the immature blood filtration barrier in newborn mice, low-level proteinuria was apparent in both genotypes. While proteinuria, given by protein/creatinine ratio, was increased in St8sia2−/−St8sia4−/− mice at P0.5, we did not observe genotype-specific proteinuria at later time points, indicating that loss of polySia did not affect the mature glomerular filtration barrier. By quantifying glomerular cell numbers, in mature kidneys (P15.5), we observed significantly reduced total glomerular cell counts in St8sia2−/− St8sia4−/− animals as compared to control mice (Fig. 2D). Staining of the three glomerular cell types (Fig. 2E) with cell type-specific nuclear markers revealed unaltered numbers of GATA3-positive mesangial cells (Fig. 2F, I) and of WT1-positive podocytes (Fig. 2G, J). In contrast, ERG-positive glomerular endothelial cells were reduced in St8sia2−/− St8sia4−/− mice by about 30% (Fig. 2H, K). As mentioned, on renal sections of Ncam−/− mice at P15.5, we observed reduced glomerular tuft areas (Fig. 2B), but overtly normal kidney function (Fig. 2C). Numbers of total glomerular cells (Fig. 2D) and glomerular endothelial cells were reduced in Ncam−/− mice compared to control mice (Fig. 2L-N). Thus, Ncam−/− mice resembled the phenotype of the St8sia2−/− St8sia4−/− mouse model. These results point towards a role of polySia presentation on NCAM for nephron development, and specifically for the formation of the glomerular microvasculature.

Fig. 2

Analysis of polySia-deficient murine kidney. (A) H&E staining of paraffin-embedded kidney sections from wildtype and St8sia2−/− St8sia4−/− mice (P15.5). Glomeruli in the renal cortex are framed. (B) Glomerular tuft areas measured in H&E stained kidney sections (P15.5) of n = 5 wildtype and mutant mice, respectively. Per section, all visible glomerular tufts were measured. Means ± SD and individual data points are depicted. Non-parametrical Kruskal–Wallis test (p = 0.0032) and Dunn’s multiple comparison indicated significant group differences (* p < 0.05). (C) Protein/creatinine ratios measured in urine from wildtype, St8sia2−/− St8sia4−/− and Ncam−/− mice collected at different postnatal time points (P0.5, 7.5 and 15.5). Means ± SD and individual data points of n = 4–16 individuals are depicted. Nonparametric Mann–Whitney test indicated significant differences between wildtype and St8sia2−/− St8sia4−/− mice at P0.5, during the maturation of the blood filtration barrier (* p < 0.05, highlighted). For clarity reasons, significant differences during kidney maturation, between P0.5-P7.5 and P0.5-P15.5, for all three indicated genotypes are not indicated. No significant differences were detected between the genotypes for the time points P7.5 and P15.5. (D) Total cell numbers per glomerulus were assessed on kidney sections from wildtype, St8sia2−/− St8sia4−/− and Ncam−/− mice (P15.5) by counting hematoxylin stained nuclei and WT-1, ERG, or GATA3 immunoreactive cells (see F, G and H) or hematoxylin-stained cell nuclei in H&E stained renal sections. N = 13 wildtype, N = 13 St8sia2−/− St8sia4−/− animals and N = 11 Ncam−/− animals and per section 30 glomeruli in comparable areas of the tissues were analysed. Means ± SD and individual data points are depicted. Statistical significance was tested with a nonparametric Mann–Whitney-U test (* p < 0.05; **** p < 0.0001). (E) Scheme of a mature glomerulus composed of three different cell types in the glomerular tuft: podocytes (blue), endothelial cells (green) and mesangial cells (orange). (F–H) Immunohistological staining of nuclear localized glomerular cell type markers on paraffin-embedded kidney sections (wildtype and St8sia2−/− St8sia4−/−, P15.5): GATA3 staining for mesangial cells (F), WT1 staining for podocytes (G) and ERG staining for endothelial cells (H). Evaluation of cell numbers of (I) GATA3+ cells, (J) WT1+ cells and (K) ERG+ cells (** p < 0.01, nonparametric Mann–Whitney test) from immunohistological staining on paraffin-embedded kidney sections (P15.5) of wildtype and St8sia2−/− St8sia4−/− mice. (L-N) Analysis of renal sections from Ncam−/− and wildtype mice at P15.5 regarding the evaluation of cell numbers of (L) GATA3+ cells, (M) WT1+ cells and (N) ERG+ cells (* p < 0.05, nonparametric Mann–Whitney test). n = 4–5 individuals were analysed per genotype and per section 30 glomeruli were selected for counting in equal areas of the section. Data is displayed as averages per mouse with means and standard deviation for the group

