Oxygen tension regulates the expression of a group of procollagen
hydroxylases
Karl-Heinz Hofbauer
1
, Bernhard Gess
1
, Christiane Lohaus
2
, Helmut E. Meyer
2
,Do¨ rte Katschinski
3
and Armin Kurtz
1
1
Institut fu
¨
r Physiologie der Universita
¨
t Regensburg, Germany;
2
Medizinisches Proteom, Center der Ruhr, Universita
¨
t Bochum,
Germany;
3
Abteilung Zellphysiologie der Martin-Luther Universita
¨
t Halle, Germany
In this study, we have characterized the influence of hypoxia
on the expression of hydroxylases crucially involved in col-

for wound healing in the skin, for the remodeling of small
muscular pulmonary arteries in hypoxia-induced pulmonary
hypertension and possibly also for cardiac hypoxia. The
formation of collagen fibers and deposits is a multi-step
event that includes procollagen protein synthesis, prolyl
hydroxylation as requirement for triple helix formation, lysyl
hydroxylation, protein folding, maturation and secretion,
and finally covalent cross-bridging between collagen fibers
through the activity of the lysyloxidase. Which of these steps
are directly triggered by hypoxia and how this is accom-
plished is not well understood. It has been reported that
hypoxia increases mRNA levels for different procollagens in
the lung [1,2] and heart in vivo [3]. In vitro studies suggest that
this effect of hypoxia on procollagen gene expression might
be isoform and cell-type specific. Thus, hypoxia stimulates
procollagen I formation in renal [4], dermal [5], and cardiac
fibroblasts [6], but neither in fetal lung fibroblasts [7] nor in
3T3 fibroblasts [8]. 3T3 fibroblasts [8], like renal mesangial
cells [9], however, increase the gene expression of procol-
lagen IV in response to hypoxia. The effect of hypoxia on the
activity of the prolyl-4-hydroxylase (PHD-4 or P4h) is
clearer; it is crucially required to enable triple helix formation
and has been found to be increased in its activity in response
to hypoxia [7,10–12]. For the PHD-4 (P4h) heterotetramer
enzyme (a
2
b
2
) there exist two isoforms with a variable
a-subunit (aIoraII) and a constant b-subunit, which is

E-mail:
Abbreviations: HIF-1, hypoxia inducible transcription factor 1; PDI,
protein disulfide isomerase; Ph4, prolyl-4-hydroxylases; PLOD, pro-
collagen lysyl-hydroxylases; SDS/PAGE, sodium dodecyl sulfate/
polyacrylamide gel electrophoresis; UPR, unfolded protein response.
(Received 31 July 2003, revised 8 September 2003,
accepted 19 September 2003)
Eur. J. Biochem. 270, 4515–4522 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03846.x
general HIF-1-related effect of hypoxia on the expression of
the critical hydroxylases for collagen fiber formation, this
study aimed to characterize the influence of hypoxia on the
gene expression of these hydroxylases in more detail.
Materials and methods
Cell cultures
Rat aortic vascular smooth muscle cells (A7r5) from BDXI
rats (ATCC CRL 1444) were cultured in 75 cm
2
flasks
(Sarstedt) with 15 mL Dulbecco’s modified Eagle’s medium
containing 10% fetal bovine serum and penicillin/strepto-
mycin (P/S; 10 U/10 lgÆmL
)1
)(Biochrom), kept in an
atmosphere of 10% CO
2
(v/v), 21% O
2
and 69% N
2
at

,10%CO
2
balance N
2
(i.e.
hypoxia) or at 21% O
2
, 10% CO
2
balance N
2
with
deferoxamine (200 lmolÆL
)1
) for 24 h.
Mouse embryonic fibroblasts with normal (+/+) and
with a disrupted (–/–) gene for HIF-1a [22] were grown
under the above-mentioned conditions. The cells were
incubated either at 0.5% O
2
(i.e. hypoxia) or at 21% O
2
with
deferoxamine (100 lmolÆL
)1
) for 24 h.
Preparation of protein samples
After removal of cell culture medium, cell were washed three
times with ice-cold NaCl/P
i

