1

Type XIX Collagen Purified from Human Umbilical Cord is

Characterized by Multiple Sharp Kinks Delineating

Collagenous Subdomains and by Intermolecular Aggregates Via Globular,

Disulfide-linked and Heparin-binding Amino Termini

Jeanne C. Myers‡*, Deqin Li‡, Peter S. Amenta¶, Charles C. Clark‡†,

Chandrasekaran Nagaswami§, and John W. Weisel§

From the Departments of ‡Biochemistry and Biophysics, †Orthopaedic Surgery, and §Cell and

Developmental Biology, University of Pennsylvania School of Medicine, Philadelphia,

Pennsylvania 19104, and the ¶Department of Pathology and Laboratory Medicine, Robert Wood

Johnson Medical School, New Brunswick, New Jersey 08903.

Running title: Molecular shape and N-terminal association of collagen XIX *To whom correspondence should be addressed: University of Pennsylvania School of Medicine,

Dept. of Biochemistry and Biophysics, 909 Stellar Chance, 422 Curie Blvd., Philadelphia, PA

19104-6059. Tel. 215 898-0712; Fax. 215 573-2085; E-mail. myers@mail.med.upenn.edu

Copyright 2003 by The American Society for Biochemistry and Molecular Biology, Inc.

JBC Papers in Press. Published on June 3, 2003 as Manuscript M304629200 by guest on February 11, 2018

http://ww

w.jbc.org/

Dow

nloaded from

2

SUMMARY

Type XIX collagen was discovered from the sequence of rhabdomyosarcoma cDNA clones.

The chain is composed of a 268-residue amino terminus, an 832-residue discontinuous

collagenous region, and a 19-residue carboxy peptide. Light microscopy immunohistochemistry

of adult human tissues demonstrated that type XIX is localized in vascular, neuronal,

mesenchymal and some epithelial basement membrane zones. It also appears to be involved in

events linked to skeletal myogenesis. In this report we have presented the first direct evidence for

the molecular structure of type XIX collagen. Using human umbilical cord, native type XIX was

purified by neutral salt extraction, and ion exchange and antibody affinity chromatography. Type

XIX was found to represent only ~10–6% of the dry weight of tissue making it by far the least

abundant collagen ever isolated. Transmission electron microscopy after rotary shadowing

revealed the appearance of rod-like structures with multiple sharp bends, a small nodule at one

end of the molecule, and a total length of 240 nm. Domain-specific antibodies were used to

identify the nodule as the noncollagenous amino terminus, while the location of most kinks

corresponds to major interruptions separating the five collagenous subdomains. More than half of

the type XIX molecules observed were present in oligomers of different size and complexity,

resulting from association of the amino-terminal domains. Biochemical analysis demonstrated

that these supramolecular aggregates are dependent upon, and/or stabilized by, intermolecular

disulfide cross-links; and that the globular amino terminus contains a high affinity, heparin-

binding site. The polymorphic conformational states of this rare collagen, and its ability to self-

assemble into a higher-order structure provide focal points for future determination of

biologically significant functions in cell-cell and/or cell-matrix interactions.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

3

INTRODUCTION

Twenty-six collagen types have currently been designated; seven of these are recent

discoveries and several have yet to be described (1–7). With the diversity that this family has

presented, one of the few defining elements is the existence of one or more triple-helical regions,

regardless of size and relative proportion to the entire protein. In the most general sense,

therefore, collagens have been divided into the classic fibrillar group, i.e., those containing the

~333 continuous Gly-X-Y triplets and involved in the formation of cross-striated fibrils, and the

nonfibrillar group, a highly heterogeneous class exhibiting a spectrum of sizes, supramolecular

assemblies, and chain organization with the one commonality being the presence of

noncollagenous sequences interrupting and/or flanking collagenous domains (1, 2, 8–10).

Understanding the complex structure and function of these many proteins has proven to be a

formidable task despite, in many instances, extensive knowledge of disease phenotypes directly

attributable to the respective collagen gene mutations (1, 11, 12). A major complication in this

process has been the scarcity of a number of collagen types and the inability to directly

characterize the in vivo form of the molecules. One of these elusive and increasingly intriguing

collagens is type XIX.

Type XIX collagen was identified from independently isolated clones representing RNA

purified from a human rhabdomyosarcoma (RMS)1 cell line (13, 14). The type XIX chain is

composed of a 268-residue, noncollagenous amino terminus, an 832-residue discontinuous

collagenous region, and a 19-residue carboxy peptide (14–16). Several features in the type XIX

sequence place this collagen in the largest subclass of the nonfibrillar group, together with types

IX, XII, XIV, XVI, XX and XXI (2–4, 10). These include an ~250-residue thrombospondin

module in the amino terminus (Tsp-N), the position of two 2-amino acid interruptions in the

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

4

collagenous subdomain closest to the carboxyl terminus, and a Cys–(Xaa)4–Cys motif situated at

the junction of the collagenous region and carboxy peptide (10, 15, 16). By virtue of four

internal 20- to 44-residue interruptions, the primary sequence of the unique type XIX collagenous

region can be divided into five 70- to 224-residue subdomains; four include a few short

interruptions and the fifth is composed solely of Gly-X-Y triplets (refs. 14–16 and illustrated in

Fig. 12A). In contrast, only two or three collagenous subdomains are found in other members of

the above-mentioned subclass, except for type XVI, which is interspersed with a heterogeneous

array of ~9–10 generally less well-defined collagenous segments (2–4, 9, 10, 17).

Although knowledge of the type XIX chain spans a decade, there are few reports focused on

the expression of this molecule, and its function remains obscure. In Western blots, polyclonal

antibodies raised against amino and carboxyl noncollagenous terminal sequences, respectively,

reacted with a 165-kDa collagenase-sensitive chain found in cultured RMS cell extracts (18).

Light microscopy immunohistochemistry demonstrated that type XIX localized to endothelial,

neuronal, mesenchymal, and most epithelial basement membrane zones (BMZ) in adult human

tissues: namely breast, colon, kidney, liver, placenta, prostate, skeletal muscle, skin, and spleen

(18, 19). Curiously, separate data revealed another aspect of type XIX expression, a correlation

with skeletal myogenesis. Type XIX RNA and protein were dramatically up-regulated in RMS

cells cultured in growth factor-depleted serum (20); this induction was restricted to a small

subpopulation of cells which also expressed structural proteins characteristic of skeletal muscle

differentiation. In a subsequent report, it was shown that during mouse embryonic development,

spatial and temporal expression of type XIX RNA was essentially coincident with the myf-5

transcription factor (21). Most recently, type XIX null mice with a perinatal lethal phenotype

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

5

were used to establish that this collagen is required for complete esophageal muscle

transdifferentiation2.

To begin direct exploration of this vital protein at the ultrastructural level, we have purified

the native tissue form of type XIX and characterized the molecule by electron microscopy and

biochemical analysis. Type XIX collagen in human umbilical cord is extremely scarce, i.e.,

~10-6% of dry weight, which is several orders of magnitude less than any other collagen so far

isolated. Electron microscope images revealed a sharply kinked and highly polymorphic

collagenous region, and the existence of higher-order complexes. The involved amino-terminal

domain is responsible for intermolecular disulfide linkages and contains a proven heparin-

binding site.

