Collagen IX is a heterotrimer of three α-chains, which consists of three COL domains (collagenous domains) (COL1–COL3) and four NC domains (non-collagenous domains) (NC1–NC4), numbered from the C-terminus. Although collagen IX chains have been shown to associate via their C-terminal NC1 domains and form a triple helix starting from the COL1 domain, it is not known whether chain association can occur at other sites and whether other collagenous and non-collagenous regions are involved. To address this question, we prepared five constructs, two long variants (beginning at the NC4 domain) and three short variants (beginning at the COL2 domain), all ending at the NC2 domain (or NC2 replaced by NC1), to study association and selection of collagen IX α-chains. Both long variants were able to associate with NC1 or NC2 at the C-terminus and form various disulfide-bonded trimers, but the specificity of chain selection was diminished compared with full-length chains. Trimers of the long variant ending at NC2 were shown to be triple helical by CD. Short variants were not able to assemble into disulfide-bonded trimers even in the presence of both conserved cysteine residues from the COL1–NC1 junction. Our results demonstrate that collagen IX α-chains can associate in the absence of COL1 and NC1 domains to form a triple helix, but the COL2–NC2 region alone is not sufficient for trimerization. The results suggest that folding of collagen IX is a co-operative process involving multiple COL and NC domains and that the COL1–NC1 region is important for chain specificity.

INTRODUCTION

Collagens are the most abundant components of the ECM (extracellular matrix). They consist of three polypeptide chains (α-chains) wound into a right-handed helix composed of Gly-Xaa-Yaa repeats, in which proline residues in the third (Yaa) position are often hydroxylated by prolyl 4-hydroxylase to stabilize the structure. Currently, the collagen superfamily consists of at least 28 homotrimeric or heterotrimeric proteins that are divided into subgroups according to their structure and function [13]. FACITs (fibril-associated collagens with interrupted triple helices; IX, XII, XIV, XVI, XIX, XX and XXI) consist of short triple helical COL domains (collagenous domains) flanked by NC domains (non-collagenous domains) and have two conserved cysteine residues in the COL1–NC1 junction [4,5].

Collagen IX is a component of hyaline cartilage, intervertebral disc and the vitreous of the eye. The collagen IX molecule is a heterotrimer of three genetically different chains, α1(IX), α2(IX) and α3(IX) and consists of three COL domains, COL1–COL3, separated by four NC domains, NC1–NC4, numbered from the C-terminus [5]. In cartilage, collagen IX resides on the surface of collagen II/XI heterofibrils, possibly connecting fibrils together or to other ECM components [68]. Stabilization of the interaction between collagen IX and II is achieved by five covalent lysine-derived cross-links between their α-chains [913].

Studies of collagen IX chain assembly have been performed with pepsin-resistant low-molecular-mass fragments comprising the COL1 domain and synthesized peptides resembling the C-terminus of the COL1 domain and the complete NC1 domain [14,15]. Low-molecular-mass fragments re-associated to form only three different triple helical forms: homotrimers of α1(IX) and α2(IX) and a heterotrimer of α1(IX)–α2(IX)–α3(IX) [15]. Similar observations were obtained from insect cells expressing FL (full length) collagen IX. In the insect cell system, collagen IX chains assembled favourably to triple helical α1(IX)α2(IX)α3(IX) heterotrimers, but also disulfide-bonded α1(IX) homotrimers were seen [16]. Synthetic peptides consisting of the C-terminus of the COL1 and NC1 domains were able to form disulfide-bonded heterotrimers, but not triple helices, suggesting that triple helix formation is not the driving force for this interaction [14].

Chain association studies with another FACIT member, collagen XII, carried out in HeLa and insect cells, suggested an important role for proline hydroxylation in stabilizing the COL1 domain in order to obtain disulfide-bonded trimers [17,18]. Also, deletion of most of the NC1 domain of chicken collagen XII, except for the seven residues of the COL1–NC1 junction, did not prevent the formation of triple helical molecules [19]. Recently, it was shown that the formation of the COL1 triple helix is the key step in the trimeric association of collagen XII α-chains, which is then stabilized by interchain disulfide bonds [20].

Chain association studies with the fibril-forming collagens I and III, carried out in semi-permeabilized cells, indicated an important role for the C-propeptide in triple helix formation [21]. The function of the FACIT NC1 domains is thought to be similar to that of the fibrillar procollagen C-propeptides in chain association. However, the C-propeptides of fibrillar procollagens are much larger, being approx. 260 amino acids compared with 15–75 residues in the FACITs [5]. Also, the short C-telopeptide in procollagen III was found to be unimportant in triple helix nucleation, while a minimum of two hydroxyproline-containing Gly-Xaa-Tyr repeats at the C-terminus of the triple helix were required. It has been suggested that interchain disulfide bond formation within the C-propeptide or the C-telopeptide is not required for chain association and triple helix formation, leaving a possible stabilizing function for these bonds [21,22]. Use of chimaeric procollagens formed of α1(III)- and α2(I)-chains identified a discontinuous sequence of 15 amino acids, which directs procollagen self-association. This 15-residue chain recognition motif, in the C-propeptide, contains the information needed for the procollagen chains to discriminate between each other and assemble in a type-specific manner [23].

α-Helical coiled-coil motifs have been shown to play an essential role in trimerization of collagen-like proteins such as collectins [24] and transmembrane collagens [2527]. Such motifs are characterized by heptad repeats abcdefg in which residues a and d are hydrophobic [28]. It has previously been shown that such motifs are present in almost all members of the collagen superfamily [29]. In the case of the fibrillar procollagens, for example, four such heptad repeats are found at the beginning of the C-propeptide domain. In the FACITs, two partially overlapping sets of heptad repeats are found in the NC2 domain, but none in the NC1 domain.

In the present study, we have addressed the questions of collagen IX chain association, assembly and triple helix formation by expressing collagen IX deletion constructs in insect cells. Our results demonstrate that collagen IX α-chain variants starting at the NC4 domain and ending at the NC2 domain (or NC2 replaced by NC1) can associate into trimers and form a triple helix starting from the COL2 domain. In contrast, the α-chains of short variants starting at COL2 and ending at NC2 (or NC2 replaced by NC1) were not able to trimerize, suggesting that structural information provided by sequences N-terminal to COL2 is required for complete folding and stabilization of collagen IX molecules.

MATERIALS AND METHODS

Construction of plasmids

Previously constructed FL α1(IX), α2(IX) and α3(IX) cDNAs [16] in pVL1392 vectors (Pharmingen) were used as templates for PCRs to generate various truncated variants of collagen IX (Figure 1). All the oligonucleotides for preparing the required cDNAs for the collagen IX α-chains are shown in Table 1. PCRs were performed with thermal cycling of 45 s at 95 °C, 20–30 s at 54–62 °C and 30–45 s at 68–72 °C for 15–36 cycles followed by a final extension at 72 °C for 5 min. For constructs that were generated by consecutive PCRs, 1 μl of the first reaction was used as a template for the second reaction, and 1 μl of the second was used as a template for the third reaction. The second and third amplifications were performed under the conditions described above using 25 cycles. The oligonucleotides used to prepare the various cDNAs for the α1(IX)- and α2(IX)-chain variants contained designed NotI cleavage sites, while the oligonucleotides for α3(IX) contained designed XbaI or EagI cleavage sites. PCR products were electrophoresed and purified (QIAEXII Gel Extraction kit; Qiagen) and then ligated into the pGEM-T-Easy vector (Promega). Amplified cDNAs together with pVL1392 vector were digested with NotI and EagI (COL9A1 and COL9A2) and with XbaI or EagI (COL9A3), and the cDNAs were ligated into pVL1392 vectors. All the constructed plasmids were sequenced (ABI Prism BigDye Sequencing kit, ABI Prism 377 XL; Applied Biosystems).

