Temporally and Spatially Regulated Collagen XVIII Isoforms Impact Ureteric Patterning Through Their TSP1-like Domain

Collagen XVIII (ColXVIII) is a component of the extracellular matrix implicated in embryogenesis and control of homeostasis. We provide evidence that ColXVIII has a specific role in kidney ontogenesis by regulating the interaction between mesenchymal and epithelial tissues as observed in analyses of total and isoform-specific knockout embryos, mice, and ex vivo organ primordia. ColXVIII deficiency, both temporally and spatially, impacts the 3D pattern of ureteric tree branching morphogenesis via its specific isoforms. Proper development of ureteric tree depends on a tight control of the nephron progenitor cells (NPCs). ColXVIII-deficient NPCs are leaving the NPC pool faster than in controls. Moreover, the data suggests that ColXVIII mediates the kidney epithelial tree patterning via its N-terminal domains, and especially the Thrombospondin-1-like domain, and that this morphogenetic effect involves ureteric epithelial integrins. Altogether, the results propose a significant role for ColXVIII in a complex signalling network regulating renal progenitors and kidney development.


INTRODUCTION 28
The extracellular matrix (ECM) plays a critical role in embryonic development as well as in maintaining organ 29 homeostasis and regulating stem cell fate (Andrew & Ewald, 2010; Lu, Takai, Weaver, & Werb, 2011; Lu, Figure 1. Schematic picture of the structure of ColXVIII and mouse lines used in the study. All three isoforms have a common collagenous domain, C-terminal endostatin domain (ES) and N-terminal TSP1-like domain (TSP1). In addition, the medium and long isoforms have MUCL-C18 domain in the N-terminus, but only the longest isoform has also the N-terminal Frizzled-domain (Fz). The total ColXVIII KO mice lack all three isoforms, whereas the P1 mice lack only the short isoform and the P2 mice lack only the two longer isoforms. The picture also shows the N-terminal fragment and the TSP1-like fragment used in the kidney organ culture studies. HS = heparan sulphate side chain. the endostatin primers, ColXVIII expression was detected both in the MM and the UB of WT and P2 kidneys 75 ( Fig. 2a,b), while this expression was lacking in the KO and P1 kidneys (Fig. 2a,b). This expression pattern 76 was confirmed with the short isoform-specific primers (Fig. 2c,d). The results of the qPCR analyses indicate 77 that only the short ColXVIII isoform is expressed in both the MM and UB and that there is no expression of 78 the two longer forms during this early phase of development. 79 To visualise the expression of the ColXVIII in the different knockout mice, the whole-mount in situ 80 hybridisation of WT and ColXVIII mutant kidneys was carried out that revealed a well-regulated expression 81 of ColXVIII isoforms. At E12.5, the ColXVIII expression was detected in the epithelium of the ureter 82 extending through the kidney in both the WT and the P2 kidneys expressing the short form, while in the P1 83 kidneys expressing the medium and long forms, the expression was only detectable in the ureteric stalk area 84 ( Fig. 2figure supplement 1a). At E13.5 (Fig. 2figure supplement 1b) and E14.5 (Fig. 2efigure  85 supplement 1c) the P1 kidneys expressing only the two longer isoforms had a less positive signal in the cortex 86 compared with the WT and P2 kidneys expressing the short isoform. Furthermore, at E16.5, the P1 kidneys 87 expressing only the two longer isoforms were devoid of a ColXVIII expression on the surface of the organ, 88 similarly to the KO kidneys, whereas WT and P2 kidneys expressing the short isoform showed a strong 89 ColXVIII expression in the ureter tips (Fig. 2f). 