Biotechnology of Extracellular Matrix

9 Vuorela, A., Myllyhaju, J., Nissi, R, Pihlajaniemi,T. and

Kivirikko, K I. ( 1997) EMBO J. 16, 6702-67 I 2
10 Lamberg, A., Helaakosk, T., Myllyharju, J., Peltonen, S.,
Notbohm, H., Pihlajaniemi, T. and Kivirikko, K. I. ( I 996)
J. Biol. Chem. 271, I 1988-1 I995
I I Fertala, A., Sieron, A. L, Ganguly, A,, Li, S.-W., Ala-Kokko, L.,
Anumula, K R and Prockop, D. J. ( 1994) Biochem. J. 298,
3 1-37
I 2 Geddis, A. E. and Prockop, D. J. (I993) Matrix 13, 399405
I 3 Fichard, A., Tillet E., Delacoux, F., Garrone, R and Ruggiero,
F. (I 997) J. Biol. Chem. 272, 30083-30087
I 4 Imamura, Y., Steiglitz, B. M. and Greenspan, D. S. ( I 998)
J. Biol. Chem. 273,275 I 1-275 I 7
15 Fertala, A., Holmes, D. F., Kadler, K. E., Sieron, A. L. and
Prockop, D. J. ( 1996) J. Biol. Chem. 27 I, 14864- I4869

16 Colman, A. ( I 998) in Biochemical Society Symposium, vol.

63 (Rudland, P. S., Femig, D. G., Leinster, S. and Lunt, G. G.,
eds), pp. I 4 I - 147, Portland Press, London
17 Carver, A. S., Dalrymple, M. A., Wright, G., Cottom, D. S.,
Reeves, D. B., Gibson, Y. H., Keenan, J. L., Banass, J. D., Scott,
A. R, Colman, A. and Gamer, I. ( 1993) Biotechnology I I,
I 263- I270
18 Prunkard, D., Cottingham, I.,Gamer, I., Bruce. S., Dalrymple,
M., Lasser, G., Bishop, P. and Foster, D. ( I 996) Nat.
Biotechnol. 14, 867-87 I
19 John, D. C. A., Watson, R, Kind, A. J., Scott, A. R, Kadler,
K. E. and Bulleid, N.J. ( I 999) Nat. Biotechnol. 17, 385-389

Received 28 February 2000

Expression of recombinant human type 1-111 collagens in the yeast Pichia pastoris
J. Myllyharju', M. Nokelainen, A. Vuorela and K. 1. Kivirikko
Collagen Research Unit, Biocenter and Department of Medical Biochemistry, University of Oulu, P. 0. Box 5000,
FIN-900 I 4 Oulu, Finland

Abstract

Introduction

An efficient expression system for recombinant

human collagens will have numerous scientific
and medical applications. However, most recombinant systems are unsuitable for this purpose, as
they do not have sufficient prolyl 4-hydroxylase
activity. We have developed methods for producing the three major fibril-forming human
collagens, types I, I1 and 111, in the methylotrophic yeast Pichia pastoris. These methods are
based on co-expression of procollagen polypeptide
chains with the a- and P-subunits of prolyl 4hydroxylase. T h e triple-helical type-I, -11 and
-111 procollagens were found to accumulate predominantly within the endoplasmic reticulum of
the yeast cells and could be purified from the cell
lysates by a procedure that included a pepsin treatment to convert the procollagens into collagens
and to digest most of the non-collagenous proteins.
All the purified recombinant collagens were identical in 4-hydroxyproline content with the corresponding non-recombinant human proteins, and
all the recombinant collagens formed native-type
fibrils. T h e expression levels using single-copy
integrants and a 2 litre bioreactor ranged from 0.2
to 0.6 g/l depending on the collagen type.

