May 1, 1990 - electrophoresis and by N-terminal amino acid sequencing of its component ..... Bernard MP, Chu M-L, Myers JC, Ramirez F, Eikenberry. EF,.
CLIN. CHEM. 36/7, 1328-1332 (1990)
Radioimmunoassay of the Carboxyterminal Propeptide of Human Type I Procollagen Jukka Melkko,’ Seija NIemI,2 Lella Ristell,1 and Juha RlsteII’3 Type I collagen is the most abundant collagen type in soft established a rapid equilibrium radioimmunoassay for the carboxyterminal propeptide of human type I procollagen (PICP), to be used as an indicator of the synthesis of type I collagen. We isolated type I procollagen from the medium of primary cultures of human skin fibroblasts, digested the protein with highly purified bacterial collagenase, and purified PICP by lectin-affinity chromatography, gel filtration, and ion-exchange separation on HPLC. The purity of the protein
(4) and in metabolic bone diseases (5-7). A radioimmunc assay for this propeptide was described as early as 1974 (8 Originally the antigen was thought to be derived from th aminoterminus of the molecule, but it was subsequentl found to be the carboxyterminal propeptide (5). Despit being a promising method with respect to monitorini metabolic bone disease, the assay of PICP has unfortu nately not been available for general use. Here we describe the purification and biochemical char actenzation of PICP and the development of a radioimmuno
was verified
assay
tissues and the only type found in mineralized
by sodium
dodecyl
bone. We
sulfate-polyacrylamide
gel
electrophoresis and by N-terminal amino acid sequencing of its component chains. The final radioimmunoassay was established with polyclonal rabbit antibodies. Material antigenically related to PICP is readily detected in human serum. There is only one form of the serum antigen, its molecular size and affinity to the antibodies being similar to those of the isolated propeptide. Intra- and interassay CVs are 3% and 5%, respectively. Preliminary reference intervals for healthy adults (18 to 61 years of age) are 38-202 jg/L for men and 50-170 j.s/L for women; in men the concentration is inversely related to age. The serum antigen is stable during storage and after repeated thawing. Additional interval
Keyphrases: skin fIbro blasts sex and age-related effects
bone
reference
Type I collagen is the most abundant collagen species in many soft tissues and accounts for >90% of the organic matrix of bone (1). The synthesis and (or) breakdown of this collagen type can be altered during the pathogenesis of many kinds of disease. Because bone is a metabolically active tissue throughout life, indicators of type I collagen turnover could be particularly useful as biochemical markers in metabolic bone disease. So far, the most important means of estimating the metabolic rate of bone collagen has been to quantif’ the urinary excretion of the amino acid hydroxyproline, which is derived from collagenous proteins. However, in clinical work this method is tedious, associated with several sources of error, and not specific for type I collagen (2). Type I collagen is synthesized in the form of a larger protein, type I procollagen, which contains additional sequences at both ends (3); these sequences are removed by specific proteinases before the collagen molecules are assembled into fibers. The part removed from the carboxyterminal end of the molecule, known as the carboxyterminal propeptide of type I procollagen (PICP), can be found in blood, where its concentration changes, e.g., during growth
‘Collagen Research Unit, and Clinical
Chemistry,
Departments
University
of Medical Biochemistry of Oulu, Kajaanintie 52 A,
SF-90220 Oulu, Finland. 2Fos
3Address
Diagnostica, correspondence
SF-90460 to this
Oulunsalo,
Finland.
author, at the Department
Medical Biochemistry, University of Oulu. Received February 28, 1990; accepted May 1, 1990.
1328 CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990
of
for it. The
monitoring
assay
is rapid
and
suitable
for clinica
of bone metabolism.
