Low-density Lipoprotein Receptor-related Protein 5 (LRP5)-deficient Rats Have Reduced Bone Mass and Abnormal Development of the Retinal Vasculature

Humans carrying homozygous loss-of-function mutations in the Wnt co-receptor LRP5 (low-density lipoprotein receptor–related protein 5) develop osteoporosis and a defective retinal vasculature known as familial exudative vitreoretinopathy (FEVR) due to disruption of the Wnt signaling pathway. The purpose of this study was to use CRISPR/Cas9-mediated gene editing to create strains of Lrp5-deficient rats and to determine whether knockout of Lrp5 resulted in phenotypes that model the bone and retina pathology in LRP5-deficient humans. Knockout of Lrp5 in rats produced low bone mass, decreased bone mineral density, and decreased bone size. The superficial retinal vasculature of Lrp5-deficient rats was sparse and disorganized, with extensive exudates and decreases in vascularized area, vessel length, and branch point density. This study showed that Lrp5 could be predictably knocked out in rats using CRISPR/Cas9, causing the expression of bone and retinal phenotypes that will be useful for studying the role of Wnt signaling in bone and retina development and for research on the treatment of osteoporosis and FEVR.


Introduction
Wnt signaling plays key roles in development, and alterations in the Wnt pathway are among the most common events associated with human disease [1,2]. Wnts initiate cellular responses by binding to a receptor complex that includes a member of the Frizzled family of seven-transmembrane spanning receptors and either LRP5 or LRP6 (low-density lipoprotein receptor-related protein 5 or 6) [1]. Activation of this receptor complex inhibits the phosphorylation of the β -catenin protein by glycogen synthase kinase 3 (GSK3), which normally targets β -catenin for ubiquitin-dependent proteolysis. Thus, β -catenin is stabilized in the cytoplasm, and it can subsequently translocate to the nucleus and activate target gene transcription.
Almost 20 years ago, homozygous inactivating mutations in Lrp5 were causally linked to the human syndrome osteoporosis pseudoglioma (OPPG) [3,4]. Patients with OPPG develop osteoporosis in early childhood. Subsequent work, using genetically engineered mouse models, demonstrated that LRP5 most likely acts within the osteoblast lineage to regulate bone mass [5][6][7][8][9]. A large body of additional work further linked alterations in genes whose protein products interact with LRP5 (or with the highly homologous LRP6) to changes in human bone mass [1].
Primary among these products is sclerostin, a protein which is lost by genetic inactivation in patients with high bone mass [10].
OPPG patients have severely impaired vision at birth, associated with microphthalmia, retinal hypovascularization, and retrolental fibrovascular tissue (pseudoglioma) [11]. A related human hereditary disorder, familial exudative vitreoretinopathy (FEVR), can also be caused by inactivating mutations in LRP5 [12]. In this case, progressive vision loss is due to decreased vascularization at the periphery of the retina [13]. This results in neovascularization in response to oxygen deprivation and leads to retinal traction and subsequent retinal detachment. The new vessels are leaky due to an impaired blood-retina barrier, resulting in exudates. Mutations in other genes linked to the regulation of Wnt signaling, including Frizzled 4 and norrin, also cause FEVR and two related retinal vascular disorders, Norrie disease, and Coats disease [13,14].
The laboratory rat at one time was commonly used for physiologic studies, and genetic strains were developed through selective phenotypic breeding. The development of genetically engineered mouse models using embryonic stem cells led to the dominance of mice in biomedical studies, but these mouse models have some limitations. The relatively small size of mice confers a significant advantage in housing and husbandry costs, but while both mice and rats are rodents, functional differences exist. Further, it has been observed that genetically, mice and rats may be no more closely related than humans are to old world monkeys [19,20]. On the other hand, a large physiologic data base suggests that in several fields, including learning and memory, neurologic function, and cardiovascular physiology, the rat is a better model for the study of human physiology and pathophysiology [19,20]. Rats also have larger organs than mice, providing larger tissue samples and easier dissection and surgical manipuations.
The gene editting technology, CRISPR/Cas9, has allowed the reemergence of the rat as a key model organism in studying human disease, arguing for the expansion of genetically engineered rat models [21]. The development of genetically engineered mice using embryonic stem cell-based methods can take over 6 months. In contrast, CRISPR/Cas9 technology allows development of genetically modified rodents in only 2 months with greater predictability, and the ability to modify genes in zygotes makes it feasible to use rats for such studies. Already, genetically modified rats are being used in studies of neurophysiology and cardiovascular disease [22][23][24][25].
In the present study, we used CRISPR-Cas9-mediated methods to create three strains of rats carrying inactivating mutations in the second exon of the Lrp5 gene. Studies were conducted to confirm knockout of the Lrp5 gene and demonstrate lack of expression of functional LRP5 protein. The LRP5-deficient rats developed the low bone mass previously seen in LRP5-deficient humans and mice. The validity of the model is also supported by imaging and quantitative analysis of the retinal vasculature, showing that adult Lrp5 knockout rats with impaired bone development also had a FEVR-like phenotype.

