CreER activation transiently impairs angiogenesis by slowing endothelial proliferation

Angiogenesis; retinal vascularisation; endothelial cell; CreER toxicity 11 Non-standard Abbreviations and Acronyms: Endothelial Cell (EC), Cre recombinase estrogen 12 fusion protein with ligand binding mutation (CreER) 13 14

15 Tamoxifen-inducible gene targeting in mice with estrogen receptor -dependent Cre (CreER) 16 recombinase has enormously advanced vascular biology research. However, CreER activation 17 under the control of vascular endothelial promoters is now recognized to cause off-target effects that 18 impair angiogenesis in the widely used perinatal mouse retina model. Although ubiquitously 19 expressed CreER is also used to study retinal angiogenesis, it remains unknown whether it causes 20 similar or more severe off -target effects compared to endothelial -selective CreER activation. 21 Moreover, the cellular processes disrupted by CreER -induced endothelial toxicity remain to be 22 identified. Here, we demonstrate that ubiquitous CreER activation in postnatal mice decreases body 23 growth throughout the period of retinal angiogenesis and impairs retinal angiogenesis in a tamoxifen 24 dose-dependent manner. We further show that CreER activation from both endothelial and 25 ubiquitously expressed CreER transgenes suppresses endothelial cell proliferation downstream of 26 p21/CDKNA1 upregulation. By contrast, we find that p21/CDKNA1 is not upregulated in quiescent 27 adult retinal endothelium, and that CreER -induced postnatal angiogenesis defects recover two 28 weeks after tamoxifen discontinuation. Altogether, our findings indicate that ubiquitous promoters 29 should be avoided for CreER expression when studying endothelial genes, and that short -term 30 retinal angiogenesis studies require endothelial CreER toxicity controls that may be less critical for 31 adult vascular studies. 32 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 2 -

33 Cre recombinase -mediated gene ablation offers an invaluable genetic tool to uncover 34 molecules that regulate vascular biology in the mouse, the major mammalian model system [1]. 35 Commonly, this method involves the use of Cre fused to a human estrogen receptor variant (ER); 36 this fusion protein, termed CreER, is retained in the cytoplasm until it binds 4 -hydroxy-tamoxifen, 37 thereby enabling its nuclear translocation to induce gene recombination [2]. Administering 4 -38 hydroxy-tamoxifen or its precursor tamoxifen to promote Cre -mediated recombination therefore 39 allows to control the time of deletion for genes that are flanked by loxP sites (floxed). By activating 40 CreER after birth in vascular endothelial cells, this method circumvents the lethality caused by 41 ablating essential vascular genes in utero [3]. Accordingly, this method allows to uncover molecular 42 pathways that contribute to the growth, remodeling and function of postnatal blood vessels [1,4]. For 43 example, the CreER system has been used to demonstrate the role of VEGF and notch signalling in 44 postnatal angiogenesis [5-11]. 45 Despite its usefulness for genetic investigation, a prior study demonstrated that CreER 46 activation in endothelial cells with the commonly used Cdh5-CreER [11] or Pdgfb-CreER [12] 47 transgenes can impair angiogenesis [13]. However, it has not yet been addressed whether 48 ubiquitously expressed transgenes such as Cagg-CreER or Rosa26-CreER cause similar or worse 49 CreER toxicity during angiogenesis, even though these transgenes have also been used to delete 50 genes in endothelial cells or cells in their environment (e.g., [14-16,9]). Moreover, the mechanism 51 that underlies CreER toxicity to impair angiogenesis remain unknown. Prior studies in non -52 endothelial cell types have suggested that Cre recombinase has toxic effects because it recognizes 53 endogenous genomic DNA sequences that resemble the engineered loxP sequences used to mark 54 genes for recombination [17-19]. Thus, off-target endonuclease activity directed at pseudo-loxP sites 55 has been linked to DNA breaks, chromosomal aberrations, reduced cell proliferation, increased 56 apoptosis and pro-inflammatory responses in non-endothelial cell types [20,21,4]. However, whether 57 CreER activation also perturbs endothelial cell health in this manner has not been addressed. 58 Moreover, it remains unknown whether the toxic effects of endothelial CreER activation are restricted 59 to angiogenic endothelium or similarly affect the quiescent adult endothelium. 60 Considering that the mouse perinatal retina is widely used to identify molecules that regulate 61 postnatal angiogenesis [22-24], we have used this model to investigate the emergence, mechanisms 62 and resolution of CreER toxicity in endothelial cells . In the mouse, blood vessels emerge from 63 vessels in the optic nerve head shortly after birth to form a superficial vascular plexus in the retina. 64 Led by filopodia studded tip cells, vessels grow forwards to cover the retina and fuse laterally, thereby 65 establishing a planar network that is exquisitely suited for high-resolution imaging and quantification 66 of angiogenic processes [24,22,25,3]. Prior work showed that CreER activation in endothelial cells 67 impaired both vessel extension and branching to significantly reduce vascular complexity as key 68 angiogenic parameters in the retina [13]. Here, we show that CreER activation (but not tamoxifen or 69 the vehicle used to administer tamoxifen) impairs retinal angiogenesis downstream of cell cycle 70 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 3 - stalling. Upregulation of the p21 cyclin dependent kinase inhibitor A1 (CDKNA1), which mediates 71 both the TP53-induced and TP53-independent response to DNA damage, was the first indicator of 72 CreER toxicity in endothelial cells. By contrast, CreER activation did not cause p21 upregulation or 73 reduce the density of quiescent adult retinal vasculature. Our findings inform on the confounding 74 mechanisms that might skew data interpretation when using the CreER -loxP system to assess 75 candidate angiogenesis regulators whilst also demonstrating that targeting of the adult vasculature 76 does not induce similar off-target effects. 77 78

79 Tamoxifen or its vehicle do not reduce weight gain or retinal angiogenesis in neonatal mice 80 Vegetable oils are commonly used as vehicle for tamoxifen administration but can cause 81 inflammation in adult mice when delivered via intraperitoneal injection [26]. To determine whether 82 the intraperitoneal injection of vegetable oil as a vehicle affects growth or angiogenesis in postnatal 83 mice, we delivered two doses, one on perinatal day (P) 2 and one on P4, and examined both male 84 and female C57Bl6/J wild type pups on P7 (Fig. 1A). We found that vegetable oil on its own did not 85 affect pup weight gain or retinal vascular network extension when compared to untreated pups (Fig. 86 1A,B). As tamoxifen has known vascular effects [27] and can reduce body weight independently of 87 CreER activation in adult mice [28], we also investigated the effect of a range of tamoxifen doses 88 commonly used to activate CreER for retinal gene targeting in neonatal pups. For this experiment, 89 we tested three tamoxifen doses commonly used for gene targeting in the neonatal period, 50, 100 90 or 150 µg (e.g., [14-16,29,30]). However, intraperitoneal injection of tamoxifen in vegetable oil on P2 91 and P4 did not decrease body weight or vascular extension in male or female C57Bl6/J pups on P7 92 (average body weight: 50 μg, 3.93 g; 100 μg, 3.94 g; 150 μg, 3.80 g; average vascular extension 93 relative to retinal radius: 50 μg, 0.80; 100 μg, 0.87; 150 μg, 0.86). 