A new Col1a1 conditional knock-in mouse model to study osteogenesis imperfecta

Abstract

Osteogenesis imperfecta (OI) constitutes a family of bone fragility disorders characterized by both genetic and clinical heterogeneity. Several different mouse models reproduce the classic features of OI, and the most commonly studied carry either a spontaneous or genetically induced pathogenic variant in the Col1a1 or Col1a2 gene. When OI is caused by primary alterations of type I collagen, it represents a systemic connective tissue disease that, in addition to the skeleton, also affects several extra-skeletal tissues and organs, such as skin, teeth, lung, heart, and others, where the altered type I collagen is also expressed. Currently, existing mouse models harbor a disease-causing genetic variant in all tissues and do not allow assessing the primary vs secondary consequences of the mutation on a specific organ/system. Here, we describe the generation of the first conditional knock-in allele for Col1a1 that can express a severe OI-causing glycine substitution (p.Gly1146Arg) in the triple helical region of α1(I) but only after Cre-driven recombination in the tissue of choice. We called this new dominant allele Col1a1^{G1146R-Floxed/+} and introduced it into the murine model. We describe its validation by crossing mice carrying this allele with EIIA-Cre expressing mice and showing that offspring with the recombined allele reproduce the classic features of a severe form of OI. The new mouse model will be useful to study the tissue-specific impact of this severe mutation on organs, such as the lung, the heart, and others.

Introduction

Osteogenesis imperfecta (OI or brittle bone disease) is a heterogeneous group of congenital bone fragility disorders that can severely affect the skeleton. It is characterized by generalized low bone mass, recurrent fractures after minor trauma, bowing of the long bones, vertebral compression, kyphoscoliosis, bone pain, and stunted growth.^1,2 Its severity ranges from mild to perinatal lethal. OI is a rare disease with an estimated incidence of 1:10 000-20 000 live births,³ and it is most commonly caused by autosomal dominant pathogenic variants in the type I collagen genes, COL1A1 or COL1A2. Rarer forms of the disease can be caused by variants in an increasing number of genes often involved with type I collagen intra- or extra-cellular post-translation modifications, transport, secretion, osteoblast differentiation/function, or bone matrix mineralization.^4,5 These rarer forms can be inherited in a dominant, recessive, or even X-linked form. While the most prevalent clinical classification of OI dates back to the late 1970’s,⁶ a panel of international experts periodically revises the nosology of skeletal dysplasias, including OI, and provides updated recommendations and revisions.^7,8 Currently, there is no cure for OI, although a number of treatment regimens are available and primarily borrowed from the treatment of osteoporosis.^9,10 The goal of these therapies is to increase bone mass, mobility, and quality of life while reducing fractures and bone pain with the support of a multidisciplinary team of medical specialists.

