Deep intronic mutation in CRTAP results in unstable isoforms of the protein to induce type I collagen aggregation in a lethal type of osteogenesis imperfecta type VII

Abstract

Genetic mutations are involved in Mendelian disorders. Unbuffered intronic mutations in gene variants can generate aberrant splice sites in mutant transcripts, resulting in mutant isoforms of proteins with modulated expression, stability, and function in diseased cells. Here, we identify a deep intronic variant, c.794_1403A>G, in CRTAP by genome sequencing of a male fetus with osteogenesis imperfecta (OI) type VII. The mutation introduces cryptic splice sites in intron-3 of CRTAP, resulting in two mature mutant transcripts with cryptic exons. While transcript-1 translates to a truncated isoform (277 amino acids) with thirteen C-terminal non-wild-type amino acids, transcript-2 translates to a wild-type protein sequence, except that this isoform contains an inframe fusion of non-wild-type twenty-five amino acids in a tetratricopeptide repeat sequence. Both mutant isoforms of CRTAP are unstable due to the presence of a unique ‘GWxxI’ degron, which finally leads to loss of proline hydroxylation and aggregation of type I collagen. Although type I collagen aggregates undergo autophagy, the overall proteotoxicity resulted in death of the proband cells by senescence. In summary, we present a genetic disease pathomechanism by linking a novel deep intronic mutation in CRTAP to unstable mutant isoforms of the protein in lethal OI type VII.

1. Introduction

Osteogenesis imperfecta (OI), also known as ‘brittle bone disease,’ is a group of genetically heterogeneous skeletal dysplasias characterized by soft and fragile bones [1]. Individuals affected by OI exhibit a loss of bone mineral density (osteopenia) that leads to bone fragility and fractures [2]. Other features of OI include Wormian bones, blue sclera, scoliosis, bending of the long bones, and developmental delays [3]. The ‘Sillence Classification’ divides OI into four types (type I-IV), describing OI conditions as mild, lethal, severely deforming, and mildly deforming [4]. While the majority of OI cases are caused by mutations in pro-α1(I) and pro-α2(I) (component proteins of type I collagen) [5,6], a spectrum of mutations in 20 other genes also results in OI [7,8]. Together, mutations in 22 different genes form the basis for the ‘genetic classification’ of OI [9]. In addition to COL1A1 and COL1A2, several protein quality control genes (e.g., P3H1, PPIB, CRTAP, FKBP10, SERPINH1, MESD, etc.), signaling proteins (e.g., BMP1, WNT1), receptors (KDELR2, MBTPS2), and transcription factors (e.g., SP7, CREB3L1) regulate bone formation in humans, and mutations in these genes cause OI. Mechanistically, collagenopathy in OI is primarily the result of a failure of proteostasis in the proper expression, localization, and function of type I collagen [10].

Type VII is a specific form of OI caused by autosomal recessive mutations in CRTAP [11]. Type VII OI is usually fatal or severe due to bone hypo-mineralization, resulting in multiple rib fractures, deformities fractures of the long bones, lack of diaphyseal modeling, and progressive physiological deformities [12–17]. Although not very common, one individual affected by OI type VII also exhibited dentiogenesis imperfecta and white to light blue sclera [15]. OI type VII is reported to cause decreased hydroxylation of proline of pro-α1(I) and pro-α2(I) of type I collagen [18], resulting in lower secretion of type I collagen into the extracellular matrix [14].

CRTAP is a scaffold protein that holds two other proteins, P3H1 and PPIB, in a heterotrimeric complex involved in post-translational modification (PTM) of the pro-α1(I) and pro-α2(I) chains of type I collagen [18,19]. While P3H1 catalyzes the prolyl-3-hydroxylation of proline-986 of pro-α1(I) and pro-α2(I) [20], PPIB is involved in the cis-trans isomerization of the proline of pro-α1(I) and pro-α2(I) [21]. In addition to its function as a PTM complex of collagen, the CRTAP-P3H1-PPIB complex also functions as a collagen chaperone [22,23]. Moreover, this complex is also reported to function as a disulfide isomerase (with oxidation-reduction activity) because of the presence of a ‘CxxxC’ motif in both CRTAP and P3H1 [24]. Although the complex is located in the endoplasmic reticulum (ER) [18], CRTAP is also secreted into the extracellular matrix [23,25]. Interestingly, CRTAP shares sequence and structural homology with the N-terminal region of P3H1. CRTAP and P3H1 stabilize each other, and loss of either protein results in impaired collagen fibrillation and secretion [23]. CRTAP has an N-terminal signal peptide followed by a series of tetratricopeptide (TPR) repeats. CRTAP is also subject to glycosylation [26], although the effects of glycosylation on CRTAP stability and function are unknown.

OI type VII-related CRTAP mutations are heterogeneous [13–17]. The CRTAP mutations in different probands occur in both intron and exon regions of the gene and result in frameshift, splice site, stop-gain, start-loss, single nucleotide variations, and small and large deletion variants in the transcript [13–17]. Some of the mutant CRTAP transcripts, e.g., those with start-loss [13] and stop-gain [13,27] mutations, produce very little protein product due to the lack of translation initiation [13] and nonsense-mediated decay of the transcripts [11,13]. Other mutant CRTAP transcripts, such as those with frameshift, splicesite, and deletion, generate mutant CRTAP isoforms that are intrinsically unstable and rapidly degraded. However, the precise sequence signatures and molecular mechanisms of degradation of mutant CRTAP isoforms with respect to induction and progression of OI type VII is not well understood.

In this study, we present a clinical case of OI type VII associated with a deep intronic mutation in CRTAP. The mutation generates two alternative transcripts of CRTAP that eventually result in two mutant isoforms that are unstable and rapidly degraded due to the presence of a unique and noncanonical degron in the protein. Loss of CRTAP in proband cells results in decreased hydroxylation of proline of pro-α1(I), leading to aggregation of type I collagen. Although the type I collagen aggregates are degraded by autophagy, the overall proteotoxicity leads to senescent death of the proband cells.

2. Results

2.1. A deep intronic mutation, c.794-1403A>G, in CRTAP results in osteogenesis imperfecta type VII

Clinical report: A 20-week-old male fetus from a third-degree consanguineous couple (Fig. 1A) showed bone developmental abnormalities. The fetus was medically aborted. At autopsy, the fetus’ height, weight, and head circumference were 19 cm (-3 SD), 280 g, and 17 cm, respectively. Clinically, the fetus had short bowed lower limbs, protruding thoracic cage, depressed nasal bridge with anteverted nares, and long philtrum (Fig. 1B). Radiological findings of the fetus also showed bowing of the bones (Fig. 1B). In addition to the bone developmental defects resembling osteogenesis imperfecta (OI), the fetus had a narrow thorax and soft calvarium, whereas the visceral organs of the brain, thoracic, and abdominal visceral organs were normal (Fig. 1B).

Fig. 1. Genome sequencing identified a c.794-1403A>G variant in the deep intronic-3 region of the CRTAP gene of a medically aborted male fetus with osteogenesis imperfecta.

(A) Three generation pedigree of the proband. (B) Clinical photographs and radiographs of the proband. (C) Sanger sequencing and segregation analysis show homozygous state of c.794-1403A>G variant in the proband and carrier status in his parents. (D) Schematic representation of the wild-type and mutant CRTAP gene, showing wild-type and variant nucleotides in the wild-type and mutant gene.

Exome sequencing of the proband revealed no mutations in the exons of the known OI genes. However, whole genome sequencing identified a novel and homozygous variant, c.794-1403A>G, in intron-3 of CRTAP (NM_006371.4) (Fig. 1C, D). This variant was not reported in the control population databases. Sanger sequencing confirmed the homozygous state of the mutation in the proband (Fig. 1C) and also confirmed that the proband’s parents were carriers of the mutation (Fig. 1C). In silico splicing prediction algorithms, such as ‘Splice AI’ and ‘Human Splicing Finder’ predicted the variant as a splice site donor (i.e., the variation being ‘splice donor gain’ or ‘new donor splice site’). The variant was asserted with uncertain significance according to the ACMG criteria.

Based on the clinical features and genome sequencing results, the fetus’ condition was diagnosed as OI type VII.