Schematic representation of the collagen IX α-chain variants used in the present study

Figure 1
Schematic representation of the collagen IX α-chain variants used in the present study

FL collagen IX (A) consists of three COL domains (COL1–COL3) and four NC domains (NC1–NC4), named from the C-terminus of the molecule. Two cysteine residues (Cys) located in the COL1–NC1 junction are conserved in FACITs. NC4–NC2 collagen IX (B) consists of NC4, COL3, NC3, COL2 and NC2 domains, while NC4–COL2/NC1 collagen IX (C) consists of NC4, COL3, NC3, COL2 and NC1 domains. In COL2/NC1 collagen IX (D), the α1(IX)-chain SS is attached to the COL2 domain followed by the NC1 domain, which contains one of the conserved cysteine residues. COL2/C/NC1 collagen IX (E) is composed of SS and COL2 and NC1 domains with an engineered cysteine residue at the C-terminus of the COL2 domain in addition to the conserved cysteine residue in the NC1 domain. COL2/NC2 collagen IX (F) consists of SS and COL2 and NC2 domains.

Figure 1
Schematic representation of the collagen IX α-chain variants used in the present study

FL collagen IX (A) consists of three COL domains (COL1–COL3) and four NC domains (NC1–NC4), named from the C-terminus of the molecule. Two cysteine residues (Cys) located in the COL1–NC1 junction are conserved in FACITs. NC4–NC2 collagen IX (B) consists of NC4, COL3, NC3, COL2 and NC2 domains, while NC4–COL2/NC1 collagen IX (C) consists of NC4, COL3, NC3, COL2 and NC1 domains. In COL2/NC1 collagen IX (D), the α1(IX)-chain SS is attached to the COL2 domain followed by the NC1 domain, which contains one of the conserved cysteine residues. COL2/C/NC1 collagen IX (E) is composed of SS and COL2 and NC1 domains with an engineered cysteine residue at the C-terminus of the COL2 domain in addition to the conserved cysteine residue in the NC1 domain. COL2/NC2 collagen IX (F) consists of SS and COL2 and NC2 domains.

Table 1
Oligonucleotides used in the PCRs to generate the various collagen IX deletion constructs

Designed restriction sites are underlined.