was performed at 50 V for 15 h. For protein separation, a
step-voltage protocol was applied (1 h 150 V, 3 h 500 V,
1 h 1000 V, gradient to 8000 V within 0.5 h). A total volt–
hour product of 60 kVh was used for 150 lgprotein
and 110 kVh for 600 lg protein. Afterwards, the strips
were incubated in 50 mmolÆL
)1
Tris/HCl (pH 6.8), urea
6molÆL
)1
, glycerol 30%, dithiothreitol 65 mmolÆL
)1
,2%
sodium dodecyl sulfate (SDS) for 20 min at room tempera-
ture followed by incubation in 50 mmolÆL
)1
Tris/HCl
(pH 8.8), urea 6 molÆL
)1
, glycerol 30%, iodoacteamide
140 mmolÆL
)1
, and 2% SDS for another 20 min. For the
second dimension, a vertical gradient slab gel of 8–18%
acrylamide was used and SDS/PAGE was performed at
8mApergelat13°C for 4 h followed by 30 mA for 12 h.
At the end of the second dimension, the gels were removed
from the glass plates.
Staining of two-dimensional PAGE
The gels were fixed and stained with silver according to

Mix (containing buffer, dNTPs, SYBR Green and hotstart
Taq polymerase). The primers used are summarized in
Table 1.
The amplification program consisted of one cycle at
95 °C for 10 min, followed by 40 cycles with a denaturing
phase at 95 °C for 15 s, an annealing phase of 5 s at 60 °C
and an elongation phase at 72 °C for 15 s. A melting curve
analysis was carried out after amplification to verify the
4516 K H. Hofbauer et al. (Eur. J. Biochem. 270) Ó FEBS 2003
accuracy of the amplicon. For verification of the correct
amplification, PCR products were analyzed on an ethidium
bromide stained 2% agarose gel.
In each real-time PCR run for each gene product under
investigation and for b-actin a calibration curve was
included, that was generated from serial dilutions (1 : 1,
1 : 10, 1 : 100, 1 : 1000) of a cDNA generated from the
pooled RNA of the normoxic (control) cultures (at the
different time points) of the respective experimental series
(standard cDNA). Analysis of the individual unknowns
therefore yielded values relative to this pool. Data are
presented as the relative mRNA/b-actin mRNA ratio. The
mRNA/b-actin mRNA ratio of the time standards (pools)
cDNA was set to 1.0 (i.e. normoxia, 21% oxygen). Data
are therefore expressed as relative values related to
normoxia.
Statistics
Levels of significance between groups were calculated using
ANOVA
test followed by Bonferoni’s reduction for multiple
comparisons. P < 0.05 was considered significant.

Antisense:
5¢-GGAATTCCAAGCAGTCCTCAGCTGT-3¢
Mouse (gi:6754969) prolylhydroxylase alpha II
Sense:
5¢-CGGGATCCTGCAGGCAGAATTCTTCA-3¢
Antisense:
5¢-GGAATTCCCAGTCTGTGTTCAACCG-3¢
Rat (gi: 6754969) prolylhydroxylase alpha II
Sense:
5¢-CGGGATCCTGCAGGCAGAATTCTTCA-3¢
Antisense:
5¢-GGAATTCGCTGAACTGAGAGGTTAG-3¢
Mouse (gi: 20913928) and rat (gi: 6981323) protein disulfide isomerase
Sense:
5¢-CGGGATCCAGCAGTATGGTGTCCGTG-3¢
Antisense:
5¢-GGAATTCACCGTCACTTCGCTTGAG-3¢
Mouse (gi: 6755105) lysylhydroxylase I
Sense:
5¢-AACTGGTGGCCGAGTGGG-3¢
Antisense:
5¢-GCAGGGTGTCATAGGCCA-3¢
Rat (gi: 409058) lysylhydroxylase I
Sense:
5¢-AACTGGTGGCCGAGTGGG-3¢
Antisense:
5¢-GCCCATTTCAAACTTGAG-3¢
Mouse (gi: 6755107) lysylhydroxylase II
Sense:
5¢-GCACATTGGGAAACGCTA-3¢