2.2. c.794-1403A>G mutation generates cryptic splice sites leading to two mutant transcripts of CRTAP

Because the c.794-1403A>G mutation in intron-3 of CRTAP is predicted to result in a splice site variant, we examined whether the mutation resulted in one or more mutant transcripts in the fibroblasts of the proband. A primer set (F1/R1) defined for PCR amplification of the region between exon-2 and exon-6 of the mature transcripts/cDNAs of CRTAP (Fig. 2A), resulted in a single band of higher DNA size (bp) from the proband cell cDNA pool than the band from the control cell cDNA pool in the 1D-1 % agarose electrophoresis (Fig. 2A), indicating cryptic exon retention in the CRTAP transcripts of the proband cells. This result also clearly indicated that the mutation generated splice site variant(s) of CRTAP, particularly by introducing a possible splice site donor in intron-3. Although the PCR-amplified dsDNA products from the proband cells showed a single band in 1D-1 % agarose electrophoresis, the trace peaks in the Sanger sequencing spectrum of the PCR products showed a primary and secondary peak at each nucleotide position from the mutation (Fig. 2A), indicating the presence of two CRTAP transcripts of nearly equal size in the proband cells. To individually assort the mutant CRTAP transcripts of the proband cells, the dsDNA PCR products previously obtained with F1/R1 primers were cloned into the TOPO vector using TOPO TA cloning. Screening of at least 10 individual colonies harboring a transformed clone revealed the existence of two distinct mutant CRTAP clones, designated as transcript-1 and transcript-2 (Fig. 2B). Transcript-1 contained a cryptic exon of 68 nucleotides of a contiguous sequence from intron-3 (Fig. 2C). Transcript-1 arose from two alternative splicing events within intron-3 with the following splice site donors and acceptors: splice site donor-1: c.793+1, splice site acceptor-1: c794-1472; splice site donor-2: c.794-1404, splice site acceptor-2: c794-1. Transcript-2 also contained a cryptic exon of 75 nucleotides derived from two separate contiguous sequences (7 nt + 68 nt) from intron-3 (Fig. 2C). Similar to transcript-1, transcript-2 was also generated by two alternative splicing events within intron-3 with the following splice site donors and acceptors: splice site donor-1: c.793+8, splice site acceptor-1: c794-1472; splice site donor-2: c.794-1404, splice site acceptor-2: c794-1.

Fig. 2. Mutant CRTAP results in two alternatively spliced mutant transcripts arising due to the mutation.

(A) Top: representative diagram showing mutation harboring region amplified by PCR using forward [F1] and reverse [R1] primers. Bottom left: representative gel electrophoresis image of the amplicons derived from the mature CRTAP transcripts obtained from control and proband fibroblasts, showing an upper shift [higher size] of the amplicon derived from the mutant CRTAP transcripts compared to the amplicon generated from the wild-type CRTAP transcripts of control cells. Bottom right: Sanger sequencing showing heterozygous (double) peaks in the mutant CRTAP transcripts, whereas wild-type CRTAP transcript shows clean chromatogram. (D) Left: representative gel electrophoresis images of the amplicons derived from two different mutant transcripts of CRTAP obtained from individual colonies of Topo TA cloning. Sequences were validated by Sanger sequencing. Right: representative diagrams of the two mutant CRTAP transcripts, showing transcript-1 and transcript-2 retained 68 nt and 75 nt cryptic exons (black regions). Splice site donors and acceptors are indicated. (D) Sequences of the regions of mature mutant CRTAP transcripts showing cryptic exon sequences and the different primers used for determining the quantity of the two mutant CRTAP transcripts in the proband fibroblasts. Sequences in – red nucleotides: 68 nt cryptic exon in transcript-1, red+orange nucleotides: 75 nt cryptic exon in transcript-2, blue highlighted nucleotides: forward [P1] and reverse [P2] primers to amplify cDNAs of both transcript-1 and transcript-2, grey highlighted nucleotides: forward [P3] and reverse [P4] primers to amplify cDNAs of transcript-2. (E) Representative gel electrophoresis image of the reverse-transcription-PCR amplicons derived from the wild-type and mutant CRTAP transcripts using P1, P2, P3 and P4 primers. Amplicons of GAPDH are controls for relative quantification. (F) q-PCR quantification of the level of the two mutant transcripts of CRTAP in proband fibroblasts. (G) q-PCR quantification of the level of the two mutant transcripts of CRTAP in proband cells that are either untreated or treated with 10 μM cycloheximide for 24 h. O1 and O2 primers are designed on the sequence of exon-2 of mutant transcripts; P1/P2 and P3/P4 primers are same as described in 2E. (H) Up: the 68 nt sequence present in the cryptic exon of both mutant transcript-1 and transcript-2 of CRTAP. The yellow highlighted region is the putative AUF1 binding ARE. Bottom: mFold-predicted secondary structure of the 68 nt sequence present in the cryptic exon of both mutant transcript-1 and transcript-2 of CRTAP. (I) q-PCR quantification of the level of the two mutant transcripts of CRTAP in proband cells that are transiently transfected with control-siRNA or AUF1-siRNA for 24 h. O1/O2, P1/P2 and P3/P4 primers are same as described in 2G.

Quantifications are shown as mean ± S.D.; P values are indicated. GAPDH is the loading control in the agarose gel electrophoresis and qPCR. Electrophoresis data are representative of at least three independent experiments.

Next, we examined the relative abundance of the two mutant CRTAP transcripts in proband cells. For this purpose, we selected two primer sets for PCR amplification of two regions of the CRTAP transcript (Fig. 2D). The first primer set (P1/P2) was designed to anneal to CRTAP transcript regions to generate PCR products from the wild-type and mutant CRTAP transcripts of control and proband fibroblasts (Fig. 2D). This primer set was also selected for PCR amplification of the region containing the cryptic exons in both of the mutant CRTAP transcripts of the proband cells. For this reason, the PCR products of the mutant CRTAP transcripts had a higher band size (bp) compared with the PCR product of the wild-type CRTAP transcript. The second primer set (P3/P4) was designed so that the forward primer (P3) of this set can only anneal to the mutant transcript-2 of CRTAP (Fig. 2D). The seven nucleotides at the 3′ end of P3 (5′....GTTGGTG-3′) are the part of the intron-3 sequence retained only in the cryptic exon of mutant transcript-2 of CRTAP. Therefore, the P3 could not hybridize with the wild-type transcript and mutant transcript-1 of CRTAP, resulting in no PCR products from these transcripts when the P3/P4 primer set was used. Therefore, a quantitative PCR (qPCR) approach using the two primer sets for the mutant CRTAP transcripts from the proband would result in a separate estimate of the abundance of transcript1 + transcript-2 and transcript-2 and eventually yield the relative abundance of mutant transcript-1 and transcript-2. Although we observed a lower abundance of mutant transcript-1 compared with abundance of mutant transcript-2 (Fig. 2E, F), the difference in abundance of transcript-1 and transcript-2 in proband cells was not significant. Interestingly, the cumulative abundance of mutant transcript-1 + transcript-2 of CRTAP in proband cells was significantly lower than that of wild-type CRTAP transcript in control cells (Fig. 2E), suggesting that the mutant CRTAP transcripts in proband cells might be degraded by specific mechanisms.

The relatively lower amount of mutant transcript-1 + transcript-2 of CRTAP in proband cells compared with wild-type CRTAP transcripts in control cells indicated that the mutant CRTAP transcripts were rapidly degraded in proband cells. Therefore, we checked the likely mechanisms of decay of the mutant CRTAP mRNAs in proband fibroblasts. Because retention of the 68-nucleotide cryptic exon resulted in a frameshift and generation of a premature termination codon (PTC) in mutant transcript-1 of CRTAP, this mutant transcript-1 of CRTAP might be subject to nonsense-mediated degradation (NMD). Therefore, we checked the possibility of NMD of the mutant transcript-1 and transcript-2 of CRTAP. qPCR examination of mutant CRTAP transcripts from untreated and NMD inhibitor (10 μM cycloheximide)-treated proband fibroblasts with a primer set (O1/O2-designed on the sequence of exon-2, detecting both transcript-1 and transcript-2) showed no significant change in the cumulative level of transcript-1 + transcript-2 in cycloheximide-treated proband cells compared with untreated proband cells (Fig. 2G). In addition, we checked the relative individual levels of transcript-1 and transcript-2 using primer sets P1/P2 and P3/P4 (according to the procedure described in the previous section) in the untreated and cycloheximide-treated proband fibroblasts. Again, no significant difference was observed between the levels of transcript-1 and transcript-2 in cycloheximide-treated proband cells and untreated proband cells (Fig. 2G). These observations indicate that neither mutant CRTAP transcript was degraded by NMD in the proband cells.

The retained cryptic exon in the mature mutant CRTAP transcripts had putative AU-rich elements (ARE) (Fig. 2H). The 68-nucleotide sequence of the cryptic exon present in both mutant transcript-1 and transcript-2 of CRTAP was not only predicted to maintain a secondary structure (Fig. 2H), but the stem of this structure also showed a putative AUF1-binding ARE (5’-AUUUU-3′) (Fig. 2H). Because AUF1 binding to AREs can destabilize the corresponding mRNAs, we tested whether AUF1 played a role in the rapid degradation of mutant CRTAP transcripts. Suppression of AUF1 expression by AUF1-siRNA resulted in a significant increase in the levels of transcript-1 and transcript-2 in the proband cells compared with the levels of transcript-1 and transcript-2 in the proband cells treated with control-siRNA (Fig. 2I). This result suggested that the rapid degradation of the mutant CRTAP transcripts was due to the destabilizing effect of AUF1, possibly by binding of AUF1 to the ARE of the cryptic exon of the mutant CRTAP transcripts.