MATERIALS AND METHODS

Type XIX Polyclonal and Monoclonal Antibodies—The type XIX polyclonal antibody

recognizing the last 14 residues of the carboxyl terminus (COOH-Ab) has been described

previously (18). The antiserum was purified using Affi-Gel 10 resin (Bio-Rad, Hercules, CA) to

which the peptide antigen was covalently bound. It was eluted in 1 M glycine, pH 2.7, and

neutralized. The monoclonal antibody recognizing the amino-terminal domain of type XIX

(NH2-Ab) was prepared by conventional methods (Genetics Core Facility, University of

Pennsylvania and ref. 22). The antigen was a 67-residue recombinant protein used to generate

polyclonal antibodies (18). The type XIX COOH-Ab used for purification of type XIX protein

was eluted from Affi-Gel 10 in 0.1 M citric acid, pH 2.7, and neutralized with 1 M HEPES, pH 9

(Sigma, St. Louis, MO). Peak fractions (1–1.2 mg) were dialyzed against 0.1 M HEPES, pH 7.0,

concentrated to 1.5 ml (Centricon-30, Millipore, Bedford, MA), and bound to 1 ml of Affi-Gel

10. Western blotting showed that almost all of the antibody bound to the resin.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

6

Purification of Type XIX Collagen—Human umbilical cords were obtained from the

Hospital of the University of Pennsylvania. About 400 cord specimens totaling ~8 kg and

weighing 3.5 to 56 g each were collected for this project. Cords were frozen temporarily at –20

°C and transferred to –80 °C. For large-scale preparations, 10–15 specimens (~250 g wet

weight) were pooled and processed as described below; for rotary shadowing, two preparations

were combined before the antibody affinity column. Processed tissue, protein samples, and

buffers were maintained in an ice slurry when possible, or else kept at 4 °C. Tissue was thawed

on ice, washed several times in buffer (50 mM Tris-HCl, 4.5 M NaCl, 20 mM EDTA, pH 7.5, 10

mM NEM, and 0.5 mM PMSF), cut into 0.5-cm pieces and added to extraction buffer (10:1 v/w,

50 mM Tris-HCl, 1.0 M NaCl, 10 mM EDTA, pH 7.5) with protease inhibitors (10 mM NEM,

0.5 mM PMSF, 5 µg/ml leupeptin, 1 µg/ml aprotinin, all obtained from Sigma). The tissue was

homogenized at speed 5 (speed 10 = 27,000 rpm) for 8 × 1 min (Polytron, Brinkman Instruments,

Westbury, NY). The tissue suspension was stirred slowly for 21–24 h and centrifuged at 32,000 ×

g for 30 min. Supernatant proteins were precipitated overnight with the addition of ammonium

sulfate to 40% saturation, centrifuged as above, resuspended in 200 ml of extraction buffer

containing 0.1% Triton X-100, stirred overnight, and dialyzed for 4 h against 2 l of buffer

containing 25 mM Tris-HCl, 0.4 M NaCl, 2.0 mM EDTA, pH 7.4, 0.1% Triton X-100, and the

inhibitors listed above. The sample was then dialyzed against the same buffer except with a

lower pH (7.2) and NaCl (0.1 M) concentration. The mixture was clarified at 20,000 × g for 15

min and the supernatant was incubated in a batch procedure with 40 ml of SP Sepharose Fast

Flow resin (Amersham Biosciences, Piscataway, NJ) pre-equilibrated in the final dialysis buffer

containing 0.1 M NaCl. The protein solution and resin were incubated on a rocker overnight,

centrifuged at 480 × g for 5 min and the resin washed/eluted stepwise at 30-min intervals using 3

× 140 ml of the binding buffer, and 4 × 80 ml, 4 × 40 ml and 2 × 40 ml of the same buffer

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

7

containing 0.3 M NaCl, 0.6 M NaCl, and 1.0 M NaCl, respectively. Type XIX collagen in the

first two portions of the 0.6 M NaCl eluate was concentrated (~16–20 fold) (Ultracell Amicon

YM 30 Ultrafiltration Discs, Millipore), and then brought to a final concentration of 0.4 M NaCl.

The BCA reagent (Pierce, Rockford, IL) was used for protein assays.

The type XIX pool was mixed with Affi-Gel 10 resin (4:1 v/v) to which the purified COOH-

Ab was bound. The resin was pre-equilibrated in 0.4 M NaCl buffer containing 25 mM Tris-HCl,

2.0 mM EDTA, pH 7.2, 10 mM NEM, 0.5 mM PMSF, 1 µg/ml leupeptin, 1 µg/ml aprotinin (or 1

µg/ml pepstatin A) and 0.1% Triton X-100. The sample was incubated with the resin on a rocker

platform for 8 to 16 h. The column was washed with 4 × 3 ml of 0.4 M NaCl binding buffer, and

type XIX collagen was eluted stepwise in 0.8–1 ml fractions of 1.0 M glycine, pH 2.7,

immediately neutralized with 1 M tris base. Fractions were collected by gravity in polypropylene

tubes pre-coated with Triton X-100 (0.1%). Before elution, Triton X-100 and EDTA, pH 7.2,

were added to each tube in order to adjust the final concentration in each fraction to 0.1% and 2.0

mM, respectively. Protease inhibitors were added to each fraction immediately after collection.

Fractions were aliquotted, frozen in dry ice, and stored at –80 °C, or the peak fraction was

dialyzed against 0.1 M ammonium formate, pH 7.4, 1.0 mM EDTA, 0.1% Triton X-100 and

protease inhibitors for 16 h for rotary shadowing. The micro-dialyzer (Spectrum, Gardena, CA)

and membrane (MWCO 50 kDa) were pre-treated with 0.1% Triton X-100.

Bacterial Collagenase Digestion—Collagenase digestion was carried out for 75 min at 37–

38 °C in a 15-µl reaction containing 50 mM Tris-HCl, pH 7.2, 10 mM calcium acetate, and 2–5

units of bacterial collagenase form III (Advance Biofactures, Lynbrook, NY). Control samples

(un-digested) were incubated in the same buffer without collagenase.

Gel Electrophoresis and Western Blot Analysis—Samples were boiled for 2 min in 60 mM

Tris-HCl, pH 6.8, 4% SDS, 10% glycerol, 50 mM EDTA, 0.025% bromophenol blue, 200 mM

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

8

DTT, and electrophoresed in 6, 7, 10 or 12% polyacrylamide-SDS gels. Proteins were transferred

to Immobilon-P membranes (Millipore) (18, 23). Membranes were incubated for 90 min with the

primary antibody (0.1–1 µg/ml of the COOH-Ab, or a 1:100–200 dilution of serum-free medium

from the clone producing the monoclonal NH2-Ab), washed, and incubated with secondary

antibodies (anti-rabbit IgG, peroxidase-linked F(ab’)2 fragment from donkey or anti-mouse IgG,

peroxidase-linked whole antibody from sheep). Membranes were developed using ECL reagents

(Amersham).

Silver Staining—Type XIX protein was stained using the SilverXpress Silver Staining Kit

from Invitrogen (Carlsbad, CA) following the company’s procedure for “samples reduced with

DTT” with the following modifications. Samples were brought to a final concentration of 20

mM DTT, boiled for 2 min, and cooled to room temperature. Iodoacetamide was added to 0.1 M

and samples were incubated at 37 °C for 10 min before being applied to the gel. In addition,

during the gel staining process, the second sensitizing step was increased from 30 to 60 min.

Immunohistochemistry—Human umbilical cord was obtained from Robert Wood Johnson

University Hospital, snap-frozen in OCT compound in methylbutane at liquid nitrogen

temperature and sliced into 4-µm sections. The immunoperoxidase staining procedure has been

detailed earlier (18, 19). Sections were incubated with the type XIX COOH-Ab (18) or a type IV

collagen polyclonal antibody (DAKOpatts, Carpentaria, CA). Swine anti-rabbit secondary

antibody and peroxidase-conjugated streptavidin were also obtained from DAKOpatts.