94 95 Ubiquitous CreER activation reduces weight gain and retinal angiogenesis in neonatal mice 96 Next, we compared tamoxifen-injected neonatal C57BL6/J mice carrying or lacking the Cagg-CreER 97 transgene [31] without floxed genes that could regulate angiogenesis to distinguish toxic effects of 98 CreER activation from the effects of gene targeting. After delivering two doses of 50, 100 or 150 µg 99 tamoxifen on P2 and P4 each, we examined pups on P7 (Fig. 1C). We found that male and female 100 P7 pups with activated CreER (CreER+) had lower body weight compared to control mice lacking 101 CreER (CreER-) across all tamoxifen doses examined (Fig. 1C). By contrast, the retina radius was 102 similar in pups with or without activated CreER at all three doses examined (Fig. 1C), which allowed 103 us to compare the effect of tamoxifen -induced CreER activation on retinal angiogenesis in the 104 absence of confounding effects of eye size. Both vascular network extension and retinal vascular 105 branch density, key angiogenic parameters, were significantly reduced in retinas of CreER+ 106 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 4 - compared to CreER- mice at all three tamoxifen doses examined (Fig. 1D,E). Although the reduction 107 in vascular branchpoints was consistent across all three tamoxifen doses ( P > 0.48), vascular 108 extension was significantly lower with 50 μg tamoxifen compared to both 100 μg (P = 0.02) and 150 109 μg (P < 0.001; no significant difference observed between the 100 and 150 μg doses). Further, we 110 found no evidence for treatment-by-sex interaction (χ² = 2.61, df = 2, P = 0.27), suggesting similar 111 effects of CreER toxicity on retinal vascularisation in both sexes. Notably, the presence of loxP sites 112 (introduced via a floxed Rosa26tdTom reporter cassette) did not ameliorate the detrimental effect of 113 CreER activation on vascular extension (Fig. S1A). 114 Together, these findings demonstrate that a ubiquitously expressed CreER transgene used 115 to target retinal angiogenesis, impairs retinal angiogenesis at the lowest dose examined and even 116 when the genome contained engineered loxP sites as intended CreER targets. Additionally, 117 ubiquitous Cagg-CreER activation impaired body growth, which was not seen with endothelial -118 selective CreER activation. 119 120 CreER activation-induced retinal vascular defects recover over time 121 In adult mice, tamoxifen is metabolized to 4-OHT, whereby both compounds have a serum half-life 122 of less than 10 hours and are barely detectable by 24 - 48 hours after administration [32,33]. We 123 therefore examined whether tamoxifen discontinuation allowed body weight and retinal angiogenesis 124 to recover from the defects caused by earlier tamoxifen -induced CreER activation after pups had 125 received two doses of 100 μg tamoxifen on P2 and P4. This dose was chosen, because it disrupts 126 both vascular extension and branching and has been used to examine gene deletion in multiple prior 127 retinal angiogenesis studies. Then, we examined retinal vasculature at P21 two weeks after 128 tamoxifen had been discontinued ( Fig. 2A). At this time, tamoxifen -injected CreER+ mice still had 129 significantly lower body weight compared to control littermates (Fig. 2B). We next acquired confocal 130 z-projections that spanned the depth of all three vascular plexi of the maturing retinal vasculature 131 (Fig. 2C) and grouped individual slices of the z-projection according to their representation of each 132 plexus (Fig. 2D). Although the complexity of the superficial vascular plexus was impaired at P7 (see 133 Fig. 1E), we observed normal vascular coverage and branchpoint complexity in the superficial retinal 134 vascular plexus at P21 ( Fig. 2D). Moreover, the deep plexus, which sprouts from the superficial 135 plexus after P7, and the intermediate plexus, which forms from the deep plexus after P12 [23], also 136 had normal vascular complexity ( Fig. 2D ). Accordingly, tamoxifen discontinuation allowed for 137 recovery of a vascular plexus that was initially impaired by CreER toxicity. Additionally, we found 138 that past CreER activation did not affect the growth of new vasculature after tamoxifen had been 139 cleared. Together, these findings suggest that CreER activation induces vascular growth defects 140 that can be compensated over time, whereas systemic consequences, such as reduced body weight, 141 persist beyond the treatment period. 142 143 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 5 - CreER activation impairs cell cycle progression during retinal angiogenesis 144 To understand the mechanisms underlying CreER-induced angiogenesis defects, we administered 145 one or two doses of 100 μg tamoxifen and examined retinas 24 hours after the single or the second 146 tamoxifen injection for altered endothelial cell behavior that might explain subsequent angiogenesis 147 defects (Fig. 3A). Although the body weight of CreER - and CreER+ pups was still similar at this 148 time (Fig. S1B,C), quantitative analysis of retinas stained for IB4 together with an antibody for the 149 endothelial-specific transcription factor ERG [34] revealed 35% fewer ERG ⁺ endothelial cells in 150 CreER+ compared to CreER - retinas after the second tamoxifen dose, but not after a single 151 tamoxifen dose or after vehicle treatment (Fig. 3A). Consistent with these observations, quantitative 152 RT-PCR analysis of retinas from the opposing eye of the same mice showed a 30% decrease of Erg 153 transcript levels after two but not one tamoxifen dose (Fig. 3A ). These findings suggest that 154 prolonged CreER activation reduces vascular growth. 155 To determine whether the reduction in ERG+ cells resulted from decreased proliferation, we 156 examined P5 retinas after tamoxifen treatment on P2 and P4 by staining for IB4 together with two 157 proliferation markers. Thus, we used an antibody for the proliferation marker Ki 67, which is 158 expressed during all active cell cycle phases [35], and antibodies specific for phospho -histone H3 159 (pHH3), which is detected in cells during the G2/M phase (but not in telophase; [36]). Quantitative 160 analysis of the stained retinas ( Fig. 3B ) revealed a 49% reduction in Ki67 ⁺ endothelial cells in 161 CreER+ compared to CreER - retinas, indicating that fewer endothelial cells were actively cycling 162 after CreER activation ( Fig. 3B). Consistent with the reduced number of Ki67 ⁺ endothelial cells in 163 retinas 24 hours after the second tamoxifen dose, Mki67 transcript levels were reduced by 36% in 164 CreER+ compared to CreER - retinas ( Fig. 3B ). Agreeing with the reduced number of Ki67+ 165 endothelial cells, we also observed 45% fewer pHH3+ endothelial cells in CreER+ compared to 166 CreER- retinas (Fig. 3B). Contrasting the findings for endothelial cells, the number of Ki67+ non -167 endothelial cells was similar whereas the number of pHH3+ non-endothelial cells was increased in 168 CreER+ versus CreER- retinas (Fig. S1D). Together, these findings suggest that a subset of non -169 endothelial cells had arrested in mitosis after CreER activation, which was not seen for endothelial 170 cells, which instead had stopped cycling without an obvious mitotic arrest. 