DEVELOPMENT OF A FLUORESCENT ASSAY FOR COLLAGENHETEROGENEITY USING MDPF-SDS-PAGE

Purified collagen were made fluorescent by coupling with 2-methoxy-2,4-diphenyl-3(2H)-furanone, MDPF, and the fluorescent products were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, SOS-PAGE. The profiles of Coomassie blue stained and MDPF-labeled collage.Tl bands on the gels were si."nilar except .for a slight increase in mobility with the MDPF coupled protein. · The relationship between the amount and the area under the peaks recorded from fluorometric scanning of purified al(I) ,a2 anda.l(III) was linear from io-5g to l08g. The standard curves for all three a chains were similar. Results from non-replicate determinations had an experimental error of + 6% SE. A mixed sample of cyanogen bromide (CB) peptides from type I and III collagen were quantitated by .measuring t.l-ie peak areas of known peptides from each type. The quantitation of collagen by the coupling of MDPF before electrophoresis is an improvement over staining with Coomassie blue after electrophoresis; since it provides a wider range of linearity, greater sensitivity, peak area independent of distance migration and less variability. The fluorescent method (MDPF-SDS-PAGE) also permits observation of bands during electrophoresis and quantitation immediately after electrophoresis.

l II.

LITERATURE SURVEY
Collagen, structure, function, and biosynthesis.
Collagen is the most abundant protein in the human body (Kivirikko and Risteli, 1976). Collagen fibrils are the fibrous connective tissue that provide structural rigidity to the body. The structure and biosynthesis of collagen require study because collagen has important roles in growth, aging, wound healing, fibrosis, and vascular disease.
The molecule collagen is a trimer consisting of three helical polypeptide chains, a chains, each consisting of 1000 amino acids residues, a molecular weight of 95,000 0 daltons, and a left-handed twist, pitch 9A. The three chains are braided together with a right-handed twist making a rigid helical structure, pitch 2.9ft, width 14A and length 2800~ (Ramachandran and Ramakrishnan;Timpl, 1976). A short sequence at the amine and carboxy terminals of each et chain are non-helical.
In the helical region, every third amino acid is glycine (Miller, 1976a); this is necessary for proper packing and possibly for hydrogen bond interaction within the helix (Ramachandran and Ramakrishnan, 1976). The imino acids praline (PRO) and hydroxyproline (HYP) occupy twenty-five percent of the collagen amino acid residues. Whereas in globular proteins imino acids serve to break helical conformations, 2 in collagen these imino acids serve to stabilize the helical conformation (Ramachandran and Ramakrishnan, 1976).
Collagen biosynthesis appears to involve more extensive post-translational reactions than for any other proteins examined to date (Prockop, 1976). Collagen is first synthesized as a precursor molecule procollagen which is 40-50% larger with globular register peptides at both the NH~and COCH-terminals. At the COOH terminal, disulfide '" bonds form bringing together three pro a chains for assembly as a triple helix (Byers et al., 1975).
The growing nascent pro a chain is hydroxylated and glycosylated in the rough endoplasmic reticulum cisternae while still on the ribosomes. Some hydroxylation and glycosylation of procollagen is present in the smooth endoplasmic reticulum, however, no hydroxylation and glycosylation is found in the golgi complex, and these modifications cannot occur after helix formation (Kivirikko and RisteIIi, 1976).
Formation of the helix is essential for normal secretion of procollagen into the extracellular matrix (Olsen et al., 1975). The procollagen extensions are cleaved extracellularly by procollagen peptidases (Goldberg et al., 1975), which permits collagen aggregation and a fibre formation in the theorized quarter stagger arrangement . Another extracellular enzyme L~portant in the processing of collagen is lyslyoxidase.
This enzyme oxidatively deaminates certain LYS and HYS residues forming reactive aldehydes (allysine and hydroxyallysine) . The aldehydes spontaneously react by the Schiff Base reaction with the s-NH 2 of a LYS and HLY from adjacent collagen molecules forming covalent intermolecular crosslinks. The crosslinks give tensile strength to the fibre (Tanzer, 1973), and increase resistance to proteolysis by collagen, therefore decreasing turnover.