As with other genetic conditions, the study of OI has benefited tremendously from the generation of animal models that mimic the cardinal features of the disease. These models allow adequately powered experimental approaches, especially difficult to do in human studies of rare conditions, and often provide important insights into disease pathogenesis. Several, well characterized mouse models are available to study OI, including mice carrying pathogenic variants in Col1a1 (eg, Col1a1^G349C/+ (Brtl IV mouse), or the Col1a1^Jrt/+11,12), Col1a2 (eg, Col1a1^G610C/+, or oim/oim^13,14) or in genes causing recessive forms of the disease (eg, Crtap−/−, Ppib−/−, P3h1−/−, and many others^15–17). However, consistent with the widespread expression of type I collagen, OI is a systemic connective tissue disease and thus affects extra-skeletal tissues, such as skin, teeth, lung, heart, and others. Current mouse models harboring a disease-causing genetic variant in all tissues do not allow assessing the primary vs secondary consequences of the mutation on a specific organ/system. A good example of this would be the study of respiratory distress and pulmonary complications, which are leading causes of mortality in patients with OI.¹⁸ Although type I collagen is expressed in the lung, the relative contribution of skeletal malformations (eg, abnormalities of the chest wall, kyphoscoliosis, fractures of the ribs, and short stature) vs that of intrinsic lung tissue defects to the respiratory co-morbidities observed in OI has remained difficult to assess. Moreover, the specific contribution of osteocytes to the skeletal disease manifestations in OI have remained difficult to determine in mice with global expression of a type I collagen mutation. To address these questions, we designed and generated a conditional knock-in allele for Col1a1 that would allow the expression of a severe OI-causing glycine substitution in the triple helical region of α1(I), but only after Cre-driven recombination in a mouse organ or tissue or cell lineage of choice. We selected the glycine substitution c.3436G>C, p.Gly1146Arg, which was identified by colleagues at the University of Washington, Seattle, in a stillborn with fractures on prenatal ultrasound and radiographic findings consistent with a diagnosis of OI type II, and in a young infant (<30 d old) with a history of prenatal and perinatal fractures and a tentative diagnosis of OI type II or severe type III and who died in the perinatal period (Drs. Peter Byers and Ulrike Schwarze, personal communication). The new murine allele, which we officially called Col1a1^{G1146R-Floxed/+} (and nicknamed it Col1a1^Morello), will provide new insights into OI disease pathogenesis, for instance by studying its specific effects on the lung in the context of a healthy skeleton. This may allow the identification of new tissue- or cell-specific targets for therapeutic intervention. We describe the generation of the first mouse model carrying a conditional knock-in allele for Col1a1 and the validation of this model by crossing it with a transgenic mouse that expresses Cre-recombinase from the zygote level (EIIA-Cre), causing recombination in all tissues and thus reproducing the characteristic features of a severe form of OI disease.

Materials and methods

Mice

This study was approved by the University of Arkansas for Medical Sciences (UAMS) IACUC committee, and it was carried out following the local, state, and US federal regulations. Mice were housed in a pathogen-free facility with a 12-hr light/dark cycle with ad libitum access to water and a standard chow diet. Mice were euthanized according to the AVMA Guidelines on Euthanasia and the recommendations of the Guide for the Care and Use of the Laboratory Animals.

Genetic construct, generation of mice, and genotyping strategy

The generation of the genetic construct to create a conditional knock-in allele for Col1a1 is described in the result section. A single base pair mutation c.3436G>C (GGT to CGT) causing a glycine substitution (p.Gly1146Arg) close to the C-terminal end of the triple helical region was selected (reference sequence for the Col1a1 gene was ENSMUSG00000001506 and for the transcript Col1a1-201—ENSMUST00000001547.8). This is equivalent to the p.G1157R pathogenic variant in COL1A1 identified in two unrelated cases of severe OI, which were diagnosed by colleagues at the University of Washington. PCR genotyping was performed with the GoTaq G2 Hot Start Polymerase reagent (cat# M7423 Promega) and a Master Cycler thermocycler (Eppendorf). Primers used for the PCR reaction were:

Wildtype allele: 159bp F3: 5′-CCATTACTTCTTCCTGGGTTCCTC-3′; R5: 5′-CCAGGTTTGCAAGGGGGCACGG-3′.

Col1a1—Floxed allele: 294bp F5: 5′-CAAGCTAATTCCTGCAGGTCGAG-3′; R1: 5′-CCCTAGGGATAGAGAGATGGTTAGA-3′.

The PCR program conditions were as follows: 94 °C ×3 min, 94 °C ×30 s, 60 °C ×30 s, 72 °C ×30 s for 33 cycles; and then 72 °C ×5 min, and then 15 °C infinite.

Col1a1—Recombined allele: 294bp F6: 5′-CACCCGTGCCCCCTTATAACTTC-3′; R6: 5′-ATGAGGCAGAGGATGTGGGGC-3′.

The PCR program is as follows: 94 °C ×1 min, 94 °C ×30 s, 61.5 °C ×30 s, 72 °C ×30 s for 31 cycles; and then 72 °C ×5 min, and then 15 °C infinite.

Cre-recombinase—700bp: CRE-UP2: 5′-GCTAAACATGCTTCATCGTCGG-3′; CRE-DN2: 5′-GATCTCCGGTATTGAAACTCCAGC-3′.