2.3. Mutant isoforms of CRTAP are unstable due to the presence of unique degron sequence

In the next phase, we investigated the stability and expression of the mutant isoforms of CRTAP in proband cells. Because of a premature termination codon, the open reading frame (ORF) of mutant transcript-1 would result in a 277 amino acid mutant isoform of CRTAP (CRTAP-Mut1) in which the thirteen C-terminal amino acids are a non-wild-type sequence (Mut1Seq) (Fig. 3A). Mutant transcript-2 would result in a wild-type protein, except that this mutant isoform (CRTAP-Mut2) contains an in-frame fusion of 25 non-wild-type amino acids (Mut2Seq) after the 264th residue (A264) (Fig. 3A). Although both mutant transcripts were theoretically translated into two mutant isoforms with different molecular weights (Mw of CRTAP-Mut1: 32 kDa, Mw of CRTAP-Mut2: 49 kDa), we could not detect either of these mutant isoforms of CRTAP in the fibroblasts of the proband (Fig. 3B) under physiological conditions. This observation suggested that either the mutant CRTAP transcripts were not translated or the translated mutant CRTAP isoforms were rapidly degraded. Inhibition of proteasomal degradation of proteins in proband fibroblasts by MG132 resulted in immunoblot detection of both mutant CRTAP isoforms in fibroblasts (Fig. 3B), indicating that the mutant CRTAP isoforms were subject to rapid proteasomal degradation in proband fibroblasts. Moreover, ectopic expression of FLAG-tagged CRTAP-Mut1 or CRTAP-Mut2 in proband fibroblasts also showed no detectable expression of these clones (Fig. 3C, D), confirming that the mutant isoforms of CRTAP were rapidly degraded in cells.

Fig. 3. The two alternatively spliced mutant transcripts of CRTAP translate to two mutant isoforms of CRTAP that are rapidly degraded due to presence of degron in the mutant sequences of the two mutant isoforms of CRTAP.

(A) Representative diagram of the wild-type and two mutant isoforms of CRTAP. The non-wild-type sequences [Mut1Seq and Mut2Seq] in the two mutant isoforms are shown in black. (B) Left: representative immunoblot of CRTAP from whole cell lysate of control and proband fibroblasts. Middle: control and proband fibroblasts were treated with 10 μM MG132 for 24 h. Representative immunoblot of CRTAP from whole cell lysate of MG132-treated control and proband fibroblasts. Right: densitometric quantification of the level of cellular CRTAP relative to ACTB of the immunoblot of 3B. (C) Up: control and proband fibroblasts were immunostained for CRTAP. Representative confocal microscopy images showing the expression of CRTAP in control fibroblasts, but no expression in proband fibroblasts. Bottom: FLAG nucleotide-tagged DNA constructs of mutant isoforms of CRTAP [CRTAP-Mut1 and CRTAP-Mut2] were transfected to proband fibroblasts. Transfected proband fibroblasts were immunostained for FLAG. Representative confocal microscopy images showing no expression of FLAG-CRTAP-Mut1 and FLAG-CRTAP-Mut2 in proband fibroblasts. (D) Percentage of cells showing expression of CRTAP, FLAG-CRTAP-Mut1, and FLAG-CRTAP-Mut2 in control and proband fibroblasts as described in 3C [100 cells/group]. (E) Alignment of non-wild-type sequences of the CRTAP-Mut1 and CRTAP-Mut2. Potential degron is highlighted in yellow and G, W and I residues in the degron are marked red. (F) Schematic diagram of the empty EGFP-mCherry construct, and EGFP-mCherry-Mut1Seq and EGFP-Mut2Seq-mCherry constructs. (G) HeLa cells were transfected separately with empty EGFP-mCherry construct, and EGFP-mCherry-Mut1Seq and EGFP-Mut2Seq-mCherry constructs. Representative confocal microscopy images showing expression of EGFP-mCherry fusion protein in the empty EGFP-mCherry construct transfected HeLa cells, but no expression of EGFP-mCherry-Mut1Seq and EGFP-Mut2Seq-mCherry proteins in the EGFP-mCherry-Mut1Seq and EGFP-Mut2Seq-mCherry constructs’ transfected HeLa cells. (H) Percentage of cells showing expression of EGFP-mCherry, EGFP-mCherry-Mut1Seq and EGFP-Mut2Seq-mCherry in HeLa cells as described in 3G [100 cells/group]. (I) Proband fibroblasts were transfected with FLAG-nucleotide tagged DNA constructs of different deletion and alanine-scanning degron mutants of CRTAP-Mut1 and CRTAP-Mut2 as described in the result. Representative confocal microscopy images of the expression of FLAG-tagged deletion and alanine scanning mutants of CRTAP-Mut1 and CRTAP-Mut2. (J) Percentage of cells showing expression of FLAG-tagged mutants of CRTAP-Mut1 and CRTAP-Mut2 in proband fibroblasts as described in 3I [100 cells/group].

Quantifications are shown as mean ± S.D.; P values are indicated. ACTB is the loading control in the immunoblots. Scale bar in confocal microscopy images: 5 μm. Microscopy and immunoblot data are representative of at least three independent experiments.

Mutant proteins can undergo rapid degradation in the cell due to presence of degron signals that are not present in the wild-type proteins. Therefore, we investigated whether the mutant isoforms of CRTAP contain degron sequence that triggers their rapid proteasomal degradation in cells. Because both mutant isoforms of CRTAP contained non-wild-type amino acid sequences, we examined whether these non-wild-type sequences in the mutant contained a possible degron. Interestingly, we observed a “GWxxI’ motif in both mutant isoforms of CRTAP (Fig. 3E). In order to understand if the ‘GWxxI’ motif could function as a degron signal in the protein, we generated a series of bicistronic constructs of enhanced green fluorescent protein (EGFP) and mCherry with different positioning of the degron motif (Fig. 3F). The EGFP-mCherry control construct has a bicistronic arrangement of EGFP and mCherry with a flanking linker sequence (Fig. 3F). In this construct, EGFP did not contain a stop codon and mCherry contained a C-terminal stop codon, which allowed simultaneous expression of EGFP and mCherry. In the EGFP-mCherry-Mut1Seq construct (Fig. 3F), the mCherry did not contain a stop codon, but the Mut1Seq was located immediately downstream of the mCherry, followed by a stop codon. The EGFP-mCherry-Mut2Seq construct (Fig. 3F) resembled the EGFP-mCherry control construct, except that the linker region of the control construct was replaced by the Mut2Seq. The different positions of Mut1Seq and Mut2Seq were chosen to mimic the C-terminal and internal degron position in the constructs similar to the degron positioning in the mutant CRTAP isoforms. Whereas the control construct showed abundant expression of EGFP and mCherry in a significant number of transfected HeLa cells (Fig. 3G, H), in the HeLa cells transfected with the EGFP-mCherry-Mut1Seq and EGFP-mCherry-Mut1Seq constructs, we observed no detectable expression of EGFP and mCherry (Fig. 3G, H), suggesting that Mut1Seq and Mut2Seq contained a degron signal that led to degradation of the EGFP-mCherry protein.

Finally, we tested whether ‘GWxxI’ is the degron and whether the glycine, tryptophan, and isoleucine residues were required for the functionality of the degron. We generated several constructs of the FLAG-tagged CRTAP-Mut1 and CRTAP-Mut2 in which either the degron motif ‘GWxxI’ was deleted, or the glycine/tryptophan/isoleucine was mutated to alanine. Not surprisingly, all degron-deleted and alanine-scanning mutants showed clearly detectable expression of the constructs (Fig. 3I, J). This observation clarified that not only is the ‘GWxxI’ motif the degron in the mutant isoforms of CRTAP, but also showed that the presence of glycine, tryptophan, and isoleucine is required for the motif to function as an active degron.

2.4. Deficiency of CRTAP leads to aggregation of type I collagen in proband fibroblasts

CRTAP is the scaffolding protein that mediates the formation of a heterotrimeric complex of CRTAP, P3H1, and PPIB [28]. Although CRTAP is catalytically inactive and P3H1 and PPIB catalyze the hydroxylation and isomerization of the proline residues of pro-α1(I) and pro-α2(I) of type I collagen [29], loss of CRTAP would prevent the functional association of P3H1 and PPIB. Consequently, hydroxylation and isomerization of the proline residues of pro-α1(I) and pro-α2(I) would be inhibited, leading to possible conformational and structural deregulation of type I collagen. Therefore, we tested whether loss of CRTAP resulted in aberrant expression and localization of type I collagen in proband fibroblasts. We observed aggregates of type I collagen in the cytoplasm of proband fibroblasts (Fig. 4A), whereas non-aggregated type I collagen was localized in the endoplasmic reticulum (ER) and at the edges of control fibroblasts (Fig. 4A). Whereas loss of chaperone function of MESD was shown to lead to aggregation of type I collagen in our previous study, aggregation of type I collagen by loss of CRTAP, as observed in this study, has not been reported previously. To confirm that loss of CRTAP was indeed associated with aggregation of type I collagen, we knocked down CRTAP by CRTAP-siRNA in control fibroblasts. Whereas control fibroblasts treated with control-siRNA showed no structural or localization abnormalities of type I collagen (Fig. 4B-E), control fibroblasts treated with CRTAP-siRNA showed significant accumulation of cytosolic type I collagen aggregates like proband fibroblasts (Fig. 4B-E).