Collagen heterogeneity
Until the late 1960s, only one type of collagen molecule was known. It was found in bone, tendon, ligament, and skin. This triple stranded collagen molecule consists of two identical chains and another which is genetically distinct. The chains are designed al and ~2 according to their elution order from carboxy methyl cellulose ion exchange column chromatography (CMC). In the early 1970s, a genetically distinct a. chain was found in cartilage·.
This consisted of a. chains that eluted from CMC only in the al position. The amino acid composition and the cyanogen bromide mapping (CB) peptide composition, demonstrated that this was a genetically distinct al molecule and therefore was designated as al(II), the more ubiquitous al was designed as al(I). Later, it was found that fetal skin has collagen CB peptides from another genetically distinct a chain; this was designated as cr l(III).
Basement membrane collagen was found to have another chain; this was designated as a l(IV). To date therefore, five genetically different a chains have been characterized for four different types of collagen (Kivirikko and Ristelei, 1976). Other putative a chains and collagen types have been found but they have not been well characterized (Benya et -al., 1977;Little et al., 1977;Lichtenstein et al., 1976).
Type I, the most abundant collagen has a chain composition consisting of two a l(I) chains and an et 2 chain; _by common convention this would be written as [al(I)] 2 a2.
The~ l(I) chain has one dissaccharide moiety and the a 2 chain one monosaccharide and one disaccharide (Miller, 1976a and is found predominantly in fetal tissue, blood vessels and to a lesser extent wherever Eype I is found (Miller, 1976a;Chung et al., 1974). In contrast to type I collagen, type III collagen contains more 4-hydroxyproline and it cystines which form intramolecular disulfide bonds that hold the collagen molecule as a trimer even under denaturing conditions. Antibodies to type III collagen have shown this type of collagen is associated with reticulin fiber . During the development of human skin, from fetal to adult, there is a decrease in type III collagen content and an increase in type I (Epstein, 1974;Chung and Miller, 1974). This is also true during the maturation of a wound from granulation tissue to a mature scar (Miller, E.J., 1976a). Type III collagen is found in arterial walls and may play an important role in vascular disease (Mccullah and Balian, 1975).
Type IV collagen [al(IV)J 3 is basement membrane collagen (Clark et al., 1976). In contrast to type I there is more hydroxylation of LYS and PRO (contains 3 hydroxyproline), more glycosylation and contains cystines. This collagen type has not been as well characterized as the other three. Breakdown of type IV collagen in basement membrane is an important step allowing metastasis to occur (Jaffe et al., 1976;Liotta et al., 1977).