P4ha2 mRNA being stronger than that of P4ha1 (Fig. 2A).
In view of this concordant regulation, we further considered
the possibility that also the expression other hydroxylases
involved in collagen fiber formation might be regulated by
the cellular oxygen tension. In fact, it turned out that also
the mRNAs for lysyl hydroxylases I and II (PLOD1 and -2)
increased clearly during hypoxia, in a very similar fashion to
the mRNAs for the prolyl hydroxylases. In addition, also
the mRNA for protein disulfide isomerase (PDI), as the
b-subunit of prolyl hydroxylases, increased in A7r5 cells,
although significantly delayed and to a lesser extent. After
24 h of hypoxia the mRNA abundance was five-, 12-, six,
seven- and fivefold increased for P4ha1, P4ha2, PDI,
PLOD1 and PLOD2, respectively. Notably the abundance
of procollagen Ia was not changed by hypoxia (Fig. 2B).
Very similar results to those with hypoxia were obtained,
when A7r5 cells were incubated with the iron-chelator
deferoxamine (100 lmolÆL
)1
) at ambient oxygen tension
(21% O
2
). After 24 h mRNA abundance was increased
five-, 16-, two-, five- and 10-fold, for P4ha1, P4ha2, PDI,
PLOD1 and PLOD2, respectively (Fig. 3A), whilst the
mRNA abundance for procollagen Ia was unchanged.
Also, cobalt(II) chloride (100 lmolÆL
)1
) moderately
increased the mRNAs of the hydroxylases, but not of

Hepa1C4 cells (Fig. 5B), supporting the assumption that
the expression of these genes was driven by HIF. Unfor-
tunately, mRNA levels for the PLOD mRNAs were too low
to allow reasonable semiquantification in both Hepa1 and
Hepa1C4 cells.
This first indication about an essential role of HIF in the
triggering of prolylhydroxylase gene expression was further
Fig. 1. Two-dimensional electrophoresis of proteins isolated from the rat vascular smooth muscle cell line A7r5. The indicated protein spots were
up-regulated by exposure of the cells to hypoxia (1% O
2
) for 12 h.
4518 K H. Hofbauer et al. (Eur. J. Biochem. 270) Ó FEBS 2003
corroborated in mouse embryonic fibroblasts lacking the
HIF-1a subunit. As shown in Fig. 6 hypoxia, deforoxamine
and cobalt stimulated the expression of P4ha1, P4ha2, PDI,
PLOD1 and PLOD2 mRNAs in wild-type embryonic
fibroblasts, but not in embryonic fibroblasts lacking the
HIF-1a subunit.
As the enzymes involved in collagen formation are
important for the correct folding of the protein, it is in
principle conceivable, that their expression is also triggered
Fig. 3. P4ha1, P4ha2, PDI and PLOD1, PLOD2, and procollagen
Ia mRNA in A7r5 cells after exposure to cobalt(II) chloride
(100 lmolÆL
-1
) (A) or to deferoxamine (100 lmolÆL
-1
) (B) for 12 h at
21% O
2

. Data are means ± SEM of
five experiments each. *P <0.05 vs. control (21% O
2
). Controls are
the means of five experiments and the average mRNA/b-actin mRNA
ratio is set to 1 (dotted line).
Fig. 2. Time course of P4ha1, P4ha2 mRNA (A) and PLOD1, PLOD2
mRNA (B) and PDI, procollagen Ia mRNA (C) in A7r5 cells after
exposure of the cells to 1% O
2
. Data are means ± SEM of five
experiments. *P < 0.05 hypoxia (1% O
2
) vs. normoxia (21% O
2
).
Controls are the means of five experiments and the average mRNA/b-
actin mRNA ratio is set to 1 (dotted line).
Ó FEBS 2003 Hypoxia and procollagen hydroxylases (Eur. J. Biochem. 270) 4519
by a disturbance of protein folding due to energy depletion
in the course of cellular hypoxia. This so-called unfolded
protein response (UPR) can also be elicited by tunicamycin
at normal oxygen tensions [28], as shown in Fig. 7.
Tunicamycin gave a clear twofold increase of PDI mRNA,
a more moderate increase of the mRNAs for prolyl
hydroxylases, and did not increase the mRNA abundance
of the other enzymes.
Discussion
Our data show that the expression of a functional cluster of
hydroxylases enzymes crucially required for procollagen