171 As the above data suggest that CreER activation stalls endothelial cell cycle progression, we 172 examined retinas for the expression of p21, which is a known regulator of both DNA replication and 173 DNA damage repair [37] and was previously shown to be upregulated after Cdh5-CreER activation 174 [38]. We therefore performed quantitative analysis of the whole-mount retinas stained for IB4 and an 175 antibody specific for p21 retinas 24 hours after the second of two tamoxifen injections (Fig. 3C). We 176 found that the number of p21⁺ endothelial cells was increased 4-fold in CreER+ compared to CreER- 177 retinas (Fig. 3C). Strikingly, p21 upregulation was confined to the IB4+ area despite ubiquitous Cagg-178 CreER activation (Fig. S1E). Quantitative RT-PCR analysis corroborated that p21 was significantly 179 upregulated in CreER+ compared to CreER- retinas after two doses of tamoxifen (Fig. 3C). No such 180 transcriptional upregulation was observed in P5 retinas of vehicle -treated mice or mice that had 181 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 6 - received only a single tamoxifen dose on P4 (Fig. 3C). Nevertheless, we observed an increase in 182 p21⁺ endothelial cells in CreER+ compared to CreER - retinas after a single tamoxifen dose ( Fig. 183 S2), which suggests that p21 upregulation initially occurred independently of transcriptional 184 induction. A post-transcriptional mechanism is consistent with the known regulation of p21 protein 185 stability through ubiquitin -mediated degradation pathways [37]. Together, these results imply that 186 p21 protein levels rapidly increase in endothelial cells following CreER activation and precede overt 187 vascular defects, such as fewer ERG+ cells or reduced vascular extension and branching. This early 188 response identifies p21 as a sensitive marker of CreER-mediated endothelial toxicity. 189 190 CreER toxicity and p21 upregulation occurs in angiogenic not quiescent endothelium 191 To determine whether CreER -induced cell cycle effects were also observed with endothelial -192 selective CreER activation, we next examined mice carrying the Cdh5-CreER transgene, for which 193 endothelial CreER toxicity was first described [13]. Similar to Cagg-CreER mice, P5 CreER+ mice, 194 which had received two tamoxifen injections on P2 and P4, had 22% fewer ERG + cells, a 59% 195 decrease in the number of pHH3+ endothelial cells, and a threefold increase in the number of p21+ 196 endothelial cells compared to Cre -negative littermates ( Fig. 4A,B). In contrast, the retinas of 5/5 197 adult CreER+ mice treated with three consecutive tamoxifen injections (1 mg per injection every 24 198 hours) exhibited no obvious changes in vascular density compared to CreER- littermates, and p21+ 199 endothelial cells were not observed ( Fig. 4C ) Similar findings were observed when CreER was 200 activated in Cagg-CreER adult mice (data not shown) . Together, these findings suggest that both 201 ubiquitous and endothelial-specific CreER activation impairs endothelial cell proliferation following 202 p21-induced cell cycle arrest, likely as a response to off-target Cre endonuclease activity. 203 204

205 CreER-loxP mutagenesis in the mouse has considerably advanced our understanding of 206 vascular biology and remains the primary approach for investigating angiogenesis regulators. For 207 example, the spatiotemporal control of gene recombination in the developing retina has enabled 208 pivotal insights into key regulatory pathways such as VEGF, TGFB and notch signaling, as well as 209 their intricate cross talk during angiogenesis (e.g., [8,7,5,39,11]). However, recent studies have 210 highlighted a significant caveat with this widely used system. Specifically, we and others have shown 211 that endothelial cells in the angiogenic retina are highly susceptible to the toxic effects of CreER 212 activation independently of targeted gene deletion [13,38], thereby creating risk that the 213 interpretation of angiogenesis effects in knockdown studies may be inaccurate unless CreER toxicity 214 is accounted for. Here, we demonstrate that CreER activation does not affect the quiescent 215 endothelium and that Cre -induced vascular defects in the angiogenic retina are attributable to 216 reduced endothelial proliferation. Importantly, our findings also reveal that CreER toxicity is not 217 mitigated by the presence of loxP sites and that the angiogenic endothelium exhibits a capacity for 218 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 7 - recovery over time. 219 Ubiquitously expressed transgenes have been used to investigate the angiogenic effects of 220 molecules expressed in endothelial cells (e.g., [16,9,40,14,15]). These transgenes are also valuable 221 for reporting on angiogenic requirements of molecules secreted by other cell types in the vascular 222 environment that affect endothelial cell behavior. While expected, our study shows that CreER 223 activation has toxic effects on endothelial cells also when using a ubiquitous promoter, thereby 224 reinforcing the importance of considering CreER-associated toxicity as a confounding variable in 225 angiogenesis studies. Furthermore, we have made two additional notable observations, which 226 suggest that endothelial-selective CreER activation should be prioritized when investigating potential 227 angiogenesis regulators in endothelial cells, in order to minimize systemic confounding factors that 228 arise from using a ubiquitously expressed CreER transgene. First, we detected reduced postnatal 229 weight gain following Cagg-CreER activation across all tamoxifen doses tested, which was not 230 observed for endothelial-specific CreER activation [13]. This effect was also observed when pups 231 were assessed two weeks after tamoxifen discontinuation, when vascular density had normalized. 232 Widespread CreER activation therefore imposes a broader burden on early postnatal development 233 that cannot be attributed to endothelial toxicity alone, and suggests that ubiquitously expressed 234 CreER should be avoided for CreER expression when studying endothelial genes. Second, 235 activating a ubiquitously and highly expressed CreER transgene impaired vascular extension even 236 at the lowest tamoxifen dose tested (50 µg), which falls into the lower range commonly reported in 237 retinal angiogenesis studies. Nevertheless, the use of ubiquitous promoters to drive CreER 238 expression remains necessary for targeting candidate pro-angiogenic modulators expressed by non-239 endothelial cells and will therefore require appropriate controls to account for endothelial toxicity [4]. 240 We also found that the repeated delivery of even low doses of tamoxifen to activate CreER 241 becomes increasingly toxic and therefore does not provide a reliable strategy to mitigate CreER -242 associated toxicity. Under our experimental conditions, a single tamoxifen injection had no 243 detectable impact on retinal endothelial cell number 24 hours later; however, two injections caused 244 a ~35% reduction in endothelial cell density 24 hours after the second injection. Notably, the most 245 commonly employed protocol for gene targeting in retinal angiogenesis studies involves three or four 246 consecutive tamoxifen injections [3]. Our findings suggest that such repeated injections may 247 exacerbate CreER -associated toxicity beyond the effect we have observed with two injections, 248 although the magnitude of the effect will likely depend on CreER expression levels. In this context, 249 a better understanding of the molecular mechanisms that underlie CreER toxicity for endothelial cells 250 may help to identify dosing thresholds that have fewer off -target effects without compromising 251 recombination of floxed target genes. 