Colla.gen heterogeneity in pathophysiology
There have been numerous reports of abnormal accumulation of particular collagen types in disease processes {Lapiere and Nugens, 1976). In Ehlers Danlos syndrome type IV, there is an absence of type III =ollagen in the extracellular matrix (Pope et al., 1975). Osteogenesis imperfecta is a deficiency of type I with an excess of type III collagen (Mueller et al., 1975). Kuttan et al. (1978) and Weiss et al. (1975) both found an elevation of type III collagen in synovial tissue of rheumatoid patients. In liver cirrhosis, Rodjkin et al. (1976) found an increase in both I and III type collagen but no difference in ratios . Kent et al. (1976) found first an increase in type III collagen followed by an increase in type I. Hypertropic scars and keloids have a high proportion of type III collagen (Bailey et al., 1975).
Vascular smooth muscles have been shown to increase total collagen synthesis in atherosclerosis (Ross and Gomsett, 1973;Mccullogh and Ehrhart, 1974) and in hypertension (Ooshima et al., 1974). Type I and III collagen are found in the arterial wall. Only type III is found in regions immediately adjacnet to the endothelial layer but type I is the predominate extractable collagen in atheromatous plaques (Mccullagh and Balian, 1975) • A problem with studying collagen heterogeneity is that unless extraction is complete, the recovery can give problems in ratio determinations. The results obtained are dependent upon the species used, the cell line, cell growth conditions, cell passages (or doublings), method of extraction and method of analysis. For example, using porcine medial smooth muscle cells, Scott et al . (1977) found type III in greater abundance than type I but Barnes (1976) found type I collagen as the more abundant type. dye cannot fully penetrate compacted protein bands (Fishbein, 1972) and the effect of alcohol on the selective removal of stain from protein bands has not been described in detail (Bertolini et al., 1976); deviation from Beer's Law (Gorovsky et al., 1970;Bennett and Scott, 197lf; error found with densiometric measurements are 10-15% (Fishbein, 1972 andScott et al., 1976a); several hours to several days are necessary to have gels destained in order to do sensitive and quantitative gel scans; and the area beneath the curve is not independent of distance migrated (Fishbein, 1972). Distance migrated can affect sensitivity and linear range. scanning in polyacrylarnide gels immediately after electrophoresis. Pace et al. (1974) showed that migration of fluorescamine-labeled proteins is log linear to molecular weight by SDS-PAGE. The use of dansyl-labeled proteins has not been useful with SDS-PAGE because of the preliminary treatment required to remove dansic acid and dansyl a.~ine (Eng and Parkes, 1974). In this procedure, the excess fluorescamine is rapidly converted to nonfluorescent products. With fluorescamine-labeled proteins, there is a postulated ring rearrangement to a nonfluorescent product that is more rapid in gels than in solution (Barger et al., 1976). The more recent use of 2-methoxy-2,4diphenyl-3(2H)-furanone (MDPF) (Weigle et al., 1973) as the fluor for detecting proteins in SDS-PAGE has advantages over fluorescamine in that the MDPF-labeled protein fluorescence is stable and more intense. Barger et al. (1977) reported quantitative fluorometric scans of MDPFlabeled proteins with linearity from 50-SOOng and a sensitivity of 1 ng for some proteins (e.g., myoglobin).
A. Purification The procedure adopted for the purification of collagen was that of Fujii and Kuhn (1975) as shown in (adjusted with HCl), 25.0 g pepsin was added slowly with · stirring, after 24 hr 17.5 g pepsin was added for a further 24 hr digestion at room temperature.
The solubilized material was filtered through sintered glass wool. Sodium chloride crystals were added to the filtrate to give a final concentration of 0.9M sodium chloride; the mixture was allowed to stand overnight at 4~c.
The coll~gen that precipitated out was collected by continuous centrifugation at 15,000Xg for l hr. The pellet was resuspended in 0.5M Tris-HCl pH 7.5, to inactivate the pepsin for four days, and then centrifuged at 35,000Xg for l hr. The supernatant was lyopholized.
Differential salt precipitation (DSP) was performed to harvest the pepsin released Type I and Type III collagen.
30g of the freeze-dried supernatant were dissolved in 10 1 of .05M Tris-HCl pH 7.5, lM sodium chloride. 4M sodium chlor:!..de was added to make a final salt solution of 1. 7!1 sodium chloride. The volume of 4M NaCl added was determined by the following equation: where Ci is the initial salt concentration (lM), Vi is the initial volume (10 1 The precipitant was washed by centrifugation three times in .05M Tris-HCl pH 7.5, l.7M NaCl. The precipitant was resuspended in 1% acetic acid and dialyzed against 1% acetic acid, and then centrifuged at 55,000Xg for 1 hr.
The supernatant was lyopholized and characterized as type III collagen by further purification gel electrophoresis; and cyanogen bromide peptide mapping. The salt concentration of the supernatant was then increased as described previously, again this time from l.7M sodium chloride to 2.5M sodium chloride with 4M sodium chloride. The resulting precipitant was allowed to settle for 24 hr and harvested by centrifugation at 35,000Xg for 2 hr. The pellet was resuspenaed in 1% acetic acid dialyzed against :% acetic acid and then centrifuged at 55,000Xg 1 hr. The    and  .
. Improved purifications of al(I) and a2 were obtained by using a gradient hold system (Figure 2) . A discontinuous ~ _..  Miller (1976b) method. A refractive index meter monitored the salt concent ration and a spectrophotometer, A=206 nm, monitored the eluted protein (collagen} . When the absorbance, as measured by the U.V. monitor exceeded a set level the T-valve closed the channel to the gradient and opened the channel for recirculating the salt solution. When the peak cleared, and the absorbance fell below the set level , the T-valve opened the channel to the gradient and the linear salt gradient continued.

HEATED
salt gradient was formed when a LKB Ultrograd valve closed in response to a protein peak sensed by the Uvicord III.
This automatically held the salt concentration constant until the peak was cleared. The fractions containing the chains were desalted by dialysis and then freeze-dried.

MclecUlar sieve chromatography
The collagen components obtained from CMC chromatography were further purified by gel filtration according to the procedure of Fujii and Kuhn (1975 These reaction mixtures were · stirred at room temperature under nitrogen for 4 hr. The reaction was terminated by diluting tenfold with distilled water followed by lyophilization (Scott and Veiss, 1976 , 1973). Figure 3 shows the reaction between a MDPF molecule and a e:-NH 2 of a.