Materials and Methods Materials Culture
medium
was collected
from primary
cultures
o
human skin fibroblasts, both confluent and subconfluent The cells (5-10th passage) were grown in Dulbecco’s mod ified Eagle’s medium supplemented with newborn cal serum (100 mL/L), ascorbic acid (50 mg/L), penicillin (10( kilo-units/L), and streptomycin (100 mg/L). We used 10 ml of medium per 9-cm-diameter culture dish. Typically, w processed the conditioned medium in 15-L batches. In the radioimmunoassay, which is available commer cially from Farmos Diagnostica, Oulunsalo, Finland, th tracer and the antiserum were diluted with phosphati buffer (0.1 mol/L, pH 7.2) containing 9 g of NaCl, 5 g o bovine serum albumin, and 0.5 g of NaN3 per liter. Th separation agent (second antibody) was a suspension a goat anti-rabbit antibody covalently bound to solid des in 0.1 mol/L Tris HC1 buffer, pH 7.4, containing
parti 9 go:
NaCl, 2 g of bovine serum albumin, 1 mL of Tween 20, anc 1 g of NaN3 per liter. We obtained this suspension, as wel as a corresponding goat anti-mouse antibody reagent, fron Farmos Diagnostica, Turku, Finland. The monoclonal antibody M-38 (9) was purchased fo research
purposes
oma Bank University munoassays
from the Developmental
Studies
Hybrid
(do Dr. Thomas August, The Johns Hopki School of Medicine, Baltimore, MD). Radioim with this antibody
were carried
out as will
described for the rabbit anti-PICP, but with an anti-mo separation reagent substituted for the anti-rabbit reagen Procedures Purification of the propeptide. We first purified type procollagen from the culture medium of human skin fibr blasts, essentially as described previously (10). The proco lagen was digested for 16 h at 30#{176}C with highly purifi bacterial collagenase (Worthington Biochemicals, grad CLSPA), 0.3 mg of enzyme per liter of original medi volume. PICP was purified from the digest by lectin-affini chromatography on concanavalin A-Sepharose (Pharm cia, Uppsala, Sweden) in 50 mmol!L Tris HC1 buffer, p 7.4, containing 29.22 g of NaC1, 1.25 g of N-ethylmale mide, 52.2 mg of phenylmethanesulfonyl fluoride, and 73. mg of CaC12 - 2H20 per liter. We eluted the propeptid which was bound to the column, with the above solutio
)lus 97.1 g of a-methylmannoside per liter. For further )urlfication we used gel filtration with Sephacryl S-300 Pharmacia), eluting with a solution containing 15.8 g of .TH4HCO3 per liter, and anion-exchange HPLC (column:
E’rotein-Pak DEAE-5PW; Waters, Milford, MA; mobile )hase: 50 mmollL Ti-is acetate buffer, pH 8.0, flow rate 1 uL/min). In the latter purification, the bound propeptide was eluted with a linear gradient of NaC1 (0-0.5 mol/L in
z
-
iO mm).
-
Production of antiserum. Polyclonal antibodies against he propeptide antigen were raised in New Zealand White rabbits by injecting the purified propeptide in complete ?reund’s adjuvant, essentially as described previously for mother procollagen propeptide antigen (11). Chemical characterization of the propeptide. We verified ±ie purity of the propeptide by sodium dodecyl sulfate,olyacrylamide electrophoresis (SDS-PAGE)in 125 g/L acrylunide gels. The concentration of the final propeptide prepiration was determined by quantitative analysis of amino icids after acid hydrolysis. For N-terminal amino acid ;equencing we used a liquid-phase sequencer (Applied 3iosystems, Foster City, CA). lodination of the tracer. We labeled the purified antigen with im1 by the Chloramine-T method, separating the abeled antigen from free iodine by gel ifitration on a 1 x 30 m column of Sephacryl S-300 (Pharmacia) equilibrated at oom temperature in the eluent: phosphate buffer (0.1 nolJL, pH 7.2) containing 2 g of bovine serum albumin and g of NaN3 per liter. The calculated specific activity of the abeled antigen was about 15-20 Ci/g. Assay procedure. Incubate 50- or 100-.tL aliquots of ;tandards or serum samples with 200 j.L of the tracer olution (about 50 000 counts/mm) and 200 L of diluted intiserum for 2 h at 37 #{176}C. Then add 500 L of the olid-phase second-antibody suspension to each tube and iortex-mix. After 30 mm at room temperature, separate he bound fraction by centrifugation (2000 x g, 15 mm, #{176}C). Decant the supernate containing the unbound tracer, tnd count the radioactivity of the precipitate containing he bound tracer (we used a Clinigamma 1272 counter; harmacia-Wallac, Turku, Finland). Gel filtration of serum samples. To test the antigen-form ipecificity of the assay developed, we separated the antiiens in serum by gel filtration, using a 1.5 x 110 cm olumn of Sephacryl S-300 equilibrated in phosphate)uffered saline containing 0.4 mL of Tween 20 per liter. Comp uter analysis of radioimmunoassay data. The slopes f the inhibition curves obtained with the reference prepaation of PICP and with serum samples were calculated frith a microcomputer (12; see 10). We used logit and log ransformations to obtain a linear dose-response curve, Ihen used iterated weighted-least-squares regression analsis.