7 91 Figure 2. Only the short ColXVIII isoform was expressed during the early stages of kidney development and in the developing structures. The qPCR analyses of separated (a) metanephric mesenchyme (MM) and (b) ureteric bud (UB) at E11.5 using primers from the common endostatin domain of ColXVIII were conducted. The expression of only the short form in the (c) MM and the (d) UB at E11.5 was confirmed with qPCR using specific primer for the short ColXVIII form in KO, P1, and P2; n(WT)=20, n(KO)=18, n(P1)=14, and n(P2)=22 pooled samples. Whole-mount in situ hybridisation of (e) E14.5 (bar: 500 µm) and (f) E16.5 (bar: 500 µm) WT, P1, and P2 kidneys with an antisense ColXVIII RNA probe. Arrows indicate the ColXVIIIpositive structures. (g) Section in situ hybridisations of E16.5 WT, P1, and P2 kidneys for ColXVIII expression (bar: 500 µm) and magnifications of the E16.5 section in situ hybridisations (h) (bar: 100 µm). In (g) black arrowheads indicate ColXVIII-positive glomeruli, black arrows indicate collecting ducts, red arrows indicate branching tubules, and yellow stars indicate ColXVIII-positive comma-and S-shape bodies. In (h), red arrows indicate branching tubules, and dashed areas indicate glomeruli (G) and S-shape bodies (S). In graphs, columns indicate mean±s.d. tip BMs lining the NPC population, we went on to later developmental stages and stained the E15.5 kidneys 138 as whole-mount with the NPC biomarker Six2 and the tubular epithelial marker TROMA-1 (Keratin-8) 139 (Djudjaj et al., 2016;Skinnider et al., 2005). From multiphoton microscope imaged kidneys, each ureteric tip-140 cap domain, indicating individual nephrogenic niches, were separately analysed (Fig. 5f, Movie 3). This 141 revealed that in the KO embryonic kidneys and those that lack specifically the short (P1) ColXVIII isoform, 142 or alternatively the two longer ColXVIII isoforms (P2) all had less Six2+ cells in foci of nephrogenic niches 143 when compared with WT embryos (27%, 36% and 34% less, respectively, Fig. 5g). 144 Since we found fewer Six2+ NPCs in all ColXVIII mutant mice at E15.5, we studied also the earlier 145 developmental stage (E13.5) aiming to define the initial signs of deficiencies in null ColXVIII embryonic 146 kidneys (Fig. 5h). Indeed, the NPC amount as judged by Six2+ NPCs had decreased by 39% in the E13.5 KO 147 kidneys when compared with the WTs (Fig. 5i). The results depicted that the Six2+ NPCs are decreased in the 148 ColXVIII null embryonic kidneys at E13.5 onwards, as was the case for the isoform specific ColXVIII mutants 149 at E15.5. 150 Figure 5. ColXVIII regulated NPC population during kidney development. (a-d) Examples of the aortagonad-mesonephros (AGM) cultures of E10.5 Rosa YFP ; Wnt4 Cre -positive WT and KO embryos. In the beginning, the Wnt4+ cells (green, indicated by arrow heads) were seen in the mesonephros/gonad area (outlined, G) and in the limb area in the WT (a) and KO (c). Neither the mesenchyme nor the ureter bud of the kidney had yet formed. The kidneys (K) formed during the culture and became clearly positive for Wnt4 both in the WT (b) and KO (d). Bars: 200 µm. (e) Analyses of the AGM cultures showed that the appearance of the Wnt4+ cells in the kidney was delayed in the ColXVIII-deficient embryos (n=16) when compared with WT embryos (n=15). Positivity was calculated from the moment when the first positive cell was detected in the developing kidney. P-value for the Mann-Whitney U test was 0.0844 (not significant). Multiphoton microscopy maximum intensity projection images of the cortex of Six2 (green)-TROMA-1 (red) whole-mount stained E15.5 (f) and E13.5 (h) kidneys. A yellow circle indicates a typical nephrogenic niche, all of which were separately calculated. Bar: 100 µm. The number of Six2+ NPCs in nephrogenic niches was decreased in the E15.5 (g) ColXVIII mutant mice lacking all forms (KO), the short form (P1), and the medium and long forms (P2), and in the E13.5 (i) KO kidneys. In the E15.5 analyses: n(WT)=8 (87 tips), n(KO)=6 (70 tips), n(P1)=5 (57 tips), and n(P2)=2 (25 tips), and in the E13.5 n(WT)=6 (28 tips) and n(KO)=9 (58 tips). In graphs, the lines in (e) indicate mean±s.d., and in (g) and (i), the boxes show 1 st and 3 rd quartiles with mean and whiskers indicate min and max values. ***P<0.0001 (Mann-Whitney U -test). (a) In the WT E13.5 kidneys, the Wnt4+ cells (YFP-positive, arrows) were detected around the