COL IX construct/primer Primer sequence (5′–3′) Restriction enzyme 
9A1 NC4–NC2   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A1NC2R TGCAGCCGGCCGTTATGAGTCTGGACGCTTAAGACTGGCAG EagI 
9A2 NC4–NC2   
 9A2F TCTGCCGTCGGTGCGGCCGCGGACACGC NotI and EagI 
 9A2NC2R TGCAGCCGGCCGTTACAGGGCTTCCCGCTTGGCACTC EagI 
9A3 NC4–NC2   
 9A3F CCCGACGCCGCAGTCTAGACTCCGCCACGC XbaI 
 9A3NC2R TTCCTCTCTAGATTAGGGTGCCAAAGGCTTCCTTAGGT XbaI 
9A1 NC4–COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A1NC1R1 CTGCATGGTGCAGGAGGCTGGCTCTGCTCTACCCGGAGGGCCCTG  
 9A1NC1R2 CCCTTTGTTAAATGCTCGCTGACCAGCCTGCATGGTGCAGGAGGCTG  
 9A1NC1R3 AGTCATCGGCCGTCAAGGGTCAGGCCCTTTGTTAAATGCTCGCTG EagI 
9A2 NC4–COL2/NC1   
 9A2F TCTGCCGTCGGTGCGGCCGCGGACACGC NotI and EagI 
 9A2NC1R1 GGCCGAAGCTCCAAGGCAGGCGGCAGGTTCATCCCGGCCCTCCACGCCCTG  
 9A2NC1R2 TCCAGGCTCTGTAAGGCGGGCAGAGGCATAGGCCGAAGCTCCAAGGCAGGC  
 9A2NC1R3 AGTCATCGGCCGTCAAGGCCCCTTGATGGATCCAGGCTCTGTAAGGCGGGCA EagI 
9A3 NC4–COL2/NC1   
 9A3F CCCGACGCCGCAGTCTAGACTCCGCCACGC XbaI 
 9A3NC1R1 CACGGCTCCTTGGCAGGCTGAGGTGTCCTCCTTCCCCGGAACTCCCGG  
 9A3NC1R2 GCCTGATTTCTCCCCGACCCCTCCTAACACGGCTCCTTGGCAGGCTGA  
 9A3NC1R3 AGTCATTCTAGATTATGAGCTTCGAGAGCCTGATTTCTCCCCGACCCCT XbaI 
9A1 SS   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 SSR AGTCATACTAGTGACAGCTGCAGATGCCCA SpeI 
9A1 COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A1NC1R1 CTGCATGGTGCAGGAGGCTGGCTCTGCTCTACCCGGAGGGCCCTG  
 9A1NC1R2 CCCTTTGTTAAATGCTCGCTGACCAGCCTGCATGGTGCAGGAGGCTG  
 9A1NC1R3 AGTCATCGGCCGTCAAGGGTCAGGCCCTTTGTTAAATGCTCGCTG EagI 
9A2 COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A2NC1R1 GGCCGAAGCTCCAAGGCAGGCGGCAGGTTCATCCCGGCCCTCCACGCCCTG  
 9A2NC1R2 TCCAGGCTCTGTAAGGCGGGCAGAGGCATAGGCCGAAGCTCCAAGGCAGGC  
 9A2NC1R3 AGTCATCGGCCGTCAAGGCCCCTTGATGGATCCAGGCTCTGTAAGGCGGGCA EagI 
9A3 COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A3NC1R1 CACGGCTCCTTGGCAGGCTGAGGTGTCCTCCTTCCCCGGAACTCCCGG  
 9A3NC1R2 GCCTGATTTCTCCCCGACCCCTCCTAACACGGCTCCTTGGCAGGCTGA  
 9A3NC1R3 AGTCATCGGCCGTTATGAGCTTCGAGAGCCTGATTTCTCCCCGACCCCT EagI 
9A1 COL2/C/NC1   
 9A1COL2F AGTCATACTAGTGGTCGCTCAGGATATCCAGGCCT SpeI 
 9A1NC1R1C CTGCATGGTGCAGGAGGCTGGCTCACATCTACCCGGAGGGCCCTGGAC  
 9A1NC1R2 CCCTTTGTTAAATGCTCGCTGACCAGCCTGCATGGTGCAGGAGGCTG  
 9A1NC1R3 AGTCATCGGCCGTCAAGGGTCAGGCCCTTTGTTAAATGCTCGCTG EagI 
9A2 COL2/C/NC1   
 9A2COL2F AGTCATACTAGTGGAATGAAAGGTCCCCCAGGGCT SpeI 
 9A2NC1R1C GGCCGAAGCTCCAAGGCAGGCGGCAGGTTCACACCGGCCCTCCACGCCCTGTCT  
 9A2NC1R2 TCCAGGCTCTGTAAGGCGGGCAGAGGCATAGGCCGAAGCTCCAAGGCAGGC  
 9A2NC1R3 AGTCATCGGCCGTCAAGGCCCCTTGATGGATCCAGGCTCTGTAAGGCGGGCA EagI 
9A3 COL2/C/NC1   
 9A3COL2F AGTCATACTAGTGGTCCCCCAGGGCCCCCTGGAATGC SpeI 
 9A3NC1R1C CACGGCTCCTTGGCAGGCTGAGGTGTCGCACTTCCCCGGAACTCCCGGCT  
 9A3NC1R2 GCCTGATTTCTCCCCGACCCCTCCTAACACGGCTCCTTGGCAGGCTGA  
 9A3NC1R3 AGTCATCGGCCGTTATGAGCTTCGAGAGCCTGATTTCTCCCCGACCCCT EagI 
9A1 COL2/NC2   
 9A1COL2F AGTCATACTAGTGGTCGCTCAGGATATCCAGGCCT SpeI 
 9A1NC2R1 AGTCATCGGCCGTCATGAGTCTGGACGCTTAAGAC EagI 
9A2 COL2/NC2   
 9A2COL2F AGTCATACTAGTGGAATGAAAGGTCCCCCAGGGCT SpeI 
 9A2NC2R1 AGTCATCGGCCGTCACAGGGCTTCCCGCTTGGCAC EagI 
9A3 COL2/NC2   
 9A3COL2F AGTCATACTAGTGGTCCCCCAGGGCCCCCTGGAATGC SpeI 
 9A3NC2R1 AGTCATCGGCCGTCAGGGTGCCAAAGGCTTCCTTAG EagI 
COL IX construct/primer Primer sequence (5′–3′) Restriction enzyme 
9A1 NC4–NC2   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A1NC2R TGCAGCCGGCCGTTATGAGTCTGGACGCTTAAGACTGGCAG EagI 
9A2 NC4–NC2   
 9A2F TCTGCCGTCGGTGCGGCCGCGGACACGC NotI and EagI 
 9A2NC2R TGCAGCCGGCCGTTACAGGGCTTCCCGCTTGGCACTC EagI 
9A3 NC4–NC2   
 9A3F CCCGACGCCGCAGTCTAGACTCCGCCACGC XbaI 
 9A3NC2R TTCCTCTCTAGATTAGGGTGCCAAAGGCTTCCTTAGGT XbaI 
9A1 NC4–COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A1NC1R1 CTGCATGGTGCAGGAGGCTGGCTCTGCTCTACCCGGAGGGCCCTG  
 9A1NC1R2 CCCTTTGTTAAATGCTCGCTGACCAGCCTGCATGGTGCAGGAGGCTG  
 9A1NC1R3 AGTCATCGGCCGTCAAGGGTCAGGCCCTTTGTTAAATGCTCGCTG EagI 
9A2 NC4–COL2/NC1   
 9A2F TCTGCCGTCGGTGCGGCCGCGGACACGC NotI and EagI 
 9A2NC1R1 GGCCGAAGCTCCAAGGCAGGCGGCAGGTTCATCCCGGCCCTCCACGCCCTG  
 9A2NC1R2 TCCAGGCTCTGTAAGGCGGGCAGAGGCATAGGCCGAAGCTCCAAGGCAGGC  
 9A2NC1R3 AGTCATCGGCCGTCAAGGCCCCTTGATGGATCCAGGCTCTGTAAGGCGGGCA EagI 
9A3 NC4–COL2/NC1   
 9A3F CCCGACGCCGCAGTCTAGACTCCGCCACGC XbaI 
 9A3NC1R1 CACGGCTCCTTGGCAGGCTGAGGTGTCCTCCTTCCCCGGAACTCCCGG  
 9A3NC1R2 GCCTGATTTCTCCCCGACCCCTCCTAACACGGCTCCTTGGCAGGCTGA  
 9A3NC1R3 AGTCATTCTAGATTATGAGCTTCGAGAGCCTGATTTCTCCCCGACCCCT XbaI 
9A1 SS   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 SSR AGTCATACTAGTGACAGCTGCAGATGCCCA SpeI 
9A1 COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A1NC1R1 CTGCATGGTGCAGGAGGCTGGCTCTGCTCTACCCGGAGGGCCCTG  
 9A1NC1R2 CCCTTTGTTAAATGCTCGCTGACCAGCCTGCATGGTGCAGGAGGCTG  
 9A1NC1R3 AGTCATCGGCCGTCAAGGGTCAGGCCCTTTGTTAAATGCTCGCTG EagI 
9A2 COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A2NC1R1 GGCCGAAGCTCCAAGGCAGGCGGCAGGTTCATCCCGGCCCTCCACGCCCTG  
 9A2NC1R2 TCCAGGCTCTGTAAGGCGGGCAGAGGCATAGGCCGAAGCTCCAAGGCAGGC  
 9A2NC1R3 AGTCATCGGCCGTCAAGGCCCCTTGATGGATCCAGGCTCTGTAAGGCGGGCA EagI 
9A3 COL2/NC1   
 9A1F ACTCCCTTGCGGCCGCTTCTTCATAGG NotI and EagI 
 9A3NC1R1 CACGGCTCCTTGGCAGGCTGAGGTGTCCTCCTTCCCCGGAACTCCCGG  
 9A3NC1R2 GCCTGATTTCTCCCCGACCCCTCCTAACACGGCTCCTTGGCAGGCTGA  
 9A3NC1R3 AGTCATCGGCCGTTATGAGCTTCGAGAGCCTGATTTCTCCCCGACCCCT EagI 
9A1 COL2/C/NC1   
 9A1COL2F AGTCATACTAGTGGTCGCTCAGGATATCCAGGCCT SpeI 
 9A1NC1R1C CTGCATGGTGCAGGAGGCTGGCTCACATCTACCCGGAGGGCCCTGGAC  
 9A1NC1R2 CCCTTTGTTAAATGCTCGCTGACCAGCCTGCATGGTGCAGGAGGCTG  
 9A1NC1R3 AGTCATCGGCCGTCAAGGGTCAGGCCCTTTGTTAAATGCTCGCTG EagI 
9A2 COL2/C/NC1   
 9A2COL2F AGTCATACTAGTGGAATGAAAGGTCCCCCAGGGCT SpeI 
 9A2NC1R1C GGCCGAAGCTCCAAGGCAGGCGGCAGGTTCACACCGGCCCTCCACGCCCTGTCT  
 9A2NC1R2 TCCAGGCTCTGTAAGGCGGGCAGAGGCATAGGCCGAAGCTCCAAGGCAGGC  
 9A2NC1R3 AGTCATCGGCCGTCAAGGCCCCTTGATGGATCCAGGCTCTGTAAGGCGGGCA EagI 
9A3 COL2/C/NC1   
 9A3COL2F AGTCATACTAGTGGTCCCCCAGGGCCCCCTGGAATGC SpeI 
 9A3NC1R1C CACGGCTCCTTGGCAGGCTGAGGTGTCGCACTTCCCCGGAACTCCCGGCT  
 9A3NC1R2 GCCTGATTTCTCCCCGACCCCTCCTAACACGGCTCCTTGGCAGGCTGA  
 9A3NC1R3 AGTCATCGGCCGTTATGAGCTTCGAGAGCCTGATTTCTCCCCGACCCCT EagI 
9A1 COL2/NC2   
 9A1COL2F AGTCATACTAGTGGTCGCTCAGGATATCCAGGCCT SpeI 
 9A1NC2R1 AGTCATCGGCCGTCATGAGTCTGGACGCTTAAGAC EagI 
9A2 COL2/NC2   
 9A2COL2F AGTCATACTAGTGGAATGAAAGGTCCCCCAGGGCT SpeI 
 9A2NC2R1 AGTCATCGGCCGTCACAGGGCTTCCCGCTTGGCAC EagI 
9A3 COL2/NC2   
 9A3COL2F AGTCATACTAGTGGTCCCCCAGGGCCCCCTGGAATGC SpeI 
 9A3NC2R1 AGTCATCGGCCGTCAGGGTGCCAAAGGCTTCCTTAG EagI 

Construction of NC4–NC2 plasmids

9A1F and 9A1NC2R, and 9A2F and 9A2NC2R oligonucleotides were used to prepare the COL9A1 and COL9A2 NC4–NC2 constructs. 9A3F and 9A3NC2R oligonucleotides were used to prepare the COL9A3 NC4–NC2 construct (Figure 1, construct B).

Construction of NC4–COL2/NC1 plasmids

9A1F and 9A1NC1R1 oligonucleotides for COL9A1, 9A2F and 9A2NC1R1 oligonucleotides for COL9A2, and 9A3F and 9A3NC1R1 oligonucleotides for COL9A3 were used in the first PCR reaction to build the NC4–COL2 domains. The NC1 domain for all the chains was constructed by two consecutive PCRs using 9A1F/9A2F/9A3F and 9A1NC1R2/9A2NC1R2/9A3NC1R2 oligonucleotides in the first reaction, and 9A1F/9A2F/9A3F and 9A1NC1R3/9A2NC1R3/9A3NC1R3 oligonucleotides in the second reaction (Figure 1, construct C).