PolySia interacts with VEGF-A188 in vitro

The reduced endothelial cell number in glomeruli of polySia-deficient mice resembles the compromised development of the glomerular microvasculature in mouse models with impaired VEGF-A expression [5, 43]. However, immunoblotting revealed that all three pro-angiogenic isoforms VEGF-A120, -A164 and -A188 (Fig. 3A), were present at similar levels in kidney homogenates from wildtype and St8sia2−/− St8sia4−/− mice (Fig. 3B). This observation was corroborated by qPCR analysis, showing similar mRNA levels of all three isoforms as well as the corresponding receptor VEGFR2 (Supplemental Fig. S3). Since VEGF-A and VEGFR2 at the gene expression and protein level was not altered by polySia-deficiency, we hypothesized that the interaction of polySia with VEGF-A itself modulates VEGF-A signalling and regulates glomerular endothelial cell numbers. To investigate the binding of VEGF-A to polySia in vitro, we incubated all three VEGF-A isoforms with purified polySia and analysed the formation of protein-polysaccharide complexes by horizontal native PAGE (Fig. 3C, D). Specifically, we used size-fractionated polySia with a degree of polymerization (DP) of 24–30 [33], which reflects the most abundant DP of polySia on perinatal brain NCAM [44, 45], and polySia with an average DP of 50 (avDP50). In the given experimental setup, binding to the polyanionic polysaccharide polySia is indicated by an increased electrophoretic mobility of the bound protein towards the anode ( +). In Fig. 3C and D, this is exemplified for the known polySia-histone interaction [46], which served as a positive control. Bovine serum albumin (BSA) was used as a non-binding control protein. Notably, VEGF-188 showed an electrophoretic mobility similar to histones. While VEGF-188 alone migrated towards the cathode, it showed increased mobility towards the anode when pre-incubated with polySia of avDP50 or DP24-30 (Fig. 3C), indicating the formation of a polySia-VEGF-A188 complex. For VEGF-A120 and -A164, however, we observed no significant change in electrophoretic mobility upon pre-incubation with polySia (Fig. 3D). The suitability of our horizontal native PAGE approach for monitoring protein–polysaccharide interactions [25], was further confirmed with heparan sulfate (HS), a polysaccharide that is known to sequester VEGF-A188 and -A164, but not VEGF–A120, to the cell surface [14]. In the presence of HS, VEGF-A188 showed an increased mobility to the anode (Supplemental Fig. S4A), which was seen to a lesser extend for VEGF–A164, but not for VEGF-A120 (Supplemental Fig. S4B). Importantly, the electrophoretic mobility of VEGF-A isoforms was not affected by the addition of mono-Sia, α2,8-linked di-Sia or oligo-Sia with DP4 (Supplemental Fig. S4C, D). When we analysed VEGF-A binding in a photometric ELISA with immobilized bacterial polySia of avDP50 as shown schematically in Supplemental Fig. S5, we observed binding for VEGF-A188, but not for VEGF-A164 or -A120 (Fig. 3E). This result confirmed an isoform-specific binding and indicates that polySia-VEGF-A interaction can be attributed to the VEGF-A188 specific 24 amino acids that are encoded by exon 6 (Fig. 3A). A comparative analysis of VEGF-A188 binding to polySia of DP > 40 and HS was performed by microscale thermophoresis (MST) (Fig. 3F). This method measures protein-glycan interactions by tracking the molecule movement within a temperature gradient. The dissociation constants (KD) calculated from the binding curves shown in Fig. 3F, revealed a 20-fold lower affinity for the polySia-VEGF-A188 binding (KD 66.1 ± 3.17 µM) compared to the HS-VEGF-A188 binding (KD 3.38 ± 0.78 µM). This newly identified interaction between polySia and VEGF-A188 provides a potential mechanism how this glycan promotes VEGF-A signalling and glomerular vasculature development.