Neville System
The samples were electrophoresed as described by Neville (1971). This is a discontinuous Tris-

Definitions for %T and %C
%T is the percent acrylamide of the gel, defined where a is the amount of acrylamide, b is the amount of bisacrylamide and ~ is the total volume of gel solution.
%C is the percent of cross-linking defined as:

Conversion from Absorbance to Relative Fluorescent
Intensity -1

RFI = log (2-Abs)
where RFI is the relative fluorescent intensity as mea- (~} is when the MDPF solution was added to t..11e samples. ( 0) is the relative fluorescence of the buffer and lo-5 g/ml collagen before coupling.
(d is 1Q-5g/ml collagen after coupling and <•> is buffer blank after coupling.  weights ~ 5% as determined by amino analysis (Miller, 1976a and Fujii;and Epstein, 1974). Svojtkova et al. (1973) The polyacrylamide gels were run according to the Neville (1971) procedure (T=S.5%, C=l%). After electrophoresis, the gels were stained with Coornassie blue and destained in 7% HAc. Rf values are the mean of two gels (range).
but raises additional questions. For example, since lysine content is similar for all the c: chains (Miller, 1976a) why is there a preferential increase in mobility?

C. Quantitation of a Chains
In order to quantitata different collagens, a chains from Type I and III were purified by CMC and agarose gel chromatography. The single fluorometric band obtained from each purified a chain was used later to establish linearity and assay limits. The scanned area recorded for each fluorometric peak was measured by planimetry three times with the average area under the curve having a +3% SE.
The same values for the areas were obtained from gels whether the gels were scanned immediately after electrophoresis or stored at 4°C in gel tubes and then scanned.
These scans show, however, that the peaks were sharper immediately after electrophoresis than those stored for several days. A scan of 10,000 ng MDPF-al(I) is shown in Figure 7D. The area is expressed as RFI X mm which takes into account the relative sensitivity (set by the fluorescent control unit) and the ratio of gel scan rate to chart speed. Therefore, the area term is the product of relative fluorescent intensity and relative band width. Each point shown in the standard curve of Type I and Type III collagens because these contain (dimers) and higher molecular weight forms. This is important since the collagen crosslinks are derived from lysines and hydroxylsines and some ligand sites would be unavailable for MDPF coupling. Estimating the amount of higher molecular weight forms of collagen using chains as standards, could result in an underestimate.
This most likely would be an insignificant difference because only 3-12% of the lysine hydroxylsines are thought to participate in crosslinking, dependent upon tissue, animal age and pathology (Tanzer, 1976;Scott, P.G., et al. 1976). This error is less than the error incurred l by the estimation of collagen hydroxyproline because there is 35% more 4-hydroxyproline in Type III collagen than Type I (Click and Bornstein, 1970;Chung and Miller, 1974). Also Clq may contribute some non-collagen 4-hydroxyproline.
In the estimation of collagen types by radioactive praline a correction factor is not considered for the 15% difference in imino content of a.l (I) and CL2 {Click and Bornstein, 1970).
Since there is approximately the same primary amine content in the collagen chains (~5%), (Fujii and Kuhn, 1975), the standard curves are similar and there is therefore no correction factor necessary for quantitating these three different genetic types of a.chains against a single standard. The quantitation:o:f collagen by MDPF-SDS-PAGE has the following advantages over staining with Coomassie blue: 1) it is more rapid, the fluorescent bands can be observed during electrophoresis with a long wave u.v. lamp and quantitated inunediately after electrophoresis; 2) there is less variability +6% SE vs 10-15%; .
there is a wider range of linearity, 10 ng to 10 µg and  (Table II).