branching tubules (B) in the NPC population and in the renal precursors (renal vesicles (R) and comma-(C) and S-shape (S) bodies) closely localised with ColXVIII expression which was detected in the BMs of the branching tubules and the renal precursors. (b) Some of the Wnt4+ cells were also positive for Six2 indicating that these cells present NPCs. Some of the Six2 and Wnt4 positive cells are indicated with arrows. (c) A magnification of one branch surrounded by Wnt4+ cells (arrows) and Wnt4+ renal vesicles. (d) A magnification of the same branch indicating that some of the Wnt4+ cells express also Six2 and represent NPCs (arrows) in the cap mesenchyme (CM) area. (e) E13.5 ColXVIII KO kidneys stained against YFP and ColXVIII showed that the Wnt4+ cells (arrows) localised around the branching tubules and renal precursors similarly with the WT kidneys. (f) In the KO kidneys, some of the Wnt4+ cells expressed also Six2 (arrows) in the CM as seen in the WTs. (g) A magnification of a CM area of the E13.5 KO kidneys, where Wnt4+ cells were detected in the renal vesicles (R), comma-shape body (C) and in the CM area around the branches (B). (h) A magnification of the same area than in (g) stained against Six2 and YFP indicated that some of the Wnt4+ cells express also Six2 (arrows) in the CM area. Bar(a, b, e, and f): 100 μm and bar(c, d, g, and h): 50 μm.
The OPT studies revealed in all ColXVIII mutant mouse lines in comparison to control a significant reduction 161 in the number of terminal branch points, which reflects the number of the ureteric tips ( Fig. 7a,b - figure  162 supplement 1a). The reduction in the ureteric tip count was most notable in the kidneys of P1 mice that were 163 deficient of the short ColXVIII isoform. The same trend was also seen in total KO kidneys and in the P2 mice 164 lacking the two longer isoforms (Fig. 7b)  Specifically, the bifurcational branching events and ureteric tree length were reduced in all mutant mice when 169 compared with controls at E15.5 (Fig. 7c,d). The kidneys of all mutant mice also had a slightly increased 170 branching angle between the root branch and extending branch points, reflecting sparser branching of the 171 ureteric tree (Fig. 7e). 172 To analyse when the branching defect begins in the ColXVIII-deficient mice, the E11.5, E12.5 and E13.5 WT 173 and KO kidneys were stained against TROMA-1 and imaged either with confocal microscopy (E11.5 and 174 E12.5) or OPT (E13.5) (Fig. 8a,c,e). The results showed that the branching of the tubular tree from E11.5 to 175 E13.5 was comparable between the WT and KO kidneys (Fig. 8b,d,f). 176  per kidney) were noted (Fig. 7f). Since only a few of the glomeruli are mature at this time point, the number 183 of mature and developing glomeruli including renal vesicles, comma-and S-shape bodies, were next calculated 184 from E16.5 serial sectioned whole kidneys. In this case, the number of glomeruli and glomerular precursors 185 were decreased in all mutant mice, but the difference was statistically significant only for KO and P1 compared 186 with WT (Fig. 7g). No change was detected in the S phase (b) between the WT and KO kidneys, but the Six2 cells were significantly more in G2-M phase in the KOs at E13.5 (c). At E15.5 the number of Six2+ cells were significantly more in the G1 phase (d) and significantly less in the G2-M phase (f) in the KO kidneys when compared with the WT but the difference in the S phase (e) was not significant. Similarly, at E17.5, the number of Six2+ cells was significantly higher in the KO kidneys in the G1 phase (g) and significantly lower in the G2-M phase (i) in comparison with the WT kidneys. No difference was detected in the S phase (h). N(WT E13.5)=4 , n(KO E13.5)=7 litters analysed separately. N(WT E15.5)=6, n(KO E15.5)=11, n(WT E17.5)=5, n(KO E17.5)=6 independent acquisitions in flow cytometry analysis per sample collected from three different litters. In graphs, the boxes show 1 st and 3 rd quartiles with mean, and the whiskers indicate min and max values. ***P<0.0001 (Mann-Whitney U -test).