252 To determine which angiogenesis-relevant endothelial cell functions are affected by CreER 253 activation, we considered previous reports showing that CreER activation lowers proliferation rates 254 in mouse embryo fibroblasts and mouse keratinocytes in vitro [21]. Of note, we established that 255 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 8 - vascular defects can be detected earlier than previously reported [13] and shown here ( Fig. 1); 256 specifically, we observed a lower number of ERG+ cells already on P5, two days before obvious 257 angiogenesis defects were detected on P7. Using Ki67 and pHH3 staining, we attributed this early 258 growth defect to reduced endothelial cell proliferation after CreER activation. By contrast, CreER 259 activation in the quiescent adult vasculature did not affect vascular density, thereby suggesting that 260 endothelial cell proliferation is an angiogenic parameter susceptible to CreER toxicity. Consistent 261 with our findings in retinal endothelium, a prior study using a Cx3Cr1-CreER transgene demonstrated 262 that CreER activation adversely affected microglia proliferation in early postnatal brains, whereas no 263 change in the microglia number was detected when Cx3Cr1 -Cre was activated in adult mice [41]. 264 Similarly, activating the ubiquitous Rosa26-CreER transgene reduced the proliferation of immature 265 hematopoietic cells in the thymus and spleen of adult mice [42], further reinforcing the notion that 266 rapidly proliferating cells are more vulnerable to the toxic effects of CreER activation. 267 According to our results, and those reported in a prior study [38], p21 upregulation is an early 268 response to CreER activation during retinal vascularisation. Although p21 is a well-known effector of 269 the DNA damage response and also serves as a cell cycle regulator [37], our experiments do not 270 directly demonstrate causality between DNA damage and reduced endothelial proliferation. 271 Nevertheless, the observed p21 upregulation is consistent with cell cycle arrest as a response to 272 DNA damage. Considering that pseudo-loxP sites are unpaired, their cleavage is expected to cause 273 single stranded breaks [17]. In non-proliferative cells, such breaks can be readily repaired with the 274 opposing DNA strand as a template. By contrast, the single-stranded DNA present in S -phase is 275 expected to be more vulnerable to DNA damage than double-stranded DNA and would require non-276 homologous end joining for repair. Supporting a model in which cells in S -phase are particularly 277 vulnerable to CreER toxicity due to DNA damage, CreER activation did not upregulate p21 in 278 endothelial cells of the quiescent retinal vasculature. Therefore, it is conceivable that CreER 279 activation-induced p21 upregulation could stall cell cycle progression to allow for single stranded 280 DNA repair. Restoring endothelial health by imposing a break in the cell cycle, including in DNA 281 synthesis, might therefore allow p21 to pave the way for the vascular normalization we observed two 282 weeks after tamoxifen discontinuation. Nevertheless, we cannot exclude that some endothelial cells 283 may also die after p21 upregulation, because it is known to trigger apoptosis in other cell types in 284 response to unrepaired DNA damage [37]. 285 Notably, p21 upregulation was restricted to endothelial cells, not only after endothelial -286 selective CreER activation but also after ubiquitous CreER activation in the retina. Consistent with a 287 protective role for p21 in promoting vascular resilience to CreER -induced stress, analysis of pHH3 288 and Ki67 staining in the retina revealed fewer cycling endothelial cells after CreER activation, without 289 evidence of mitotic arrest. In contrast, a large number of non-endothelial cells, which showed no p21 290 upregulation, appeared to have stalled in mitosis, potentially indicating a cell cycle block. Importantly, 291 p21 upregulation in retinal endothelial cells was detectable as early as 24 hours after a single 292 tamoxifen dose and preceded the detection of effects on cell cycle progression. These findings raise 293 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 9 - the question why endothelial cells in the retina are uniquely primed to mount a rapid p21 -mediated 294 response to CreER activation. A plausible explanation might arise from the prior finding that p21 295 serves a physiological function in retinal vascular development. Specifically, it has been reported 296 that p21 is selectively upregulated in a subset of endothelial cells at the vascular front to balance 297 endothelial cell sprouting and proliferation in the context of high mitogenic stimulation [43]. It is 298 therefore conceivable that CreER-mediated p21 upregulation and its associated function in 299 regulating endothelial cell proliferation co-opts a physiological mechanism that normally operates at 300 the angiogenic front at low levels, and thus serendipitously confers vascular resilience. 301 In summary, our results support the concept that CreER toxicity is a significant confounder 302 in studies of postnatal angiogenesis and reveal that these effects also occur when a ubiquitously 303 expressed transgene is employed to target endothelial cells. These findings emphasize the need for 304 tamoxifen-activated CreER controls to disentangle CreER -induced effects from gene -specific 305 phenotypes due to loxP-mediated recombination. Although our evidence suggests that quiescent 306 vasculature in adult animals appears less affected by CreER toxicity, further work is required to 307 understand implications for mouse models of adult neovascular diseases, in which endothelial cell 308 proliferation is reactivated, for example, in choroidal neovascular disease, tumor angiogenesis, or 309 ischemic vascular diseases such as peripheral artery disease or myocardial infarction. 310 311

We thank the staff of the Biological Resources and the Microscopy Facility at 312 the UCL Institute of Ophthalmology for their valuable assistance; James Brash and Laura Denti for 313 technical help in the initial phase of this study; and Tiago V. Pereira for statistical analysis assistance. 314 315 Sources of Funding: This work was supported by funding from the British Heart Foundation to CR 316 and MR [PG/23/11342], CR [PG/24/11119] and VR [FS/16/61/32740] and from Moorfields Eye 317 Charity to CR [GR001659]. 318 319 Disclosures: None. 320 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 10 -