The PCR program is as follows: 95 °C ×1 min, 95 °C ×30 s, 54 °C ×30 s, 72 °C ×60 s for 32 cycles; and then 72 °C ×5 min, and then 15 °C infinite.

All PCR products were resolved on a 2% agarose gel by horizontal electrophoresis.

DXA and digital X-ray imaging

BMC and BMD of the femur, lumbar spine, and body composition were determined in all mice using a DEXA scanner (PIXIMUS2, Lunar). Male and female 9-wk-old mice were weighted and then euthanized right before the DEXA scan. Each mouse was placed on the scanner bed in the prone position, with the limbs and tail stretched away from the body. One scan per mouse was performed and analyzed with PIXImus software (2.1; GE/Lunar). The head and the neck were excluded from calculation of the whole body parameters using a manual ROI. The PIXImus was calibrated with a phantom (corresponding to BMD = 0.0584 g/cm² and 15.5% fat) on each day of testing according to the manufacturer’s instructions. High-resolution imaging of the skeleton was acquired after the DEXA procedure using a Faxitron instrument (Faxitron Biovision Ultrafocus DXA Specimen Imaging System).

Micro-CT analysis

Left femurs were collected at sacrifice, cleaned from muscle tissue, wrapped in a 0.9% saline solution-soaked gauze (0.9% NaCl in DD2), and frozen at −20 °C. Bones were allowed to thaw for at least 2 hr at room temperature before scanning. All micro-CT scan analysis and 3D reconstructions were performed on a Scanco 40 instrument (Scanco Medical) using a slice resolution of 12 μm isotropic voxel size (only the microCT 3D rendering of the femurs was acquired at a high resolution of 6-8 μm). Full-length femurs were scanned with an effective energy of 55 kVp, X-ray tube current of 114 mA, and 275 ms integration time. For the quantification of trabecular bone in the distal femur metaphysis, the region of interest comprised 150 contoured transverse slices extending 1.8 mm above the distal growth plate, applying a grayscale threshold (lower threshold 220, upper threshold 1000) and Gaussian noise filter (sigma 0.8, support 1). The quantification of the cortical thickness was performed in 20 contoured transverse slices at the femoral midshaft region, using the previous settings but with a lower threshold of 260. Moment of inertia (MOI) values were derived from micro-CT scans of the same femur. Standard nomenclature guidelines were followed to report all micro-CT measurements (Bouxsein et al., 2010).¹⁹

Biomechanics

Femurs from the 9-wk-old male mice were collected at sacrifice wrapped in saline-soaked gauze and frozen. After micro-CT scanning, these femurs were analyzed in a 3-point bending test using an ElectroForce 5500 Test Instrument (TA Instruments) with a ramp rate of 0.05 mm/s, support span of 8 mm, and running WinTest software version 8.2. Stiffness, yield force, maximum load, and post-yield displacement were calculated for each sample.

Spine and lung histology and morphometry

Lumbar spines were harvested and fixed into 10% buffered formalin for 12-24 hr and then processed for methyl-methacrylate embedding and sectioning at 5 μm, according to standard procedures. Lungs were collected from a subset of mice (3-5 from each genotype and sex). After euthanasia, the mouse trachea was cannulated (with an 18-gauge cannula) and then attached to a reservoir containing 10% buffered formalin, and the lungs were fixed in situ for 30 min at a constant pressure of 25 cmH₂O. The lungs were then tied under that pressure and carefully detached from the thoracic cavity. They were further fixed overnight in 10% buffered formalin. The next day they were cleaned from the heart and connective tissue, and then, utilizing the Archimedes’ principle of water displacement, the lung fixed volume was measured.²⁰ Three volume measurements for each pair of lungs were taken and then averaged for the calculation of the internal lung surface area (see below). The lungs were then sequentially dehydrated in increasing concentrations of ethanol until 100% and finally embedded in paraffin. Sections of 5 μm were obtained and stained with H&E for morphological analysis. From a good-quality histological section of each mouse, 10 non-overlapping fields were acquired to quantify the mean linear intercept (MLI)—images taken at 20× magnification using a Nikon Microscope (Eclipse E400). The ImageJ software plug-in grid analysis was used to overlay a grid over each image of size equal to 745.28 × 558.96 μm with 8 horizontal line and 10 vertical lines. The ImageJ software was also used to count the intersection of each alveolar wall with the grid line. The MLI was calculated using the following equation: Lm = horizontally (N) × (L) + Vertically (N) × (L)/m, where N is the number of times the transverse was placed on the tissue, L is the length of the transverses, and m is the sum of all the intercepts from each field. The internal lung surface area (ISA) was also calculated using the following formula: ILSA = 4 × (lung fixed volume)/MLI.²¹