T h e collagen family consists of about 20 proteins

formally defined as collagens and more than 10
additional proteins with collagen-like domains
[l-31. Type-I collagen is now used as a biomaterial
in numerous medical applications and as a delivery
system for various drugs [4-61. In addition, all
gelatins are made from collagens. T h e collagens
used in all these applications have been isolated
from animal tissues and are liable to cause allergic
reactions in some subjects and carry a risk of
disease-causing contaminants. T h e various collagen types have different properties, and therefore
some of the other collagens might be more suitable
for certain applications than type I. However, it
has been difficult or impossible to isolate sufficient
quantities of the other collagens from animal
tissues. It is obvious, therefore, that an efficient
large-scale recombinant system for the production
of human collagens would have numerous applications in medicine.
Most recombinant systems now available for
large-scale production of proteins cannot be used
as such for the production of recombinant collagens, as bacteria and yeast have no prolyl 4hydroxylase activity, and insect cells [7] and the
mammary gland [8] have insufficient levels of this
enzyme activity. Prolyl 4-hydroxylase, an aJI2
tetramer in vertebrates, plays a central role in the
synthesis of all collagens, as the 4-hydroxyproline
residues formed are essential for the folding of the
newly synthesized collagen polypeptide chains
into triple-helical collagen molecules [2,9,10].

Key words: methylotrophic yeast, procollagen, prolyl 4-hydroxylase.

Abbreviations used: aMF, a matingfactor; proa I (I), proa I (11) and
proa I (Ill) chains, proa I chains of type-I, -11 and -111 procollagen,
respectively; proa2(1) chain, proa2 chain of type-I procollagen.
'To whom correspondence should be addressed (e-mail
johanna.myllyharju@oulu.fi).

353

0 2000 Biochemical Society

Biochemical Society Transactions (2000) Volume 28, part 4

culture medium. Even in this P. pastoris strain,

however, the vast majority of the a- andp-subunits
were found in unassembled forms.
T o study the expression of recombinant
human collagens in P. pastoris, cDNAs for the
proal chains of type-I, -11 and -111 procollagens
[proal(I), proal(I1) and proal(III)] were cloned
separately into the expression vector pPICZB and
transformed into a recombinant P. pastoris strain
expressing human prolyl 4-hydroxylase subunits
in which the 8-subunit had the S. cerevisiae a M F
pre-pro sequence ([13,14] and M. Nokelainen, A.
Vuorela, K. I. Kivirikko and J. Myllyharju, unpublished work). A highly unexpected finding was
that co-expression of the prolyl 4-hydroxylase
subunits with any of these procollagen polypeptide chains led to an up-to-about- 10-fold increase
in the amount of the enzyme tetramer with no
increase in the total amounts of its subunits
([13,14] and M. Nokelainen, A. Vuorela, K. I.
Kivirikko and J. Myllyharju, unpublished work).
Pulse-chase experiments indicated that the halflives of the recombinant enzyme tetramers expressed in P. pastoris without co-expression with
collagen polypeptide chains were only about
50 min [14], while co-expression with the proal(II1) chains increased this half-life to 12.5 h and
co-expression with the proal( I) chains gave a halflife of 6.5 h, i.e. 8 times that of the strain expressing
the enzyme alone but 50% of that of the strain
co-expressing prolyl 4-hydroxylase with the
proal (I1I) chains [141. The difference in half-life
between the strains co-expressing the proal (I)
and proal(II1) chains is likely to be related to the
level of procollagen expression, that of type-I
procollagen homotrimers being 35-70 yo of that of
type-I11 procollagen. The data thus indicate that
collagen synthesis in P . pastoris, and probably also
in other cell types, involves a highly unusual
control mechanism, in that the production of a
stable prolyl 4-hydroxylase tetramer requires the
expression of collagen polypeptide chains, whereas
the production of collagen molecules with stable
triple helices requires the expression of active
prolyl 4-hydroxylase [13,141.

Therefore, the recombinant collagen polypeptide

chains expressed in most systems will remain as a
non-triple-helical, non-functional protein or, if
the cells are grown at low temperatures, the chains
may form molecules with unstable triple helices.
We have demonstrated that co-expression of
polypeptide chains of various types of human
collagen with the two types of subunit of human
prolyl 4-hydroxylase can be used for efficient
recombinant expression of human collagens in
insect cells [7,11,12]. The recombinant type 1-111
collagens produced have been very similar if not
identical with the corresponding non-recombinant
proteins in their 4-hydroxyproline contents and
various other properties, and the highest expression levels obtained in suspension cultures
have ranged up to about 50mg/l [7,11,12]. In
addition, it has been demonstrated that this same
principle can be used for the high-level production
of an engineered form of human type-I collagen in
mouse milk [8]. We have recently applied the
principle to the high-level production of recombinant human type-1-111 collagens in the methylotrophic yeast Pichia pastoris.