Heparin Sepharose Chromatography—An aliquot (220 µg) of the SP Sepharose pool was

diluted to 0.2 M NaCl, and digested to completion with 100 units of bacterial collagenase for 3 h

at 37 °C in 0.5 ml. EDTA was added to a final concentration of 20 mM and the sample was

diluted to 1 ml and applied to a 1-ml Heparin Sepharose HiTrap column (Amersham) in a Tris-

HCl, pH 7.4, buffer containing 0.1 M NaCl, 10 mM EDTA and protease inhibitors. The column

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

9

was washed with 7 ml each of 0.1 M and 0.3 M NaCl buffers. The bound material was eluted in

1 M NaCl buffer in 0.3 ml-fractions.

Rotary-Shadowing Electron Microscopy—Type XIX collagen was visualized by electron

microscopy of rotary shadowed samples prepared by modifications of published methods (24–

26). Type XIX, eluted from the antibody affinity column, was dialyzed in ammonium formate as

stated above, and autoclaved glycerol (99.5+% spectrophotometric grade, Aldrich, Milwaukee,

WI) was added to give a final concentration of 50%. The sample was sprayed onto 0.5-cm2

pieces of freshly cleaved mica using an EFFA spray mount device (E.F. Fullam Inc, Latham,

NY). The sheets of mica were then placed in a Denton DV-502 vacuum evaporator (Denton

Vacuum, Cherry Hill, NJ) and pumped until the vacuum was about 3–4 × 10-7 Torr. Tungsten

was evaporated at an angle of about 4–7° while the stage containing the mica was rotating, and

carbon was evaporated on top of the tungsten as a support layer. The replicas were floated off

the mica onto a water surface and picked up onto 400-mesh copper grids and examined with a

Philips EM 400T transmission electron microscope (Philips, Hillsboro, OR) at 80 kV. Many

grids were examined and micrographs were taken from a variety of areas at a magnification of

generally 60,000. In an alternative method (Nagaswami technique) modified from previous

reports (27), about 10 µl of the sample was applied to a piece of Parafilm on ice. A 0.5-cm2 piece

of freshly-cleaved mica was placed on the top of the droplet and allowed to remain on the ice for

about 2 h. The mica was removed from the Parafilm, and the buffer described above (50%

glycerol in ammonium formate buffer) was placed onto the mica. The excess solution was

removed by blotting and the mica was allowed to remain in the cold room overnight. The next

day, rotary shadowing was carried out as described above.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

10

RESULTS

Identification of Type XIX Collagen in Human Umbilical Cord Tissue by Western Blotting—

Purification of the in vivo form of type XIX collagen depended upon two immediate factors. One

was the ability to detect the protein with characterized antibodies raised against the terminal

noncollagenous domains (ref. 18 and Methods), and the second was finding a tissue source that

could be obtained in sufficient amounts to isolate what was expected to be an extremely scarce

protein. This limited human tissue to placenta and umbilical cord, whereas a number of bovine

tissues were potentially available if the human antibodies cross-reacted. Immunoblots of crude

extract prepared from human placenta and umbilical cord were negative for type XIX, so

conventional procedures were employed to enrich the preparation for matrix proteins (28).

Soluble material obtained from a 5 M NaCl precipitate following homogenization and 1 M NaCl

extraction showed a convincing type XIX signal in human umbilical cord extract (Fig. 1, lanes 2

and 5). Both the type XIX NH2- and COOH-Abs identified the expected 165-kDa collagenous

protein (observed in Western blots of RMS cell extracts, refs. 18, 20). No type XIX collagen

was found using similarly treated human placenta or bovine cord samples (Fig. 1, lanes 1 and 4,

3 and 6, respectively). The former was consistent with the weak placenta immunostaining

reported earlier (18), and the latter suggested that neither human antibody recognized bovine

XIX.

Immunohistochemical Staining of Type XIX in Umbilical Cord—To corroborate the Western

blotting result and to identify what structures type XIX was associated with in cord tissue,

immunohistochemical analysis was conducted. The umbilical cord is covered by a simple

amniotic epithelium, and contains one vein and two arteries surrounded by a mucous connective

tissue matrix, Wharton’s jelly (29–31). Type XIX was prevalent in epithelial, smooth muscle

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

11

and endothelial BMZs (Fig. 2). The amniotic epithelial BMZ (upper left panel) and the BMZ

surrounding the muscle cells of the vessels (upper right panel) were strongly reactive with the

type XIX COOH-Ab. Within Wharton’s jelly (upper left panel), type XIX staining showed a

diffuse localization throughout the outer region, whereas in the portion of the cord adjacent to the

large vessels, it appeared more condensed (data not shown). The BM/BMZ localization of type

IV collagen (Fig. 2, lower panels) was comparable to type XIX, and in Wharton’s jelly, the type

IV staining was more uniform throughout the matrix (lower left panel).

Purification of Type XIX Collagen by Neutral Salt Extraction, and Ion Exchange and

Antibody Affinity Chromatography—A number of different size umbilical cord pieces (usually

10–15) were pooled for each of the type XIX preparations; the maximum amount of tissue

processed at one time was 250 g. Almost all of the type XIX collagen was extracted from

homogenized tissue in a 1 M NaCl buffer, and subsequent concentration and fractionation using

40% saturated ammonium sulfate resulted in an excellent recovery (Methods and data not

shown). Following precipitation, the pellet was resuspended, dialyzed, and applied to an ion

exchange resin. The choice was dictated by the 8.6 pI of the protein, and as predicted, type XIX

exhibited a high affinity for the cation exchange resin, SP Sepharose (Methods and data not

shown). The protein was bound in a batch procedure at 0.1 M NaCl, completely retained after a

0.3 M NaCl wash, and eluted using 0.6 M NaCl where >95% of the type XIX was recovered in

the first two buffer applications. The ionic strength of the 0.6 M pools 1 and 2 corresponded to

0.44 M and 0.55 M NaCl, respectively, reflecting mixture with the wash buffer. Western blotting

of a final 1.0 M NaCl treatment of the resin was negative for type XIX.

In a silver-stained gel of the SP Sepharose 0.6 M NaCl pool, a probable type XIX band

appeared as a very minor species in a heterogeneous array of various sized proteins (data not

shown). This result verified the scarcity of type XIX and the absolute requirement for a specific

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

12

mode of purification only afforded by an antibody affinity column. The SP Sepharose eluate was

concentrated, diluted to 0.4 M NaCl and mixed overnight with the type XIX COOH-Ab

covalently bound to a gel matrix. The column elution profile was evaluated by Western blotting

using both the NH2- and COOH-Abs (Fig. 3A and 3B). A minimal amount of type XIX was

found in the flow-through and wash (Fig. 3, lanes 2–3 and 9–10); type XIX eluted sharply in

fraction 2 where the pH 2.7 glycine displaced the pH 7.2 equilibration and wash buffer (Fig. 3,

lanes 5 and 12).

Different type XIX cleavage fragments were detected depending upon the antibody used.

Noticeable throughout the procedure, but somewhat increased in the purified fraction, was a

~120-kDa minor fragment identified by the NH2-Ab (Fig. 3A, lanes 1 and 5). A much more

pronounced cleavage product was evident using the COOH-Ab; about one-third to one-half

(depending upon the preparation) of type XIX was present in the form of a 140-kDa band that

was seen before the column elution as a comparatively minor fraction (Fig. 3B, lanes 8 and 12).

Despite immediate addition of a variety of freshly-prepared protease inhibitors to the fractions,

the 140-kDa species (later visualized by electron microscopy) remained a significant portion of

the type XIX recovered.

The yield and purity of type XIX was also determined by silver staining (Fig. 4). The 165-

kDa intact chain and the 140-kDa degradation fragment were clearly seen; the 120-kDa fragment

was faintly evident (Fig. 4, lane 1). All bands were digested with bacterial collagenase (Fig. 4,

lane 2) and their identity and relative proportions verified by Western blotting of aliquots

electrophoresed on parallel lanes (data not shown). The total amount of type XIX at this final

step, too low for conventional protein assays to be employed, was estimated by comparing the

intensity of the bands to a range of known amounts of type I collagen electrophoresed on adjacent

lanes (data not shown). The results showed that ~1.5 µg of type XIX protein was purified from

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

13

230 to 250 g of umbilical cord tissue, translating to a representation of 6 × 10-6%. A flow

diagram of the procedure, together with absolute and relative amounts of protein recovered at

each step, is shown in Fig. 5.

The Unique Shape of Type XIX Collagen as Visualized by Electron Microscopy—The

purified native type XIX collagen was examined by transmission electron microscopy following

rotary shadowing with tungsten in a vacuum evaporator. Eight independent experiments were

carried out using type XIX collagen alone, and/or type XIX incubated with the COOH- or NH2-

Ab. Visualization of the grids containing type XIX alone revealed that most of the individual

particles were long, rod-like structures with multiple kinks or bends and a small nodule at one

end (Fig. 6), indicating that this is the shape of the type XIX molecule. The dimensions of the

structures observed were determined. The mean diameter of the globular region was 19.7 ± 2.6

nm (n = 40). The mean length of the rods could not be established directly because of the many

different combinations and angles of the kinks. Instead, individual segments were measured, and

a histogram was constructed to show the average positions of the kinks relative to the end of the

molecule as defined by the globular region (Fig. 7). Six peaks (P1–P6) representing the sites of

the major kinks were identified; they were located at 52 (P1), 70 (P2), 85 (P3), 135 (P4), 170

(P5), and 215 (P6) nm. While these distances are the averages of all molecules containing each of

these kinks, it should be noted that not all molecules contain all of the kinks (Fig. 6). Some peaks

are broad, suggesting that there may be variability in the location of the kinks. Such differences

can be accounted for if portions of some molecules were not lying flat on the surface when they

were sprayed. In this case, the images observed would be projections of the individual rod

segments with correspondingly shorter lengths depending upon how steeply they were angled.