RNA preparation and gene expression data

Skin, lung, tail, and tibia were collected from a WT and a Col1a1^{G1146R-Floxed/+} adult mouse and flash frozen in NO₂ liquid and stored at −80 °C. A portion of the tissues was homogenized in 1 mL TriPure Isolation Reagent (Roche REF11667157001) and processed for total RNA extraction according to the manufacturer’s instructions. RNA pellets were resuspended in DEPC water and then re-precipitated in 2M final concentration of ammonium acetate and EtOH 100%. Synthesis of cDNA was performed using 100-500 ng of total RNA with the Transcriptor First Strand cDNA Synthesis Kit (Roche REF 04379012001), according to the manufacturer’s instructions. The relative abundance of Col1a1 mRNA was measured using multiplex quantitative real-time PCR (qRT-PCR) using the FastStart DNA Master PLUS Syber Green I kit (Roche Ref. 03515885001) and following the manufacturer’s instruction. The Col1a1 gene expression was normalized to the housekeeping gene Gapdh. The relative mRNA levels were calculated using the comparative cycle threshold (ΔCt) method.²² Primer sequences that were used: for Col1a1 F—5′-TTGGGGCAAGACAGTCATCGAAT-3′ and R—5′TTGGGGTGGAGGGAGTTTACACGAA-3′; for Gapdh F—5′-GCAAGAGAGGCCCTATCCCAA-3′ and R—5′-GCAAGAGAGGCCCTATCCCAA-3′.

Statistical analysis

All measurements are presented as mean ± SD. Outcomes were analyzed by genotype by one-way ANOVA. Residuals were inspected for normality by Shapiro–Wilk test and for equal variance by Levene’s test. p-values were adjusted for multiple comparisons by the false discovery rate (FDR) procedure and were considered statistically significant if ≤.05. All statistical analysis was performed in SAS version 9.4.

Results

Strategy for the generation of mice carrying a new Col1a1 conditional knock-in allele

Because of the complexity of the Col1a1 gene, which contains 51 exons, and the inherent feasibility challenges to generate a conditional knock-in allele construct for this gene, with our strategy we were not able to select any of the Col1a1 mutations used to generate previous mouse models of OI. Therefore, we designed a knock-in construct that introduces a new single base pair change and a classic OI glycine substitution c.3436G>C, p.Gly1146Arg (p.Gly979Arg in reference to the main triple helix) into exon 47 of the endogenous Col1a1 gene (Figure 1A). Colleagues at the University of Washington in Seattle (WA) identified this pathogenic variant in two unrelated cases of severe/lethal OI (Drs. Peter Byers and Ulrike Schwarze, personal communication). To make expression of this variant conditional on Cre recombination, we also inserted a loxP-flanked minigene, encoding WT exons 47-51, upstream from the inserted mutation in the endogenous exon 47 and followed by an Frt-flanked Neomycin resistance (Neo) cassette (Figure 1A). The targeting vector, identification of recombinant embryonic stem (ES) cell clones, and the generation of mice with the desired allele were outsourced to Cyagen, Inc. Founder mice with this targeted allele were mated, and germline transmission was confirmed. These mice were then crossed with Flpe mice (expressing FLP recombinase in germ cells) to remove the FRT-flanked Neo cassette and obtain mice with the floxed allele. In absence of Cre-recombinase, this allele should express WT Col1a1 and have no phenotype. Upon Cre-mediated excision of the sequence flanked by the loxP sites, the targeted Col1a1 allele should express the dominant severe OI glycine substitution in the original exon 47 in all cells targeted by the Cre and their descendants (Figure 1A). For practical purposes, the new allele called Col1a1^{G1146R-Floxed/+} is going to be abbreviated as Col1a1^FL/+ in the rest of the manuscript and all figures.