common regulation of a functional cluster of genes by the
oxygen tension has already been found for the enzymes
involved in the glycolytic cascade [29] and also for the key
players of angiogenesis [30].
Considering the parallel up-regulation of enzymes
involved in collagen fiber formation raises the question of
whether this up-regulation is physiologically primarily
meant to increase collagen formation in hypoxic tissues,
or if it reflects more a compensatory change of the
concentration of enzyme molecules to maintain a given
normal hydroxylation rate at altered substrate (oxygen)
concentrations (Fig. 8). The explanation as a compensatory
increase of gene expression was previously also presented
for the increased expression of the endoplasmic oxido-
reductase Ero1-L during hypoxia which transfers oxidizing
equivalents onto PDI [27]. Such a view would be supported
by the observation that the expression of the procollagen
(Ia) gene itself was not regulated by the oxygen tension,
which is also in accordance with data obtained by others [7].
As the genes for P4ha1, P4ha2, PDI, and PLOD 1,2 are
localized on different chromosomes the question arises
concerning the mechanisms underlying the orchestrated
expression of these enzymes by the oxygen tension.
The findings that the effect of hypoxia on the gene
expression of the hydroxylases was mimicked by the iron
Fig. 6. P4ha1, P4ha2, PDI, PLOD1 and PLOD2 mRNA in mouse
embryonic fibroblasts with intact and with disrupted HIF-1a gene after
exposure to hypoxia (0.5% O
2
) or to deferoxamine (100 lmolÆL

tension, in the way that the a-subunit is more stable at low
oxygen tensions. The reason for this behavior is an oxygen
dependent prolyl-hydroxylation of the a-subunit, which
finally directs the protein to proteasomal degradation [32].
The assumption that HIF could in fact be a main trigger of
the procollagen hydroxylases is further corroborated by the
findings that the stimulatory effect of hypoxia on gene
expression was absent in cells with a functional inactive HIF
or lacking the HIF-1a protein in general. In fact, for the
P4ha1I the involvement of HIF in the activation of gene
expression during hypoxia has recently been directly dem-
onstrated [14]. A search for the HIF-binding consensus
sequence CGTG revealed six, six, two, nine and 10
theoretical HIF binding sites within the first 1 kb of the
5¢-promoter region of mouse P4ha1, P4ha2, PLOD1,
PLOD, and PDI, respectively.
Although prolyl hydroxylation is a critical event for both
procollagen triple helix stabilization on the one hand and
for the stability of HIF-1a protein, different prolyl
hydroxylases appear to be required for these processes.
HIF-a prolyl hydroxylation is managed by PHD-1, -2 and
-3 [33,34], whilst procollagen prolylhydroxylation is
performed by P4h [13], which does not accept HIF-a as
a substrate [31]. Interestingly, the expressions of PHD-3
[35,36] and eventually of PHD-2 itself are also oxygen
sensitive [35,36]. They are up-regulated by hypoxia by a
process critically involving HIF, which in turn is also the
substrate of PHD-2 and PHD-3. Thus, PHD-2, PHD-3 and
P4h expression appear to be subject to a common control by
oxygen, whilst the expression of PHD-1 is not. The

is known to be induced by the UPR [41], however, none of
the other enzymes was relevantly induced by the UPR,
supporting the assumption that they are more directly
triggered by HIF.
References
1. Berg,J.T.,Breen,E.C.,Fu,Z.,Mathieu-Costello,O.&West,J.B.
(1998) Alveolar hypoxia increases gene expression of extracellular
matrix proteins and platelet-derived growth factor B in lung par-
enchyma. Am. J. Resp. Crit. Care Medical 158, 1920–1928.
2. Durmowicz, A. G., Parks, W.C., Hyde, D.M., Mecham, R.P. &
Stenmark, K.R. (1994) Persistence, re-expression, and induction
of pulmonary arterial fibronectin, tropoelastin, and type I pro-
collagen mRNA expression in neonatal hypoxic pulmonary
hypertension. Am. J. Pathol. 145, 1411–1420.
3. Ostadal, B., Kolar, F., Pelouch V. & Widimsky, J.(1995) Onto-
genetic differences in cardiopulmonary adaptation to chronic
hypoxia. Physiol. Res. 44, 45–51.
4. Norman, J. T., Clark, I.M. & Garcia, P.L. (2000) Hypoxia pro-
motes fibrogenesis in human renal fibroblasts. Kidney Int. 58,
2351–2366.
5. Falanga, V., Martin, T.A., Takagi, H., Kirsner, R.S., Helfman, T.,
Pardes J. & Ochoa, M.S. (1993) Low oxygen tension increases
mRNA levels of alpha 1 (I) procollagen in human dermal fibro-
blasts. J. Cell. Physiol. 157, 408–412.
6.Tamamori,M.,Ito,H.,Hiroe,M.,Marumo,F.&Hata,R.I.
(1997) Stimulation of collagen synthesis in rat cardiac fibroblasts
Fig. 8. Feedback regulation of procollagen-hydroxylase expression by
oxygen to maintain procollagen hydroxylation at variable cellular oxy-
gen tensions.
Ó FEBS 2003 Hypoxia and procollagen hydroxylases (Eur. J. Biochem. 270) 4521