Such behavior would be expected for rigid collagen rods. Some of the variation could also be

explained because the bend could occur at different sites in the four large noncollagenous (NC)

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

14

segments (20–44 residues) as well as within the nine smaller interruptions (1–6 residues) that

occur in four of the five collagenous subdomains (schematic diagram of the chain is shown in

Fig. 12A). The mean total length of the whole molecule was estimated at 240 ± 23 nm.

Other structures commonly seen in the micrographs were long, thin, kinked rods without any

globular domains at the end, and globular nodules alone (data not shown). Many of the nodules

were identical in size and shape to those at one end of the most commonly observed structures

shown in Fig. 6. Other particles had the appearance of two or more nodules associating with

each other. The rods were identical to the rod-portion of the structures that also contained a

nodule at one end (Fig. 6). It was apparent that the separated nodules and rods are, as described

above in the purification process (Figs. 3 and 4), cleavage products of the larger, intact structures.

Polyclonal and monoclonal antibodies to the type XIX carboxyl- and amino-terminal ends,

respectively, were used to identify those portions of the structures. Rotary shadowed IgG

antibodies appear by electron microscopy to be three-lobed structures in which the relationship

among the lobes is highly variable because of their flexibility (32). Many different views of

various antibody conformations are visible. After incubation of the antibodies with type XIX

collagen, the complexes were rotary shadowed and studied by electron microscopy. The type

XIX polyclonal COOH-Ab bound to the end of the collagen opposite from the nodule, verifying

the location of the short carboxyl-terminal peptide (Fig. 8A). In contrast, images of type XIX

collagen incubated with the monoclonal NH2-Ab show that this IgG molecule bound to the

nodule itself, corroborating that this is the amino-terminal noncollagenous domain (Fig. 8B).

Identification of Type XIX Amino-terminal Linked Complexes—In the electron micrographs

of all preparations of type XIX, there were many aggregates of the collagen molecules. These

almost invariably interacted with each other via their globular amino-terminal ends (Fig. 9). In

most cases, the number of molecules in each aggregate could be extrapolated from the number of

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

15

rod-like tails extending away from the core containing the interacting nodules. All molecules,

both those interacting with each other and the individual ones, were counted in many

micrographs, yielding an estimate of the proportion of multimers. Total percentages for each size

oligomer (n = 300) were: monomer, 43.7%; dimer, 25.9%; trimer, 8.9%; tetramer, 7.4%;

pentamer, 7.4%; hexamer or larger, 6.7%. In some instances, these structures were too intricate

to accurately assess the number of constituent molecules. Also seen in the micrographs were

examples of a higher level of type XIX organization; a classic representative is shown in Fig. 9

(large panel). These featured clusters of the aggregates described above, and comprised a great

many individual type XIX molecules.

To further ascertain the nature of the amino-terminal interactions, Western blotting of

purified type XIX was conducted under different conditions using the NH2-Ab as probe (Fig. 10).

As reported earlier for type XIX synthesized by RMS cells (18), in the absence of reducing agent,

all the intact type XIX molecules were found in the stacking gel or at the top of the separating gel

(Fig. 10, lane 1). No individual 165-kDa chains or lower molecular weight species were detected

showing that the bands revealed in lanes 2 and 3 were a function of the treatment indicated. Type

XIX, digested with collagenase and electrophoresed under nonreducing conditions, appeared

mainly in the form of several bands ranging in size from ~110 to ≥250 kDa (Fig. 10, lane 2). A

band of ~60 kDa was also seen. In the presence of DTT (Fig. 10, lane 3), the amino fragment

migrated as a 34-kDa band (previously shown for collagenase-treated, RMS type XIX, ref. 18).

[The minor band of 29 kDa is due to a secondary cleavage site for collagenase.] The 110-kDa

band (3 × 34 kDa) would represent the cleaved amino terminus of a monomer (linked by

interchain bonds) and the higher molecular weight bands correspond to this domain in the

oligomers. The intensity of the larger fragments will be underrepresented compared to smaller

ones due to the decrease in transfer efficiency. Consistent with the electron microscopy images,

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

16

these results independently show that the type XIX molecules associate via their amino-terminal

domains, and furthermore reveal that this complex occurs by formation of intermolecular

disulfide cross-links.

The Type XIX Amino Terminus Contains a Heparin-Binding Site—The type XIX amino

terminus has been characterized as a Tsp-N module (10, 33, 34), and specific functions of

thrombospondin-1 (TSP-1) have been mapped to this domain (reviewed in refs. 35, 36).

Alignment of the modules in type XIX and TSP-1 revealed about 15% absolute identity as

dispersed single residues or doublets (data not shown). The most conspicuous homology was the

TSP-1 motif {R x x K K x R} (37) embedded in a longer basic-residue sequence {R x R R x x K

K x R} in type XIX (16). The {K K x R} peptide is a known TSP-1 heparin-binding site (38).

To ascertain whether this site in type XIX is functional, the SP Sepharose peak was digested

with collagenase and applied to a heparin Sepharose column. Western blotting demonstrated that

all of the type XIX amino-terminal fragment bound to the column in 0.1 M NaCl, was retained

after a 0.3 M NaCl wash, and eluted sharply in a 1 M NaCl buffer (Fig. 11). Prior

chromatography showed that the fragment gradient-eluted at 0.70–0.75 M NaCl.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

17

DISCUSSION

We have performed the first direct biochemical and structural characterization of type XIX

collagen. Prior to this report, all information has been deduced from the DNA sequence and

immunoblots of type XIX found in RMS cell extracts. The identity of type XIX purified from

human tissue was conclusively established using antibodies raised against the carboxyl-terminal

peptide and the middle of the amino-terminal noncollagenous domain.

Type XIX is a Rare Protein in Human Tissue—Type XIX is by far the least abundant

collagen so far purified, with a representation of ~10–6 % of the dry weight of umbilical cord

(even less than type VII collagen which is ~0.001% of pepsin-digested skin tissue, 39). This

value should be reasonably accurate, since the relative amount of type XIX at each stage was

monitored to assess the recovery. Once optimized, the isolation procedure was straightforward

and designed to minimize loss by capitalizing on favorable type XIX properties. Type XIX was

readily extracted from tissue using mild conditions, showing that it is not incorporated into an

insoluble matrix complex. At the next stage, an inefficient 5 M NaCl precipitation was

effectively replaced by use of 40% saturated ammonium sulfate, and type XIX in the pelleted

material was completely soluble in neutral salt buffer. Strong affinity for a cation exchanger

allowed for a stringent wash and complete disassociation from the resin using a high salt buffer.

The rapid batch procedure was not replaced by gradient column elution since it was clear that ion

exchange chromatography just served as a prerequisite for specific capture by a type XIX

antibody. The former step afforded an ~2000-fold enrichment and the latter an additional ~7000

fold (totaling ~1.4 × 107). The strategy exemplified here (Fig. 5) may provide a blueprint for

purification of other extremely rare collagens. Only the COOH-Ab bound to native type XIX,

consistent with its ability to also recognize the protein by immunohistochemistry (18). The NH2-

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

18

Ab was ineffective in both respects, likely due to inaccessibility of the epitope within the clusters

visualized by rotary shadowing.

Proteolysis Results in Discrete Type XIX Cleavage Fragments—Two particularly labile sites

resulted in distinct cleavage products that were detected in Western blots and silver-stained gels.

The type XIX NH2-Ab reacted with a minor 120-kDa form that is likely generated from cleavage

in NC3, the 23-amino acid interruption (identified as flexible site P5 in Fig. 12C) separating

COL3 from the uninterrupted COL2 subdomain. Although NC3 is 301 residues from the

carboxyl terminus, this length corresponds to ~45 kDa considering that the 1119-residue type

XIX chain migrates as 165-kDa by SDS-PAGE, 1.5 times the estimated size. [This retarded

mobility is characteristic of collagen chains.]