321 322 Animals 323 All procedures were conducted in accordance with the Institutional Animal Welfare Ethical Review 324 Body and the United Kingdom Home Office guidelines under the Animals (Scientific Procedures) Act 325 1986. All mice were bred at the UCL Institute of Ophthalmology’s Biological Resources Unit and 326 housed in individually ventilated cages at 20-24°C ambient temperature, 55±10% relative humidity, 327 and a 12-hour light/dark cycle. All mice had access to a standard diet and water ad libitum. In this 328 study, Tg(CAG -cre/Esr1*)5Amc ( Cagg-CreER) (MGI:212767) [31] or Tg(Cdh5 -cre/ERT2)#Ykub 329 (Cdh5-CreER) (MGI:5705396) [11] males a C57BL6/J background were crossed with wild -type 330 females a C57BL6/J background to generate mice carrying or lacking ubiquitous and endothelial 331 cell-specific tamoxifen -inducible CreER, respectively. In some experiments, Cagg-CreER males 332 were crossed with GT(Rosa)26Sortm14(CAG-tdTomato)Hze (Rosa26tdTom) (MGI:3813512) females on a 333 mixed genetic background (C57BL6/J:129/Sv). 334 335 Tamoxifen treatments 336 Tamoxifen powder (Sigma, T5648) was dissolved in vegetable oil (peanut oil, Sigma P2144, or corn 337 oil, Sigma C8267) by agitation at 37°C for 2 hours. To activate CreER during the perinatal period, 338 mice received two intraperitoneal injections of 25 µL tamoxifen (2, 4 or 6 mg/mL) on postnatal day 339 (P) 2 and P4, and eyes were collected on P5, P7 or P21. For experiments assessing the effects of 340 a single tamoxifen injection, mice received one 25 µL tamoxifen (4 mg/mL) intraperitoneal injections 341 on P4, and eyes were collected on P5. To activate CreER in adult mice (6 -8 weeks old), 100 µL 342 tamoxifen (10 mg/mL) was administered via intraperitoneal injections once daily for three 343 consecutive days, and eyes were collected 24 hours after the final injection. Littermate animals 344 lacking CreER received intraperitoneal tamoxifen or vehicle (corn oil) injections to serve as controls. 345 Both male and female mice were included in all analyses. 346 347 Whole-mount retina staining 348 Eyes were enucleated at the indicated time points and fixed in 4% formaldehyde in PBS for 10 349 minutes at room temperature (RT). Retinas were immediately micro-dissected and prepared for flat 350 mounting by making four radial incisions. Freshly dissected retinas were stored in 96 -well plates at 351 -20°C in 100% methanol until staining. Before immunostaining, retinas were rehydrated in PBS and 352 permeabilized in PBS containing 0.5% Triton X -100 (PBST) for 30 minutes at RT. P5 retinas were 353 incubated in blocking solution comprised of 10% heat-inactivated normal goat serum and 1% bovine 354 serum albumin (Sigma-Aldrich) in PBST for 1 hour at RT. Primary antibodies were diluted 1:400 in 355 blocking solution and incubated for 90 minutes at RT; the following primary antibodies were used: 356 anti-phospho-histone H3 (anti -phH3; H9908, Sigma -Aldrich), anti -Ki-67 (ab16667, Abcam), anti -357 ERG (ab92513, Abcam) and anti-p21 (ab188224, Abcam). Following four 15 -minute PBS washes, 358 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 11 - retinas were incubated with fluorescein -labeled G. simplicifolia lectin (GSL I –BSL I; FL -1101-5, 359 Vector Labs) and the following secondary antibodies, diluted 1:200 in PBS for 1 hour at RT: Alexa 360 Fluor® 647 AffiniPure® Fab Fragment Donkey Anti -Rabbit IgG (H+L) (711 -607-003, Stratech 361 Jackson), Cy™3 AffiniPure® F(ab')₂ Fragment Donkey Anti-Rat IgG (H+L) (712-166-150, Stratech 362 Jackson), Cy ™3 AffiniPure® F(ab') ₂ Fragment Donkey Anti -Rabbit IgG (H+L) (711 -166-152, 363 Stratech Jackson). After four 15-minute washes in PBS, retinas were mounted on glass slides using 364 VECTASHIELD® Vibranceä Antifade Mounting Medium (H -1700-10, 2BScientific). P7 and P21 365 retinas were blocked in 10% serum-free protein block (DAKO; X0909, Agilent) in PBST for 1 hour at 366 RT. Biotinylated IB4 (L2140, Sigma) was diluted 1:200 in the blocking solution and incubated 367 overnight at 4°C with gentle rocking. After washing three times for 10 minutes in PBS, retinas were 368 incubated for 2 hours at RT with streptavidin conjugated to Alexa Fluor® 488, 647 or 350 (S11223, 369 S21374 or S11249, respectively, Thermo Fisher Scientific), diluted 1:200 in the blocking solution. 370 Stained retinas were washed three times for 10 minutes in PBS and mounted on glass slides using 371 Fluoromount-G™ Mounting Medium (00-4958-02, Thermo Fisher Scientific). 372 373 Image acquisition and analysis 374 IB4-stained P7 retinas were imaged using an SZX16 fluorescent stereomicroscope (Olympus) 375 equipped a C4742 -95 camera (Hamamatsu). Retinal radius and retinal vascular extension were 376 measured in each leaflet using the FIJI (NIH Bethesda) and averaged by retina. Retinal radius was 377 defined as the distance from the optic disc to the retinal margin, whereas retinal vascular extension 378 was defined as the distance from the optic disc to the front of the vascular network divided by the 379 retinal radius. Vascular branchpoints were quantified in 4 fields of view per retina (150 x 150 pixel 380 each) behind the angiogenic front in a region between an artery and a vein using Angiotool v.0.9226 381 (NIH Bethesda) and averaged by retina. IB4-stained P5 retinas co -stained for ERG, pHH3, Ki -67 382 and P21 imaged using a widefield inverted Ti2 fluorescent microscope (Nikon) equipped with a DS-383 Qi2 camera (Nikon). The entire leaflet (or whole retina for ERG) was analyzed using FIJI. 384

fluorescence was subtracted using a rolling ball algorithm with a radius of 10 pixels for 385 ERG and p21, 30 pixels for Ki-67, and 50 pixels for pHH3 and IB4. The number of positively stained 386 cells was automatically quantified using the “Analyze Particle” function in FIJI after fluorescent signal 387 thresholding and then confirmed by visual inspection of each image. Co-localization of pHH3+ or Ki-388 67+ cells with IB4 was assessed manually. The values for each leaflet were summed to obtain the 389 value for that retina. Images of IB4 -stained P21 retinas were acquired with a Stellaris 5, confocal 390 microscope equipped with LAS X software (Leica) and comprised a 581 µm 2 region between an 391 artery and a vein, with the scan beginning at the surface of the retina and extending through the 392 ganglion cell layer into the outer plexiform layer. Z-projections of the vasculature were generated 393 using a temporal color-coding method in FIJI. Confocal slices containing the superficial, intermediate 394 or deep vascular plexus were extracted from each z projection and processed separately for 395 quantification of vascular coverage; the number of branchpoints relative to the retinal area was 396 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 12 - determined using Angiotool v0.9226 (NIH Bethesda) in four regions of interest (ROI) for P21 and 397 one ROI for adults . When four ROIs were acquired per retina, it was one for each leaflet, and the 398 values were averaged for each plexus in each retina. 399 400 Gene expression analysis 401 Total RNA was isolated from P5 retinas using the RNeasy Mini Kit (Qiagen Inc.) and reverse -402 transcribed using the Superscript IV reverse transcription kit (Thermo Fisher Scientific). cDNA was 403 analyzed by real-time PCR on the QuantStudio 6 Flex Real-Time PCR system (Applied Biosystems) 404 using the SYBR Green master mix (Applied Biosystems) and oligonucleotide primer pairs specific 405 for Erg (forward 5’ -CCGGATACTGTGGGGATGAG-3’ and reverse 5’ -406 TCTGCGCTCATTTGTGGTCA-3’), Cdkn1a (forward 5’ -TCGCTGTCTTGCACTCTGGTGT-3’ and 407 reverse 5’-TCGCTGTCTTGCACTCTGGTGT-3’), mKi67 (5’-GAGGAGAAACGCCAACCAAGAG-3’) 408 and Actb (forward, 5’ -CACCACACCTTCTACAATGAG-3’ and reverse 5’ - 409 GTCTCAAACATGATCTGGGTC-3’). Expression of each target gene was normalized to Actb levels, 410 and quantification was performed using the 2-ΔΔCT method [44]. Values are expressed as fold change 411 in the CreER+ relative to the CreER- control group. 