Fig. 3

VEGF-A isoform expression in murine kidney, interaction with polySia in vitro and spatial relationship between VEGFA and polySia during nephron development (A) Scheme of the exon structure and the three major murine pro-angiogenic VEGF-A isoforms 120, 164 and 188. VEGFR and heparan sulfate (HS) binding sites are indicated. 24 amino acid sequence of exon 6, which is only present in VEGF-A188 is displayed and basic residues are underlined. (B) Western blot analysis of wildtype (WT) and St8sia2−/− St8sia4−/− (KO) kidney homogenates prepared at different postnatal time points. Protein concentrations of the samples were determined and 25 µg were loaded per lane. Staining with a VEGF-A antibody results in multiple bands. Actin staining serves as loading control. (C, D) Binding analysis of (C) VEGF-A188 and (D) VEGF-A120 and -A164 with polySia of different chain length (average DP50 (avDP50) and DP24-30) in horizontal native PAGE. Bovine serum albumin (BSA) is shown as a negative control and histone as a known polySia interaction partner is shown as a positive control. Complex formation of histone with both polySia pools increased the negative charge of the protein reflected by a reversion of the migration direction in the native PAGE (pH = 8.1). (C) Complex formation of VEGF-A188 (pI = 9.25) with avDP50 and DP24-30 reverted the migration direction towards the anode ( +) in horizontal native PAGE (pH = 7.4), as observed for histone. (D) In contrast, the mobility of BSA (pI ~ 5) and the two VEGF isoforms -A120 (pI = 6.48) and -A164 (pI = 7.93) was not grossly affected by either long chain (avDP50) or shorter chain polySia (DP24-30). Both polySia pools did not increase the covered migration distance of the VEGF-A isoforms to the anode, indicating a lack of interaction with polySia. Native PAGE conditions allowing optimal migration in PAGE were used for all three proteins. ( +) anode, (-) cathode. (E) ELISA of VEGF-A isoforms binding to polySia. PolySia of avDP50, derived from E. coli K1 capsule polysaccharide, was immobilized on a plate. Mean and standard deviation calculated from three individual experiments are shown with horizontal axis in log scale. The dotted lines show a non-linear fit to the mean values of the individual isoforms. Binding is observed only for the VEGF-A isoform 188. (F) Protein-glycan interaction analysis by microscale thermophoresis. Binding curves illustrate the interaction of fluorescently labelled VEGF-A188 with polySia (DP ≥ 40) and heparan sulfate (HS) as positive control. Normalized thermophoresis fluorescence averaged from 3 to 5 independent experiments is plotted against ligand concentration and KD values were calculated based on the Hill equation (KD(polySia) = 66.1 ± 3.17 µM, KD(HS) = 3.38 ± 0.77 µM). The error bars report the standard deviation. (G) Illustration of endothelial (precursor) cells migration into the S-shaped body of a developing nephron. Podocyte precursors (blue) secrete VEGF-A (red) and thereby attract VEGFR-2 (dark purple) expressing endothelial cells (green) into the vascular cleft of the S-shaped body. (H)Vegfa188 base scope assay was performed on paraffin-embedded kidney sections from newborn wildtype mice. Vegfa188 mRNA is expressed in developing (black arrowheads) and mature (green arrowheads) glomeruli in the renal cortex. Micrographs with higher magnification clearly show intense red staining in (I) developing nephrons (black arrowheads) and (J) mature glomeruli, in which the outer cell layer (podocytes) is intensively stained. (K) Immunohistochemical staining for polySia on renal section (wildtype P0.5) with 735 antibody. Intense polySia staining is visible in the centre of an S-shaped body (circled with dashed line). (L) Immunofluorescence co-staining for polySia and VEGF-A on murine kidney section (wildtype P0.5). PolySia (magenta) and VEGF-A (yellow) staining is visible in the developing glomerulus (S-shaped body) and overlapping signals are shown in white in the merged image. Cell nuclei are stained with DAPI (blue)