positive NPC cells using immunostaining against Ki67, Six2 and tubular marker TROMA-1. At E13.5, the KO 210 kidneys had less Ki67+ cells in the Six2+ NPC population when compared with the WT (Fig. 10a,c). The 211 proliferating Ki67+ cells localised less often to the cortex side of the nephrogenic niche (Fig. 10a,  stopped after a few branches, but the WT kidneys grew and branched robustly (Fig. 11a,b). In contrast, when 231 the kidney rudiments were cultured with 500 ng/ml or 1000 ng/ml of the N-terminal fragment, the phenotype 232 of the KO kidneys was rescued, and the branching occurred normally (Fig. 11a,b). Interestingly, the WT 233 kidneys also seemed to benefit from the fragment by forming more branches, but the effect was moderate 234 tubulogenesis and strikingly rescue the branching defect of the ColXVIII-deficient kidneys. 237 The TSP1-like domain is the only common N-terminal domain for all ColXVIII isoforms. Since the branching 238 analyses indicated that especially the short ColXVIII isoform is important for ureteric branching, we performed 239 the kidney organ cultures with the recombinant TSP1-like fragment (Fig. 1) to evaluate its role in the branching 240 morphogenesis. The kidney cultures were treated either with 200 ng/ml or with 500 ng/ml of the TSP1-like 241

fragment. 242
The treatment of the WT kidneys with the TSP1-like fragment did not cause any changes in the branching 243 morphogenesis when compared with the untreated kidneys (Fig. 11c,d). In contrast, the treatment of the KO 244 kidneys with 200 ng/ml of the TSP1-like fragment caused a similar rescue effect of the ureteric branching than 245 was seen with the full N-terminal fragment (Fig. 11c,d). The treatment with 500 ng/ml of the TSP1-like 246 fragment also benefitted the branching of the KO kidneys but less than the treatment with less of the TSP1-247 like fragment (Fig. 11c,d). These results suggest a novel role for the TSP1-like domain of ColXVIII as the 248 main branching promoting part of the N-terminus. 249 250 Integrin α3β1 is a potential receptor to mediate the branching promoting effect of the TSP1-like domain 251 The TSP1 protein has been shown to bind integrin α3β1 and, interestingly, the kidney phenotypes of Itga3-252 null mice, as well as Itgb1 knockout in UB, resemble the one described here for the ColXVIII-deficient mice 253 detected in close proximity ( Fig. 12 and 13). Moreover, lack of ColXVIII did not affect the expression patterns 257 of the α3 and β1 integrin subunits. 258  treated with integrin α3 or β1 function blocking antibodies. Inhibiting integrin α3 or β1 functions with the 265 respective antibodies strongly inhibited the tubular branching of the cultured WT kidneys, which was seen as 266 a reduction of the number of tips when compared with IgG-treated controls (Fig. 14a-d). In the KO cultures, 267 the branching appeared further reduced when the integrin antibodies were added (Fig. 14a-d). The rescue of 268 the branching defect of the KO kidneys with the N-terminal fragment (Fig. 14a-d) was inhibited by the addition 269 of the integrin α3 or β1 blocking antibodies, and neither was there a rescue of the integrin antibody-induced 270 branching defect of the WT kidneys (Fig. 14a-d). Thus, the results suggest that integrin α3β1 could mediate 271 the function of the N-terminal fragment on the developing ureteric tree. 272 To study the TSP1-like domain of ColXVIII in more detail, we performed a sequence alignment and protein the TSP1-like domain of ColXVIII (Fig. 14f,g). deficiency leads to altered tubular branching, reduced nephron formation, and kidney hypoplasia. In addition, 290 our results indicated different expression patterns of ColXVIII isoforms in the developing kidney and identified 291 a novel role for the ColXVIII N-terminal TSP1-like domain as a driver of tubular branching which probably 292 functions through integrins (Fig. 15). 