Expression of an active recombinant

human prolyl4-hydroxylase tetramer
and the effect of co-expression with
collagen polypeptide chains
In order to study whether subunits of human
prolyl 4-hydroxylase are able to form an active
enzyme tetramer in yeast cells, cDNAs for the
human a- and 8-subunits were cloned into the
Pichia expression vectors pARG815 (complementing for arg4 in the host) and PA0815 (complementing for his4 in the host), respectively, and
co-transformed into the GS200 (his4, arg4) P .
pastoris host train [13]. Initial attempts to express
an active human prolyl4-hydroxylase tetramer in
P. pastoris were only partially successful, as only a
minor fraction of the recombinant polypeptides
expressed were found in the form of the tetramer,
whereas the vast majority were present in unassembled forms [13]. A much higher tetramer
assembly level was obtained [13] when the signal
peptide of the /?-subunit was replaced by the
Saccharomyces cerevisiae a mating factor (aMF)
pre-pro sequence by cloning the P-subunit cDNA
into the expression vector pPIC9 (generating
vector pPIC9p). This signal sequence gave the
highest amount of tetramer among the various
constructs studied, even though it also markedly
increased the secretion of the P-subunit into the

0 2000 Biochemical Society

Expression of human type-I, -11 and -111

collagens in shaker flasks
The strains described above were used to study
the expression of recombinant type-I, -11 and -111
procollagen homotrimers in P . pastoris. In order
to express type-I procollagen heterotrimers, a
cDNA for the proa2 chain of human type-I

354

Biotechnology of Extracellular Matrix

15 mg/l, whereas the levels obtained for type-I

and -11 collagens were approx. 35-70 yo of that of
type-I11 collagen ([13,14] and M. Nokelainen,
A. Vuorela, K. I. Kivirikko and J. Myllyharju,
unpublished work).
T h e triple-helical type-I, -11 and -111 procollagen molecules produced in P. pastoris were
found to accumulate predominantly inside the
yeast cell, only about 10% being found in the
culture medium. This is surprising, as triplehelical procollagen molecules are rapidly secreted
into the extracellular space from various animal
cells. Replacement of the signal sequence of the
human proal(II1) chain with the S. cerevisiae
a M F pre-pro sequence led to only a slight improvement in secretion, and the total expression
level of type-I11 procollagen with the a M F
pre-pro sequence was lower than that with the
authentic signal peptide [16]. Immunoelectron
microscopy indicated that the recombinant procollagen molecules accumulated within the endoplasmic reticulum and did not proceed any further
in the secretory pathway [16]. T h e lackof secretion
may have been related to the large size of the
procollagen molecule.

procollagen [proa2(1)] was cloned into the

pBLADE I X vector (complementing for adel in
the host; M . Nokelainen, H. T u , A. Vuorela,
H. Notbohm, K. I. Kivirikko and J . Myllyharju,
unpublished work). A strain expressing prolyl 4hydroxylase was first generated to a yJC300 (his4,
arg4, adel) P. pastoris host strain by cloning a
cDNA for the a-subunit into the pBLARG I X
vector (complementing for arg4 in the host), and
this construct was co-transformed with the
pPIC9/3 expression vector (see above) into the
yJC300. This was followed by subsequent transformations of the pPICZB vector containing the
proal(1) cDNA and pBLADE I X vector containing the proa2( I) cDNA.
All the P. pastoris strains expressing procollagen were found to produce full-length proa
chains ([13,14] and M. Nokelainen, A. Vuorela,
K. I. Kivirikko and J. Myllyharju, unpublished
work). T h e p r o d (I) chains, when expressed
alone, and the proal(I1) and proal(II1) chains,
each formed triple-helical molecules with collagen
domains that were resistant to pepsin digestion,
whereas no pepsin-resistant chains were obtained
when the proa2(1) chains were expressed alone.
Co-expression of the proal(1) and proa2(I) chains
led to the formation of heterotrimeric type-I
procollagen molecules with the correct 2 : 1 chain
ratio (M. Nokelainen, H. T u , A. Vuorela, H.
Notbohm, K. I. Kivirikko and J . Myllyharju,
unpublished work). Studies by SDS/PAGE
under reducing and non-reducing conditions
indicated that all the proa chains and also the
a1(111) chains produced by pepsin digestion
from the corresponding procollagen molecules
formed disulphide-bonded trimers ([13] and M .
Nokelainen, A. Vuorela, K. I. Kivirikko and J.
Myllyharju, unpublished work).
T h e thermal stability of the pepsin-resistant
recombinant collagens was studied using digestion
with a mixture of trypsin and chymotrypsin after
heating to various temperatures [15]. T h e T,
values of the recombinant type 1-111 collagens
were approx. 38 "C, which is 2-3 "C lower than
that found in vivo ([13] and J. Myllyharju, M.
Nokelainen, A. Vuorela and K. I. Kivirikko, unpublished work). Amino acid analysis of the
recombinant type-I I I collagen purified from
shaker-flask cultures showed that the degree of 4hydroxylation of the proline residues was 44.2 yo,
whereas the corresponding value for non-recombinant human type-I I I collagen was 5 1.6 % [ 131.
T h e best level of type-I11 collagen expression
obtained in shaker-flask cultures was approx.