VEGF-A188 and polySia are concurrently expressed in the developing nephron

During nephron development, presumptive podocytes secrete VEGF-A to induce endothelial precursor cell migration into the vascular cleft of the developing nephron in the S-shaped body (Fig. 3G) to later form the glomerular tuft [47]. A crucial prerequisite for a possible interaction between VEGF-A188 and polySia in vivo is the spatiotemporal colocalization of the two molecules. Using the BaseScope in situ hybridization approach (Supplemental Fig. S6), we observed VEGF-A188 expression in both, comma- and S-shaped bodies of the developing nephron (Fig. 3H, I, black arrowheads) and additionally in mature glomeruli (Fig. 3H, J, green arrowheads). Immunohistological staining revealed the presence of polySia on cells of the S-shaped body as well as on interstitial cells (IC) surrounding the developing nephron (Fig. 3K). Co-staining with a pan-VEGF-A antibody and the anti-polySia antibody 735 confirmed the co-localization of both factors in developing nephrons of newborn wildtype kidney (Fig. 3L).

Identification of polysialyltransferase-expressing cell types during kidney development

Publicly available neonatal kidney single-cell RNA sequencing data from C57BL/6N mice [37] was used to evaluate the cell types with putative polySia presence, by the analysis of St8sia2 and St8sia4 gene expression at P0.5 (Fig. 4A, Supplemental Fig. S7A, B). Expression of the two polysialyltransferase genes was detected in five selected cell types with a score above a threshold of > 5% of cells in the respective group. In line with polySia-positive cells of the S-shaped body detected in immunostained renal sections (Fig. 1C and E, Fig. 3K and L), approximately 10% of the nephron progenitor 1 (NP1) cluster displayed the expression of St8sia2. The NP clusters are built by Six2-positive nephron progenitor cells found in the cap mesenchyme and in all cells of the nascent nephron [48]. Immunofluorescent costaining of polySia with NCAM, which is found to be expressed in all stages of nephron development [49], we observed co-localization of polySia and NCAM on the surface of the cells in the cap mesenchyme as well as in comma- and S-shaped bodies of renal sections from P0.5 wildtype mice (Supplemental Fig. 7C). In contrast to St8sia2, St8sia4 was present in about 10% of endothelial cells (EC) and blood cells (BC). Corresponding to the known heterogeneity of renal interstitial cells (IC) [50, 51] four IC clusters were identified, with St8sia2 expression in about 13% of cells in the clusters IC1 and IC2 (Fig. 4A). Low levels of St8sia2 expression were assigned to a subset of about 10% of cells of the loop of Henle in mature nephrons. Analysis of the respective scRNAseq data set from embryonic day 15 (E15) [37] revealed a similar gene expression pattern of the two polysialyltransferases St8sia2 and St8sia4 as seen at P0.5 (Supplemental Fig. S7D, E). In line with this, polySia was detected on renal sections from wildtype mice at E14.5, showing co-localization with NCAM on cells of S-shaped bodies (Supplemental Fig. S7F). Altogether, a highly cell type-specific expression of St8sia2 and St8sia4 was detected in the developing kidney of wildtype mice.