Construction of COL2/C/NC1 and 9A1 SS (signal sequence) plasmids

Three consecutive PCRs were performed to generate COL2/C/NC1 cDNAs for the three collagen IX α-chains (Figure 1, construct E). COL9A1 cDNA was generated with 9A1COL2F and 9A1NC1R1C, 9A1COL2F and 9A1NC1R2, and 9A1COL2F and 9A1NC1R3. 9A2COL2F and 9A2NC1R1C, 9A2COL2F and 9A2NC1R2, 9A2COL2F and 9A2NC1R3 were used for COL9A2 COL2/C/NC1 and 9A3COL2F and 9A3NC1R1C, 9A3COL2F and 9A3NC1R2, and 9A3COL2F and 9A3NC1R3 for COL9A3 COL2/C/NC1 respectively. 9A1NC1R1C, 9A2NC1R1C and 9A3NC1R1C contained a codon for cysteine instead of alanine, aspartic acid or glutamic acid at the 3′-end of COL2 respectively.

9A1F and SSR oligonucleotides were used to amplify the α1(IX)-chain SS (9A1 SS). The 9A1 SS and COL9A1 cDNAs were obtained by digestion with NotI and SpeI, and the cDNAs were ligated into the pGEM-T vector. The plasmids with the 9A1 SS and COL9A2 cDNAs were digested with SpeI, and the cDNAs were ligated into the pGEM-T-Easy vector. SpeI- and EagI-digested COL9A3 cDNA was ligated into NotI- and SpeI-digested pGEM-T vector. This plasmid and the 9A1 SS plasmid were then digested with SpeI and SacI and the inserts were ligated into the pGEM-T-Easy vector. All the resulting plasmids and the pVL1392 vector were digested with NotI and EagI, and the cDNAs were ligated into pVL1392.

Construction of COL2/NC1 plasmids

COL2/C/NC1 cDNAs in pVL1392 were used as templates for PCR to amplify the 9A1 SS, and the COL2 and NC1 domains for COL9A1 and COL9A2. The PCR was performed in three consecutive steps using 9A1F and 9A1NC1R1, 9A1F and 9A1NC1R2, 9A1F and 9A1NC1R3 for COL9A1 COL2/NC1, 9A2F and 9A2NC1R1, 9A2F and 9A2NC1R2, 9A2F and 9A2NC1R3 for COL9A2 COL2/NC1. The cDNAs were transferred to the pVL1392 vector as indicated above for the COL2/C/NC1 construct. The corresponding domains for COL9A3 were amplified in three steps using FL α3(IX) cDNA as a template with the following primer pairs: 9A3COL2F and 9A3NC1R1, 9A3COL2F and 9A3NC1R2, and 9A3COL2F and 9A3NC1R3 (Figure 1, construct D). COL9A3 COL2/NC1 and 9A1 SS cDNAs were ligated together as shown above for the COL2/C/NC1 construct.

Construction of COL2–NC2 plasmids

To amplify the α1(IX), α2(IX) and α3(IX) COL2 and NC2 cDNAs, 9A1COL2F together with 9A1NC2R1, 9A2COL2F with 9A2NC2R1, and 9A3COL2F with 9A3NC2R1 were used for PCR (Figure 1, construct F). The products were ligated into the pGEM-T vector. The resulting plasmids and the 9A1 SS plasmid were digested with SpeI. The COL2/NC2 and 9A1 SS cDNAs were ligated into the pGEM-T-Easy vector. The resulting plasmids and pVL1392 vector were digested with NotI and EagI and the inserts were ligated into pVL1392.

Generation of the recombinant viruses

The cDNAs constructed above (15 total) were co-transfected into Spodoptera frugiperda (Sf9; Invitrogen) insect cells with a modified Autographa californica nuclear polyhedrosis virus by using the BaculoGold™ Transfection kit (Pharmingen). Sf9 cells were cultured in TNM-FH medium (Sigma) supplemented with 10% (v/v) FBS (fetal bovine serum; Atlanta Biologicals or Bioclear) at 27 °C and were seeded at ∼60% confluency in tissue culture plates for the transfection. Viral pools were plaque purified and amplified three to four times, using either monolayer (∼70% confluency) or suspension cultures, before use in the expressions.

Expression of recombinant proteins

Trichoplusia ni (High Five; Invitrogen) insect cells were cultured in monolayers or in suspension in Sf-900 II SFM medium (Gibco) with or without 5% FBS at 27 °C. Prior to the infection, the insect cells were seeded at ∼70% confluency in monolayers and (1–1.5)×106 cells/ml for expression in suspension. The cells were co-infected with one to three viruses containing cDNAs for the various α1(IX)-, α2(IX)- and α3(IX)-chains and a double promoter virus 4PHαβ for the α- and β-subunits of human prolyl 4-hydroxylase cDNAs [30], with an MOI (multiplicity of infection) of 2 for each virus. Ascorbate (80 μg/ml) was added daily to the culture medium.

Isolation and purification of the recombinant protein

After 72 h of infection, the High Five cells were detached from the culture plates by pipetting and harvested by centrifugation at 600–1000 gfor 5–10 min. The medium samples from monolayer cultures were analysed by SDS/PAGE followed by Western blotting with the pan collagen-specific monoclonal antibody 95D1A [31].

Cells cultured in suspension were harvested by centrifugation at 1000 g for 30 min at 4 °C and the medium was further clarified at 10000 g for 30 min. Collagen IX molecules and their variants were precipitated from the culture medium by adding solid ammonium sulfate to 26% saturation and placing the mixture on ice while stirring for 1 h. The precipitate was collected by centrifugation at 10000 g for 30 min at 4 °C and dissolved in 0.4 M NaCl, 2 M urea, 10 mM EDTA and 0.1 M Tris/HCl (pH 7) buffer at 4 °C overnight. The protein solution was clarified by centrifugation at 12000 gfor 30 min at 4 °C and then purified by gel filtration (Superdex™ 200; Amersham Biosciences) in the same buffer. The protein was further purified by cation-exchange chromatography (HiTrap™ SP HP or SP FF; Amersham Biosciences) in 50 mM Pipes, 20 mM NaCl and 2 M urea (pH 6.5) buffer, eluting with an increasing NaCl concentration (from 0.02 to 1 M NaCl). Collagen IX-containing fractions were pooled and dialysed against 50 mM acetic acid and then concentrated. The concentrated samples were analysed by SDS/PAGE followed by staining with Gelcode® Blue stain (Pierce) or Bio-Safe™ Coomassie (Bio-Rad) or Western blotting with the monoclonal antibody 95D1A. The purified material was then subjected to amino acid analysis in an Applied Biosystems 421 analyser. The composition of the material usually corresponded well to the calculated values for human collagen IX [16].

Chemical cross-linking

Covalent cross-linking of the short variants COL2–NC1, COL2/C/NC1 and COL2–NC2 was performed using BS3 [bis(sulfosuccinimidyl) suberate; Sigma]. Increasing amounts of BS3 (0, 0.1, 0.5, 1 or 2 mM final concentration) prepared in 5 mM sodium citrate (pH 5) were added to conditioned medium dialysed against 1×PBS for 1 h at room temperature (21 °C). Addition of SDS/PAGE loading buffer containing Tris/HCl (0.5 M) inhibited the reaction. Samples were subjected to SDS/PAGE, which was carried out without thiol reduction. Proteins were transferred on to PVDF membranes by Western blotting, and the presence of cross-linked dimers or trimers was identified by using the 95D1A monoclonal antibody.

CD analysis of purified recombinant proteins

Purified heterotrimeric FL collagen IX (α1–α2–α3) and heterotrimeric NC4–NC2 (α1–α2–α3) were analysed for their triple helicity using CD (J810; Jasco). Proteins in 50 mM acetic acid were placed in a rectangular cuvette (1 mm path length) at 5 °C and their far-UV spectra were measured in the 190–250 nm region with a scan speed of 20 nm/min and two accumulations per spectrum. The spectrum of 50 mM acetic acid was subtracted from the collagen IX spectra, followed by smoothing.