Fig. 4. CRTAP deficiency leads to type I collagen aggregation in proband fibroblasts.

(A) Control and proband fibroblasts were immunostained for pro-α1(I) and CANX. Representative confocal microscopy images showing the localization of pro-α1(I) in CANX-positive endoplasmic reticulum region of control fibroblasts, whereas aggregated pro-α1(I) [type I collagen] was scattered in cytoplasm of proband fibroblasts. (B) Control fibroblasts were transiently transfected with control-siRNA, CRTAP-siRNA, P3H1-siRNA, and PPIB-siRNA. siRNA transfected cells were immunostained for pro-α1(I) 48 h after transfection. Representative confocal immunofluorescence microscopy images showing cytosolic pro-α1(I) [type I collagen] aggregates in CRTAP-siRNA and P3H1-siRNA treated control fibroblasts. Control-siRNA and PPIB-siRNA treated control fibroblasts did not show aggregates of pro-α1(I) [type I collagen]. (C) Pearson’s correlation coefficient of colocalization of pro-α1(I) and CANX in control and proband fibroblasts as described in 4A and 4B [50 cells/group]. (D) Percentage of cells showing pro-α1(I) [type I collagen] aggregates in control and proband fibroblasts as described in 4A and 4B [100 cells/group]. (E) Control fibroblasts were transiently transfected with control-siRNA, CRTAP-siRNA, P3H1-siRNA, and PPIB-siRNA as described in 4B. Representative immunoblot of CRTAP, P3H1 and PPIB from lysate of control-siRNA, CRTAP-siRNA, P3H1-siRNA, and PPIB-siRNA treated control fibroblasts.

Quantifications are shown as mean ± S.D.; P values are indicated. TUBA is the loading control in the immunoblots. Scale bar in confocal microscopy images: 5 μm. Microscopy and immunoblot data are representative of at least three independent experiments.

Whereas loss of CRTAP triggered aggregation of type I collagen, loss of the heterotrimer of CRTAP, P3H1, and PPIB was thought to cause loss of spatial functionality of P3H1 and PPIB, and this phenomenon might be responsible for aggregation of type I collagen. To understand which of the two losses of function of P3H1 and PPIB caused aggregation of type I collagen, we knocked down P3H1 and PPIB separately in control fibroblasts. Whereas control fibroblasts treated with control-siRNA and PPIB-siRNA did not show type I collagen aggregation (Fig. 4B-E), control fibroblasts treated with P3H1-siRNA showed typical type I collagen aggregation similar to that observed in CRTAP-deficient proband fibroblasts (Fig. 4B-E). These results suggest that hydroxylation of pro986 is required for pro-α1(I) stability and that lack of hydroxylation of pro-α1 (I) prolines triggers aggregation of type I collagen because of loss of expression and spatially defined localization of P3H1. Therefore, we confirmed that the rapid degradation of mutant CRTAP isoforms in proband cells caused the loss of the CRTAP-P3H1-PPIB heterotrimer, resulting in a spatially functional loss of P3H1 in the hydroxylation of pro986 of pro-α1(I) and pro-α2(I). This in turn leads to aggregation of type I collagen and severe proteotoxicity in proband fibroblasts.

2.5. Loss of CRTAP leads to endoplasmic reticulum stress and senescence in proband fibroblasts

Loss of CRTAP stability and function led to deregulation of P3H1-mediated hydroxylation of prolines of pro-α1(I), resulting in aggregation of type I collagen in proband cells. Because the heterotrimeric complex of CRTAP, P3H1, and PPIB is an important post-translational modifier complex in the endoplasmic reticulum (ER) involved in proline hydroxylation and isomerization of substrate proteins, we speculated that the effects of loss of CRTAP might extend beyond collagen aggregation to global proteotoxic stress in the ER of proband cells. Indeed, we observed ER stress manifested by increased expression of ER stress-responsive chaperones such as HSPA5, PDIA2, and HSP90B in proband fibroblasts compared with control fibroblasts (Fig. 5A-C). In addition, we observed increased levels of proteostat-positive and ubiquitinated protein aggregates in proband fibroblasts compared with control cells (Fig. 5D-G). The presence of such protein aggregates in the proband cells further described the proteotoxic stress in the proband fibroblasts. To test whether the loss of CRTAP was associated with the induction of proteotoxicity in the cells, we knocked down CRTAP in control fibroblasts. Compared with control-siRNA-treated control fibroblasts, control fibroblasts treated with CRTAP-siRNA had quantitatively higher proteostat-positive and ubiquitinated protein aggregates (Fig. 5D-G), similar to proband fibroblasts. These results suggest that CRTAP, in association with P3H1 and PPIB, is required for the maintenance of ER proteostasis and its loss, as observed in proband fibroblasts, can lead to severe proteotoxicity in cells.

Fig. 5. Proteostasis imbalance leads to senescence-mediated death of proband fibroblasts.

(A) Relative quantification of the transcripts of endoplasmic reticulum-resident chaperones, HSPA5, PDIA2, and HSP90B, in control and proband fibroblasts. (B) Representative immunoblots of PDI, HSPA5, and HSP90B from the cell lysate of control and proband fibroblasts. (C) Densitometric quantification of level of HSPA5, PDI and HSP90B relative to TUBA of the immunoblot of 5B. (D) Up: control and proband fibroblasts were stained with Proteostat. Representative confocal fluorescence microscopy images showing high Proteostat-positive protein aggregates in proband fibroblasts. Bottom: control fibroblasts were transiently transfected with control-siRNA or CRTAP-siRNA for 48 h. Transfected cells were stained with Proteostat. Representative confocal fluorescence microscopy images showing CRTAP knockdown control fibroblasts showing higher number of Proteostat-positive protein aggregates. (E) Left: quantification of number of protein aggregates per cell in control and proband fibroblasts (100 cells/group). Right: quantification of number of protein aggregates per cell in control-siRNA and CRTAP-siRNA treated control fibroblasts (100 cells/group). (F) Up: control and proband fibroblasts were immunostained for ubiquitin. Representative confocal immunofluorescence microscopy images showing more ubiquitinated puncta in proband fibroblasts. Bottom: control fibroblasts were transiently transfected with control-siRNA or CRTAP-siRNA for 48 h. Transfected cells were immunostained for ubiquitin after 48 h. Representative confocal fluorescence microscopy images showing CRTAP knockdown control fibroblasts showing higher number of ubiquitinated protein aggregates. (G) Left: quantification of number of ubiquitin-positive puncta per cell in control and proband fibroblasts (100 cells/group). Right: quantification of number of ubiquitin-positive puncta per cell in control-siRNA and CRTAP-siRNA treated control fibroblasts (100 cells/group). (H) Precent of control and proband fibroblasts cell survival measured by MTT assay. (I) Control and proband fibroblasts were subjected to CellEvent Green staining. Representative confocal microscopy images showing higher number of CellEvent Green-positive senescent proband fibroblasts compared to control fibroblasts. (J) Representative immunoblot of p16^INK4a and p21^WAF1/Cip1 from lysate of control and proband fibroblasts. (K) Densitometric quantification of p16^INK4a and p21^WAF1/Cip1 level relative to TUBA of immunoblot of 5 J. (L-O) Control and proband fibroblasts were immunostained for p16^INK4a and p21^WAF1/Cip1. L and N: representative confocal immunofluorescence microscopy images of p16^INK4a and p21^WAF1/Cip1 in control and proband fibroblasts. M and O: representative confocal immunofluorescence microscopy images of p16^INK4a and p21^WAF1/Cip1 in control and proband fibroblasts. (H) Quantification of percentage of control and proband fibroblasts showing higher expression of p16^INK4a and p21^WAF1/Cip1 [100 cells/group].

Quantifications are shown as mean ± S.D.; P values are indicated. TUBA is the loading control in the immunoblots. Scale bar in confocal microscopy images: 5 μm. Microscopy and immunoblot data are representative of at least three independent experiments.

It is well known that failure of proteostasis triggers cellular senescence. Therefore, we investigated whether proteotoxicity led to senescence-mediated death of proband cells. The proband fibroblasts showed significantly lower cell viability than the control fibroblasts in the MTT assay (Fig. 5H). Many of the proband fibroblasts were also positive for the CellEvent senescence green fluorophore (Fig. 5I), indicating that the proband fibroblasts were undergoing senescent cell death. In addition, it was observed that the expression of the senescence marker proteins p16^INK4a and p21^WAF1/Cip1 was significantly upregulated in the proband fibroblasts than in the control fibroblasts (Fig. 5J-O).