The 140-kDa degradation product, identified only by the COOH-Ab, increased considerably

after elution from the antibody column where type XIX, at an ~2 µg/ml concentration, is sensitive

to even a trace amount of protease. The cleavage is predicted to be near the junction of the 30-

kDa amino terminus (not to be confused with the 34-kDa collagenase-generated fragment) and

COL5 (see Fig. 12A). Accordingly, globular domains separated from the kinked rods were

prominent in rotary shadowed images and their shape was the same as the corresponding regions

in the intact molecule. Three lysines are located within the last 14 residues of the type XIX amino

terminus (14); the middle one is a {K D} sequence found at the equivalent position in the TSP-1

Tsp-N module (37) where it is highly susceptible to trypsin cleavage (40). Identification of the

type XIX site will require a more abundant source of protein to obtain the amount needed for

sequencing. Characterization of these labile sites may prove important in studies of type XIX

catabolism and matrix remodeling.

The Type XIX Molecule Can Assume Many Conformations—Electron microscopy

established that individual type XIX molecules span 240 nm, terminate with a small nodule at the

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

19

amino end and contain rod-like collagenous domains interrupted by sharp bends. The 220 nm

length, estimated for the sum total of the collagenous rods, agrees well with the established value

of 0.286 nm per residue (41); 712 residues × 0.286 = 204 nm, which would be a minimum since

the residues in NC5–2 could not be measured.

A wide array of different type XIX images was observed (Fig. 6 gallery). The histogram

(Fig. 7) indicated that there are up to six highly flexible regions. Some type XIX molecules

display all six kinks while others have fewer; not all hinge regions are always apparent. There are

examples of extended molecules with one or two kinks; these contrast with others that are highly

bent and assume a “zigzag” shape (Fig. 6a, e, k, l). In some cases, the angles are so pronounced

that a part of the molecule folds back on itself (Fig. 6b, d, g–j). Another flexible site may flank

the amino terminus where cleavage of exposed residues could give rise to the 140-kDa fragment;

however, this remains unresolved since a rod emanating from the nodule radially or tangentially

cannot be easily differentiated.

A high correlation exists between the positions of the six major kinks and the NC segments.

The five predicted type XIX collagenous subdomains are composed of 144, 224, 108, 168 and 70

amino acids (Fig. 12A), which would represent 20%, 31%, 15%, 23.5%, and 10%, respectively,

of 220 nm. The length of each of these rods would therefore be 44, 68, 33, 52, and 22 nm,

(discounting the NC5–2 segments), and the cumulative distance to the bends would be 64 (i.e.,

44 + 20 nm), 132, 165, 217, and 240 nm (Fig. 12B). An illustration depicting the spatial

relationship of the kinks (P1–P6 peaks in Fig. 7) to the type XIX domain structure reveals a

remarkable coincidence of P2, P4, P5 and P6 with NC5, NC4, NC3 and NC2, respectively. The

other two bends, P1 and P3, closely align with three-residue interruptions in COL5 and COL4.

Extreme flexibility at such sites is not surprising since in osteogenesis imperfecta even a single

glycine substitution in type I collagen was shown to cause kinks (42–44).

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

20

The nonfibrillar collagens contain various size interruptions in the collagenous region;

electron microscopy has been carried out to visualize some of these molecules (10 and refs.

therein, and refs. 45–47). The kinks/bends that have been observed are often difficult to map to

specific interruptions when the collagenous subdomains are not as defined as they are in type

XIX and related family members (see Introduction). Alternatively, it is known that the presence

of even large noncollagenous segments does not necessarily extrapolate into kinks as seen for the

three-subdomain type IX collagenous region. Only one kink, equivalent to the 12–17-residue

NC3 and the site of glycosaminoglycan chain attachment, is seen by electron microscopy despite

the presence of a 30-residue NC2 segment (10, 48). In type XIX, manifestation of all the large,

and at least several of the small, interruptions into highly flexible sites results in the most

polymorphic collagen so far characterized. The interruptions support numerous spatial

configurations (illustrated in Fig. 12D) and, in this regard, can confer a high degree of

adaptability to the microenvironment by differentially positioning the rigid collagenous

subdomains. Moreover, the long, extensible NC segments may, in particular, represent binding

sites whose availability can be modulated by the degree to which the residues are exposed to

other matrix molecules. Taken together, the permutations in the type XIX structure are

considerable, and even subtle differences may influence many biological processes.

The Supramolecular Assembly of Type XIX is Dependent Upon Amino-terminal

Interactions—More than half the type XIX collagen molecules were present in oligomers

interacting via their amino-terminal ends, suggesting that these complexes are physiologically

significant. It is very unlikely that they could be attributed to a technical artifact since the

inherent low density of the type XIX molecules sprayed onto the surface never favored a crowded

environment. There are many examples of self-assembly of biological systems, but structural

proteins in particular are "built to assemble”. In other words, the binding interactions that are

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

21

required for organization of some biological structures (exemplified in fact by collagens, refs. 8,

10) are often manifested in the purified protein(s), and therefore supramolecular aggregates

observed by electron microscopy commonly reflect aspects of their in vivo assembly.

Nonfibrillar collagens—with most still to be characterized—are known to organize into

diverse higher order structures, e.g., polygonal networks, hexagonal lattices, beaded filaments,

antiparallel dimer filaments, and direct association with fibrillar collagens (reviews 1, 2, 8 and

10). They occur through end-to-end and/or side-by-side interactions, and many, as shown here

for type XIX, are stabilized by intermolecular disulfide cross-links. The type XIX arrangement

represents a previously unknown structure. The only visual parallel is the type X collagen

aggregates, which were found in cultured cells expressing the endogenous or recombinant protein

(49, 50). The individual type X molecules radiate from a globular carboxyl (not amino)-terminal

core and these complexes appear to interconnect through antiparallel overlap of rod-like triple

helices (8, 10, 49). As demonstrated in the type X studies (49, 50), immunoelectron microscopy

and molecular aggregation using a recombinant form may prove useful to further elucidate the

type XIX structure.

Potential Biological Significance of the Type XIX Amino Terminus—Knowledge of the Tsp-

N domain in other systems may provide important clues about the role of the type XIX amino

terminus. The Tsp-N module is found in about 100 proteins, particularly those multi-domain

adhesive proteins that act as molecular bridges between cells and matrix, and participate in cell-

cell communication (51). Tsp-N contains patterns of alternating hydrophobicity characteristic of

anti-parallel β strands, and the presence of the predicted sandwiches appears, at least in part, to

dictate the conserved, known and predicted, structural homology (34). A high propensity for β

sheet formation is also found in the Tsp-N/amino terminus of type XIX (14). Functions ascribed

to Tsp-N of TSP-1 include cell adhesion, spreading and migration, binding to glycosaminoglycan

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

22

chains, disruption of focal contacts, regulation of proliferation, endocytosis and platelet

aggregation (35). Statistical analysis corroborated by the crystal structure has also shown a

similarity of Tsp-N to the laminin G module which shares a number of the aforementioned

activities in addition to signaling, assembly and differentiation (51, 52). The Tsp-N module is

found too in TSP-like NELL proteins that are involved in neurogenesis and the maintenance of

neuronal plasticity (53).

The Tsp-N module is also present in collagens IX, XII, XIV and XVI (10, 17, 33, 34), but

the {K K x R} heparin-binding site in TSP-1 is only conserved in type XIX. It has been shown

that TSP-1, mediated by the Tsp-N module, binds to cells and is internalized and degraded in a

process that requires heparan sulfate proteoglycans (HSPGs) (54–56). One can speculate that the

type XIX oligomers may function as a nidus to increase the local concentration of

signaling/structural molecules by associating with multiple extracellular matrix HSPGs, and/or

type XIX could serve as a cell-cell/cell-matrix adhesive protein by binding to transmembrane

HSPGs, like the syndecans, as has been established for TSP-1 (35, 36). Taken together, the

results shown here provide a solid foundation for approaching functional studies of type XIX in

the basement membrane/zone and in myogenic processes.