Experimental animals
Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) were maintained in accordance with institutional animal care and use guidelines, and experimental protocols were approved by the Institutional Animal Care and Use Committee of the Van Andel Institute.
The generation of Lrp5 knockout rats using CRISPR/Cas9 Three rat lines were created with a deletion in Lrp5 by using a modified CRISPR/Cas9 protocol [26]. Two sgRNAs targeting exon 2 of Lrp5 were designed using the MIT guide Genotyping by PCR-HMA and amplicon sequencing Genomic DNA from rat tail biopsies was isolated by alkaline digestion. To genotype Lrp5 knockout rats, we used the following primers-Lrp5-E2-Fwd (CCTCACCACTCCTGTTGTTT) and Lrp5-E2-Rev (CCTGCCAGAAGAGAACCTTAC)-to amplify a 354-bp product. Anticipating that the founders would have small deletions, a heteroduplex mobility assay [27] was used to define genotypes. Amplicons were subjected to denaturation-slow renaturation to facilitate formation of heteroduplexes using a thermocycler.
Samples were resolved on polyacrylamide gels (10%), and mobility profiles were used to define genotypes. PCR using the genotyping primers above was performed to amplify indels for each founder for sequencing. The amplified products were cloned using the NEB PCR cloning kit (New England Biolabs, Ipswich, MA). Clones from each founder were Sanger-sequenced using the NEB analysis primers to define indels.
Immunoblot analysis of LRP5 protein expression The lack of expression of functional LRP5 protein was confirmed by immunnoblotting.

Plasmid construction
As will be shown, a gene product was expressed by the rat strain that had an 18-bp deletion in the Lrp5 gene. To assess its potential signaling activity, site-directed mutagenesis was used to generate a human LRP5 construct with the same 18-bp deletion. Using a previously generated human LRP5-V5 plasmid [28], we produced a human LRP5Δ131-136-V5 plasmid using the Q5 site-directed mutagenesis kit (New England Biolabs) with forward primer of (CTCAATGGCACATCCCGG) and a reverse primer of (GTTGGTCTCTGAGTCCGTC). The plasmid was sequenced to confirm the correct modification.

Cell culture, transient transfections, and reporter studies
Twelve-well plates coated with poly-d-lysine were seeded with equal numbers of human HEK293-Super-TOPflash (STF) cells [29]. for transfection efficiency by measuring β -galactosidase activity as previously described [30].
Luciferase and β -galactosidase activity were measured in duplicate.
A linear mixed-effects model with random intercepts to account for repeated sampling via technical repeats was used to analyze these data via lmer in R v3.6.0 (https://cran.rproject.org/). Luciferase measures were log2-transformed to improve model fit based on normality of residuals. A plasmid directing the expression of B-galactosidase under the control of a CMV promoter was included as a fixed-effect to adjust for varying levels of success for the transfection. Linear contrasts with a Benjamini-Hochberg adjustment for multiple testing were used to test specific two-sided hypotheses of interest while maintaing a 5% false discovery rate.
To check protein expression from the various expression constructs, the pellets were resuspended in 1× SDS sample buffer, boiled for 10 min, passed through a 26-gauge needle five times, and analyzed by SDS-PAGE followed by western blotting.
Femoral bone mineral density analysis by DXA Femoral aerial bone mineral density (BMD) of 6-month-old rats was measured by dualenergy X-ray absorptiometry (DXA) using a PIXImus II bone densitometer (GE Lunar) [5]. Rats were anesthetized via inhalation of 2% isoflurane (Henry Schein, Melville, NY) with oxygen (1.0 L/min) for 10 min prior to imaging and during the procedure (≤ 5 min). The right hindlimb was placed on a specimen tray in the densitometer for analysis. Bone mineral density was calculated by the PIXImus software based on the defined active bone area of the femur.