Figure 1.

Generation of the new mouse model. (A) Schematic diagram of the strategy for the generation of mice with the knock-in allele into the Col1a1 gene. The annotation of the various genetic elements is provided at the bottom of the figure. (B) Example of the identification of the four different mouse genotypes by PCR genotyping. (C) Comparison of body weight among male (n = 6-9) and female mice (n = 4-8) of the four genotypes of interest. p-values are indicated in the dot-plots and in red bold fonts when statistically different (one-way ANOVA).

Crossing with EIIa-Cre mice and analysis of the offspring phenotype

To validate this new genetic model, heterozygous mice for the floxed allele were crossed with EIIA-Cre mice (JAX stock No. 3724) to induce recombination at the zygote level and express the p.Gly1146Arg substitution ubiquitously. The mouse breeding strategy is shown in Figure S1. Assuming that this severe mutation is compatible with life in the mouse model, offspring carrying both the floxed allele and Cre-recombinase should reproduce a systemic OI phenotype. At weaning, we genotyped the mice using primer sets able to identify Col1a1 WT, floxed, and Cre-recombined alleles in addition to Cre-recombinase, as shown in Figure 1B. Consistent with our expectations, mice that were positive for the Cre-recombinase and were originally heterozygous for the floxed allele showed the presence of the recombinant allele. Importantly, these animals were viable although they were not obtained at the expected Mendelian ratio at weaning, suggesting some degree of in utero/perinatal lethality, which is commonly observed in other mouse models of OI.^11,23,24 At 2 mo of age, we noticed that mice with the recombined allele were runted and thus harvested them for study at this age. The body weight of both male and female Col1a1^FL/+;EIIA-Cre+ mice was low compared to the other three genotype control mice (Figure 1C). Interestingly, some phenotype variability was noted among the Col1a1^FL/+;EIIA-Cre+ male mice, and 1 out of 6 males was not different in body weight or other skeletal parameters compared to other control groups (discussed further below).

Skeletal characterization of the mice at 9 wk of age

After euthanasia, we performed an in-depth skeletal phenotyping that included X-ray imaging, DXA, micro-CT (microCT), 3 point-bending of the femur, and histology. Whole-body X-ray imaging confirmed the reduced size of the Col1a1^FL/+;EIIA-Cre+ mice and also revealed important skeletal alterations, including thoracic scoliosis and bone fractures, in particular affecting the pelvis (Figure 2A and Table 1). These were not present in control mice. DXA analysis showed a consistent and highly significant reduction in BMD and BMC in Col1a1^FL/+;EIIA-Cre+ male mice (5 out of 6) compared to control genotypes (n = 6-9, p < .0001). We showed this for the whole body, as well as for the lumbar spine and femur (Figure 3 and Figure S2). Female Col1a1^FL/+;EIIA-Cre+ mice showed a similar reduction in both BMD and BMC compared to their control groups (n = 3-8, Figure 3 and Figure S2). High-resolution microCT imaging of the femur (male mice) showed reduced length and a striking visible reduction in trabecular bone in Col1a1^FL/+;EIIA-Cre+ mice compared to control groups (Figure 2B). Consistent with this, the quantification of cancellous bone parameters at the distal femur of male mice showed a significant reduction in bone volume/tissue volume (BV/TV), trabecular number (Tb.N), connectivity density, and apparent bone density (n = 6-9, p < .0001 for most parameters) (Figure 4). These were accompanied by an increase in trabecular separation (Tb.Sp), while trabecular thickness (Tb.Th), structure model index (SMI), and material density were not different among the four genotypes (Figure 4 and data not shown). The analysis of cortical bone parameters at the femur mid-shaft showed a significant reduction in cortical thickness and estimated moment of inertia (n = 6-9, p = .0013 and p = .0046, respectively) in Col1a1^FL/+;EIIA-Cre+ mice compared to control groups (Figure 5). Apparent density was not different with some minor differences observed in material density but only among some of the genotypes (Figure 5). Very similar, if not identical results, were observed from the study of female distal and mid-shaft femurs, although some parameters did not reach statistical significance, most likely due to the smaller sample set that was analyzed (n = 3-5, Figures S3 and S4). Consistent with the reduced cortical thickness and moment of inertia, a biomechanical assessment of the femur from male mice using a 3-point bending test showed a trend towards reduced stiffness, yield force, and ultimate load in Col1a1^FL/+;EIIA-Cre+ mice compared to the other genotypes (n = 6-8). However, a statistically significant difference among these extrinsic bone parameters was not detected in all comparisons, again likely due to the variability of some of these parameters and the need to test a higher number of samples (Figure 6A). No differences were observed in the intrinsic bone parameters, including Young’s modulus, yield stress, and ultimate stress (Figure 6B). Female bones were not tested for biomechanics.