ACKNOWLEDGEMENTS

These studies were supported by NIH grants GM64777, AR20553, and HL30954. We thank

Dr. Francesco Ramirez for providing information prior to publication. We have had valuable

assistance/advice in different aspects of this project from Donna Hardy (Bio-Rad Laboratories)

and Dr. Marcos Milla, as well as from Fangping Zhou, Kelly Walton, and Nicole Scivoletti. We

appreciate the kind gift of highly purified bacterial collagenase from Dr. Bo Yu at Advance

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

23

Biofactures. We are especially indebted to the many people (residents, attending physicians,

nurses, house staff) in the Department of Obstetrics and Gynecology, Hospital of the University

of Pennsylvania, for saving us > 400 cord specimens over a period of two years. The enthusiastic

support of Dr. Matheu Beshara, who initiated the effort, and Dr. Jerome F. Strauss, who first

approved the arrangement, is very gratefully acknowledged.

REFERENCES

1. Myllyharju, M., and Kivirikko, K. I. (2001) Ann. Med. 33, 7-21

2. Kielty, C. M., and Grant, M. E. (2002) in Connective Tissue and Its Heritable Disorders

(Royce, P.M., and Steinman, B., eds), pp. 159-221, Wiley-Liss, Inc., New York

3. Koch, M., Foley, J. E., Hahn, R., Zhou, P., Burgeson, R. E., Gerecke, D. R., and Gordon, M.

K. (2001) J. Biol. Chem. 276, 23120-23126

4. Fitzgerald, J., and Bateman, J. F. (2001) FEBS Lett. 505, 275-280

5. Hashimoto, T., Wakabayashi, T., Watanabe, A., Kowa, H., Hosoda, R., Nakamura, A.,

Kanazawa, I., Arai, T., Takio, K., Mann, D. M. A., and Iwatsubo, T. (2002) EMBO J. 21,

1524-1534

6. Sato, K., Yomogida, K., Wada, T., Yorihuzi, T., Nishimune, Y., Hosokawa, N., and Nagata,

K. (2002) J. Biol. Chem. 277, 37678-37684

7. Banyard, J., Bao, L., and Zetter, B. R. J. Biol. Chem., Mar 2003 online; 210616200

8. van der Rest, M., and Garrone, R. (1991) FASEB J. 5, 2814-2823

9. Brown, J. C., and Timpl, R. (1995) Int. Arch. Allergy Immunol. 107, 484-490

10. Ricard-Blum, S., Dublet, B., and van der Rest, M. (2000) Unconventional Collagens: Types

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

24

VI, VII, VIII, IX, X, XII, XIV, XVI and XIX, 1st edn., Oxford University Press, Oxford

11. Prockop, D. J., and Kivirikko, K. I. (1995) Annu. Rev. Biochem. 64, 403-434

12. Bruckner-Tuderman, L., and Bruckner, P. (1998) J. Mol. Med. 76, 226-237

13. Yoshioka, H., Zhang, H., Ramirez, F., Mattei, M.-G., Moradi-Ameli, M., van der Rest, M.,

and Gordon, M. K. (1992) Genomics 13, 884-886

14. Myers, J. C., Sun, M. J., D’Ippolito, J. A., Jabs, E. W., Neilson, E. G., and Dion, A. S.

(1993) Gene 123, 211-217

15. Myers, J. C., Yang, H., D’Ippolito, J. A., Presente, A., Miller, M. K., and Dion, A. S. (1994)

J.

Biol. Chem. 269, 18549-18557

16. Inoguchi, K., Yoshioka, H., Khaleduzzaman, M., and Ninomiya, Y. (1995) J. Biochem.

(Tokyo)

117, 137-146

17. Pan, T.-C., Zhang, R.-Z., Mattei, M.-G., Timpl, R., and Chu, M.-L. (1992) Proc. Natl. Acad.

Sci. USA 89, 6565-6569

18. Myers, J. C., Li, D., Bageris, A., Abraham, V., Dion, A. S., and Amenta, P. S. (1997) Am. J.

Pathol. (1997) 151, 1729-1740

19. Amenta, P. S., Hadad, S., Lee, M. T., Barnard, N., Li, D., and Myers, J. C. (2003) J. Pathol.

199, 298-308

20. Myers, J. C., Li, D., Rubinstein, N. A., and Clark, C. C. (1999) Exp. Cell Res. 253, 587-598

21. Sumiyoshi, H., Laub, F., Yoshioka, H., and Ramirez, F. (2001) Dev. Dyn. 220, 155-162

22. Harlow, E., and Lane, D. (1988) Antibodies: A Laboratory Manual. Cold Spring Harbor

Laboratory, New York

23. Li, D., Clark, C. C., and Myers, J. C. (2000) J. Biol. Chem. 275, 22339-22347

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

25

24. Fowler, W. E., and Erickson, H. P. (1979) J. Mol. Biol. 134, 241-249

25. Weisel, J. W., Stauffacher, C. V., Bullitt, E., and Cohen, C. (1985) Science 230, 1388-1391

26. Weisel, J. W., Nagaswami, C., Woodhead, J. L., DeLa Cadena, R., Page, J. D., and Colman,

R. W. (1993) J. Biol. Chem. 269, 10100-10106

27. Weisel, J. W., Nagaswami, C., Woodhead, J. W., Higazi, A. A.-R., Cain, W. J., Marcovina,

S. M., Koschinsky, M. L., Cines, D. B., and Bdeir, K. (2001) Biochemistry 40, 10424-

10435

28. Furuto, D. K., and Miller, E. J. (1987) Meth. Enzymol. 144, 41-61

29. Eyden, B. P., Ponting, J., Davies, H., Bartley, C., and Torgersen, E. (1994) Submicrosc.

Cytol. Pathol. 26, 347-355

30. Meyer, F. A. (1996) in Extracellular Matrix (Comper, W. D., ed.) pp. 443-456, Harwood

Academic Publishers, The Netherlands

31. Nanaev, A. K., Kohnen, G., Milovanov, A. P., Domogatsky, S. P., and Kaufman, P. (1997)

Placenta 18, 53-64

32. Veklich, Y. I., Gorkun, O. V., Medved, L. V., Nieuwenhuizen, W., and Weisel, J. W. (1993)

J. Biol. Chem. 268, 13577-13585

33. Bork, P. (1992) FEBS Lett. 307, 49-54

34. Moradi-Ameli, M., Deleage, G., Geourjon, C., and van der Rest, M. (1994) Matrix Biol. 14,

233-239

35. Chen, H., Herndon, M. E., and Lawler, J. (2000) Matrix Biol. 19, 597-614

36. Adams, J. C. (2001) Annu. Rev. Cell Dev. Biol. 17, 25-51

37. Lawler, J., and Hynes, R. O. (1986) J. Cell Biol. 103, 1635-1648

38. Lawler, J., Ferro, P., and Duquette, M. (1992) Biochemistry 31, 1173-1180

39. Bruckner-Tuderman, L., Schnyder, U. W., Winterhalter, K. H., and Bruckner, P. (1987) Eur.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