293 The NPC population is crucial for kidney formation as it produces inductive signals for the UB to invade the 294 Six2+ NPC population was decreased in the ColXVIII-deficient mice at E13.5 and E15.5, and the cell cycle 306 progression of these cells was affected in the ColXVIII mutants. We found that at E13.5 there were more fast 307 cycling Six2+ cells, representing cells committed to differentiation, which in turn might lead to the observed 308 decrease of the Six2+ NPC population. At E15.5 and E17.5 the KO kidneys had more slow cycling Six2+ cells 309 than the WTs suggesting that the reduced NPC population in the KO kidneys tried to retain this population to 310 ensure the continuity of the development. We detected a similar number of the mature glomeruli at E15.5 311 between the WT and the KO that could be explained by the fastened NPC differentiation at E13.5 in the KOs. 312 However, the overall decrease in the number of Six2+ NPCs, and later a slowing down in the cell cycle 313 probably leads to the detected reduction in the total number of the forming nephrons. The lack of ColXVIII 314 was shown to also delay the time when the Wnt4+ cells appeared in the developing kidney that might indicate 315 also a wider defect in the NPC induction in the ColXVIII mutants. The results also revealed that the NPC 316 population is decreased before the branching defect begins in the ColXVIII mutant fetuses suggesting that this 317 reduction is at least partly a cause for the branching defect.  The Rosa26 LacZ , Rosa26 YFP , and Wnt4 EGFPCre mice were in the C57Bl/NCrl background. The vaginal plug 373 appearance of the inbred mice was used as a criterion for mating and the right age of the embryos was 374 confirmed with the somite counting of the E10.5-E11.5 and the limb bud staging in the older embryos. 375

In situ hybridisation 376
Section and whole-mount in situ hybridisations were performed as described earlier (Lin et  ColXVIII antibodies and was found to be highly specific. 395

Immunofluorescence and β-galactosidase staining 396
For immunostaining, paraffin and cryosections were cut to a 5-or 7-µm thickness. Rabbit anti-ColXVIII 397 antibody (1 μg/ml) was produced and purified as described above. Primary antibodies against TROMA-1 398 (DSHB, 1:500), Six-2 (11562-1-AP; Proteintech, 1:200), podocin (P0372; Sigma, 1:300), Ki67 (Ab15580; 399 Abcam, 0.25-0.5 μg/ml), integrin subunits α3 (AB1920; Millipore, 1:1000), and β1 (553715; BD Bioscience, To perform fragment cultures with E11.5 kidneys, a nucleopore filter with the kidney primordium was placed 522 on top in a trowel grid, cultured for 96 h in medium (DMEM, 10% FBS, 1% streptomycin, penicillin), and 523 incubated in a humidified atmosphere at 37°C with 5% CO2. To titrate the ColXVIII fragment concentration 524 for the experiment, 25, 50, 100, 500, and 1000 ng/ml concentrations of the fragment were used in culture 525 medium and changed every 48 h. A clear effect in the growth was detected at 500 ng/ml, and it was used 526 together with 1000 ng/ml in the final experiments. The function blocking studies were performed similarly to 527 the fragment cultures by adding either integrin α3 or β1 antibody alone or together with N-terminal fragment 528 1000 ng/ml, and for controls, adding IgG alone or with the N-terminal fragment 1000 ng/ml. After culture, the 529 kidney primordia were fixed in 4% paraformaldehyde (PFA) and processed for whole-mount-immunostaining In the WT, the ColXVIII expression was seen in the tubules (black arrows), tips (arrowheads) and forming glomerular structures (yellow stars), and the same pattern was seen in the P2 kidneys. In the P1 mice, the ColXVIII expression was mainly seen in the tubules (black arrows). No expression was seen in the KO kidneys. Bar: 100 μm.  The NB graphs are shown separately since the background of the mice was changed during the study, and the kidneys in the C57B/6NCrl line were bigger than those in the C57Bl/JOlaHsd line, but the ratio in the tip number between WT and mutants remained the same. In graphs, means±s.d. are shown. *P<0.05, **P<0.01, and ***P<0.001 (Mann-Whitney Utest).