Fig. 4

PolySia expression in distinct renal cell types and contribution of the polysialyltransferases ST2SIA2 and ST8SIA4 during postnatal renal development. (A) Evaluation of polysialyltransferase gene expression in different renal cell types (Leiden clusters) in P0 wildtype mice. Single cell RNAseq data (Accession no. GSM4648414) was obtained from Naganuma et al. (2021). LOH: loop of Henle, IC: interstitial cell, EC: endothelial cell, NP: nephron progenitor (describing cells of the cap mesenchyme and the nascent nephron), PT: proximal tubule, UB: ureteric bud tip, CD: collecting duct, POD: podocyte, BC: blood cell. (B) Normalized histograms of spectral flow cytometry analysis from kidney single cell suspensions of newborn wildtype and polySia-deficient (St8sia2−/− St8sia4−/−) mice. Cells were stained for polySia with the 735 antibody and renal cell type markers for interstitial cells (PDGFRβ), (C) endothelial cells (CD31) and (D) immune cells (CD45). (E) Immunofluorescence co-staining of NCAM (yellow), polySia (magenta) and the endothelial cell marker ERG (cyan) on renal section of newborn wildtype mice. Partial colocalisation is observed in cells of an S-Shaped body. Overlapping signals are coloured white in the merged image. The white arrowhead indicates a polySia-positive endothelial cell. (F–H) Spectral flow cytometry analyses of renal single cell suspensions from newborn wildtype and polysialyltransferase knockout mouse strains St8sia2−/−, St8sia4−/− and St8sia2−/− St8sia4−/−. Cell suspensions were stained for polySia with inactive endosialidase (iEndo) and for interstitial cells (PDGFRβ, F), endothelial cells (CD31, G) and immune cells (CD45, H) to discriminate between different cell types. The calculated median fluorescence intensities (MFI) are shown with each data point representing one biological replicate (n = 4 for St8sia4−/−, n = 3 for others). Means and standard deviations are depicted. Non-parametrical Kruskal–Wallis test indicated significant differences (PDGFRβ, CD31: p < 0.0001, CD45: p = 0.0001). Uncorrected Dunn’s post hoc tests were applied and significant group differences are indicated (* p < 0.05, ** p < 0.01). Representative histograms for the results depicted in B-D are shown in Supplemental Figure S8G-I. (I) PolySia (735) staining on renal sections of newborn St8sia2−/− and (J)St8sia4−/− mice. Cortical areas of the tissue are depicted to appreciate polySia staining in the nephrogenic zone. Residual polySia signal in St8sia2−/− kidney on interstitial cells is marked with arrowheads

PolySia is expressed on endothelial precursor cells in spatial proximity to developing nephrons

Polysialylation of the different renal cell types was further assessed by spectral flow cytometry from whole kidney cell isolation, in newborn wildtype mice. Samples from St8sia2−/− St8sia4−/− mice were also analysed (Fig. 4B-D, gating strategy in Supplemental Fig. S8A-E). Platelet Derived Growth Factor Receptor beta (PDGFRβ+) was used as a cell surface marker for the heterogeneous group of renal interstitial cells including several types of interstitial fibroblasts, vascular smooth muscle cells, glomerular mesangial cells, parietal epithelial cells and pericytes. In line with strong polySia signals detected in the interstitium of renal sections (Fig. 1C and 3K), we observed a strong polySia signal on all cells of the PDGFRβ+ population (Fig. 4B). Endothelial cells (CD31+) were also identified as polySia-positive (Fig. 4C), but with a lower signal intensity compared to the PDGFRβ+ cell population. Finally, a subset of cells that were positive for the pan-leukocyte marker CD45 displayed polySia (Fig. 4D). PolySia signals were lost in St8sia2−/− St8sia4−/− mice (Fig. 4B-D). The identification of polySia-positive endothelial cells led to the hypothesis that polysialylation fosters the VEGF-A guided migration of endothelial precursor cells into the vascular cleft of developing nephrons and/or promotes differentiation processes that are essential for the formation of the glomerular microvasculature. Using confocal immunofluorescence microscopy, we observed polySia on ERG-positive endothelial cells that are in close proximity to developing nephrons, as shown exemplarily for an S-shaped body at P0.5 (Fig. 4E) and at E14.5, exemplarily shown for a comma-shaped body and developing renal epithelium (Supplemental Fig. S8F). In line with this, the polysialyltransferase ST8SIA4 is expressed in endothelial cells in both developmental stages (Fig. 4A, Supplemental Fig. S7E). Moreover, we detected polySia in the vascular cleft of S-shaped bodies (Fig. 4E). The assignment to a specific cell type(s) however, is challenging due to the close proximity of polySia-positive cells of the developing nephron and infiltrating endothelial precursor cells. Together, our combined data set obtained by flow cytometry, histological stainings and the analysis of scRNAseq data demonstrated polySia in interstitial cells, as well as subtypes of nephron progenitors and endothelial cells in the developing kidney and illustrate how cell-type specific expression of St8sia2 and St8sia4 determine the polySia pattern in the developing kidney.