26

J. Biochem. 163, 607-611

40. Misenheimer, T. M., Huwiler, K. G., Annis, D. S., and Mosher, D. F. (2000) J. Biol. Chem.

275, 40938-40945

41. Piez, K. A. (1984) in Extracellular Matrix Biochemistry (Piez, K. A., and Reddi, A. H., eds.)

pp. 1-40, Elsevier Science Publishing Co., Inc., New York

42. Vogel, B. E., Doelz, R., Kadler, K. E., Hojima, Y., Engel, J., and Prockop, D. J. (1988) J.

Biol. Chem. 263, 19249-19255

43. Lightfoot, S. J., Holmes, D. F., Brass, A., Grant, M. E., Byers, P. H., and Kadler, K. E.

(1992) J. Biol. Chem. 267, 25521-25528

44. Forlino, A., Keene, D. R., Schmidt, K., and Marini, J. C. (1998) Matrix Biol. 17, 575-584

45. Timpl, R., Wiedemann, H., Van Delden, V., Furthmayr, H., and Kühn, K. (1981) Eur. J.

Biochem. 120, 203-211

46. Hirako, Y., Usukura, J., Nishizawa, Y., and Owaribe, K. (1996) J. Biol. Chem. 271, 13739-

13745

47. Tu, H., Sasaki, T., Snellman, A., Gohring, W., Pirila, P., Timpl, R., and Pihlajaniemi, T.

(2002) J. Biol. Chem. 277, 23092-23099

48. Vaughan, L., Mendler, M., Huber, S., Bruckner, P., Winterhalter, K. H., Irwin, M. I., and

Mayne, R. (1988) J. Cell Biol. 106, 991-997

49. Kwan, A. P. L., Cummings, C. E., Chapman, J. A., and Grant, M. E. (1991) J. Cell Biol. 114,

597-604

50. Frischholz, S., Beier, F., Girkontaite, I., Wagner, K., Poschl, E., Turnay, J., Mayer, U., and

von der Mark, K. (1998) J. Biol. Chem. 273, 4547-4555

51. Beckmann, G., Hanke, J., Bork, P., and Reich, J. G. (1998) J. Mol. Biol. 275, 725-730

52. Hohenester, E., Tisi, D., Talts, J. F., and Timpl, R. (1999) Mol. Cell 4, 783-792

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

27

53. Kuroda, S., Oyasu, M., Kawakami, M., Kanayama, N., Tanizawa, K., Saito, N., Abe, T.,

Matsuhashi, S., and Ting, K. (1999) Biochem. Biophys. Res. Commun. 265, 79-86

54. Chen, H., Sottile, J., Strickland, D. K., and Mosher, D. F. (1996) Biochem. J. 318, 959-963

55. Merle, B., Malavai, L., Lawler, J., Delmas, P., and Clezardin, P. (1997) J. Cell. Biochem. 67,

75-83

56. Mikhailenko, I., Krylov, D., McTigue Argraves, K., Roberts, D. D., Liau, G., and

Strickland, D. K. (1997) J. Biol. Chem. 272, 6784-6791

FOOTNOTES

1 The abbreviations used are: RMS, rhabdomyosarcoma; TSP-N, thrombospondin-amino

terminal domain; BMZ, basement membrane zone; COOH-Ab, type XIX antibody recognizing

the noncollagenous carboxy peptide; NH2-Ab, type XIX antibody recognizing the amino-

terminal noncollagenous domain; TSP-1, thrombospondin-1; NC, noncollagenous; COL,

collagenous subdomain; P, peak; HSPG, heparan sulfate proteoglycan.

2 Francesco Ramirez, Personal communication.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

28

LEGENDS

Fig. 1. Identification of type XIX collagen in human umbilical cord by Western blotting.

Total protein was extracted from homogenized human placenta (lanes 1 and 4), and human and

bovine umbilical cord (lanes 2 and 5, 3 and 6, respectively) in a neutral pH, 1 M NaCl buffer,

precipitated using 5 M NaCl, and resuspended in extraction buffer. Sixty micrograms per lane

were electrophoresed in a 6% polyacrylamide-SDS gel, Western blotted and reacted with the type

XIX NH2- (lanes 1–3) or COOH-Ab (lanes 4–6) as indicated. The 165-kDa type XIX band was

not seen when the sample was incubated with bacterial collagenase prior to electrophoresis (data

not shown).

Fig. 2. Immunohistochemical localization of types XIX and IV collagen in human umbilical

cord. The amniotic epithelial layer (e) BMZ (arrows) and Wharton’s jelly (*) stain positive for

types XIX (upper left) and IV collagen (lower left). Original magnification = × 82. Type XIX

(upper right) and type IV (lower right) are also present in the BMZ of the endothelium

(arrowhead) and smooth muscle cells (curved arrows) of the vein as shown here and the two

arteries (data not shown). l = vascular lumen. Original magnification = × 41 for the type XIX

illustration and × 82 for type IV.

Fig. 3. Complete purification of type XIX collagen using a COOH-Ab affinity column.

Pools 1 and 2 from the SP Sepharose eluate (Methods) were concentrated, diluted to 0.4 M NaCl,

and incubated with Affi-Gel resin to which the type XIX COOH-Ab was covalently bound. The

aliquots used in lanes 1–2/8–9 and 3–7/10–14 were 0.2% and 0.4%, respectively, of the total

volume. After Western blotting, the filters were incubated with the type XIX NH2-Ab (panel A),

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

29

and the COOH-Ab (panel B). The 165-kDa band is the intact type XIX chain, and the 120- and

140-kDa bands appear to be cleavage fragments identified by the NH2- and COOH-Abs,

respectively. Lanes 7 and 14 show an aliquot of column fraction 2 incubated with bacterial

collagenase before gel electrophoresis.

Fig. 4. Silver-stained gel of purified type XIX collagen. About 40 ng of purified type XIX

collagen was electrophoresed under reducing conditions on a 7% SDS-polyacrylamide gel after

incubation in the (-) absence (lane 1) or (+) presence of bacterial collagenase (lane 2). The gel

was silver-stained as described in Methods. In lane 1, the 165-kDa intact type XIX chain is seen

along with the dominant 140-kDa cleavage fragment and the barely visible 120-kDa fragment.

The identity of each band were corroborated by Western blotting (data not shown but comparable

to the fraction 2 profile in Fig. 3). The bands seen in lane 2 (arrowheads) are the collagenase

enzyme.

Fig. 5. Schematic diagram of the type XIX collagen purification. Details of the purification

are provided in Methods. The values shown are averaged from five preparations with little

variation. At each step following the 1 M NaCl extraction, > 90% of the starting material was

accounted for (e.g. SP Sepharose flow-through of 1255 mg, 0.3 M NaCl pool of 92 mg and 0.6 M

NaCl pool of 11 mg = 1358 mg, which is 93% of the 1460 mg column load). The percent

recovery at each step was calculated using the original dry weight value of umbilical cord tissue.

Boxes outlined in bold show the fraction in which type XIX collagen was recovered. The yield of

purified type XIX, ~6 × 10-6%, was estimated by comparing the intensity of type XIX samples

(20–50 ng total for all three bands) with a range of known amounts of type I collagen

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

30

electrophoresed on the same gel and silver stained. The minimum amount of type XIX required

for the standard protein assay, at least 5 µg, was prohibitive to sacrifice for this purpose.

Fig. 6. Transmission electron micrographs of rotary shadowed type XIX collagen

molecules. This gallery of images illustrates the most commonly observed particles, which are

long rod-like structures with a nodule at one end and several sharp bends or kinks. Some of the

molecules are sharply kinked (a, d, f, j, l) to such an extent that one portion of the rod folds back

on the remainder of the molecule (g). Other molecules are kinked and more elongated (b, e, i, k).

Magnification bar, 50 nm.

Fig. 7. Histogram showing the location of the sharp bends or kinks in the rod regions of

type XIX collagen. Molecules were measured from the globular end, and the positions of the

kinks are shown as peaks in the plot of the frequency of kinks versus the cumulative lengths.

Molecules have up to six kinks, peaks (P1–P6) in the histogram, but most molecules display

fewer. The distance from the globular domain to each kink allows the identification of that kink

with one of the histogram peaks. The numbers beneath P1–P6 represent the average distances in

nm from the end of the molecule to each kink measured in the molecules displaying that kink(s).

Fig. 8. Electron micrographs of rotary shadowed type XIX collagen complexed with

antibodies to the carboxyl- and amino-terminal domains. Purified type XIX was incubated

for 60–75 min at room temperature with 1:1 or 1:2 ratio of collagen to antibody. Panel A. The

COOH-Ab binds to the rod at the end opposite to the small nodule. In some cases, the three

lobes of the IgG are visible (arrowheads). Panel B. The NH2-Ab (arrowheads) binds to the

globular region, identifying it as the amino-terminal end. Magnification bar, 50 nm.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

31

Fig. 9. Electron microscope images of rotary shadowed aggregates of type XIX molecules.

The five small panels show interaction of two to six or more collagen molecules via their

globular domains, so that the tails extend radially from the heads. More than half of the

molecules were present in such aggregates. As seen in Fig. 6, the tails often contain multiple

kinks, hindering a precise count of the number of molecules. The large image illustrates a higher

order structure that appears to consist of smaller individual aggregates joined at the amino

termini, and not just a larger version of one aggregate. None of the fields included an abundance

of collagen molecules; therefore, the structures seen here are not the result of random

juxtaposition on the surface and thus may well represent in vivo forms. Magnification bar, 50

nm.