Taken together, these results indicate that CRTAP loss in the proband fibroblasts triggers a critical level of proteotoxicity that leads to senescent death of the proband cells.

2.6. Type I collagen aggregates are degraded by aggrephagy in proband fibroblasts

Protein aggregates are degraded by cells through autophagy [30]. Therefore, we tested whether type I collagen aggregates in proband fibroblasts were also undergoing autophagic degradation (i.e., aggrephagy). Indeed, compared with control fibroblasts, the proband cells exhibited a higher number of SQSTM1 and LC3B puncta (Fig. 6A, B), indicating the presence of a higher number of polyubiquitinated protein aggregates and autophagosomes in the proband cells. Several of the SQSTM1 and LC3B puncta were also colocalized with type I collagen aggregates in a significant number of proband fibroblasts (Fig. 6C-F). Autophagy flux in proband fibroblasts was also higher, as they showed higher conversion of LC3B-I to LC3B-II and lower level of SQSTM1 in proband fibroblasts compared with control fibroblasts (Fig. 6G, H). In addition, inhibition of autophagy by the inhibitor bafilomycin-A1 prevented the degradation of pro-α1(I) [type I collagen] in proband fibroblasts (Fig. 6G, H). Taken together, these results indicate that the induction and flux of autophagy was higher in proband cells, and that type I collagen aggregates were subjected to poly-ubiquitination-dependent aggrephagy in proband fibroblasts.

Fig. 6. Type I collagen aggregates undergo autophagic degradation [aggrephagy] in proband fibroblasts.

(A) Control and proband fibroblast cells were immunostained for LC3B and SQSTM1. Representative confocal immunofluorescence microscopy images showing higher number of LC3B and SQSTM1 puncta, representing higher autophagy/aggrephagy, in proband fibroblasts. (B) Quantification of LC3B and SQSTM1 puncta in control and proband fibroblasts (100 cells/group). (C) Control and proband fibroblasts were immunostained for pro-α1(I) and SQSTM1. Representative confocal immunofluorescence microscopy images showing higher number of SQSTM1 puncta colocalized with pro-α1(I) aggregates in proband fibroblasts. (D) Pearson’s correlation coefficient of colocalization of SQSTM1 and pro-α1 (I) in control and proband fibroblasts. (E) Control and proband fibroblasts were immunostained for pro-α1(I) and LC3B. Representative confocal immunofluorescence microscopy images showing higher number of LC3B puncta colocalized with pro-α1(I) aggregates in proband fibroblasts. (F) Pearson’s correlation coefficient of colocalization of LC3B and pro-α1(I) in control and proband fibroblasts. (G) Control and proband fibroblasts were untreated or treated with 1 μM bafilomycin A₁ for 48 h. Up: representative immunoblots of LC3B, SQSTM1 and pro-α1(I) from lysate of untreated and bafilomycin A₁ treated control and proband fibroblasts. Bottom: densitometric quantification of the LC3B, SQSTM1 and COL1A1 bands relative to TUBA of the blots of 6G.

Quantifications are shown as mean ± S.D.; P values are indicated. TUBA is the loading control in the immunoblots. Scale bar in confocal microscopy images: 5 μm. Microscopy and immunoblot data are representative of at least three independent experiments.

3. Discussion

A systemic network of protein quality control mechanisms plays a critical role in maintaining homeostasis of synthesis, stability, and secretion of collagen molecules [31]. Deregulation of one or more of these quality control proteins results in abnormal expression and accumulation of type I collagen in several collagenopathies, such as OI [32,33]. Based on the mutant genes involved, OI is classified into different types. The type VII of OI results from mutations in the CRTAP gene.

To date, 39 pathogenic variants causing OI type VII have been reported [12–17], involving truncating, missense, and splice variants in the CRTAP gene [12–17] (Table 1). The major findings in patients reported to date are multiple fractures at birth or gestational age [12–17]. The pathogenic variants of CRTAP are either fatal or severe in context of OI manifestations [12–17]. Some of the patients from previous studies showed narrow thorax and rhizomelia with shortening of the upper and lower extremities [12–17]. Radiological studies in many of the CRTAP mutant probands showed bending of the extremities and/or fractures throughout fetal and neonatal development [12–17]. Among the CRTAP variants in OI type VII, ten different splice variants have been identified in CRTAP, including the first deep intronic variant that activates a cryptic exon due to a single nucleotide change (c.472-1021C>G) in intron 1 [18]. Among the reported variants, two (c.471+1G>C; c.471+2C>A) have been classified as pathogenic, four (c.471+2C>G; c.473_621+1del; c.794-2A>G; c.794-1G>C) have been classified as likely pathogenic and one variant (c.471+2C>T) remains of uncertain significance according to the ACMG classification. Another splice variant, c.1153-3C>G, was found to cause a significant reduction in CRTAP mRNA expression due to a chromosomal deletion of one allele and NMD of mRNA from the splicing variant.

Table 1. Clinical features of probands affected with OI type VII.

Clinical features	Reported patients (n = 29)	Present study	Total number of patients (n = 30)
Large open anterior fontanelle	1	−	1/30
Proptosis	2	−	2/30
Blue sclera	3	−	3/30
Depressed nasal bridge	1	+	2/30
Anteverted nares	−	+	1/30
Long philtrum	1	+	2/30
Narrow thorax	2	+	3/30
Pectus excavatum	3	−	3/30
Multiple rib fractures	7	−	7/30
Hydronephrosis	1	−	1/30
Multiple factures	27	+	28/30
Wormian bone	6	−	6/30
Soft calvarium	4	+	5/30
Scoliosis	1	−	1/30
Platyspondyly	3	−	3/30
Severe osteopenia	14	+	15/30
Shortening of upper extremities	16	+	17/30
Shortening of lower extremities	21	+	22/30
Rhizomelia	9	+	10/30
Popcorn epiphysis	2	−	2/30

+ is phenotype observed. – is phenotype not observed.

In the current study, antenatal ultrasonography of the proband (male fetus) showed short and bent lower limbs. The fetus also had a depressed nasal bridge with anteverted nostrils and a long philtrum. Other characteristic features included narrow thorax, soft calvarium, normal brain size, normal thorax, and normal abdominal visceral organs. Radiological findings of the fetus showed severe osteoporosis, rhizomelic shortening of both upper and lower extremities with severely bent femur, tibia, and fibula. Although solo exome sequencing of the fetus did not identify a significant candidate variant of OI, genome sequencing identified a deep intronic variant g.33128536A>G (c.794-1403A>G) in intron-3 of the CRTAP gene, suggesting that the affected fetus had OI type VII.

The deep intronic mutation in CRTAP is a type-II splice site mutation, which results in the inclusion of a cryptic exon or pseudoexon in the mature transcripts of mutant CRTAP. In addition to the existing splice site donor at the end of exon-2, the mutation creates two new donor sites, one of which is 7 nucleotides downstream of the end of exon-2 and the other is the mutant nucleotide itself, while the 68th nucleotide up-stream of the mutation site functions as a cryptic splice site acceptor in intron-3. As a result of the formation of these new donor and acceptor sites, two mutant alternative CRTAP transcripts are formed. Mature transcript-1 exhibits a frameshift after the 792nd nucleotide of ORF, resulting in a premature termination codon at 832–834 nucleotides of ORF. In contrast, the mature transcript-2 shows an in-frame insertion of 75 nucleotides after the 792nd nucleotide of ORF. The mutant transcripts translate two mutant isoforms of CRTAP with lengths of 277 amino acids and 426 amino acids.