RESULTS

Association of FL, NC4–NC2 and NC4–COL2/NC1 collagen IX α-chains

To study the chain association of FL collagen IX (Figures 1 and 2) and the ability of the NC2 domain to direct chain association, a cDNA construct NC4–NC2, consisting of cDNAs for the NC4, COL3, NC3, COL2 and NC2 domains was prepared for each collagen IX α-chain (Figures 1 and 2). A corresponding construct, NC4–COL2/NC1, was also prepared in order to determine whether NC1 could direct the chain association at COL2 (Figures 1 and 2). These constructs were prepared by PCR using primers described in Table 1. NC4–NC2, NC4–COL2/NC1 and FL cDNA constructs in the baculovirus expression vectors were co-transfected into insect cells with a nuclear polyhedrosis virus, and resulting viral pools were plaque purified and amplified. Viruses were tested for expression on adherent insect cell cultures by performing SDS/PAGE and Western blotting, using 95D1A antibody, for the conditioned medium. The best virus for each chain of the three variants of collagen IX was used for expression, on adherent cultures of insect cells, together with virus for prolyl 4-hydroxylase with an MOI of 2 for each virus. After expression (for 72 h), conditioned medium of all possible combinations of FL, NC4–NC2 and NC4–COL2/NC1 chains were subjected to SDS/PAGE and Western blotting under reducing (Figures 3A, 4A and 5A) and non-reducing (Figures 3B, 4B and 5B) conditions. Each chain of the three variants could be detected in every combination whenever the corresponding virus was used in the expression. All the chains were of the expected sizes (Figures 3A, 4A and 5A). When individual chains were expressed together with prolyl 4-hydroxylase, both FL and NC4–NC2 chains were found mainly as monomeric and dimeric molecules (Figures 3B and 4B). Small amounts of FL α1-chain homotrimers were observed, but not for the α2- and α3-chains. In contrast, for the NC4–NC2 variants, both the α1- and α3-chains formed homotrimers, but not α2. The tendency to form homotrimers increased even more when the NC2 domains were replaced by NC1 in the NC4–NC2 variants, thereby forming the NC4–COL2/NC1 variants, where homotrimers were observed for all three chains (Figure 5B). When FL chains were co-expressed in pairs (α1/α2, α1/α3 or α2/α3), trimers were only observed when the α1-chain was present (Figure 3B). When all three chains were co-expressed, as previously described [16], a new trimer band was observed (indicated by the arrow in Figure 3B) which corresponds to the heterotrimeric molecule consisting of all three chains. In contrast, when different NC4–NC2 chains were co-expressed (Figure 4B), heterotrimers were observed with the pairwise combinations but there was no indication of a new heterotrimeric band when all three chains were co-expressed. This was also the case for mixtures of NC4–COL2/NC1 chains (Figure 5B). These results demonstrate that trimerization can occur when the COL1 and NC1 domains are deleted, but they also indicate that collagen IX α-chains lacking NC1 and COL1 domains can assemble in every possible chain combination and form disulfide-bonded trimers. Thus the COL1 and NC1 domains are not essential for trimerization, although chain selection specificity is significantly reduced in the absence of COL1 and NC1.

Amino acid sequences of the N- and C-terminal regions of the collagen IX α-chain variants

Figure 2
Amino acid sequences of the N- and C-terminal regions of the collagen IX α-chain variants

Collagenous regions are presented in boldface to separate them from non-collagenous regions. Extra amino acids generated to the C-terminus of the SS are presented in italics. The generated cysteine residues at the COL2–NC1 junction are presented in boldface italics.

Figure 2
Amino acid sequences of the N- and C-terminal regions of the collagen IX α-chain variants

Collagenous regions are presented in boldface to separate them from non-collagenous regions. Extra amino acids generated to the C-terminus of the SS are presented in italics. The generated cysteine residues at the COL2–NC1 junction are presented in boldface italics.

Analysis of FL collagen IX chain association

Figure 3
Analysis of FL collagen IX chain association

Chain association of FL collagen IX was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for the collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at MOI 2 for each virus. All possible chain compositions were subjected to SDS/10% PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Chains are identified by the notation 9A1, 9A2 and 9A3, corresponding to the α1-, α2- and α3-chains respectively. Lane 1, 9A1; lane 2, 9A2; lane 3, 9A3; lane 4, 9A1/2; lane 5, 9A1/3; lane 6, 9A2/3; lane 7, 9A1/2/3. The migration position of the FL α1/α2/α3 heterotrimer is indicated by an arrow (B). The migration positions of monomers, dimers and trimers are also indicated (B).

Figure 3
Analysis of FL collagen IX chain association

Chain association of FL collagen IX was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for the collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at MOI 2 for each virus. All possible chain compositions were subjected to SDS/10% PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Chains are identified by the notation 9A1, 9A2 and 9A3, corresponding to the α1-, α2- and α3-chains respectively. Lane 1, 9A1; lane 2, 9A2; lane 3, 9A3; lane 4, 9A1/2; lane 5, 9A1/3; lane 6, 9A2/3; lane 7, 9A1/2/3. The migration position of the FL α1/α2/α3 heterotrimer is indicated by an arrow (B). The migration positions of monomers, dimers and trimers are also indicated (B).

Analysis of NC4–NC2 collagen IX chain association

Figure 4
Analysis of NC4–NC2 collagen IX chain association

Chain association of the long variant NC4–NC2 was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for the collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at an MOI of 2 for each virus. All possible chain compositions were subjected to SDS/10%PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Chains are identified by the notation 9A1, 9A2 and 9A3, corresponding to the α1-, α2- and α3-chains respectively. Lane 1, 9A1; lane 2, 9A2; lane 3, 9A3; lane 4, 9A1/2; lane 5, 9A1/3; lane 6, 9A2/3; lane 7, 9A1/2/3. The migration positions of monomers, dimers and trimers are indicated (B).

Figure 4
Analysis of NC4–NC2 collagen IX chain association

Chain association of the long variant NC4–NC2 was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for the collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at an MOI of 2 for each virus. All possible chain compositions were subjected to SDS/10%PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Chains are identified by the notation 9A1, 9A2 and 9A3, corresponding to the α1-, α2- and α3-chains respectively. Lane 1, 9A1; lane 2, 9A2; lane 3, 9A3; lane 4, 9A1/2; lane 5, 9A1/3; lane 6, 9A2/3; lane 7, 9A1/2/3. The migration positions of monomers, dimers and trimers are indicated (B).

Analysis of NC4–COL2/NC1 collagen IX chain association

Figure 5
Analysis of NC4–COL2/NC1 collagen IX chain association

Chain association of NC4–COL2/NC1 collagen IX was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at MOI 2 for each virus. All possible chain compositions were subjected to SDS/10% PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Lane 1, 9A1; lane 2, 9A2; lane 3, 9A3; lane 4, 9A1/2; lane 5, 9A1/3; lane 6, 9A2/3; lane 7, 9A1/2/3. Monomers, dimers, and trimers of the chains are indicated (B).

Figure 5
Analysis of NC4–COL2/NC1 collagen IX chain association

Chain association of NC4–COL2/NC1 collagen IX was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at MOI 2 for each virus. All possible chain compositions were subjected to SDS/10% PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Lane 1, 9A1; lane 2, 9A2; lane 3, 9A3; lane 4, 9A1/2; lane 5, 9A1/3; lane 6, 9A2/3; lane 7, 9A1/2/3. Monomers, dimers, and trimers of the chains are indicated (B).