Figure 2.

X-ray and microCT imaging. (A) Representative X-ray imaging of male mice from the four different genotypes in prone position at 9 wk of age. The far-right panel is a blow-up of the mouse with the recombined allele and shows reduced size in addition to clear skeletal abnormalities and fractures, as labeled by arrows. (B) High-resolution microCT 3D renderings of mouse femurs from the four genotypes. Note how the femur from the mouse expressing the OI genetic variant is shorter and has drastically reduced trabecular bone.

Table 1.

Table reporting kyphoscoliosis and fractures (fx) in male and female mice at 9 wk of age.

	Sex	Kyphoscoliosis	Long bone fx	Pelvic fx
Col1a1FL/+;EIIA-Cre+ #6642	M	Severe	Humerus	Bilateral
Col1a1FL/+;EIIA-Cre+ #6668	M	Severe	Humerus	Bilateral
Col1a1FL/+;EIIA-Cre+ #6768	M	Severe	-	Mono-lateral
Col1a1FL/+;EIIA-Cre+ #6798	M	Moderate	-	Mono-lateral
Col1a1FL/+;EIIA-Cre+ #6942	M	Mild	Femur	-
Col1a1FL/+;EIIA-Cre+ #6907	M	Moderate	-	Mono-lateral
Col1a1FL/+;EIIA-Cre+ #6809	F	-	-	-
Col1a1FL/+;EIIA-Cre+ #6811	F	Mild	-	Mono-lateral
Col1a1+/+ n = 5	M	-	-	-
Col1a1+/+ n = 6	F	1 mild scoliosis	-	-
Col1a1FL/+ n = 2	M	-	-	-
Col1a1FL/+ n = 1	F	-	-	-

Figure 3.

DEXA analysis. BMD calculated from the whole body, the spine, and the femur in male (left panels) and female (right panels) mice at 9 wk of age (n = 6-9 and n = 3-8—one-way ANOVA).

Figure 4.

MicroCT analysis of cancellous bone at the distal femur. Bone volume/total volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), connectivity density (conn-density), and apparent density in male mice at 9 wk of age (n = 6-9—one-way ANOVA). p-values that are statistically different are reported as such and in red fonts.

Figure 5.

MicroCT analysis of cortical bone at the mid-shaft femur. (A) Cortical thickness (Ct.Th), polar moment of inertia, material density, and apparent density in male mice at 9 wk of age (n = 6-9—one-way ANOVA). p-values that are statistically different are reported in bold red fonts. (B) Representative 3D rendering of the bone cortical cross sections from the four genotypes.

Figure 6.

Biomechanic assessment of the femur. (A) Extrinsic and intrinsic (B) parameters calculated from the 3 pt bending test of the femur in male mice at 9 wk of age (n = 6-8—one-way ANOVA). p-values that are statistically different are reported in red fonts.