esu its Purification and characterization of the antigen. The ICP obtained after the purification procedure gave one iajor band in SDS-PAGE, which upon reduction resolved ito two components with an approximate ratio of 2:1 E’igure 1). The molecular mass of the unreduced compound ras about 100 kDa and those of the reduced components rere around 30 kDa, as determined by comparison with Landard
proteins.
In N-terminal amino acid sequencing ropeptide, we identified two sequences
of the purified (Figure 2). The
94000
-
-
-R
67000 43000 30000
+R
Fig. 1. SOS-PAGE of the purified carboxyterminal propeptide of human type I procollagen -A, unreduced; +R, reducedwith 2-mercaptoethanol.The arrows indicatethe positions of the standard proteIns (94 000 phosphorylase b, 67 000 = bovine serum albumin, 43 000 = ovalbumin, and 30 000 = carbonic anhydrase) A:
PRO1(
I)
R AD D A N V V(R)DCR)D
B:
PRO2(
I)
S A P(S) L R P K 0 V E V D A T L K CS)
L E V 0 T CT) L
Fig. 2. Aminoterminal sequences obtained from the purified propeptide (upper part) and the locations of the cleavage sites in the procollagen chains (lower part) The amino acid residues shown in parentheses could not be identified. The black arrowheads indicate the cleavage obtained here with purified bacterial collagenase; the white arrowheads show the sites where the specific Cproteinase removes the propeptide. The predicted secondary structures are (-) random coil where a change in direction indicates fl-turn, (#{149} #{149})
a-helical region, and HELIX the C-terminal region of triple helical collagen sequence. The lower part, modified from (18). is reprinted with permission from the publisher and the authors
origins of the component chains of the isolated PICP can be traced to the carboxyterminal region of type I procollagen, as also shown in Figure 2, the amino terminals being
amino acids 1169 (13) and 1102 (14) for the proal(I) and proa2(I) chains, respectively. The same sequences were obtained for the two different batches of PICP analyzed, which were prepared with two different concentrations of bacterial collagenase. Characteristics of the radioimmunoassay. The titration of a rabbit anti-PICP antiserum with iodine-labeled PICP is shown in Figure 3 (left), together with the binding of the monoclonal antibody M-38 to the labeled antigen. As illustrated in Figure 3 (right), only the polyclonal antibodies could serve as a basis for a sensitive inhibition assay. In the latter, 50% inhibition of binding was achieved by using about 100 g of PICP per liter. For most hunian serum samples studied, the inhibition curve given for serial dilutions of a sample was superimposable with that of the standard antigen (Figure 3), CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990 1329
C.-.) 6.3
SERUM 25
(pL) 100 -J
-
100
100
80
80
z
0 ‘
S
0
S
o z
z 40
20
z
Ui Q
2.0
.
1.0
Ui
10
0 I
20
I-
z
40
0
z
0 Ui
‘C
80
80
1
15
0 0
5
0 0. 0
5
0.
