Purification and characterization of the thyrotropin-releasing
hormone (TRH)-degrading serum enzyme and its identification
as a product of liver origin
Stephanie Schmitmeier
1,
*, Hubert Thole
1,
†, Augustinus Bader
2
and Karl Bauer
1
1
Max-Planck-Institut fu
¨
r experimentelle Endokrinologie, Hannover, Germany;
2
Gesellschaft fu
¨
r Biotechnologische Forschung,
Abt. Organ und Gewebekulturen, Braunschweig, Germany
Previous biochemical studies have indicated that t he mem-
brane-bound thyrotropin-releasing hormo ne (TRH)-degra-
ding enzyme (TRH-DE) from brain and liver and t he serum
TRH-DE are derived from the s ame gene. These studies also
suggested that the serum enzyme is of liver origin. The
present study was undertaken to verify these hypotheses. In
different species, a close relationship between the activities of
the serum enzyme and the particulate liver enzyme was
noticed. The activity of the serum enzyme decreased when
rats were treated with thioacetamide, a known hepatotoxin.
With hepatocytes cultured i n a sandwich configuration,

found in brain and significant activities are also detected in
retina, lung and liver but not in other tissues such as heart,
kidney and muscle [5–7]. Because the membrane-bound
brain T RH-DE a nd the serum TRH-DE exhibit the same
extraordinary high degree of substrate specificity and
identical enzyme-chemical characteristics [8–14] it has b een
suggested t hat both enzymes are derived from the same
gene.
Based o n the observation that the developmental pattern
of the particulate liver TRH-DE and the serum TRH-DE
are almost identical it has been proposed that the serum
TRH-DE, like most serum enzymes and p roteins, might be
of liver origin [9,15]. This i nterpretation w as supported by
the findings t hat the activities of the particulate liver
enzyme, like the serum enzyme [16–18], is a lso regulated
by thyro id h ormones [19]. Moreover, similar enzyme-
chemical properties between the particulate liver enzyme
and the serum enzyme were noticed [9,15].
To verify the hypothesis that the serum TRH-DE is of
liver origin we analyzed the TRH-DE i n serum and
tissue ho mogenates of different species and studied the
effect of thioacetamide, a hepatotoxin, on the expression
of the serum enzyme and t he particulate liver enzyme.
Furthermore, with hepatocytes in culture we analyzed the
release of the TRH-DE into the medium. Finally, we
purified the TRH-DE from pig serum and liver to
electrophoretic homogeneity and studied th e r elationship
between these e nzymes. By s equence analysis we also
verified the hypothesis that the membrane-bound brain
TRH-DE and the serum TRH-DE a re derived from the

Germany). Insulin was from Hoechst AG (Frankfurt,
Germany), prednisone from MSD Sharp & Dohme GmbH
(Haar, Germany), and glucagon from Novo Nordisk
Pharma GmbH (Mainz, Germany). Poly(ethylene glycol)
6000 was obtained from Serva (Heidelberg, Germany).
Digoxigenin-labeled lectins, antidigoxigenin antibodies con-
jugated either to alkaline phosphatase or to horseradish
peroxidase as well as endoglycosidase F/N-glycosidase
F enzyme preparation and endoproteinase Lys-C were
purchased from Roche Diagnostics GmbH (Mannheim,
Germany). Goat anti-(mouse IgG) Ig con jugated to alkaline
phosphatase was obtained from Bio-Rad Laboratories
GmbH (Munich, Germany). The enhanced chemilumines-
cence (ECL)-Western blotting detection kit was from
Amersham Pharmacia Biotech (Freiburg, Germany).
Nitrocellulose BA-S83 was f rom S chleich er & Schuell
(Dassel, Germany). 5-Bromo-4-chloro-indolylphosphate
and Nitro blue tetrazolium were purchased from Biomol
Feinchemikalien GmbH (Hamburg, Ge rmany).
Animals
Cows (Schwarz-bunte Rasse) and pigs (Deutsche Land-
rasse) w ere raised a nd maintained at the Institut fu
¨
r
Tierzucht und-verhalten, Mariensee, Germany. Sprague–
Dawley rats were maintained at our institute according to
the animal welfare committee of the Medizinische Hochsch-
ule Hannover, Germany. All animals had access t o s tandard
chow and water ad libitum .
Preparation of tissue homogenates and serum

Induction of liver cirrhosis in rats
Adult male Sprague–Dawley rats weighing 380–400 g
were used. Over the experimental period 10 rats were
given t ap water containing 0.03% thioacetamide and 10
rats were kept as control. At given time intervals,
approximately 1 mL of blood was collected by retrobul-
bar puncture and after clotting serum was obtained by
centrifugation.
Hepatocyte isolation and culture
Hepatocytes were isolated from young male pigs
(about 7 w eeks old) as described previously [20]. Isolated
hepatocytes were adjusted to a density of 2 · 10
6
viable
cells per mL in Williams’ medium E supplemented with
fetal bovine serum (10%), insulin (0.17 IUÆmL
)1
), predni-
sone (0.85 lgÆmL
)1
), glucagon (0.015 lgÆmL
)1
), penicillin
(100 UÆmL
)1
), streptomycin (100 lgÆmL
)1
) and glutamine
(4.3 m
M