412 413 Statistical analysis and reproducibility 414 One retina from one mouse was considered one biological replicate, whereby both retinas were used 415 for different analyses (e.g., one for staining and the other for gene expression). We analyzed 4 416 different fields of view from each retina and averaged the values to obtain the value for that sample. 417 Blinding was used during treatment and outcome assessment, with genotypes being disclosed only 418 during statistical analysis. Samples or data points were excluded only in the case of technical 419 equipment or human error that caused a sample to be poorly controlled. Statistical analyses were 420 performed using Prism 10 (GraphPad Software Inc.) or Stata (version 18, StataCorp, College 421 Station, Texas, USA). Data are shown as means ± SD. A non -parametric Mann-Whitney test was 422 performed for all analyses. Interaction between tamoxifen doses and sex was tested by comparing 423 the goodness-of-fit of two linear regression models: a reduced model including only the main effects 424 of treatment and sex, and a full model that also included the treatment -by-sex interaction term. We 425 then used a likelihood-ratio test to test if the full model offered a better fit to the data than the reduced 426 model. Significance was established at P< 0.05. P values are indicated in each Figure as *P< 0.05, 427 **P< 0.01, ***P< 0.001, ****P< 0.0001. At least two independent litters were used for each analysis 428 to ensure reproducibility and robustness of findings. Sample size estimates for vascular extension 429 as the primary outcome were based on the number of animals needed to observe an effect size of 430 2.0 standard deviation units (a mean difference of 0.1 and a common SD of 0.05, based on prior 431 published data [13]; using calculations with a 2 -sided alpha 0.05 and power 80%, the required 432 sample size was calculated to be seven mice per group. 433 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 13 -

434 1. Payne, S., De Val, S., & Neal, A. (2018). Endothelial-Specific Cre Mouse Models. Arterioscler 435 Thromb Vasc Biol, 38(11), 2550-2561, doi:10.1161/ATVBAHA.118.309669. 436 2. Feil, R., Brocard, J., Mascrez, B., LeMeur, M., Metzger, D., & Chambon, P. (1996). Ligand-437 activated site-specific recombination in mice. Proc Natl Acad Sci U S A, 93(20), 10887-438 10890, doi:10.1073/pnas.93.20.10887. 439 3. Pitulescu, M. E., Schmidt, I., Benedito, R., & Adams, R. H. (2010). Inducible gene targeting in 440 the neonatal vasculature and analysis of retinal angiogenesis in mice. Nat Protoc, 5(9), 441 1518-1534, doi:nprot.2010.113 [pii]10.1038/nprot.2010.113. 442 4. Rashbrook, V. S., Brash, J. T., & Ruhrberg, C. (2022). Cre toxicity in mouse models of 443 cardiovascular physiology and disease. Nature Cardiovascular Research, 1(9), 806-816, 444 doi:10.1038/s44161-022-00125-6. 445 5. Aspalter, I. M., Gordon, E., Dubrac, A., Ragab, A., Narloch, J., Vizan, P., et al. (2015). Alk1 and 446 Alk5 inhibition by Nrp1 controls vascular sprouting downstream of Notch. Nat Commun, 6, 447 7264, doi:10.1038/ncomms8264. 448 6. Tammela, T., Zarkada, G., Nurmi, H., Jakobsson, L., Heinolainen, K., Tvorogov, D., et al. 449 (2011). VEGFR-3 controls tip to stalk conversion at vessel fusion sites by reinforcing Notch 450 signalling. Nature Cell Biology, 13(10), 1202-1213, doi:10.1038/ncb2331. 451 7. Benedito, R., Rocha, S. F., Woeste, M., Zamykal, M., Radtke, F., Casanovas, O., et al. (2012). 452 Notch-dependent VEGFR3 upregulation allows angiogenesis without VEGF-VEGFR2 453 signalling. [Research Support, Non-U.S. Gov't]. Nature, 484(7392), 110-114, 454 doi:10.1038/nature10908. 455 8. Chappell, J. C., Darden, J., Payne, L. B., Fink, K., & Bautch, V. L. (2019). Blood Vessel 456 Patterning on Retinal Astrocytes Requires Endothelial Flt-1 (VEGFR-1). J Dev Biol, 7(3), 457 doi:10.3390/jdb7030018. 458 9. Ho, V. C., Duan, L. J., Cronin, C., Liang, B. T., & Fong, G. H. (2012). Elevated vascular 459 endothelial growth factor receptor-2 abundance contributes to increased angiogenesis in 460 vascular endothelial growth factor receptor-1-deficient mice. Circulation, 126(6), 741-752, 461 doi:10.1161/CIRCULATIONAHA.112.091603. 462 10. Serra, H., Chivite, I., Angulo-Urarte, A., Soler, A., Sutherland, J. D., Arruabarrena-Aristorena, 463 A., et al. (2015). PTEN mediates Notch-dependent stalk cell arrest in angiogenesis. Nat 464 Commun, 6, 7935, doi:10.1038/ncomms8935. 465 11. Okabe, K., Kobayashi, S., Yamada, T., Kurihara, T., Tai-Nagara, I., Miyamoto, T., et al. (2014). 466 Neurons limit angiogenesis by titrating VEGF in retina. Cell, 159(3), 584-596, 467 doi:10.1016/j.cell.2014.09.025. 468 12. Claxton, S., Kostourou, V., Jadeja, S., Chambon, P., Hodivala-Dilke, K., & Fruttiger, M. (2008). 469 Efficient, inducible Cre-recombinase activation in vascular endothelium. Genesis, 46(2), 74-470 80, doi:10.1002/dvg.20367. 471 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 14 - 13. Brash, J. T., Bolton, R. L., Rashbrook, V. S., Denti, L., Kubota, Y., & Ruhrberg, C. (2020). 472 Tamoxifen-Activated CreERT Impairs Retinal Angiogenesis Independently of Gene 473 Deletion. Circ Res, 127(6), 849-850, doi:10.1161/CIRCRESAHA.120.317025. 474 14. Koo, Y., Barry, D. M., Xu, K., Tanigaki, K., Davis, G. E., Mineo, C., et al. (2016). Rasip1 is 475 essential to blood vessel stability and angiogenic blood vessel growth. Angiogenesis, 19(2), 476 173-190, doi:10.1007/s10456-016-9498-5. 477 15. Zhang, S., Liu, W., Yang, Y., Sun, K., Li, S., Xu, H., et al. (2019). TMEM30A deficiency in 478 endothelial cells impairs cell proliferation and angiogenesis. J Cell Sci, 132(7), 479 doi:10.1242/jcs.225052. 480 16. Kim, J., Oh, W. J., Gaiano, N., Yoshida, Y., & Gu, C. (2011). Semaphorin 3E-Plexin-D1 481 signaling regulates VEGF function in developmental angiogenesis via a feedback 482 mechanism. Genes & Development, 25(13), 1399-1411, doi:10.1101/gad.2042011. 483 17. Thyagarajan, B., Guimaraes, M. J., Groth, A. C., & Calos, M. P. (2000). Mammalian genomes 484 contain active recombinase recognition sites. Gene, 244(1-2), 47-54, doi:10.1016/s0378-485 1119(00)00008-1. 486 18. Pugach, E. K., Richmond, P. A., Azofeifa, J. G., Dowell, R. D., & Leinwand, L. A. (2015). 487 Prolonged Cre expression driven by the alpha-myosin heavy chain promoter can be 488 cardiotoxic. J Mol Cell Cardiol, 86, 54-61, doi:10.1016/j.yjmcc.2015.06.019. 489 19. Semprini, S., Troup, T. J., Kotelevtseva, N., King, K., Davis, J. R., Mullins, L. J., et al. (2007). 490 Cryptic loxP sites in mammalian genomes: genome-wide distribution and relevance for the 491 efficiency of BAC/PAC recombineering techniques. Nucleic Acids Res, 35(5), 1402-1410, 492 doi:10.1093/nar/gkl1108. 493 20. Wang, X., Lauth, A., Wan, T. C., Lough, J. W., & Auchampach, J. A. (2020). Myh6-driven Cre 494 recombinase activates the DNA damage response and the cell cycle in the myocardium in 495 the absence of loxP sites. Dis Model Mech, 13(12), doi:10.1242/dmm.046375. 496 21. Loonstra, A., Vooijs, M., Beverloo, H. B., Allak, B. A., van Drunen, E., Kanaar, R., et al. (2001). 497 Growth inhibition and DNA damage induced by Cre recombinase in mammalian cells. Proc 498 Natl Acad Sci U S A, 98(16), 9209-9214, doi:10.1073/pnas.161269798. 