It is important to be sure that the disulfide bonds have not been broken in the extraction procedure or preparation for electrophoresis, because a. l (III) does not separate from . a.l(I) and it could give a false result during heat denuration (preparation of sample for electrophoresis). Also, disulfide bonds will break if the pH is not returned to neutrality after MDPF coupling. Conformation of the validity of this system can be demonstrated with the cyanogen bromide peptide mapping because some· a.l(III)-CB peptide separate from ~l(I ) -CB peptides.   (Fishbein, 1971). An analysis of variance shows there was no significant difference between distance migrated and fluorescent peak area of MDPF-~l(I) (TableIII) . There was a decrease in peak height and a broadening with increased migration (Figure 13). This is one explanation why with type III collagen reduction, peak area remains the same when determined by fluorometric scanning but not by densiometric scanning.   Figure 14 shows the fluorometric scans obtained from OlBr digested CL 1 (I), a1 (III) and CL 2. Based upon the measured relative mobility (Rf) from Figure 14 and the possible molecular weight forms of the cyanogen bromide peptides (Table III) a semi-log plot of molecular weight vs Rf can be derived ( Figure 15). The coefficient of de.termination (R 2 ) for this regression analysis for all three O{chain cyanogen bromide peptides is 0.9~9.
The cyanogen bromide peptide profiles found here were similar to Scott and Weiss (1976) and Benya et al. (1977)      Fig. 14. Fluorometric gel scans of purified o1 chains CB peptides. Three purified o< chains of collagen were cleaved with CNBr. The peptides were coupled with MDPF and then electrophoresed according to Neville (1971) procedure. Identification of CB number is based on their relative mobility.  16. A standard curve for % Type III by CB peptide fluorescent areas. Type I and Type III collagen were cleaved with cyanogen bromide. The 100% Type III is 10 µg of III-CB peptides, 0% Type III is 10 µg Type I-CB peptides and 50% Type III is 5 µg III-CB peptides and 5 µg I-CB peptides. The samples were made fluorescent by coupling with MDPF before electrophoresis. The fluorescent area used for Type III was from III-CB 4,8,3,6. The fluorescent area for Type I was from I-CB6. The area ratio was obtained by dividing the RFI x mm for III-CB4,8,3,6 by the RFI x mm for I-CB6 and 8,3,6. in Figure 17. The standard curve is not linear. Crystal et al. (1977) found a similar result with the quantitation of radioactive cyanogen bromide peptides from type I and III collagen. The reason for this non-linear standard curve is that the amount of radioactivity or RFI x mn for the CB peptides is not equal for type I and III collagen as was shown with the a.chain quantitation.
An alternative method is to quantitate the cyanogen bromide peptides for type I and III collagen and plot the amount of I and III for the specific cyanogen bromide peptides. Using this method one .can obtain a plot in which each a. chain is quanti tated (Figure 18) .
The advantage of quantitating collagen types by cyanogen bromide peptides is that both the amount of type I and III collagen can be quantitated from a single gel by measuring the area beneath the curves for the peaks associated with type I and type III collagen cyanogen bromide cleavage products. The disadvantage of this method compared to quantitation by a.chains is that it is ten times less sensitive.
The cyanogen bromide digestion of each sample would require the time-consuming process of lyophilization to remove the CNBr and to concentrate ~~e sample for SDS-PAGE. The cyanogen bromide peptides profiles are very complex and also there is not always complete cleavage at each methionine residue such that some peptides are contiguous cyanogen bromide peptides. This is because the methiodine residues 10 µg of MDPF-Type I collagen (A) before cyanogen bromide digestion (the a 1 and a 2 separate with the Neville system even at 15%T and 3%C) . 20 µ g Type I collagen after CNBr digestion (B); numbers on the left are from a. 2, the numbers on the right are CB peptides from a l(I) . ~he CB pep tides obtained from 20 µg Type III (C) , the numbers are ~l(III)-CB numbers. 10 µg I-CB and 10 µg III-CB are in (D). The higher molecular weight peptides are either uncleaved products or crosslinked peptides. The coupling of MDPF to the CB peptides was · just prior tc electrophoresis.  Fig. 18. An X plot standard curve for amount of Type I and III collagen versus CB peptides peak area. The RFI x mm for I-CB6 was used for quantitating Type I collagen and the RFI x mm for III-CB9,5 was used for quantitating Type III collagen. A total of 10 µg of MDPF-CB collagen was applied on each gel. Both lines were determined by linear regression with a coefficient of determination greater than .95 for both lines.
have ~rarying degrees of sus~ptibili ty to cleavage depending upon the adjacent amino acid. There is formation of hornoserine-0 serine or 0-theonine in the peptide linkage (Scott and Veis, 1976). The cyanogen bromide method is not applicable for less than 10% type III collagen because of the closeness of the CB peaks from type I and III collagen. The cyanogen bromide method.can be used for confirmation of type III by a. chain quantitation from the decrease of Y and increase in a. after reduction mea- '' ·