Expression of human type-I, -11 and -111

collagens in a bioreactor
Many previous studies have shown that shakerflask conditions are not optimal for protein production in P. pastoris, due to the lack of sufficient
0,, and marked increases in expression levels are
usually obtained in bioreactors [17,181. As the K ,
of 0, in the prolyl4-hydroxylase reaction is about
40 p M [19], the 0, concentration within the lumen
of the endoplasmic reticulum is also likely to be
rate-limiting for hydroxylation in shaker-flask
cultures. Thus it could be expected that the
differences in 4-hydroxyproline content between
the recombinant and non-recombinant collagens
may disappear when the recombinant collagens
are produced in a bioreactor. T h e type-I procollagen homotrimers and heterotrimers and the
type-I1 and -111 procollagens were therefore
expressed in a 2 litre B. Braun Biostat C bioreactor
equipped with an 0, supply system, whereupon
their expression levels were indeed markedly
higher than in the shaker-flask cultures, ranging
from about 0.2 to 0.6 g/1 (M. Nokelainen, H. T u ,
A. Vuorela, H. Notbohm, K. I. Kivirikko and J .
Myllyharju, unpublished work). It should be
noted that all the experiments reported in this
paper were carried out with single-copy integrants. It has been demonstrated previously that

355

0 2000 Biochemical Society

Biochemical Society Transactions (2000) Volume 28, part 4

rikko and J. Myllyharju, unpublished work). Nterminal sequencing of the polypeptide chains of
the recombinant collagens showed that in most
cases pepsin digestion had removed several residues from the N-terminus of the telopeptide
domain, but in the case of the a2(I) chain only one
residue had been removed and occasionally, if
pepsin digestion was incomplete, the a2( I) chains
had two additional N-terminal amino acids (i.e.
the cleavage had left the last two amino acids of the
N-propeptide on the N-terminus of the chain;
Figure 2). All the recombinant collagens produced
in P. pastoris were found to form native-type
fibrils (M. Nokelainen, H. T u , A. Vuorela, H.
Notbohm, K. I. Kivirikko and J. Myllyharju,
unpublished work), which indicates that the differences at the N-terminus do not influence the
fibrillar properties. This conclusion is supported
by a recent study on pepsin and pronase treatment
of rat non-recombinant type-I collagen molecules,
indicating that chains with shortened N-termini
form fibrils that are identical with those formed
from full-length chains [22]. It thus seems likely
that the recombinant procollagens produced in P.
pastoris can be used for numerous applications
that currently require collagens purified from
animal tissues.

the levels of expression of various proteins in P.

pastoris increase markedly with the number of
DNA copies, at least up to 30-50 copies [17,20,21].
The present system can thus be optimized for the
production of very large amounts of various
collagens.
The recombinant collagens produced in the
bioreactor were purified by pepsin digestion and
selective salt precipitation followed by Sephacryl
S-500HR gel filtration in the AKTA explorer
system (Amersham Pharmacia Biotech). All the
recombinant collagens produced were found to be
essentially pure when analysed by SDS/PAGE
followed by Coomassie Brilliant Blue staining
(Figure 1). Amino acid analyses showed that the 4hydroxyproline contents of all the purified recombinant collagens were identical with those reported
for the corresponding non-recombinant human
proteins (M. Nokelainen, A. Vuorela, K. I. KiviFigure I

SDSlPAGE analysis of purified recombinant human

collagens expressed in P. pastoris
The long arrow indicates the a1 chains of the type-I collagen
homotrimer (lane I) and heterotrimer(lane 2), and type-ll (lane 3)
and type-Ill collagens (lane 4). The short a m w indicates the a2
chain of the type-I collagen heterotrimer (lane 2).