Purification and CD analysis of FL and NC4–NC2 collagen IX

To further study the collagen IX trimers found in the absence of COL1 and NC1, larger amounts of these proteins as well as FL collagen IX were produced in suspension cultures of insect cells with prolyl 4-hydroxylase. In addition to co-expression of the α1-, α2- and α3-chains of FL collagen IX, α1-, α2- and α3-chains of NC4–NC2 were also co-expressed. Collagen IX was precipitated from the conditioned medium (after 72 h) with 26% saturation of ammonium sulfate, and subjected to gel filtration and cation-exchange chromatography. Purified proteins were dialysed against 50 mM acetic acid and stored frozen. When proteins were analysed by SDS/PAGE under reducing conditions, FL collagen IX appeared as three distinct bands after Coomassie staining (Figure 6A), as shown previously [16]. When the samples were not reduced prior to SDS/PAGE, FL collagen IX appeared as a major band (Figure 6B) as also previously reported [16]. Collagen IX NC4–NC2 was not expressed as efficiently as the FL molecule, but monomers were detected under reducing conditions (Figure 6C) and trimeric molecules were seen under non-reducing gels (Figure 6D). The purified protein samples were then subjected to CD analysis in 50 mM acetic acid at +5 °C. Far-UV spectra were studied by performing wavelength scans from 250 to 190 nm. A typical collagen triple helical spectrum was observed for FL collagen IX (Figure 7A) with maximum ellipticity at approx. 221 nm and minimum ellipticity at approx. 198 nm [32]. The NC4–NC2 chain combination of collagen IX showed a similar spectrum (Figure 7B) as the FL protein. The CD spectrum for NC4–NC2 suggests that the trimeric molecules seen on the gels under non-reducing conditions were indeed triple helical and not only disulfide-bonded. The triple helicity of the NC4–COL2/NC1 variant was not analysed because of difficulties in purifying the protein.

SDS/PAGE of purified FL and NC4–NC2 collagen IX α-chains

Figure 6
SDS/PAGE of purified FL and NC4–NC2 collagen IX α-chains

Viruses for FL collagen IX α1-, α2- and α3-chains, or NC4–NC2 collagen α1-, α2- and α3-chains, were expressed together with virus for prolyl 4-hydroxylase in insect cell-suspension cultures. Secreted collagen IX was precipitated from the conditioned medium with ammonium sulfate and further purified with gel filtration and cation-exchange chromatography. Purified FL collagen IX was subjected to SDS/10% PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B). Purified NC4–NC2 trimers were subjected to SDS/10% PAGE under reducing (C) and non-reducing (D) conditions.

Figure 6
SDS/PAGE of purified FL and NC4–NC2 collagen IX α-chains

Viruses for FL collagen IX α1-, α2- and α3-chains, or NC4–NC2 collagen α1-, α2- and α3-chains, were expressed together with virus for prolyl 4-hydroxylase in insect cell-suspension cultures. Secreted collagen IX was precipitated from the conditioned medium with ammonium sulfate and further purified with gel filtration and cation-exchange chromatography. Purified FL collagen IX was subjected to SDS/10% PAGE under reducing conditions (A) and SDS/6% PAGE under non-reducing conditions (B). Purified NC4–NC2 trimers were subjected to SDS/10% PAGE under reducing (C) and non-reducing (D) conditions.

CD analysis of the purified FL and NC4–NC2 collagen IX α-chains

Figure 7
CD analysis of the purified FL and NC4–NC2 collagen IX α-chains

Purified collagen IX molecules formed from FL α1-, α2- and α3-chains (A) and from the NC4–NC2 regions of α1-, α2- and α3-chains (B) were studied by CD in the far UV (250–190 nm) width. Both proteins showed similar spectra at +5 °C in 50 mM acetic acid with maxima at approx. 221 nm and minima at approx. 198 nm, indicating a triple helical structure.

Figure 7
CD analysis of the purified FL and NC4–NC2 collagen IX α-chains

Purified collagen IX molecules formed from FL α1-, α2- and α3-chains (A) and from the NC4–NC2 regions of α1-, α2- and α3-chains (B) were studied by CD in the far UV (250–190 nm) width. Both proteins showed similar spectra at +5 °C in 50 mM acetic acid with maxima at approx. 221 nm and minima at approx. 198 nm, indicating a triple helical structure.

Association of COL2/NC1, COL2/NC2 and COL2/C/NC1 collagen IX α-chains

Since the above results showed that collagen IX chains could trimerize in the absence of the COL1 and NC1 domains, we asked the question whether the NC2 domain could direct trimerization of the COL2 domain, in the absence of domains NC4, COL3 and NC3. For this, short COL2/NC2 variants were prepared for all three collagen IX chains. We also prepared COL2/NC1 constructs for comparison. The COL2/NC1 construct was prepared by amplifying the COL2 domain encoding cDNA and simultaneously building the NC1 domain encoding cDNA to the 3′-end of COL2 by PCR (Figures 1 and 2 and Table 1). The COL2/NC2 construct was prepared by PCR using primers designed for the 5′-end of the COL2 domain encoding cDNA and the 3′-end of the NC2 domain encoding cDNA (Figures 1 and 2 and Table 1). The cloned SS of the α1(IX)-chain (9A1 SS) was added to the 5′-end of the constructs to ensure secretion of the resulting proteins. When analysed by SDS/PAGE and Western blotting under reducing conditions, monomers of the expected sizes were found, as shown for COL2/NC1 in Figure 8(A). Under non-reducing conditions (Figure 8B), both monomers and dimers were found, but not trimers. Similar results were obtained for COL2/NC2 (results not shown). However, it should be noted that the corresponding amino acid sequences for these constructs contained at most one cysteine residue per chain (see Figure 2). Therefore, if trimers were present, they would not be visible in non-reducing gels due to the lack of sufficient cysteine residues to form disulfide bridges linking all three chains. To circumvent this problem, we changed the last residue in the COL2 domain of each COL2/NC1 construct to cysteine, thereby mimicking the two cysteine residues separated by four amino acids found at the COL1–NC1 junction, which have been shown to be important in chain association. These constructs, called COL2/C/NC1, were prepared for each collagen IX α-chain, by amplifying the COL2 cDNA and building the NC1 cDNA by PCR and substituting a codon for alanine (α1), aspartic acid (α2) and glutamic acid (α3) by codons for cysteine in the 3′-end of the cDNA coding for the COL2 domain (Figures 1E and 2 and Table 1). As before, a cloned SS of the α1(IX)-chain (9A1 SS) was added to the 5′-end of the COL2/C/NC1 construct to ensure secretion. However, when conditioned medium was analysed by SDS/PAGE and Western blotting under non-reducing conditions, again only monomers and dimers were detected (Figure 8B).

Analysis of the collagen IX COL2/NC1 and COL2/C/NC1 chain association

Figure 8
Analysis of the collagen IX COL2/NC1 and COL2/C/NC1 chain association

Chain association of COL2/NC1 and COL2/C/NC1 was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at an MOI of 2 for each virus. All possible chain compositions of COL2/NC1 and COL2/C/NC1 were subjected to SDS/10% PAGE under reducing (A) and non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Lane 1, 9A1 COL2/NC1; lane 3, 9A2 COL2/NC1; lane 5, 9A3 COL2/NC1; lane 7, 9A1/2 COL2/NC1; lane 9, 9A1/3 COL2/NC1; lane 11, 9A2/3 COL2/NC1; lane 13, 9A1/2/3 COL2/NC1. The lanes with even numbers represent the corresponding COL2/C/NC1 variants. Monomers and dimers are indicated on the right side of the blot (B).

Figure 8
Analysis of the collagen IX COL2/NC1 and COL2/C/NC1 chain association

Chain association of COL2/NC1 and COL2/C/NC1 was studied using conditioned medium from adherent cultures of insect cells that were infected with one to three viruses for collagen IX α-chains together with the virus for prolyl 4-hydroxylase, at an MOI of 2 for each virus. All possible chain compositions of COL2/NC1 and COL2/C/NC1 were subjected to SDS/10% PAGE under reducing (A) and non-reducing conditions (B) followed by Western-blot analysis with the monoclonal antibody 95D1A. Lane 1, 9A1 COL2/NC1; lane 3, 9A2 COL2/NC1; lane 5, 9A3 COL2/NC1; lane 7, 9A1/2 COL2/NC1; lane 9, 9A1/3 COL2/NC1; lane 11, 9A2/3 COL2/NC1; lane 13, 9A1/2/3 COL2/NC1. The lanes with even numbers represent the corresponding COL2/C/NC1 variants. Monomers and dimers are indicated on the right side of the blot (B).