Histology of the spine and lungs and calculation of the lung MLI

Although we did not perform microCT scanning of the spine, we analyzed the lumbar spine vertebrae by histology. Un-decalcified mid-coronal sections of lumbar vertebrae at 2 months of age were stained with H&E and, similar to our findings in the femur, showed a striking reduction of both cortical and cancellous bone (n = 2-5, male mice) (Figure 7A). This is consistent with the severe low bone mass, typical of OI. Moreover, we and others have recently shown that several OI mouse models have primary lung defects with clear enlargement of the alveolar structures, resembling an emphysema-like phenotype.^25–28 To determine if this phenotype was also reproduced in our new mouse model, we collected lungs at 2 mo of age and performed histology followed by a quantification of the alveolar defect. Our analysis showed a lung parenchyma defect in Col1a1^FL/+; EIIA-Cre+ mice compared to the other genotypes (n = 3-5), which we quantified using the MLI method. Although with some variability (note that 1 recombinant mouse was not different from controls), MLI values in Col1a1^FL/+; EIIA-Cre+ mice were elevated (p-values between .065 and .075), which is in accordance to what we previously reported in other mouse models of OI.^25,26 As a consequence of the reduction in alveolar structures, the calculated internal lung surface area was significantly reduced in Col1a1^FL/+;EIIA-Cre+ mice (Figure 7B and C). Similar elevation in MLI values was seen in the lungs of female mice (Figure S5A).

Figure 7.

Histology of the spine and lung. (A) Representative images of H&E stained, mid-coronal sections of lumbar vertebrae from the four mouse genotypes at 9 wk of age (scale bar = 1 mm). Average trabecular BV/TV and sample size for each genotype are reported below each panel. (B) H&E staining of lung sections from the four mouse genotypes at 9 wk of age, shown at two different magnifications (5×—scale bar = 500 μm, left panels; and 20×—scale bar = 100 μm, right panels). (C) Representation of the calculated MLI and the internal lung surface area (ISA) from the lung sections of 9-wk-old male mice (n = 3-5).

Discussion

We have generated the first conditional knock-in OI allele for the Col1a1 gene and provided evidence that it is functional in the mouse model. In fact, upon in vivo Cre-recombination and expression of the p.Gly1146Arg glycine substitution in all murine tissues, we have confirmed that these mice display the typical features of OI. Because the substituted glycine is located towards the C-terminal end of the collagen triple-helical region, it was expected to cause a severe form of OI, similar to what has been observed in patients with this type of variant.²⁹ The selection of a severe mutation was done deliberately with the ultimate, future intent to express it in a tissue-restricted fashion and have the best opportunity to observe a local phenotype that we can study and possibly reproducing non-skeletal defects seen in patients with OI. Notwithstanding some in-utero/perinatal lethality, when expressed in all tissues, the p.Gly1146Arg substitution was compatible with life, and we were able to harvest and study several mice at 9 wk of age and from both sexes. These mice faithfully reproduced the skeletal features of a severe form of OI (OI type 3), including a significantly reduced body size, moderate to severe scoliosis, and spontaneous fractures as seen by X-ray imaging. DXA analysis confirmed the low BMD and BMC, and microCT scanning of the femurs clearly showed a dramatic reduction in trabecular bone and thinning of cortical bone. Cancellous bone trabecular number but not thickness was reduced, which is consistent with the known insufficient and defective bone formation by osteoblasts that is typical of OI during growth. Although the biomechanical test on the femurs did not always provide statistically significant differences among all genotypes, this was expected given the low sample number and, at least in our experience, the frequent variability of some of the calculated parameters that often requires the testing of 10-15 samples. Nonetheless, the biomechanic results together with the reduced polar moment of inertia suggest that the femurs from the Col1a1^FL/+;EIIA-Cre+ mice are weaker and more prone to fractures. Indeed, fractures were observed by X-ray and rather consistently in the pelvis (Table 1). It is interesting to note that 1 out of the 6 Col1a1^FL/+;EIIA-Cre+ male mice that we analyzed, hardly showed any phenotype, and most if not all of its parameters were not different from controls. We verified the correct genotyping of that mouse multiple times, and the presence of the Cre-recombined allele (and consequent absence of the floxed allele) was always detected in the genomic DNA prepared from a tail snip, liver, and spleen. We do not currently have an explanation for this observation other than the mouse could perhaps have been mosaic for the Cre-recombination event in the skeleton. However, this is reminiscent of the well-known inter- and intra-familial phenotype variability, even between subjects carrying the same identical OI-causing variant.^14,29,30 The variable penetrance of disease manifestations in connective tissue disorders is still poorly understood and in need of further research. While in a genetically outbred human population the presence of modifier genes has been postulated, none have been identified so far, with rare examples in the mouse model.³¹ Instead, our mouse colony was maintained in a homogeneous C57B6 genetic background, which makes it more difficult to explain the presence of an outlier. However, we have encountered similar differences in severity when studying the lung phenotype both within and between different mouse models of OI.²⁶