We thank Dr.JamesCregg, Keck GraduateInstitute of Applied Life

Sciences, for the gift of the P. pastoris host strains and the
pBLARG IX and pBLADE IX vectors, and Raija Juntunen,Anne
Kokko, Eeva Lehtimaki,Minna Siunra and Tanja VaisLnen for their
expert technical assistance. This work was supported by grants
from the Health Sciences Council ofthe Academy of Finland,from
the European Commission(B104-Cr96-0537),from the National
Institutes of Health (ROI AR45879) and from FibroGen (South
San Francisco, CA, U.S.A.).

References
1

Kielty, C. M., Hopkinson, I. and Grant., M. E. ( I 993) in

Connective Tissue and its Heritable Disorders. Molecular,
Genetic and Medical Aspects (Royce, P. M. and Steinmann,
B., eds), pp. 103- 147,Wiley-Liss, New York
Prockop, D. J. and Kivirikko, K I. ( I 995) Annu. Rev. Biochem.
64,403-434
Kadler, K E., Holmes, D. F., Trotter, J. A. and Chapman, J, A.
(I 996) Biochem.J. 3 16, I- I I
Ramshaw, J.A., Werkmeister, J. A. and Glattauer, V. (I996)
Biotechnol. Genet Eng. Rev. 13, 335-382
Pachence, J. M. (I995) J. Biomed. Mater. Res. 33, 3 5 4
Rao, K P. ( I 995) J. Biomater. Sci. Polym. Ed. 7, 623-645
Lamberg, A., Helaakoski, T., Myllyharju, J., Peltonen, S.,
Notbohm, H., Pihlajaniemi, T. and Kivirikko, K I. (I 996)
J. Biol. Chem. 27 I, I I 988- I I 995
John, D. C. A., Watson, R, Kind, A. J., Scott, A. R, Kadler,
K E. and Bulleid, N.J. ( I 999) Nat. Biotechnol. 17, 385-389
Kivirikko, K I. and Pihlajaniemi, T. ( I 998) Adv. Enzymol.
Related Areas Mol. Biol. 72, 325-398

Figure 2
N-termini of the a-chains of the purified recombinant
human collagens
The N telopeptide sequences are underlined.The amws indicate
the pepsin cleavage sites, while the N-terminalamino acid of the
a-chains of the final recombinant collagens is shown in bold.
J,
a 1(I).. . G N F ~ G G I S V P G P M G P S.. .
J,
WPMGLM
a2(I)...GNFMQYDGKGV

...

hXMQGPMGPM ...

a1@
GNFM-GGAO
I)...

a1@I)...QNYSPQYDSYDVKSGVAVGCL.AGYP...

0 2000 Biochemical Society

356

Biotechnology of Extracellular Matrix

I 7 Romanos, M. A., Scorer, C. A. and Clare, 1. J. (I 992) Yeast 8,

423-488
I 8 Cregg, J. M., Vedvick T. S. and Raschke, W. C. (I 993)
Biotechnology I I , 905-9 I0
I 9 Tuderman, L., Myllyla, R and Kivirikko, K. 1. (I 977) Eur. J.
Biochem. 80, 34 1-348
20 Buckholz, R G. and Gleeson, M. A. (I 99 I) Biotechnology 9,
1067- I072
21 Scorer, C. A., Clare, J.I.,
McCombie, W. R., Romanos, M. A.
and Sreekrishna, K (I 994) Biotechnology 12, I 8 I- I84
22 Kuznetsova, N. and Leikin, S. (I 999) 1. Biol. Chem. 274,
36083-36088