To further investigate the possible presence of trimers using the short COL2-containing variants, conditioned medium was subjected to covalent cross-linking using BS3 [29] to stabilize dimers and any trimers present. When analysed by SDS/PAGE and Western blotting however, only monomeric, dimeric or aggregated molecules with very-high-molecular-mass, but not trimers, were detected, with COL2/NC2, COL2/NC1 and COL2/C/NC1 constructs (results not shown). Thus the COL2/NC2 region alone appears unable to form a stable triple helix in the experimental conditions used, even after substituting NC2 for NC1 and introducing a cysteine residue at the end of COL2.

DISCUSSION

Collagen chain assembly is a complex multistep process in which three chains must properly associate, form a correctly aligned triple helix and hence stabilize the structure. Most chain association studies on collagens have focused on finding the minimal requirements for this process to happen and have been performed on fibrillar collagens. These studies showed a specific role for the C-propeptide region in chain recognition, while triple helix nucleation occurs at the C-terminus of the collagenous region, leaving only stabilizing roles for the C-telopeptides and disulfide bonds [2123]. Chain association studies with FACITs suggest that the NC1 domains share some of the functions of the C-propeptides, although the information for chain association and nucleation is thought to reside in the COL1–NC1 junction [14,15,1720]. Previous to the work reported here, nothing was known about whether FACITs can associate and form triple helices starting in other non-collagenous regions, e.g. at NC2, and whether the information in the COL1–NC1 junction is specific, or whether COL and NC can act independently or in different combinations.

In the present study, collagen IX chain association using FL α-chains and various deletion variant chains was studied. Collagen IX NC4–NC2 α-chains (i.e. devoid of both COL1 and NC1 domains) were found to trimerize in insect cells, as shown by SDS/PAGE and Western blotting under non-reducing conditions. Furthermore, replacement of NC2 in these variants by NC1 (to create collagen IX NC4–COL2/NC1 α-chains) also led to trimerization. This suggests that COL and NC domains may act in different combinations and that the information for association may not be junction-specific. Alternatively, association may take place at the NC3 domain and trimerization may proceed in the N to C direction, as is the case with collagen XIII [26]. In both cases, however, the results also suggested that the specificity of chain selection was diminished, since trimerization occurred with many different chain compositions, including homotrimers, unlike FL chains that preferentially formed heterotrimers. This observation is consistent with previous reports [18,19] supporting the role of the NC1–COL1 junction in chain selection of collagen IX. Thus, contrary to other regions in collagen IX, it seems that the NC1 and COL1 domains function as a pair, where both domains have possibly evolved together to nucleate the triple helix in a specific manner.

Trimers formed from a mixture of NC4–NC2 α1-, α2- and α3-chains were found to be triple helical, by CD analysis. The results are in agreement with the findings of Paassilta et al. [33] who found that a naturally occurring mutation in the COL1 domain leading to deletion of one Gly-Xaa-Yaa repeat, with normal α1(IX)- and α2(IX)-chains, did not affect trimerization and triple helix formation, suggesting that the NC2 domain of collagen IX can correct chain alignment. The fact that these deletions were neutral variants was very surprising, since similar deletions in collagen I are lethal [34,35].

In addition, we studied the role of the NC1 and NC2 domains in chain association in the context of short variants, by expressing chains composed of COL2 and NC1 as well as COL2 and NC2. COL2 was chosen as the shortest unit among the constructs because it was expected to be of sufficient length to ensure triple helical folding [30]. If NC1 has all the necessary information needed for collagen IX chains to associate, then chains composed of NC1 attached to COL2 should be able to trimerize and possibly also form triple helices. In the work carried out here, this was not the case; only dimeric or monomeric molecules were detected. Chains with NC1 appended to COL2 did not trimerize, even when the final residue in the COL2 domain was replaced by cysteine, thereby mimicking the two cysteine residues separated by four amino acids found at the COL1–NC1 junction. In addition, when COL2/NC2 chain association was analysed, only dimeric or monomeric molecules were detected. Therefore it appears that while COL1/NC1 fragments do have the ability to trimerize [1415,19], COL2/NC1 and COL2/NC2 do not.

In summary, we have shown that the COL1 and NC1 domains are not essential for trimerization of collagen IX chains. Moreover, NC2 domains appear to be unable to drive trimerization in the short COL2/NC2 variant chains. This is surprising in view of the identification of a coiled region in FACIT NC2 domains [29] and the fact that coiled-coil regions have been shown to induce trimerization in several collagens and collagen-like proteins. The FACIT NC2 coiled-coil region is unusual however, in that it consists of two partially overlapping sets of heptad repeats [32], and is therefore unable to form a continuous classical coiled coil, which might weaken trimerization ability. These observations suggest that trimerization in the absence of the COL1 and NC1 domains is a co-operative process, involving not only the COL2 and NC2 domains but also other domains such as COL3 and NC3. A similar argument holds for the long variants with NC1 following COL2 instead of NC2, since these also trimerize but not the corresponding short-chain variants. Furthermore, the observed results demonstrate that in the absence of the COL1 and NC1 domains, the specificity of chain association is diminished, which suggests that chain specificity is strongly influenced by these domains. Further research is required to determine whether the results obtained also apply for other FACITs.

We thank Helena Satulehto, Minta Lumme and Marjorie McCants for expert technical assistance.