Fig. 10. Biochemical characterization of the type XIX amino-terminal domain. Aliquots of

purified type XIX collagen were digested with bacterial collagenase (+), electrophoresed on a

10% SDS-polyacrylamide gel in the absence (- in lane 2) or presence (+ in lane 3) of 200 mM

DTT, Western blotted, and reacted with the type XIX NH2-Ab. All of the bands seen in lanes 2

and 3 appear only after collagenase digestion and reduction as illustrated in lane 1 where

undigested, unreduced type XIX is found solely in the stacking gel and at the top of the

separating gel. In the unreduced, collagenase-treated sample (lane 2), the ~110-kDa band

(arrowhead) most likely originates from a single type XIX molecule linked by interchain

disulfide bonds. The higher molecular weight bands (brace), more accurately sized on a 6% gel

(not shown), range from ~160 to ≥ 210-kDa. These would correspond to the type XIX

aggregates shown in Fig. 9, reflecting the existence of intermolecular cystine linkages. The 60-

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

32

kDa band may represent a degradation product. Note that the intensity of the bands is inversely

proportional to their size because of the decrease in transfer of high molecular weight proteins,

especially on a 10% gel. In the reduced sample (lane 3), the type XIX amino-terminal domain

migrates primarily at 34 kDa (arrow). Other data (not shown) have revealed the existence of

secondary collagenase cleavage sites that give rise to 29- and 60-kDa bands.

Fig. 11. The type XIX amino-terminal domain contains a heparin-binding site. Samples as

described below were electrophoresed in a 12% polyacrylamide-SDS gel, Western blotted and

probed with the type XIX NH2-Ab. An aliquot of the collagenase-digested SP Sepharose peak

was loaded on a heparin Sepharose column in 0.1 M NaCl (lane 1); no signal was found in the

flow-through or the 0.3 M NaCl wash (lanes 2 and 3, respectively). The 34-kDa fragment

(arrow) was eluted in 0.3 ml fractions in a 1 M NaCl buffer. [Prior chromatography showed that

the fragment gradient-eluted at 0.70–0.75 M NaCl.] An aliquot of fractions 10–14 are shown in

lanes 4–8. The 29-kDa fragment originates from a secondary collagenase cleavage site.

Fig. 12. Schematic diagram showing the shape of type XIX collagen, including comparison

with the primary structure, localization of the kinks, and illustration of individual

molecules. Part A. Schematic diagram of the type XIX polypeptide chain, predicted from the

amino acid sequence (14-16), shows the collagenous subdomains (COL1 to COL5) and non-

collagenous (NC1 to NC6) regions (numbered from carboxy to amino according to convention

for this class of collagens, see Introduction). NC1 corresponds to the carboxy peptide and NC6

to the amino terminus. The numbers of amino acids in each segment are indicated in parentheses.

Small interruptions, one to three residues in COL5, COL4, and COL1, and six amino acids in

COL3, are indicated as vertical black bars/lines. Each cysteine is designated by the letter “C”.

Part B. Predicted nm size of the collagenous subdomains is based on the percentage of total

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

33

length of the rod-like regions (220 nm = 240 nm molecular length – 20 nm globular domain)

from the number of amino acid residues shown in Part A. Part C. Schematic diagram of the

position of kinks in type XIX collagen (Fig. 7) derived from analysis of electron microscope

images (Fig. 6). The globular nodule is at the amino-terminal end, and is followed by rod-like

collagenous subdomains interrupted by noncollagenous segments at which site kinks can occur.

There are kinks at each of the long internal NC regions (P2, P4, P5, P6); two additional kinks (P1

and P3) may correspond to two 3-amino acid interruptions. Cumulative distances in nm from the

amino-terminal nodule to each of the kink regions are indicated. Part D. Illustration of some

possible conformations of type XIX collagen, from the diagram in Part C and based on electron

micrographs shown in Fig. 6. In some cases, the molecules are more extended, while in others

there are sharp kinks imparting a zigzag shape.

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

1 2 3 4 5 6

XIXNH2-Ab

XIXCOOH-Ab

Type XIX165 kDa

220

97

Hu.

Pla

cent

a

Hu.

Pla

cent

a

Hu.

U. C

ord

Bov

. U. C

ord

Bov

. U. C

ord

Hu.

U. C

ord

kDa

Figure 1

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

XIX

IV IV

XIXe

e

l

l

*

*

Figure 2

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

1 2 3 4 5 6

Load

Flow

-thro

ugh

Was

hFr

actio

n 1

Frac

tion

2

Frac

tion

3

Type XIX 165-kDa

120-kDa Frag.

XIX NH2-Ab

8 9 10 11 12 13

XIX COOH-Ab

140-kDa Frag.

208

101

kDa

kDa

208

101

A

B

Type XIX 165-kDa

14

7

Fr. 2

+ C

oll'a

se

Figure 3

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

1 2

Type XIX 165-kDa140-kDa Frag.

206

119

91

kDa

Silver stain

120-kDa Frag.

– +

Figure 4

Collagenase

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

Umbilical Cord(236 g wet wt.)23.6 g dry wt.

• Homogenize & extract in 1 M NaCl, pH 7.5 (1:10 w/v)

Insoluble Soluble

• Precipitate at 40% sat'd (NH4)2SO4

40% Pellet 40% Sup.

• Resuspend in 1 M NaCl

Insoluble Soluble

• Bind to SP-Sepharose in 0.1 M NaCl, pH 7.2

Flow-through &0.1 M NaCl wash

• Elute

0.3 M NaCl pool 0.6 M NaCl pool

• Bind to type XIX COOH-Ab affinity column

8670 mg, 37%

1682 mg, 7.1% 7055 mg, 30%

173 mg, 0.7% 1460 mg, 6.2%

1255 mg, 5.3%

92 mg, 0.4% 11 mg, 0.05%

• Elute with 1 M glycine, pH 2.7

Type XIX Collagen

• Centrifuge

• Centrifuge

• Centrifuge

~1.5 µg, 0.000006%

• Dialyze into 0.1 M NaCl, pH 7.2

• Concentrate• Dilute to 0.4 M NaCl, pH 7.2

8.45 mg, 0.036%

Flow-through &0.4 M NaCl wash

Figure 5

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

aa bb cc dd

ee ff gg hh

ii jj kk ll

Figure 6

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

0

1

2

3

4

5

6

7nu

mbe

r of

kin

ks

distance (nm)0 50 100 150 200 250 300

52 70 85 135 170 215 P1 P2 P3 P4 P5 P6

Figure 7

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

B

A

Figure 8

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

Figure 9

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

2 3

+

XIX NH2-Ab

220

97

66

45

30

DTTCollagenase+

– +

1

––

{

Figure 10

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

1 2 3 4 5 6 7 8

Bef

ore

colu

mn

Flow

-thro

ugh

0.3

M N

aCl w

ash

1 M

NaC

l-El

uted

Frac

tions

XIX NH2-Ab

Figure 11

34 kDa

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

20 nm 52 nm 70 nm 85 nm 135 nm 170 nm 215 nm 240 nm

20 nm 64 nm 132 nm 165 nm 217 nm 240 nm

44 nm 68 nm 33 nm 52 nm 23 nm

C CC C C C C CCC

CCC CCOL5 COL4 COL3 COL2 COL1(144) (224) (108) (168) (70)

NC5NC6 NC4 NC3 NC2 NC1(268) (20) (31) (23) (44) (19)

N CA

B

C

D

P1 P2 P3 P4 P5 P6

Figure 12

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from

and John W. WeiselJeanne C. Myers, Deqin Li, Peter S. Amenta, Charles C. Clark, Chandrasekaran Nagaswami

globular, disulfide-linked and heparin-binding amino terminisharp kinks delineating collagenous subdomains and by intermolecular aggregates via Type XIX collagen purified from human umbilical cord is characterized by multiple

published online June 3, 2003J. Biol. Chem. 

  10.1074/jbc.M304629200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

by guest on February 11, 2018http://w

ww

.jbc.org/D

ownloaded from