0
0.5 ANTIBODIES
2
8
32
(1/DILUTION
128 10)
31 PICP
125 STANDARD
0
(iiIL)
Fig. 3. (Left) Binding of iodinated PICP to a polyclonal anti-PICP antiserum (#{149}) and to the monoclonal antibody M-38 (0); (right) assay curves for PICP standard (#{149}) (slope -1.084, 50% intercept 101 ig/L) and a human serum sample (A) (slope -1.104, 50% intercept 25 L, PICP 420 g/L), as assayed with a polyclonal anti-PICP antibody, and an assay curve for PICP standard (0), as assayed with the monoclonal antibody M-38
indicating the presence in serum of material that is antigenically similar to the standard PICP. The intra- and interassay variations (n = 16 and 8, respectively) were tested with 100-pL human serum samples containing several different concentrations of the PICP antigen (Table 1). In our hands, the intra-assay CV is constantly around 3%, the interassay CV close to 5%. Use of 50-L serum samples from infants and children gave no significant loss in precision. The sensitivity (detection limit) of the PICP assay, defined as the detectable mass equivalent to twice the standard deviation of the zero binding value, was 1.2 pg/L. Two serum samples with different concentrations of PICP were mixed in different ratios and analyzed. The mean analytical recovery was 99.2% (SD 5.7%, n = 15) within the assay concentration range 90 to 315 tg/L. No interference was seen in hemolyzed, lipemic, or icteric sera. Size and stability of the serum antigen. The PICP radioimmunoassay detects one major peak of antigenicity in human serum (Figure 4). Its elution position in gel filtration exactly corresponds to that of the purified standard antigen. To test the stability of the PICP antigen in serum, we separated the serum of a blood sample with a PICP concen-
Table 1. Intra- and interassay Variations of PICP Assay Mean PICP, pgIL Intra -assay (n = 16) 54 88 103 127 214 451 lnterassay(n 52 89 105 132
CV, % 3.1 3.7 2.1 2.1 2.7
3.2 =
8) 5.8 3.5 4.1 4.6
134
3.6
216
6.6
435
4.0
1330 CLINICAL CHEMISTRY, Vol. 36, No. 7, 1990
I 0
30
40 FRACTION
50
60
0
NUMBER
Fig. 4. Gel-filtration analysis of a human serum sample on a columi of Sephacryl S-300 The arrow indicatesthe elution position of isolated human PICP. (-),
toti
protein
tration of 224 pg/L, froze part of the serum immediately and divided the rest into two portions for storage at 4 and at room temperature, respectively. Aliquots were re moved and frozen from both portions at different times There were no significant changes in the measurable PICI concentrations during storage for up to 15 days at eithe: temperature. At -20 #{176}C the samples could be stored fo: several months, with no loss of antigenicity upon repeate freezing and thawing. Reference values. The reference values for serum PICI were calculated for healthy adults, based on samples fron 75 blood donors (Figure 5). In men ages 20 to 60 years, th PICP concentration was inversely related to age, wherea in women no such relationship could be found. In boti sexes, the distribution of the concentrations was somewha skewed. The difference between results for serum and plasmt samples was tested by measuring both kinds of samplei from eight healthy volunteers. The mean PICP concentra tions were 124 (SD 49) g/L for serum and 119 (SD 42) pg/I for plasma, significantly correlated (r = 0.997, P
z 100
#{149}
50
#{149}
.
,.
S
75
,, 50
14/
r7112.,
#{149} .:‘
0.
I!
025
4j/
-
0 20
30
40
50
80
38
70
AGE (YEARS)
4
fl 202
120 PICP
(pg/Li
of cancellous bone formation in a group of patients with various metabolic bone diseases (7). However, despite being a promising method, the PICP assay has been available in
WOMEN 250
a
.3
200
100
only one laboratory. Procollagens are difficult to handle, which easily leads to very low yields of the purified propeptide. We have now succeeded in redevelopment of the PICP method, and the characteristics of our assay seem to be fully comparable with those of the method reported previously (4-8). The present assay, based on a highly purified
z 0 C
0 I-
150
S
II. 0
S#{149}
I-
75
SQ
z 100
#{149} S
2#{149}#{149}#{149} . #{149}#{149}#{149} 0
50
1
Si 0
50
25
and well-characterized
Si
0.
I’-
0 20
30
40
50
60
AGE (YEARS)
Fig. 5. PICP concentrations
women and 33
identical with, the cleavage sites obtained here (Figure 2). Most of the type I collagen of the body is present in bone, where it forms the scaffold of the calcifying extracellular matrix. Thus a method that can accurately and reproducibly measure its rate of synthesis could prove quite useful for monitoring metabolic bone diseases noninvasively. Earlier work has suggested that PICP could serve as such a test, e.g., in Paget’s disease of bone (5, 6), and its concentration in serum has been found to correlate with the rate
170
PICP
230
(vg/L)
in healthy Finnish blood donors, 41
men
individual values are shown in relation to age. Forwomen, r = -0.133 (not significant);for men, r = -0.546, P