499 22. Ruhrberg, C., & Bautch, V. L. (2013). Neurovascular development and links to disease. Cellular 500 and Molecular Life Sciences, 70(10), 1675-1684, doi:10.1007/s00018-013-1277-5. 501 23. Stahl, A., Connor, K. M., Sapieha, P., Chen, J., Dennison, R. J., Krah, N. M., et al. (2010). The 502 mouse retina as an angiogenesis model. Invest Ophthalmol Vis Sci, 51(6), 2813-2826, 503 doi:10.1167/iovs.10-5176. 504 24. Fruttiger, M. (2007). Development of the retinal vasculature. Angiogenesis, 10(2), 77-88, 505 doi:10.1007/s10456-007-9065-1. 506 25. Fantin, A., & Ruhrberg, C. (2015). The Embryonic Mouse Hindbrain and Postnatal Retina as In 507 Vivo Models to Study Angiogenesis. Methods Mol Biol, 1332, 177-188, doi:10.1007/978-1-508 4939-2917-7_13. 509 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 15 - 26. Alsina-Sanchis, E., Mulfarth, R., Moll, I., Mogler, C., Rodriguez-Vita, J., & Fischer, A. (2021). 510 Intraperitoneal Oil Application Causes Local Inflammation with Depletion of Resident 511 Peritoneal Macrophages. Mol Cancer Res, 19(2), 288-300, doi:10.1158/1541-7786.MCR-512 20-0650. 513 27. Montenegro, M. F., Pessa, L. R., Gomes, V. A., Desta, Z., Flockhart, D. A., & Tanus-Santos, J. 514 E. (2009). Assessment of vascular effects of tamoxifen and its metabolites on the rat 515 perfused hindquarter vascular bed. Basic Clin Pharmacol Toxicol, 104(5), 400-407, 516 doi:10.1111/j.1742-7843.2009.00377.x. 517 28. Ye, R., Wang, Q. A., Tao, C., Vishvanath, L., Shao, M., McDonald, J. G., et al. (2015). Impact 518 of tamoxifen on adipocyte lineage tracing: Inducer of adipogenesis and prolonged nuclear 519 translocation of Cre recombinase. Mol Metab, 4(11), 771-778, 520 doi:10.1016/j.molmet.2015.08.004. 521 29. Raimondi, C. (2014). Neuropilin-1 enforces extracellular matrix signalling via ABL1 to promote 522 angiogenesis. [Research Support, Non-U.S. Gov't]. Biochem Soc Trans, 42(5), 1429-1434, 523 doi:10.1042/BST20140141. 524 30. Fantin, A., Lampropoulou, A., Gestri, G., Raimondi, C., Senatore, V., Zachary, I., et al. (2015). 525 NRP1 Regulates CDC42 Activation to Promote Filopodia Formation in Endothelial Tip 526 Cells. Cell Rep, 11(10), 1577-1590, doi:10.1016/j.celrep.2015.05.018. 527 31. Hayashi, S., & McMahon, A. P. (2002). Efficient recombination in diverse tissues by a 528 tamoxifen-inducible form of Cre: a tool for temporally regulated gene activation/inactivation 529 in the mouse. Dev Biol, 244(2), 305-318, doi:10.1006/dbio.2002.0597. 530 32. Gabel, F., Aubry, A. S., Hovhannisyan, V., Chavant, V., Weinsanto, I., Maduna, T., et al. 531 (2020). Unveiling the Impact of Morphine on Tamoxifen Metabolism in Mice in vivo. Front 532 Oncol, 10, 25, doi:10.3389/fonc.2020.00025. 533 33. Zhong, Q., Zhang, C., Zhang, Q., Miele, L., Zheng, S., & Wang, G. (2015). Boronic prodrug of 534 4-hydroxytamoxifen is more efficacious than tamoxifen with enhanced bioavailability 535 independent of CYP2D6 status. BMC Cancer, 15, 625, doi:10.1186/s12885-015-1621-2. 536 34. Birdsey, G. M., Dryden, N. H., Amsellem, V., Gebhardt, F., Sahnan, K., Haskard, D. O., et al. 537 (2008). Transcription factor Erg regulates angiogenesis and endothelial apoptosis through 538 VE-cadherin. Blood, 111(7), 3498-3506, doi:10.1182/blood-2007-08-105346. 539 35. Gerdes, J., Lemke, H., Baisch, H., Wacker, H.-H., Schwab, U., & Stein, H. (1984). Cell cycle 540 analysis of a cell proliferation-associated human nuclear antigen defined by the monoclonal 541 antibody Ki-67. Journal of immunology (Baltimore, Md.: 1950), 133(4), 1710-1715. 542 36. Hans, F., & Dimitrov, S. (2001). Histone H3 phosphorylation and cell division. Oncogene, 543 20(24), 3021-3027, doi:10.1038/sj.onc.1204326. 544 37. Ticli, G., Cazzalini, O., Stivala, L. A., & Prosperi, E. (2022). Revisiting the Function of 545 p21(CDKN1A) in DNA Repair: The Influence of Protein Interactions and Stability. Int J Mol 546 Sci, 23(13), doi:10.3390/ijms23137058. 547 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 16 - 38. Garcia-Gonzalez, I., Rocha, S. F., Hamidi, A., Garcia-Ortega, L., Regano, A., Sanchez-Muñoz, 548 Maria S., et al. (2024). iSuRe-HadCre is an essential tool for effective conditional genetics. 549 Nucleic Acids Research, 52(13), e56-e56, doi:10.1093/nar/gkae472. 550 39. Zarkada, G., Heinolainen, K., Makinen, T., Kubota, Y., & Alitalo, K. (2015). VEGFR3 does not 551 sustain retinal angiogenesis without VEGFR2. Proc Natl Acad Sci U S A, 112(3), 761-766, 552 doi:10.1073/pnas.1423278112. 553 40. Schlecht, A., Leimbeck, S. V., Jagle, H., Feuchtinger, A., Tamm, E. R., & Braunger, B. M. 554 (2017). Deletion of Endothelial Transforming Growth Factor-beta Signaling Leads to 555 Choroidal Neovascularization. Am J Pathol, 187(11), 2570-2589, 556 doi:10.1016/j.ajpath.2017.06.018. 557 41. Sahasrabuddhe, V., & Ghosh, H. S. (2022). Cx3Cr1-Cre induction leads to microglial activation 558 and IFN-1 signaling caused by DNA damage in early postnatal brain. Cell Rep, 38(3), 559 110252, doi:10.1016/j.celrep.2021.110252. 560 42. Higashi, A. Y., Ikawa, T., Muramatsu, M., Economides, A. N., Niwa, A., Okuda, T., et al. (2009). 561 Direct hematological toxicity and illegitimate chromosomal recombination caused by the 562 systemic activation of CreERT2. J Immunol, 182(9), 5633-5640, 563 doi:10.4049/jimmunol.0802413. 564 43. Pontes-Quero, S., Fernandez-Chacon, M., Luo, W., Lunella, F. F., Casquero-Garcia, V., 565 Garcia-Gonzalez, I., et al. (2019). High mitogenic stimulation arrests angiogenesis. Nat 566 Commun, 10(1), 2016, doi:10.1038/s41467-019-09875-7. 567 44. Livak, K. J., & Schmittgen, T. D. (2001). Analysis of relative gene expression data using real-568 time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods, 25(4), 402-408, 569 doi:10.1006/meth.2001.1262. 570 571 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 17 - Figures and legends: Fig. 1 Cagg-CreER activation impairs weight gain and retinal angiogenesis in postnatal mice (A) Schematic representation of the experimental setup and body weight measurement of P7 wild - type C57Bl6/J pups treated with vehicle (vegetable oil, n = 8) on postnatal day (P) 2 and P4, compared to untreated controls (n = 6). (B) Representative images of P7 whole-mount retinas stained with IB4 from untreated and vehicle- treated pups, and quantification of vascular extension (n = 6 retinas per group). Vascular extension was measured as the ratio of the distance from the optic nerve head to the peripheral vascular front (red arrow and line), relative to total retinal radius (blue arrow and dotted line). (C-E) Schematic representation of the tamoxifen treatment of Cagg-CreER+ and CreER- pups (C), along with body weight and retinal radius on P7 following tamoxifen administration at the indicated doses on P2 and P4 (n = 7 -14 pups per group). ( D-E) Representative images of P7 CreER+ and CreER- whole-mount retinas stained with IB4, including quantification of vascular extension (D, n = 7-11 pups per group) and branchpoints (E, n = 7-12 pups per group). Boxes in (D) indicate regions between an artery and a vein shown at higher magnification in (E) and used for branchpoint analysis. Scale bars: 500 µm (vascular extension); 100 µm (branchpoints). Data are shown as mean ± SD. Each data point represents one mouse (body weight) or the average of four measurements from one retina from one mouse (retinal radius and vascular parameters). Retinal data are normalized to untreated controls (B), or CreER- controls (C-E). *P< 0.05; **P< 0.01; ***P<0.001; ****P0.05). Mann-Whitney U test. Fig. 2 Recovery of retinal angiogenesis but not body weight defects caused by past CreER activation. (A-D) Schematic representation of the experimental setup (A) and body weight (B) of Cagg-CreER+ and CreER- littermate pups on a C57BL6/J background at P21 after treatment with 100 µg tamoxifen on P2 and P4 (n = 6 -7 per group). (C) Representative projection of confocal z -stack images from whole-mount retinas of CreER+ and CreER- pups stained with IB4, with the vascular plexi resolved by pseudo-coloring according to retinal depth (20 z-slices; blue: superficial, green: intermediate, red: deep); (D) Representative images of the three IB4+ retinal plexi, shown in greyscale, extracted from the confocal z-projection shown in (C), and quantification of vascular coverage and branchpoints per region of interest (ROI) between an artery and vein for each plexus (n = 6-7 per group). Scale bars: 100 µm. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 18 - Data are shown as mean ± SD. Each data point represents one mouse (body weight) or the average of four regions of interest (ROI) from one retina from one mouse (vascular measurements). ** P 0.05). Mann-Whitney U test. Fig. 3 Cagg-CreER activation reduces endothelial cell proliferation and upregulates p21 (A-C) Cagg-CreER and CreER- littermate pups on a C57BL6/J background were treated with either vehicle, 100 µg tamoxifen on P2 and P4, referred to as the tam (2) condition, or 2 doses of vehicle, or with 100 µg tamoxifen on P4 only, referred to as the tam (1) condition, and then analyzed on P5; schematic representations of the experimental setup are shown in each panel. The retina from each mouse was used for staining whereas the other retina was used for gene expression analysis. Scale bars: 100 µm. (A) P5 retinas labeled with IB4 (greyscale) and stained for ERG (green), including quantification of ERG+ endothelial cells per retina 24 hours, and quantification of Erg mRNA levels relative to Actb. (B) P5 retinas labeled with IB4 (greyscale) and stained for Ki67 and pHH3, including quantification of Ki67+ and pHH3+ cells detected in IB4+ vasculature 24 hours; Ki67+ pHH3 - cells (red), Ki67+ pHH3+ cells (pink), and quantification of mRNA levels for Mki67 (encoding Ki67) relative to Actb. (C) P5 retinas labeled with IB4 (greyscale) and stained for p21 (magenta), including quantification of p21+ cells per retina and quantification of mRNA levels of Cdkna1 (encoding p21) relative to Actb. Data are shown as mean ± SD. Each data point represents one retina from one mouse (n = 4-8 per group). Gene expression results are reported as fold change relative to the average of the CreER - group. *P < 0.05; **P 0.05). Mann-Whitney U test. Fig. 4 Reduced vascular growth and p21 upregulation in angiogenic but not in quiescent endothelium after Cdh5-CreER activation (A, B) Cdh5-CreER+ and CreER- littermate pups on a C57BL6/J background were treated with 100 µg tamoxifen on P2 and P4, referred to as the tam (2) condition, and their retinas were analyzed on P5. Schematic representations of the e xperimental setup are shown in each panel . (A) P5 retinas labeled with IB4 (greyscale) and stained for ERG (green), including quantification of ERG+ endothelial cells per retina. (B) P5 retinas labeled with IB4 (greyscale) and stained for p21 (magenta) and pHH3 (cyan), quantification of p21+ cells per retina and pHH3+ cells detected in IB4+ vasculature per retina. Scale bars: 100 µm. (C) Schematic representation of the experimental setup of adult Cdh5-CreER+ and CreER- mice on a C57BL6/J background, which were treated with 1 mg tamoxifen daily for three consecutive days and analyzed 24 hours after the last injection. (D) Representative projection of confocal z -stack images from whole-mount retinas of CreER+ and CreER- mice stained with IB4; with vascular plexi resolved by pseudo -coloring according to retinal depth (25 z -slices; blue: superficial, green: intermediate, red: deep). (E) Representative images of the three IB4+ retinal plexi, shown in greyscale, extracted from the confocal z-projection shown in (D) including quantification of vascular .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint - 19 - coverage and branchpoints per region of interest (ROI) between an artery and vein in each plexus (n = 4-5 per group). Scale bars: 100 µm. Data are shown as mean ± SD. Each data point represents one ROI for one retina from one mouse. *P 0.05). Mann-Whitney U test. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint 150 µg 100 µg 50 µg C Figure 1 50 100 1500.0 0.5 1.0 1.5 Tamoxifen dose (μg) Retinal radius ns ns ns CreER- CreER+ CreER- CreER+ C57Bl6/J Cagg-CreER P2 P4 P7 analysistamoxifen (50, 100 or 150 µg) 50 100 1500 1 2 3 4 5 6Body weight (g) Tamoxifen dose (μg) ✱ ✱ ✱✱ CreER- CreER+ CreER- CreER+ CreER- CreER+ 150 µg 50 µg 100 µg CreER+ CreER- CreER- CreER+ CreER- CreER+ 500.0 0.5 1.0 1.5 Branchpoints ✱✱ 1000.0 0.5 1.0 1.5 Branchpoints ✱✱ 1500.0 0.5 1.0 1.5 Branchpoints ✱✱✱✱ 500.0 0.5 1.0 1.5 Vascular extension ✱✱✱✱ 1000.0 0.5 1.0 1.5 Vascular extension ✱✱✱ 1500.0 0.5 1.0 1.5 Vascular extension ✱✱✱ D E 0.0 0.5 1.0 1.5 Vascular extension untreated vehicle ns 0 1 2 3 4 5 6Body weight (g) untreated vehicle nsC57Bl6/J wild type untreated vehicle P2 P4 P7 analysisvehicle (no tamoxifen) A B .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint Figure 2 tamoxifen (100 µg) P2 P4 P21 analysis 100 0 3 6 9 12 15 Body weight (g) ✱✱ Legend Legend CreER- CreER+ C57Bl6/J Cagg-CreER CreER- CreER+ SuperficialIntermediateDeep 0 10 20 30 Vascular coverage (%) DeepIntermediateSuperficial ns ns ns CreER- CreER+CreER- CreER+ 0 90 180 270 Branchpoints / ROI DeepIntermediateSuperficial ns ns ns CreER- CreER+ A B C D .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint A Figure 3 ERG+ IB4+ B CreER- CreER+ p21+ IB4+ C CreER- CreER+ Ki67+ pHH3- Ki67+ pHH3+ IB4+ vehicle tam (1) tam (2)0 1 2 Erg relative to Actb Cre- CaggCre-ERTMCreER+ CreER-ns ns * vehicle tam (1) tam (2)0 1 2 3 4 5 Cdkn1a (p21) relative to Actb nsns ** vehicle tam (1) tam (2)0.0 0.5 1.0 1.5 2.0 2.5 Mki67 relative to Actb ns ns * tam (2)0 2000 4000 6000 8000 ERG+ cells / retina ✱ tam (2)0 20 40 60 80 100 Ki67+ cells in IB4+ vasculature ✱ tam (2)0 20 40 60 80 100 pHH3+ cells in IB4+ vasculature ✱ C57Bl6/J Cagg-CreER C57Bl6/J Cagg-CreER C57Bl6/J Cagg-CreER CreER- CreER+ 100 µg tam (2) P2 P4 P5 analysis or P4 P5 analysis 100 µg tam (1) 100 µg tam (2) P2 P4 P5 analysis or P4 P5 analysis 100 µg tam (1) 100 µg tam (2) P2 P4 P5 analysis tam (1)0 2000 4000 6000 8000 ERG+ cells / retina ns tam (2)0 450 900 1350 1800 p21+ cells / retina ✱ .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint Figure 4 A C57Bl6/J Cdh5-CreER 100 µg tam (2) P2 P4 P5 analysis CreER- CreER+ pHH3+ p21+ IB4+ tam (2)0 450 900 1350 1800 p21+ cells / retina ✱ tam (2)0 25 50 75 100 125 150 pHH3+ cells in IB4+ vasculature ✱ tam (2)0 2000 4000 6000 8000 ERG+ cells / retina CreER CreER+ CreER- CreER+ ✱ ERG+ IB4+ 1 mg tamoxifen -3 -2 0 analysis -1 C57Bl6/J Cdh5-CreER 100 µg tam (2) P2 P4 P5 analysisB C57Bl6/J Cdh5-CreER C CreER- CreER+ same experiment P7 control 0 10 20 30 Vascular coverage (%) CreER-CreER+CreER- CreER+ DeepIntermediateSuperficial ns ns ns 0 50 100 150 200 Branchpoints / ROI DeepIntermediateSuperficial ns ns ns CreER- CreER+ D CreER- CreER+ SuperficialIntermediateDeep E .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655564doi: bioRxiv preprint