Splice site mutations of CRTAP are not uncommon in OI type VII. Previous studies have shown that a mutation in the splice site donor (type-I splicing mutation) and a deep intronic mutation (type-II splicing mutation) in CRTAP lead to OI [12,13,15]. In both cases, the cellular level of the CRTAP transcript was very low, indicating rapid degradation of the mutant CRTAP mRNAs. Because of the lack of expression of the mutant CRTAP mRNAs, the splicing events in these probands were not studied in detail. Although the CRTAP in the proband of the current study had a deep intronic type-II splicing mutation, we observed measurable amounts of mutant CRTAP transcripts in the proband’s fibroblasts. This suggests that the mutant CRTAP transcripts are not completely degraded in the proband cells. However, because we found that the total level of both mutant CRTAP transcripts in the proband fibroblasts was significantly lower than the level of wild-type CRTAP transcript in the control fibroblasts, we suspect that the mutant transcripts were partially degraded in the proband cells. Among the mutant CRTAP transcripts, transcript-1 has a premature termination codon (PTC), and this mutant transcript could be more susceptible to rapid degradation by NMD. However, the presence of both the mutant CRTAP transcripts indicates that NMD may be less effective in eliminating mutant CRTAP transcripts. Although NMD is a mRNA surveillance mechanism that eliminates mutant mRNAs with PTC, NMD does not necessarily clears any and all mRNAs containing PTC [34–36]. Recent studies show that many of the PTC-containing mRNAs can evade NMD. Such studies have established several rules that govern the NMD evasion of several PTC-bearing mRNAs. If a PTC is far from the normal stop codon or if there are specific factor binding sequences near the PTC, then the PTC-harboring mRNA can escape NMD [37]. While the PTC in the cryptic exon of mutant transcript-1 of CRTAP is somewhat far from the normal stop codon, the cryptic exon in mutant transcript-1 of CRTAP also contains a possible AUF1 binding site near the PTC. Such binding of AUF1 to the PTC-neighboring ARE sequence makes the mutant transcript-1 of CRTAP less susceptible to NMD. However, because the ARE is present in the cryptic exon of both mutant transcript-1 and transcript-2 of CRTAP, both the mutant transcripts are optimal targets of AUF1-mediated degradation. The efficiency of ARE/AUF1-mediated degradation of mRNAs is dependent upon the qualitative and quantitative binding strength of AUF1 to the corresponding mRNA [38]. Binding of the AUF1 to the ARE (‘AUUUUU’) sequence of the mutant transcripts of CRTAP could be such strong that allows degradation of significant fraction but not all of the mutant CRTAP transcripts in the proband cells. Had the mutant transcripts been undergoing NMD, we would have observed almost complete loss of mutant CRTAP transcripts in the proband cells. Because of these reasons, we hypothesize that alternatively spliced mutant CRTAP mRNAs, such as those found in this study, may be the suboptimal targets of NMD, whereas they could be good targets of AUF1-mediated degradation.

Interestingly, we did not observe wild-type transcript of CRTAP in the proband cells, although the original splice site donors and acceptors were intact in the premature CRTAP transcript. The absence of wild-type CRTAP transcript in the proband cells may possibly be explained by the hypothesis of induction of a splicing enhancer or removal of a splicing silencer due to the mutation. The c.794-1403A>G mutation may have altered the mutation-bearing contiguous sequence such that the sequence becomes a splicing enhancer or loses its activity as a splicing silencer. Such an alteration is thought to cause either binding of splicing activators or loss of binding of splicing repressors to the mutation-bearing region. In either case, the mutation region becomes a splicing hotspot that induces alternative splicing that overrides normal splicing, leading to retention of the cryptic exon in the mutant CRTAP transcripts. Because the mutation is homozygous, the identical sequence of the premature mutated CRTAP transcript of both alleles undergoes similar alternative splicing, eliminating the chance of a wild-type splicing event in the proband cells.

Previous studies on OI type VII have shown that the expression and function of the mutant CRTAP protein is affected by the loss of the mutant CRTAP transcripts [12,13,15]. Although the transcripts of mutant CRTAP are detectable, the expression of CRTAP protein is not detectable in the proband fibroblasts of the current study. The two mature mutant CRTAP transcripts are translated into two mutant isoforms - one is a truncated isoform of 277 amino acids in which the thirteen amino acids at the C-terminus are not wild-type, and the other mutant isoform is 426 amino acids long and identical to the wild-type protein, except that this mutant isoform has an insertion of 25 amino acids in-frame that are not wild-type, downstream of 264th position. Interestingly, the amino acids that are not wild-type have a ‘GWxxI’ sequence in both mutant isoforms, which was found to be a degron in the mutant isoforms of CRTAP when examined. The detection of degron in the mutant isoforms of CRTAP is evident because suppression of the proteasome by an inhibitor restores expression of both mutant isoforms of CRTAP in proband fibroblasts. The likelihood that the ‘GWxxI’ sequence is a degron is first evident from the presence of this short linear motif in the loop region of the model structures of both mutant isoforms of CRTAP (data not shown). The ‘GWTAI’ and ‘GWCYI’ sequences of the mutant isoforms of CRTAP are confirmed to be degrons by their ability to degrade a stable bicistronic EGFP-mCherry construct. The site-directed mutations of the glycine, tryptophan, and isoleucine residues of the ‘GWxxI’ sequence by alanine scanning also diminish the degron property of the sequences, demonstrating the necessity of the glycine, tryptophan, and isoleucine residues in the sequence as a protein degron. However, this degron is not a classical N- or C-degron. Although the ‘GWTAI’ degron is located at the C-terminus of the truncated isoform of CRTAP, this degron is not a Gly/C or RxxG/C or Arg/C degron, which means that this degron is not a conventional C-degron. Since degrons can occur anywhere in a protein sequence, the ‘GWxxI’ degron is an example of a novel degron sequence that can enhance protein degradation not only at the C-terminus but also in the internal region. It would be interesting to find out whether this region is a binding site for the cullin-RING complex to promote degradation of the mutant CRTAP isoforms. Since proteasomal inhibition prevents degradation of the mutant CRTAP isoforms, we hypothesize that post-translational modifications, such as polyubiquitination, are required for turn-over of the mutant CRTAP isoforms. Therefore, the ‘GWxxI’ degron may belong to the class of ‘acquired degrons” rather than inherent degrons. Although studies have reported classical degron libraries [39] and synthetic use of degrons [40], a comprehensive understanding of degron signaling in mutant proteins of genetic disorders is very rare. Therefore, the ‘GWxxI’ degron in mutant CRTAP provides a good opportunity to study the interplay of novel protein degradation signaling with proteostasis imbalance in genetic disorders.

CRTAP is a scaffold protein that binds to P3H1 and PPIB proteins, and the heterotrimeric complex is involved in prolyl-3-hydroxylation and hydroxyproline isomerization in type I collagen. CRTAP, as part of a post-translational modifier complex, is a quality control protein that provides stability to type I collagen. Complete or partial loss of expression and function of CRTAP is therefore associated with the induction of collagenopathy in OI type VII [12–17]. Although prolyl-4-hydroxylation is more abundant in both chains of type I collagen [41] and this hydroxylation is required for the formation and stability of the triple helix structure of type I collagen [42], 3-hydroxylation of Pro986 of the pro-α1(I) and pro-α2(I) chains is also required for the structural integrity of type I collagen fiber [43]. Loss of the prolyl-3-hydroxylase complex due to mutations in CRTAP does not affect the formation of the trimeric type I collagen molecules but may disturb the fibrous architecture of the type I collagen structure. Although the fate of Pro986-unhydroxylated type I collagen remains largely unclear, we show in this study that hydroxy-Pro986-deficient type I collagen tends to aggregate. This observation is similar to that we previously reported in OI type XX [44], where mutation of the chaperone MESD leads to aggregation of unstable triple helical type I collagen [44]. With these studies, it is gradually becoming clear that aberrant and unstable trimeric type I collagen is more prone to aggregation under physiological conditions in various OIs. With further preliminary evidence, we have observed that aggregation of type I collagen in fibroblasts and squamous cells is a hallmark of a particular group of OI types.

The imbalance in proteostasis in the proband cells could have resulted because of aggregation of type I collagen and failure of hydroxylation of prolines in proteins that are substrates of CRTAP-P3H1-PPIB complex. Because prolyl-3-hydroxylation complex is involved in hydroxylation of prolines of substrates other than the chains of type I collagen, it is possible that the loss of CRTAP also perturbs the hydroxylation of prolines of such substrates, resulting in structural and functional abnormalities of those proteins and leading to proteostasis failure. Such deregulation in proteostasis is linked to upregulated expression of endoplasmic reticulum-resident chaperones, aggregation and increased ubiquitination of proteins in proband cells. Although the type I collagen aggregates undergo autophagic degradation (aggrephagy), such aggrephagy is not sufficient to substantially reduce the proteotoxicity in the proband cells. A similar kind of proteostasis failure and aggrephagy of type I collagen were previously reported by us in a case of OI type XX with mutation in MESD protein [44]. Like in that study, the deregulated proteostasis consequently causes death, specifically by senescence, of the proband cells. Collectively, these results indicate that perturbed proteostasis due to aberrant expression, stability, function and localization of quality control proteins is a major contributing factor to the bone developmental defects in certain types of OI. Therefore, it will be exciting to investigate whether proteotoxicity mediated by type I collagen aggregates is the primary pathomechanism underlying the collagenopathy in different types of OI.

Overall, this study describes genotype-phenotype correlation in OI type VII by showing how a deep intronic mutation in CRTAP leads to alternatively spliced transcripts and gives rise to mutant isoforms of the protein that are rapidly degraded due to the presence of a novel degron. Downregulation of CRTAP expression crosstalk with aberrant quality control of type I collagen, resulting in acute proteotoxicity due to type I collagen aggregation, which leads to pathological bone developmental defects and lethality in OI type VII.

4. Methods and materials

4.1. Family ascertainment, autopsy and diagnosis

A third-degree consanguineous couple with a history of fetal loss visited the Department of Medical Genetics, Kasturba Medical College, Manipal, with complications in their third fetus. The mother of the proband had a spontaneous abortion in the second month in her first pregnancy and delivered a male child with normal development in her second pregnancy. Family history, biochemical investigations, ultrasound, X-ray, autopsy and counseling of the couple were done at the Department of Medical Genetics, Kasturba Medical College, Manipal. The prenatal ultrasound and radiological findings of the third fetus (19 weeks old) showed features of bone developmental abnormalities resembling osteogenesis imperfecta. Written informed consent was obtained from the parents for medical photography, autopsy, skin biopsy, genetic analysis, and participation in the study. The study was thoroughly reviewed and approved by the Ethics Committee of Kasturba Medical College and Kasturba Hospital.