cultured cells. J. Biol. Chem. 275, 14139– 14146.
15. Tanaka,S.,Uehara,T.&Nomura,Y.(2000)Up-regulationof
protein-disulfide isomerase in response to hypoxia/brain ischemia
and its protective effect against apoptotic cell death. J. Biol. Chem.
275, 10388–10388.
16. Myllyharju, J. & Kivirikko, K.I. (2001) Collagens and collagen-
related diseases. Ann. Medical 33, 7–21.
17. Ruotsalinen, H., Sipila
¨
,L.,Kerkela
¨
, E., Pospiech, H. & Myllyla
¨
,
R. (1999) Characterization of cDNAs for mouse lysyl hydroxylase
1,2 and 3, their phylogenetic analysis and tissue-specific expression
in the mouse. Matrix Biol. 18, 325–329.
18. Sigmund, C.D., Okuyama, K., Ingelfinger, J., Jones, C.A.,
Mullins, J.J., Kane, C., Kim, U., Wu, C.Z., Kenny, L., Rustum,
Y., Dzau, V.J. & Gross, K.W. (1990) Isolation and characteriza-
tion of renin expressing cell lines from transgenic mice containing a
renin-promoter viral oncogene fusion construct. J. Biol. Chem.
265, 19916–19922.
19. Cuthill, S. & Poellinger, L. (1988) DNA binding properties of
dioxin receptors in wild-type and mutant mouse hepatoma cells.
Biochemistry 27, 2978–2982.
20. Numayama-Tsuruta, K., Kobayashi, A., Sogawa, K. &
Fujii-Kuriyama, Y. (1997) CBP/p300 functions as a possible
transcriptional coactivator of Ah receptor nuclear translocator
(Arnt). Eur. J. Biochem. 246, 486–495.

Biochem. 270, 2228–2235.
28. Benedetti, C., Fabbri, M., Sitia, R. & Cabibbo, A. (2000) Aspects
of gene regulation during the UPR in human cells. Biochem.
Biophys. Res. Commun 278, 530–536.
29. Semenza, G.L. (1999) Regulation of mammalian O
2
homeostasis
by hypoxia-inducible factor 1. Annu. Rev. Cell Dev. Biol 15,
551–578.
30. Pugh, C.W. & Ratcliffe, P.J. (2003) Regulation of angiogenesis by
hypoxia: role of the HIF system. Nat. Med. 9, 677–684.
31. Wang, G.L. & Semenza, G.L. (1995) Purification and character-
ization of hypoxia-inducible factor 1. J. Biol. Chem. 270, 1230–
1237.
32. Jaakkola, P., Mole, D.R., Tian, Y.M., Wilson, M.I., Gielbert, J.,
Gaskell, S.J., Kriegsheim, A.V., Hebestreit, H.F., Mukherji, M.,
Schofield, C.J., Maxwell, P.H., Pugh, C.W. & Ratcliffe, P.J.
(2001) Targeting of HIF-alpha to the von Hippel–Lindau
ubiquitylation complex by O
2
-regulated prolyl hydroxylation.
Science 292, 468–472.
33. Bruick, R.K. & McKnight, S.L. (2001) A conserved family of
prolyl-4-hydroxylases that modify HIF. Science 294, 1337–1340.
34. Semenza, G.L. (2001) HIF-1, O(2), and the 3, PHDs: how animal
cells signal hypoxia to the nucleus. Cell 107,1–3.
35. Epstein, A.C., Gleadle, J.M., MsNeill, L.A., Hewitson, K.S.,
O’Rourke, J., Mole, D.R., Mukherji, M., Metzen, F., Wilson,
M.I., Dhanda, A., Tian, Y.M., Masson, N., Mailton, D.L.,
Jaakola,P.,Barstead,R.,Hodgkin,J.,Maxwell,P.H.,Pugh,