Results
Three independent rat lines carrying deletions in exon 2 of Lrp5 were created by CRISPR/Cas9-mediated germline modification. After gene editing, one-cell-stage embryos were transferred to pseudopregnant females. A total of 15 pups were born, 3 of them containing indels within exon 2. These pups included a founder with an 18-bp deletion, designated Lrp5 KO1 , and a second founder with a 22-bp deletion at the sgRNA2 site, designated Lrp5 KO2 (Fig. 1A). A third founder, designated Lrp5 KO3 , had an inversion coupled with small deletions in the exon at both the sgRNA1 (11 bp) and sgRNA2 sites (3 bp) (Fig. 1B). These three founder rats were then crossed with wild-type rats, resulting in germline transmission of each of the three modified alleles. Rats heterozygous for each of these mutations were intercrossed and tail tissue was collected for evaluation of amount of LRP5 protein from each of these three alleles. Consistent with the predicted frameshifts associated with Lrp5 KO2 (22-bp deletion) and Lrp5 KO3 (containing the inverted allele), we could not detect LRP5 protein in rats homozygous for these mutations. In contrast, Lrp5 KO1 , containing the 18-bp deletion, had detectable LRP5 protein (Fig. 1C). This deletion resulted in the loss of amino acids 131-136, which encodes the amino acid sequence RIEVAN within the first β -propeller ( Fig. 2A).
To assess whether the in-frame deletion protein expressed in rats homozygous for the This quantification showed aerial BMD of 6-month-old male and female femurs was significantly decreased in the Lrp5 KO1 , Lrp5 KO2 , and Lrp5 KO3 rats compared to wild type animals (Suppl. 1). To further examine the skeletal phenotypes of Lrp5 knockout rats, femurs from the Lrp5 KO1 , Lrp5 KO2 , and Lrp5 KO3 lines were analyzed using micro-computed tomography (µCT).
Both sexes in all three lines displayed the same trends in trabecular bone, but with varying severities (Fig. 3A). In the three female and male Lrp5 knockout rat lines, we found significant decreases in bone mineral density (BMD; by 29-81%) and bone volume/tissue volume (BV/TV; by 84-92%) relative to wild-type controls (Fig. 3B-D). The decrease in bone mass was coupled with changes in trabecular morphology. The thickness of the trabeculae decreased significantly by 23-93% in all Lrp5 knockout rats, resulting in greater separation. This increase in trabecular separation was significant in all Lrp5 knockout rats with the exception of Lrp5 KO2 female rats.
Cortical bone architecture was also modified in both sexes (Fig. 4A). Cortical bone in  (Fig. 7), it was not possible to quantify the vascular pathology in this strain .
Because the large exudates in the Lrp5 knockout retinas obscure the vasculature, it was not possible to make measurements of the entire retina. Thus, rather than measuring the vascularized area of the entire retina, as reported in previous studies of young mice with an Lrp5 mutation [16,18], in this study, that area is defined as the percentage of retinal area that is vascularized within a 4-mm 2 digital explant (Fig. 8).. These explants were taken from areas of wild -type, heterozygous, and homozygous animals where the vasculature was fully visible and where there were minimal cuts in the retina that were made in the process of flat mounting.
Where possible, two explants were analyzed from the right and left retinas of each animal, although in some retinas it was possible to analyze only one explant.
In Lrp KO1 rats (homozygous 18-bp Lrp5 deletion), 25.4% of the retina was vascularized, significantly different than the 37.7% vascularization of wild-type retinas (Fig. 9). The median vessel length of wild-type retinas was 1.75 mm, but there was a significant reduction, to 0.63 mm, in Lrp5 KO1 retinas (Fig. 9B). The decreased branching of vessels was notable on images of homozygous knockout retinas (see Figs. 5, 6 and 8). Analysis showed that in rats with the 18-bp deletion, there were 86.3 branch points/mm 2 in wild-type retinas and 42.3/mm 2 in Lrp5 KO1 retinas, a statistically significantt reduction (Fig. 9C). None of the measured parameters in heterozygous Lrp5 KO1 retinas were different from those of the wild type.
The inactivation of Lrp5 in Lrp KO3 mice (allelic inversion) had a effect on the retinal vasculature similar to that of the 18-bp deletion. The vascularized area was reduced from 40.7% in wild-type rats to 32.7% in Lrp5 KO3 rats (Fig. 10A). The median vessel length was reduced from 1.9 mm to 0.9 mm (Fig. 10b), and the median branch point density in the Lrp5 KO3 retinas was only 48.8/mm 2 , compared with 97.5/mm 2 in wild-type retinas (Fig. 10C). All of these differences were statistically significant. As in rats with an 18-bp deletion, the heterozygous Lrp5 KO3 genotype had no effect on the retinal vasculature.

Discussion
In this study, we produced three strains of rats with loss-of-function mutations in the Lrp5 gene using CRISPR/Cas9-mediated gene editing. The loss of this co-receptor inhibited Wnt signaling required for normal development of both trabecular and cortical bone. In trabecular bone, this resulted in decreased bone mineral density, bone mass, trabecular thickness, and trabecular separation. Cortical bone density did not change, but bone size was decreased. The Lrp5 mutations also caused retinal pathology that closely resembled familial exudative vitreoretinopathy. Retinal vascularization was abnormal, with a reduced area of vascularization and reduced branching of vessels. Notably, the animals had extensive retinal exudates.
We have successfully developed the first rat model of osteoporosis by modifying the Lrp5 gene. The availability of this genetically modified rat increases the options available for evaluating osteoporosis therapies, and the larger size of the rat relative to the mouse may facilitate the assessment of orthopedic procedures in the context of Lrp5 deficiency.
The retinal vascular pathology relative to their control littermates associated with Lrp5 knockout was similar in all three rat lines and was similar to that of Lrp5 knockout mice [11,15] [11,[15][16][17]. In two studies that included 1-month-old mice, a reduced vasculature with vessels ending in tufts was also observed [11,17]. Further, the expression in Lrp5-null rats of a sparce, tortuous retinal vasculature with vessels that end in bulbous tufts also suggests that the rat retinal pathology may prove to be a good model of human        Exudates and abnormal development of the vasculature are evident. Images are representative of seven rats.   Suppl. Fig. 1. Lrp5-deficient rats showed a decrease in total femoral BMD. Aerial BMD of femurs was measured for Lrp KO1 , Lrp5 KO2 , and Lrp KO3 rats using DXA, n =4-9. For all graphs, * = p < 0.05.