4.2. Exome and genome sequencing, data analysis

4.2.1. Exome sequencing

Genomic DNA was isolated from parental white blood cells and proband skin fibroblasts. Singleton exome sequencing from the proband’s genomic DNA was performed by massively parallel sequencing using the NovaSeq platform (Illumina Inc., USA) with a targeted average coverage of 100×. Exonic and exonic-intronic boundary regions were captured using the Agilent SureSelect CREv3 Capture Kit. After initial quality control assessment by FastQC, raw reads were aligned to the human reference genome (GRCh38) using BWA-MEM (v0.7.15) [45]. Variant calling on the sequenced exomes was performed using GATK (v3.6) [46] for the discovery of germline SNVs and INDELs. Allele frequencies and states were identified from the variant call format (VCF) file using BCFTOOLS (v1.3.1) and custom Perl scripts. Data were annotated using ANNOVAR [47]. Allele frequencies and states were obtained from the standard databases of gnomAD [48], GenomeAsia [49], Singapore Genome Project [50], and our internal database of 1909 exomes. Allele frequencies with <1 % occurrences (rare variants) were annotated. Exonic and splice-site variants near the exonic-intronic boundaries were considered. The overall protocol and process of exon sequencing, variant annotation, and analysis were similar to those described in our previous studies [51,52]. We did not find any relevant variant during exon sequencing.

4.2.2. Genome sequencing

Genome sequencing was performed using the NovaSeq 6000 from Illumina Inc. USA, with 100–150 bp paired-end reads and an average coverage of 40×. Quality assessment of the raw reads was performed using FastQC. BWA-MEM2 was used to align the reads to the reference genome (GrCh38). After post-processing the alignment with GATK4, variants were determined using HaplotypeCaller in GVCF mode. Cohort-based joint genotyping was performed by combining available gvcfs files from in-house genome sequencing data using GenotypeGVCFs and variant recalibration. Internal allele frequencies and allele states were determined using BCFTOOLS, custom PERL scripts, and the available ten genome sequencing data. Annotation of variants was performed using ANNOVAR [47]. To extend the GrCh37-based variant datasets from the GenomeAsia [49], Singapore Genome Project [50], and the in-house variant dataset from exomes to GrCh38, they were LiftOver using CROSSMAP and merged with ANNOVAR annotation before being added with OMIM disease phenotypes and HPO terms. Population variant datasets were used for variant filtering, and known genes were selected for further investigation using the Genomics England PanelApp gene panel for osteogenesis imperfecta. SpliceAI and Human Splicing Finder (HSF) were used to predict splicing effects for prioritized coding and noncoding variants. Non-coding essential regulation (ncER) score was used to prioritize regulatory genetic variants found in known genes. Structural and copy number variants were identified using Delly, and those affecting known genes were subjected to further analysis. Authenticity of variant calls was confirmed using IGV, and regulatory regions of identified variants were screened for promoter/enhancer binding sites using the UCSC browser.

4.3. Cloning

For cloning, total RNA was extracted from control and proband fibroblasts (ThermoFischer Scientific [TFS], AM1924), and mRNA from the control and proband total RNAs was extracted using an mRNA enrichment kit (TFS, 61006). cDNA pool from the extracted mRNA was synthesized using reverse transcription (TFS, 4368814) and oligo-dT primer (Eurofin). The overall cloning procedure was similar to that described in our previous studies [53,54]. Briefly, cDNAs were subjected to polymerase chain reaction (New England Biolabs Inc. [NEB], M0531S) with the appropriate primer sets. The PCR products and plasmids were restriction digested with specific restriction enzymes (NEB), followed by ligation of the restriction-digested products with T4 DNA ligase (TFS, K1422). The ligated DNA was transformed into the ultracompetent DH5α strain of Escherichia coli. Positively selected colonies that grew on the LB-antibiotic medium were identified for the clone by colony PCR (Takara Bio, RR350A). The mutant transcript-1 and transcript-2 were also cloned into a ‘Topo TA’ cloning vector using the Topo TA cloning kit (TFS, K457501). The sequences of all clones were confirmed by sequencing. The clones and their corresponding oligonucleotides are listed in Tables 2 and 3.

Table 2. Clones and plasmids.

Clone/Plasmid	Source	Identifier
EGFP-mCherry-Mut1_pN1	Present study
EGFP-Mut2-mCherry_pN1	Present study
FLAG-CRTAP_pcDNA3.1	Present study
FLAG-CRTAP-Mut1_pcDNA3.1	Present study
FLAG-CRTAP-Mut2_pcDNA3.1	Present study
FLAG-CRTAP-Mut1-ΔGWTAI_pcDNA3.1	Present study
FLAG-CRTAP-Mut1-G273A_pcDNA3.1	Present study
FLAG-CRTAP-Mut1-W274A_pcDNA3.1	Present study
FLAG-CRTAP-Mut1-I277A_pcDNA3.1	Present study
FLAG-CRTAP-Mut1-G273A, W274A, W274A_pcDNA3.1	Present study
FLAG-CRTAP-Mut2-ΔGWCYI_pcDNA3.1	Present study
FLAG-CRTAP-Mut2-G265A_pcDNA3.1	Present study
FLAG-CRTAP-Mut2-W266A_pcDNA3.1	Present study
FLAG-CRTAP-Mut2-I269A_pcDNA3.1	Present study
FLAG-CRTAP-Mut2-G265A, W266A, W269A_pcDNA3.1	Present study
Transcript-1_Topo TA	Present study
Transcript2_Topo TA	Present study
pEGFPC3	NovoPro Biosciences	V012022
EGFP-mCherry (no stop codon)_pCDNA3.1	Present study
EGFP-mCherry (stop codon)_pCDNA3.1	Present study
pcDNA3.1	ThermoFischer Scientific	V790−20
Topo TA	ThermoFischer Scientific	K457501

Table 3. Oligonucleotides for gene cloning.

No.	Clone	Primer	Sequence
	Cloning primers
1	FLAG-CRTAP- Mutl_pcDNA3.1	Forward	CGCGGATCCATGGACTACAAAGACGATGACGACAAGGAGCCGGGGCGCCGGGGGGCCGCGGCG
		Reverse	CCGGAATTCTTATATAGCGGTCCAGCCAAAAATACACAA
2.	FLAG-CRTAP- Mut2_pcDNA3.1	Forward	CGCGGATCCATGGACTACAAAGACGATGACGACAAGGAGCCGGGGCGCCGGGGGGCCGCGGCG
		Reverse	CCGGAATTCCTAGCTGGTCTCCTCCAGTTCCAAGAGGTC
3.	EGFP_pcDNA3.1	Forward	CCCAAGCTTATGGTGAGCAAGGGCGAGGAGCTGTTCACC
		Reverse	CGCGGATCCCTTGTACAGCTCGTCCATGCCGAGAGTGAT
4.	EGFP-mCherry(no stop codon)_pcDNA3.1	Forward	CCGGAATTCATGGTGAGCAAGGGCGAGGAGGATAACATG
		Reverse	CCGCTCGAGCTTGTACAGCTCGTCCATGCCGCCGGTGGA
5.	EGFP-mCherry (stop codon) _pcDNA3.1	Forward	CCGGAATTCATGGTGAGCAAGGGCGAGGAGGATAACATG
		Reverse	CCGCTCGAGCTACTTGTACAGCTCGTCCATGCCGCCGGT
6	EGFP-mCherry(no stop codon)- Mut1_pcDNA3.1	Forward	CCGCTCGAGTTACATTCAGGTTGTGTATTTTTGGCTGGACCGCTATATAAATGTCCTCCTCAGGATATCACGTCAAGTAATCTAGAAAA
		Reverse	TTTTCTAGATTACTTGACGTGATATCCTGAGGAGGACATTTATATAGCGGTCCAGCCAAAAATACACAACCTGAATGTAACTCGAGCGG
7	EGFP-Mut2- mCherry(stop codon)_pcDNA3.1	Forward	CGCGGATCCGTTGGTGTTACATTCAGGTTGTGTATTTTTGGCTGGACCGCTATATAAATGTCCTCCTCAGGATATCACGTCAAGGAATTCAAA
		Reverse	TTTGAATTCCTTGACGTGATATCCTGAGGAGGACATTTATATAGCGGTCCAGCCAAAAATACACAACCTGAATGTAACACCAACGGATCCGCG

4.4. Site directed mutagenesis

The ΔGWTAI, G273A-W274A-I277A, G273A, W274A and I277A mutations in the mutant CRTAP-Mut1, and ΔGWCYI, G265A-W266A-I269A, G265A, W266A and I269A mutations in the CRTAP-Mut2 were generated by site-directed mutagenesis using nested PCR method and specific primer sets. Mutations in clones were verified by sequencing.