Abbreviations

     
  • BS3

    bis(sulfosuccinimidyl) suberate

  •  
  • COL domain

    collagenous domain

  •  
  • ECM

    extracellular matrix

  •  
  • FACIT

    fibril-associated collagen with interrupted triple helices

  •  
  • FBS

    fetal bovine serum

  •  
  • FL

    full length

  •  
  • MOI

    multiplicity of infection

  •  
  • NC domain

    non-collagenous domain

  •  
  • SS

    signal sequence

References

References
1
Prockop
D. J.
Kivirikko
K. I.
Collagens: molecular biology, diseases, and potentials for therapy
Annu. Rev. Biochem.
1995
, vol. 
64
 (pg. 
403
-
434
)
2
Myllyharju
J.
Kivirikko
K. I.
Collagens, modifying enzymes and their mutations in humans, flies and worms
Trends Genet.
2004
, vol. 
20
 (pg. 
33
-
43
)
3
Myllyharju
J.
Kivirikko
K. I.
Collagens and collagen-related diseases
Ann. Med.
2001
, vol. 
33
 (pg. 
7
-
21
)
4
Mayne
R.
Brewton
R. G.
New members of the collagen superfamily
Curr. Opin. Cell Biol.
1993
, vol. 
5
 (pg. 
883
-
890
)
5
Shaw
L. M.
Olsen
B. R.
FACIT collagens: diverse molecular bridges in extracellular matrices
Trends Biochem. Sci.
1991
, vol. 
16
 (pg. 
191
-
194
)
6
van der Rest
M.
Mayne
R.
Type IX collagen proteoglycan from cartilage is covalently cross-linked to type II collagen
J. Biol. Chem.
1988
, vol. 
263
 (pg. 
1615
-
1618
)
7
Vasios
G.
Nishimura
I.
Konomi
H.
van der
R. M.
Ninomiya
Y.
Olsen
B. R.
Cartilage type IX collagen-proteoglycan contains a large amino-terminal globular domain encoded by multiple exons
J. Biol. Chem.
1988
, vol. 
263
 (pg. 
2324
-
2329
)
8
Vaughan
L.
Mendler
M.
Huber
S.
Bruckner
P.
Winterhalter
K. H.
Irwin
M. I.
Mayne
R.
D-periodic distribution of collagen type IX along cartilage fibrils
J. Cell Biol.
1988
, vol. 
106
 (pg. 
991
-
997
)
9
Diab
M.
Wu
J. J.
Eyre
D. R.
Collagen type IX from human cartilage: a structural profile of intermolecular cross-linking sites
Biochem. J.
1996
, vol. 
314
 (pg. 
327
-
332
)
10
Wu
J. J.
Woods
P. E.
Eyre
D. R.
Identification of cross-linking sites in bovine cartilage type IX collagen reveals an antiparallel type II–type IX molecular relationship and type IX to type IX bonding
J. Biol. Chem.
1992
, vol. 
267
 (pg. 
23007
-
23014
)
11
Wu
J. J.
Eyre
D. R.
Covalent interactions of type IX collagen in cartilage
Connect. Tissue Res.
1989
, vol. 
20
 (pg. 
241
-
246
)
12
Eyre
D. R.
Apon
S.
Wu
J. J.
Ericsson
L. H.
Walsh
K. A.
Collagen type IX: evidence for covalent linkages to type II collagen in cartilage
FEBS Lett.
1987
, vol. 
220
 (pg. 
337
-
341
)
13
Eyre
D. R.
Pietka
T.
Weis
M. A.
Wu
J. J.
Covalent cross-linking of the NC1 domain of collagen type IX to collagen type II in cartilage
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
2568
-
2574
)
14
Mechling
D. E.
Gambee
J. E.
Morris
N. P.
Sakai
L. Y.
Keene
D. R.
Mayne
R.
Bächinger
H. P.
Type IX collagen NC1 domain peptides can trimerize in vitro without forming a triple helix
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
13781
-
13785
)
15
Labourdette
L.
van der Rest
M.
Analysis of the role of the COL1 domain and its adjacent cysteine-containing sequence in the chain assembly of type IX collagen
FEBS Lett.
1993
, vol. 
320
 (pg. 
211
-
214
)
16
Pihlajamaa
T.
Perälä
M.
Vuoristo
M. M.
Nokelainen
M.
Bodo
M.
Schulthess
T.
Vuorio
E.
Timpl
R.
Engel
J.
Ala-Kokko
L.
Characterization of recombinant human type IX collagen. Association of alpha chains into homotrimeric and heterotrimeric molecules
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
22464
-
22468
)
17
Mazzorana
M.
Snellman
A.
Kivirikko
K. I.
van der Rest
M.
Pihlajaniemi
T.
Involvement of prolyl 4-hydroxylase in the assembly of trimeric minicollagen XII. Study in a baculovirus expression system
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
29003
-
29008
)
18
Mazzorana
M.
Gruffat
H.
Sergeant
A.
van der Rest
M.
Mechanisms of collagen trimer formation. Construction and expression of a recombinant minigene in HeLa cells reveals a direct effect of prolyl hydroxylation on chain assembly of type XII collagen
J. Biol. Chem.
1993
, vol. 
268
 (pg. 
3029
-
3032
)
19
Mazzorana
M.
Giry-Lozinguez
C.
van der Rest
M.
Trimeric assembly of collagen XII: effect of deletion of the C-terminal part of the molecule
Matrix Biol.
1995
, vol. 
14
 (pg. 
583
-
588
)
20
Mazzorana
M.
Cogne
S.
Goldschmidt
D.
Aubert-Foucher
E.
Collagenous sequence governs the trimeric assembly of collagen XII
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
27989
-
27998
)
21
Bulleid
N. J.
Dalley
J. A.
Lees
J. F.
The C-propeptide domain of procollagen can be replaced with a transmembrane domain without affecting trimer formation or collagen triple helix folding during biosynthesis
EMBO J.
1997
, vol. 
16
 (pg. 
6694
-
6701
)
22
Bulleid
N. J.
Wilson
R.
Lees
J. F.
Type-III procollagen assembly in semi-intact cells: chain association, nucleation and triple-helix folding do not require formation of inter-chain disulphide bonds but triple-helix nucleation does require hydroxylation
Biochem. J.
1996
, vol. 
317
 (pg. 
195
-
202
)
23
Lees
J. F.
Tasab
M.
Bulleid
N. J.
Identification of the molecular recognition sequence which determines the type-specific assembly of procollagen
EMBO J.
1997
, vol. 
16
 (pg. 
908
-
916
)
24
Hakansson
K.
Reid
K. B.
Collectin structure: a review
Protein Sci.
2000
, vol. 
9
 (pg. 
1607
-
1617
)
25
Latvanlehto
A.
Snellman
A.
Tu
H.
Pihlajaniemi
T.
Type XIII collagen and some other transmembrane collagens contain two separate coiled-coil motifs, which may function as independent oligomerization domains
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
37590
-
37599
)
26
Snellman
A.
Tu
H.
Väisänen
T.
Kvist
A. P.
Huhtala
P.
Pihlajaniemi
T.
A short sequence in the N-terminal region is required for the trimerization of type XIII collagen and is conserved in other collagenous transmembrane proteins
EMBO J.
2000
, vol. 
19
 (pg. 
5051
-
5059
)
27
Areida
S. K.
Reinhardt
D. P.
Muller
P. K.
Fietzek
P. P.
Kowitz
J.
Marinkovich
M. P.
Notbohm
H.
Properties of the collagen type XVII ectodomain. Evidence for N- to C-terminal triple helix folding
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
1594
-
1601
)
28
Kammerer
R. A.
Alpha-helical coiled-coil oligomerization domains in extracellular proteins
Matrix Biol.
1997
, vol. 
15
 (pg. 
555
-
565
)
29
McAlinden
A.
Smith
T. A.
Sandell
L. J.
Ficheux
D.
Parry
D. A.
Hulmes
D. J.
Alpha-helical coiled-coil oligomerization domains are almost ubiquitous in the collagen superfamily
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
42200
-
42207
)
30
Nokelainen
M.
Helaakoski
T.
Myllyharju
J.
Notbohm
H.
Pihlajaniemi
T.
Fietzek
P. P.
Kivirikko
K. I.
Expression and characterization of recombinant human type II collagens with low and high contents of hydroxylysine and its glycosylated forms
Matrix Biol.
1998
, vol. 
16
 (pg. 
329
-
338
)
31
Snellman
A.
Keränen
M. R.
Hägg
P. O.
Lamberg
A.
Hiltunen
J. K.
Kivirikko
K. I.
Pihlajaniemi
T.
Type XIII collagen forms homotrimers with three triple helical collagenous domains and its association into disulfide-bonded trimers is enhanced by prolyl 4-hydroxylase
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
8936
-
8944
)
32
Piez
K. A.
Sherman
M. R.
Characterization of the product formed by renaturation of alpha 1-CB2, a small peptide from collagen
Biochemistry
1970
, vol. 
9
 (pg. 
4129
-
4133
)
33
Paassilta
P.
Pihlajamaa
T.
Annunen
S.
Brewton
R. G.
Wood
B. M.
Johnson
C. C.
Liu
J.
Gong
Y.
Warman
M. L.
Prockop
D. J.
, et al. 
Complete sequence of the 23-kilobase human COL9A3 gene. Detection of Gly-X-Y triplet deletions that represent neutral variants
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
22469
-
22475
)
34
Hawkins
J. R.
Superti-Furga
A.
Steinmann
B.
Dalgleish
R.
A 9-base pair deletion in COL1A1 in a lethal variant of osteogenesis imperfecta
J. Biol. Chem.
1991
, vol. 
266
 (pg. 
22370
-
22374
)
35
Wallis
G. A.
Kadler
K. E.
Starman
B. J.
Byers
P. H.
A tripeptide deletion in the triple-helical domain of the pro alpha 1(I) chain of type I procollagen in a patient with lethal osteogenesis imperfecta does not alter cleavage of the molecule by N-proteinase
J. Biol. Chem.
1992
, vol. 
267
 (pg. 
25529
-
25534
)

Author notes

1

Present address: Academic Department of Medical Genetics, St Mary's Hospital, Manchester, U.K.