10 Kivirikko, K I. and Myllyhaju, J. (I 998) Matrix Biol. 16,

357-368
I I Myllyhaju, I., Lamberg, A., Notbohm, H., Fietzek P. P.,
Pihlajaniemi, T. and Kivirikko, K I. ( I 997) J. Biol. Chem. 272,
2 I 8 2 4 2 I830
12 Nokelainen, M., Helaakoski,T., Myllyhaju, J., Notbohm, H.,
Pihlajaniemi, T., Fietzek P. P. and Kivirikko, K. I. (I 998)
Matrix Biol. 16, 329-338
13 Vuorela, A., Myllyhaju, I., Nissi, R, Pihlajaniemi, T. and
Kivirikko, K. I. ( 1997) EM60 J. 16, 6702-67 I 2
14 Vuorela, A., Myllyhaju, J., Pihlajaniemi. T. and Kivirikko. K. I.
(I999) Matrix Biol. 18, 5 19-522
15 Bnrckner, P. and Prockop, D. J. ( I98 I) Anal. Biochem. I 10,
360-368
16 Keizer-Gunnink I,, Vuorela, A,, Myllyhaju, I., Pihlajaniemi, T.,
Kivirikko, K I. and Veenhuis, M. (2000) Matrix Biol. 19,
29-36

Received I March 2000

Towards a fibrous composite with dynamically controlled stiffness :lessons

from echinoderms
J. A. Trotter*', J. Tipper*, G. Lyons-Levy*, K. Chino*, A. H. Heuert. Z. Liut, M. Mrksichf, C. Hodnelandl,
W. S. Dillrnoref. T. J. Koobtj, M. M. Koob-Ernundstj, K. Kadlery and D. Holrnesy
*Department of Cell Biology and Physiology, University of New Mexico School of Medicine, Albuquerque,
N M 87 I 3 I, U.S.A., +Department of Materials Science and Engineering, Case Western Reserve University,
I0900 Euclid Ave., Cleveland, OH 44 106, U.S.A., f Department of Chemistry, Universrty of Chicago,
5735 S. Ellis Ave., Chicago, IL 60637, U.S.A., SShrinen Hospital for Children, I2502 N. Pine Drive, Tampa,
FL 336 12, U.S.A., and flWellcome Trust Centre for Cell-Matrix Research, School of Biological Sciences,
university of Manchester, Stopford Building, Oxford Road, Manchester M I 3 9PT, U.K.

elastomeric matrix that contains photo- or electrosensitive reagents that reversibly form interfibrillar cross-links.

Abstract
Sea urchins and sea cucumbers, like other echinoderms, control the tensile properties of their
connective tissues by regulating stress transfer
between collagen fibrils. The collagen fibrils are
spindle-shaped and up to 1 mm long with a
constant aspect ratio of approx. 2000. They are
organized into a tissue by an elastomeric network
of fibrillin microfibrils. Interactions between the
fibrils are regulated by soluble macromolecules
that are secreted by local, neurally controlled,
effector cells. We are characterizing the non-linear
viscoelastic properties of sea cucumber dermis
under different conditions, as well as the structures, molecules and molecular interactions that
determine its properties. In addition, we are
developing reagents that will bind covalently to
fibril surfaces and reversibly form cross-links with
other reagents, resulting in a chemically controlled
stress-transfer capacity. The information being
developed will lead to the design and construction
of a synthetic analogue composed of fibres in an

Collagenous tissues
The structural materials of animals are, for the
most part, composites containing insoluble fibres
in a non-fibrous matrix. Familiar examples of such
materials include the tendons, ligaments and
dermis of mammals. The mechanical properties of
these fibrous composites are due largely to the
contributions of the protein collagen, which selfassembles into long, thin fibrils that may be
millimetres in length and nanometres in diameter
[l]. Collagen molecules (approx. 300 nm long x
1.5 nm in diameter) within the same fibril become
covalently cross-linked through enzymic action.
As a result of cross-linking, the fibrils possess
high tensile stiffness and strength (on the order
of GPa). In most cases we do not know how long
the individual collagen fibrils are; nor do we
know how stress is transferred between them. We
do know, however, that the composition and
organization of the tissues is such as to make
effective use of the tensile properties of the fibrils.
In addition to collagen fibrils, connective tissues

Key words: collagen, interfibrillar cross-links, fibrils.

'To whom correspondence should be addressed (e-mail:
jtrotter@salud.unm.edu).

357

0 2000 Biochemical Society