4.5. Cell culture

The skin fibroblasts were obtained from the fetus (proband) and a healthy control individual. Cells were cultured in advanced DMEM medium (TFS, 12491023) containing 2 mM l-glutamine (TFS, 25030164), 10 % fetal bovine serum (TFS, 26140095), and 1x penicillin-streptomycin (TFS, 11548876). Cells were grown and maintained in a humified static incubator at 37 °C and 5 % CO₂.

Mammalian cell-expressible DNA clones were transfected to the fibroblasts using lipofectamine2000 (TFS, 11668019) and opti-MEM (TFS, 31985062) in a method described in our earlier studies [51,55]. siRNAs (listed in Table 4) were transfected to fibroblasts using lip-ofectamine 3000 (TFS, L3000001) according to the manufacturer’s protocol.

Table 4. siRNAs.

Target of siRNA	Source	Identifier
Control	Sigma Aldrich	SIC001
AUF1	Sigma Aldrich	EHU094941
CRTAP	Sigma Aldrich	EHU019361
P3H1	Sigma Aldrich	EHU145841
PPIB	Sigma Aldrich	EHU110531

Cells were treated with 10 μM cycloheximide, 10 μM MG132 or 1 μM bafilomycin A₁ for 24–48 h.

4.6. Quantitative PCR

Total RNA was extracted from cultured fibroblast cells using the method described earlier. cDNA pool from the mature mRNAs were generated using reverse transcription (TFS, 11756050) with random hexamer oligonucleotides. Quantitative PCR (q-PCR) was done using SYBRTM green PCR master mix (TFS, 4309155). The list of forward and reverse primers used in q-PCR is listed in Table 5. Data generated in q-PCR was analyzed using QuantStudio™ Design and Analysis software. Relative expression of target genes was normalized to the expression of GAPDH.

Table 5. Oligonucleotides for qPCR.

No.	Transcript	Primer	Sequence
1	Mutant CRTAP transcript- 1+2	Forward (P1)	CGGGCATACAACGGTGAGAA
		Reverse (P2)	AGGAGGCTATCCGGTTGAGA
2	Mutant CRTAP transcript- 1+2	Forward (O1)	GCAAATAATCTCCCCAAA
		Reverse (O2)	TTCATATGACTTGGTTTC
3	Mutant CRTAP transcript- 2	Forward (P3)	CCTTTCCATAGCAGGTTGGTG
		Reverse (P4)	ACTTCCAGCCCAGACCTGAA
4	GAPDH	Forward	ACCTGCCAAATATGATGAC
		Reverse	TCATACCAGGAAATGAGCTT
5	HSPA5	Forward	CCCGTGGCATAAACCCAGAT
		Reverse	GGTCATGACACCTCCCACAG
6	HSP90B	Forward	TCATGTCCCTCATCATCAAT
		Reverse	TCATGCCAATGCCTGTGTCT
7	PDIA2	Forward	CGAGGAGCCCCTCGGAGGAG
		Reverse	CATTGACTCGGCCGCGAGCA

4.7. Immunoblotting

Cells were lysed with RIPA buffer (TFS, 89900) and cleared by centrifugation at 14000 rpm for 20 min (4 °C). 40–60 μg of total protein from the cleared cell lysate was separated on a 12 % SDS-PAGE. Proteins from the gel were then transferred to a polyvinylidene fluoride (PVDF) membrane (Amersham Hybond P 0.45; GE Healthcare Life Sciences, 10,600,023). After blocking the membrane with 5 % skim milk (in Tris buffer saline [TBS], TFS, 28358), the membrane was incubated overnight with primary antibodies (in TBS with 5 % skim milk) at 4 °C. The membrane was then blocked with 5 % skim milk and incubated for 2 h with secondary antibodies (in TBS with 5 % skim milk) at room temperature. The membrane was washed intermittently with TBST solution (TBS containing 0.1 % Tween20). Details of the primary and secondary antibodies are shown in Table 6. Chemiluminescence signals of protein bands were generated using SuperSignal west femto maximum sensitivity substrate (TFS, 34094). Chemiluminescence signals were detected using the iBright imaging system (TFS). Densitometric quantification of protein bands was performed using ImageJ2 software.

Table 6. Antibodies.

Primary antibodies
Target protein	Host	Supplier	Catalogue number
CRTAP	Rabbit	ThermoFischer Scientific	PA5-60431
CANX	Mouse	ThermoFischer Scientific	MA3-027
COL1A1	Rabbit	Cloud-Cone Corp.	PAA350Hu02
FLAG	Mouse	Sigma Aldrich	F3165
HSPA5	Rabbit	ThermoFischer Scientific	PA1-014A
HSP90B	Rabbit	ThermoFischer Scientific	PA3-012
LC3B	Mouse	ThermoFischer Scientific	MA5-37852
P3H1	Mouse	ThermoFischer Scientific	H00064175- B01P
p16INK4a	Rabbit	Thermo Fischer Scientific	MA5-32133
p21waf1/cip1	Rabbit	Cell Signaling Technology	2947
PPIB	Rabbit	ThermoFischer Scientific	PA1-027A
PDI	Mouse	ThermoFischer Scientific	MA3-019
SQSTM1	Mouse	Sigma Aldrich	WH0008878M1
TUBA	Mouse	Sigma Aldrich	T6199
Ubiquitin	Rabbit	ThermoFischer Scientific	PA3-16717
Secondary antibodies (Immunoblotting)
anti-rabbit IgG (whole molecule) peroxidase antibody	Goat	Sigma-Aldrich	A0545
anti-mouse IgG (whole molecule) peroxidase antibody	Rabbit	Sigma-Aldrich	A9044
Secondary antibodies (Immunocytochemistry)
anti-Mouse IgG (H+L)-Alexa Fluor Plus 488	Goat	ThermoFischer Scientific	A-11029
anti-Mouse IgG (H+L)-Alexa Fluor Plus 555	Goat	ThermoFischer Scientific	A32727
anti-Rabbit IgG (H+L)-Alexa Fluor Plus 488	Goat	ThermoFischer Scientific	A-11034
anti-Rabbit IgG (H+L)-Alexa Fluor Plus 555	Goat	ThermoFischer Scientific	A32732

4.8. Immunocytochemistry and fluorescence microscopy

Fibroblasts grown on clean glass coverslips were washed with phosphate buffer saline (PBS, pH 7.4; TFS, 10010023), followed by fixation with 4 % paraformaldehyde (SA, 158127) (in PBS) solution. Cells were permeabilized with 0.2 % Triton X-100 (SA, T8787) (in PBS) and subjected to blocking with 1 % molecular biology grade bovine serum albumin (BSA, in PBS) (SA, B6917). These cells were successively treated with primary and secondary antibodies (listed in Table 6) with intermittent washing with PBS. Finally, cells were mounted on glass coverslip with prolong antifade gold with DAPI (TFS, P36931). The overall process of immunostaining was similar to our earlier studies [56,57].

Proteostat (Enzo lifesciences, ENZ-51035-0025) staining of the adherent fibroblasts was done according to the manufacturer’s protocol.

Senescence in cells was detected using CellEvent™ Senescence Green Detection Kit (TFS, C10850) according to the manufacturer’s protocol.

Fluorescence imaging of the cells was done using LSM700 confocal laser scanning microscope (Carl Zeiss) with a 63x Plan Apochromat/1.4 NA oil/DIC M27 objective. Images were processed using Zen-Lite 2010 software (Carl Zeiss).

Number of LC3B-positive autophagosomes, ubiquitinated protein aggregates, SQSTM1 puncta and proteostat-positive protein aggregates were counted manually.

Colocalization of proteins were measured in Coloc 2 program of FIJI. Pearson’s correlation coefficient was used as a measure of colocalization of proteins.

4.9. MTT assay

5000 control and proband fibroblasts cells were separately seeded in each well of a 96-well plate and grown for 24 h at optimal cell growth conditions. For MTT assay, growth medium was removed, and cells were incubated with 0.5 mg/ml MTT (Sigma Aldrich, 475,989) (in growth medium) at 37 °C, 5 % CO₂ for 2 h. The MTT medium was then removed and 200 μl of DMSO was added to each well. The colour development in was read at 515 nm.

4.10. Bioinformatics analysis

Multiple sequence alignment of protein sequences was done in Clustal Omega [58].

Secondary structure prediction of mRNA sequence was done in mFold [59] web server.

4.11. Statistical analysis

Data are presented as mean ± standard deviation. Statistically significant difference between means of groups were analyzed by two-way ANOVA or two-tailed, homoscedastic student’s t-test. P value <0.05 was considered as statistically significant different.

4.12. Graphics

The graphics were made in Adobe Illustrator.

Data availability

All the data of the study is given in the article. Materials and protocols of this study are available from the corresponding authors upon request.