Both polysialyltransferases, ST8SIA2 and ST8SIA4, contribute to polysialylation during nephron development

Based on the finding that the two polysialyltransferases showed only partially overlapping expression profiles (Fig. 4A), we assessed the individual contribution of ST8SIA2 and ST8SIA4 to polysialylation in selected renal cell types by spectral flow cytometry. As expected, polysialylation was abolished in all analysed renal cell types of newborn St8sia2−/− St8sia4−/− mice (Fig. 4F-H, representative histograms are shown in Supplemental Fig. S8G-I). In accordance with the scRNAseq data, the major portion of polySia in PDGFRβ+ renal interstitial cells was synthesized by ST8SIA2 (Fig. 4F). In comparison to interstitial cells, we observed considerably lower polysialylation levels in CD31+ endothelial cells in wildtype kidney, but apparently, both polysialyltransferases were able to contribute to endothelial cell polysialylation (Fig. 4G). However, compensatory mechanisms in mice lacking only one polysialyltransferase cannot be excluded. PolySia levels in CD45+ leukocytes of St8sia4−/− kidneys were reduced to the level in polySia-negative mice, validating ST8SIA4 as the enzyme responsible for polysialylation in immune cells (Fig. 4H). Complementary to the spectral flow cytometry analysis we conducted immunohistological staining for polySia on kidney sections of newborn St8sia2−/− and St8sia4−/− mice. The polySia signal in St8sia2−/− mice was strongly reduced with residual staining found on a small fraction of cells in the interstitium (Fig. 4I, arrowheads). In St8sia4−/− renal sections, the polySia signal was comparable to wildtype levels with strong signal in the cortical region especially in the interstitium (Fig. 4J).

NCAM is the major polySia carrier in murine kidney

In Ncam−/− kidney, glomerular tuft areas were significantly decreased (Fig. 2B). We observed a reduced number of total glomerular cells (Fig. 2D), which is attributable to a reduced number of ERG-positive glomerular endothelial cells (Fig. 2L-N), as seen in St8sia2−/− St8sia4−/− kidneys (Fig. 2B, D, I-K). This indicates that NCAM might be the main polySia carrier in the developing kidney. Likewise, at P0.5 the polySia signal was almost completely lost in Ncam−/− kidneys. Careful inspection of immunohistological staining with DAB, however, revealed a few polySia-positive cells spread across the tissue, mainly localized in the tubulointerstitial space in groups of one or two positive cells (Fig. 5A). Spectral flow cytometry analysis of wildtype and Ncam−/− kidney cell suspensions at age P3 identified subsets of polySia-positive CD45-positive leukocytes in both genotypes (Fig. 5B), in line with previous reports demonstrating polySia on immune cells as a posttranslational modification of C–C motif chemokine receptor 7 (CCR7) [27] and Neuropilin 2 [52]. In contrast, polySia staining on PDGFRβ+ interstitial cells (Fig. 5C) was abolished in Ncam−/− kidneys, indicating that NCAM is the major polySia carrier on this cell population in newborn kidney. When we analysed CD31+- polySia+ double-positive renal cells, we observed that a subfraction of the CD31+ endothelial cells remained polySia-positive in Ncam−/− kidneys (Fig. 5D). A detailed analysis of CD31+ endothelial cells in the kidney of P0.5 wildtype mice confirmed the presence of two subpopulations of polySia+ CD31+ double-positive cells: an NCAM-positive and a smaller, NCAM-negative subpopulation (Fig. 5E, left panel). In line with this, most but not all of the CD31+ endothelial cells in Ncam−/− kidney were polySia-negative (Fig. 5E, right panel). In kidneys of P0.5 wildtype mice, less than 5% of the CD31+ polySia+ endothelial cells were NCAM-negative, and a similar percentage of CD31+ polySia+ endothelial cells was observed in Ncam−/− kidneys (Fig. 5F). About 15% of CD31+ polySia+ endothelial cells in wildtype mice were also NCAM-positive. These results, together with the phenotypic similarities between kidneys of Ncam−/− and St8sia2−/− St8sia4−/− mice, indicate that NCAM is the major but not the only polySia carrier in endothelial cells. Of the known polySia carriers neuropilin-2 (Nrp2) [52], E-selectin ligand-1/Golgi Glycoprotein 1 (Glg1) [53] and SynCAM 1/Cell adhesion molecule 1 (Cadm1) [54], only the former two are express