Limitations of our study

A limitation of our study is the fact that the floxed Col1a1 allele may not transcribe type I collagen at normal levels. In fact, when we crossed two mice that were heterozygous carriers, we could not retrieve any mice that were homozygous for the floxed allele (0 out of 32 pups generated—expected 8). This suggests that the floxed allele may be hypomorphic and express reduced levels of normal Col1a1 so that it is not compatible with life in homozygosity. We do know that loss of Col1a1 expression in mice is lethal during embryogenesis,^32–34 and it is therefore likely that a hypomorphic allele at the homozygous state may not transcribe enough Col1a1 transcript for survival. To address this issue, we performed QPCR to detect Col1a1 transcripts from skin, lung, tail, and tibia of both WT and Col1a1^FL/+ mice (n = 3-5). As expected, quantities of Col1a1 transcripts varied among mice and between different tissues, but we could not identify any significant differences between the two genotypes (Figure S5B). Therefore, we do not understand why homozygosity for the floxed allele appears incompatible with life; however, a potential reduction in Col1a1 protein did not translate into a detectable skeletal or lung phenotype in the heterozygous state.

Conclusions and future studies

We have successfully generated the first conditional knock-in allele for Col1a1 that can express a severe OI glycine substitution in the tissue of choice. Ongoing work in our laboratory aims at dissecting the relative contribution of skeletal vs lung defects to the respiratory issues in OI, by expressing, for instance, the p.Gly1146Arg mutation in the lung only and studying the resulting respiratory phenotype without the interference of skeletal chest deformities. Moreover, we are interested in expressing the glycine substitution in osteocytes only to determine if these cells contribute to some of the OI manifestations, either during skeletal development or homeostasis, as some of our data previously suggested.³⁵ There is also muscle weakness in OI, and there has been a lingering question in the field on whether the proper crosstalk between bone and muscle is affected in OI due to the skeleton being dramatically impacted. Having the opportunity to express a severe Col1a1 variant only in skeletal muscle (or in cells within the skeletal muscle that produce type I collagen) and not in bone can provide important insights into this issue and help to clarify whether the muscle weakness is secondary to the bone fragility or rather constitutes a primary defect of the muscle in OI. This would also give us the opportunity to better learn the role of type I collagen in muscle physiology and function. Additionally, other opportunities to study the tissue-specific effects of this severe variant lie ahead and will hopefully help in the identification of new cell-specific targets for therapeutic intervention that will positively impact the lives of patients with OI.

Author contributions

Milena Dimori (Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing—review & editing), Mahtab Toulany (Data curation, Formal analysis, Visualization, Writing—review & editing), Samia Sultana Lira (Investigation, Methodology, Writing—review & editing), Melda Onal (Formal analysis, Investigation, Methodology, Writing—review & editing), Jeff Thostenson (Formal analysis, Methodology, Writing—review & editing), John Carroll (Formal analysis, Investigation, Writing—review & editing), Charles O’Brien (Conceptualization, Formal analysis, Investigation, Methodology, Writing—review & editing), and Roy Morello (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing)

Funding

This work was supported by a 2022-24 Innovation Award from the American Lung Association (R.M.) and by an National Institute of Health (NIH) R01 HL166748 (R.M.). Partial support also came from P20 GM125503 from NIGMS, and U54 TR001629 from NCATS.

Conflicts of interest

All authors have nothing to disclose.

Data availability

The data underlying this article are available in the article and in its online supplementary material. The new mouse model will be made available to the scientific community upon proper MTA.