M
2-iodoacet-
amide for p ost proline cleaving enzyme and pyroglutamate
aminopeptidase, respectively). As a measure for the enzy-
matic activity, the initial rate of TRH-degradation was
determined by a four-point kinetic t est.
Purification of the TRH-DE from porcine serum
Porcineserum(1L)wasdilutedwith1LofbufferA.
Under constant stirring, 2 L of a poly(ethylene glycol) 6000
solution (dissolved 50% w/v in buffer A) were added
through a dropping funnel o ver a period of 3 h. After an
additional hour without stirring, the precipitated protein
was pelleted by centrifugation at 17 0 00 g for 3 h. The
supernatant was discarded and the protein pellet was
dissolved by stirring overnight with 3 L of buffer A. The
clear supernatant obtained a fter centrifugation at 17 000 g
for 1 h was subjected to the purification procedure as
described for the trypsin-solubilized membrane-bound
TRH-DE from pig brain [28].
Purification of the membrane-bound TRH-DE
from porcine liver
After homogenization of thoroughly perfused pig livers a nd
washing of the membranes, the membrane-bound TRH-DE
was solubilized by trypsin treatment and purified to
homogeneity by following the protocol described for the
isolation of TRH-DE from pig brain [28].
Ó FEBS 2002 Characterization of the TRH-DE from serum and liver (Eur. J. Biochem. 269) 1279
SDS/PAGE analysis
SDS/PAGE analysis was carried out according to Laemmli
[29]. The proteins were denatured under reducing co nditions

M
NaCl, 10 m
M
MgCl
2
,pH9.5)for
visualization.
Lectin blot analysis
Lectin blot an alysis was performed according to the method
described by Haselbeck et al. [31]. After Western blotting
and blocking, the membrane was cut and individual strips
were incubated for 1 h with d igoxigenin-conjugated lectins
(1 : 1000 in NaCl/Tris containing 1 m
M
MgCl
2
,1m
M
MnCl
2
,1m
M
CaCl
2
and 1 m
M
ZnCl
2
, pH 7 .5). The strips
were then washed with NaCl/Tris/Tween and s ubseq uently

HPLC on C
4
or C
8
Vydac columns using acetonitrile in
0.1% trifluoroacetic acid as eluant. Isolated f ragments were
analyzed by gas-phase sequencing using the Applied
Biosystem 477A sequenator.
RESULTS
Degradation of TRH by serum and tissue homogenates
from different species
For comparative studies the specific activities of t he TRH-
DEs we re dete rmined i n se rum as well as in brain and liver
homogenates from cow, pig and rat (Table 1). For all t hree
species, high e nzymatic activities were found in brain. In rat
and pig, high enzymatic activities were also detected in
serum and significant activities in liver. In c ontrast, very low
activities were measured in liver homogenates and serum
from cows.
Table 1. Specific activities of the TRH-DEs in serum, liver and brain
from rat, pig and cow. Serum and tissue homogenates were prepared
and analyzed as described in Materials and methods (n ¼ 3 for pig and
cow, n ¼ 8 for rats, values are me ans ± SD).
Specific activity of the TRH-degrading enzyme
(% TRH-degradedÆmin
)1
Æmg protein
)1
)
Species Serum Liver Brain

medium due to cell leakage (Fig. 2). After 2 days of
cultivation and restoration of the cell membrane integrity,
the c oncentration of a lbumin a nd the activity of t he
TRH-DE in the culture medium increased in a correlative
manner (Fig. 2).
Purification of the TRH-DEs from serum and liver
The membrane-bound live r TRH-DE was s olubilized and
purified by following exactly the procedure elaborated for
the purification of the particulate TRH-DE from rat and
pig brain [28]. For the purification of the serum TRH-
DE fractionation by poly(ethylene glycol) precipitation
was used as the first step not only to partially purify the
enzyme but also to reduce the ionic strength due to salt.
At a poly(ethylene glycol) c oncentration of 25% the
serum enzyme completely precipitated. The enzyme was
recovered almost completely (97%) from the protein
pellet and could be applied directly to the Q-Sepharose
column. Elution from this column and further purifica-
tion followed again the procedure described previously
[28].
Characterization of the TRH-DEs from serum and liver
Molecular mass estimation. An approximate molecular
mass of 260 000 Da has been determined before for the
serum TRH-DE by gel filtration of porcine serum [9]. By
the same method, a molecular mass of % 250 000 Da
could be e stimated for t he trypsin-solubilized and purified
membrane-bound liver TRH-DE (Fig. 3). When the
purified enzymes were s ubjected to SDS/PAGE under
reducing conditions a molecular mass of % 125 000 Da
could be estimated for both enzymes, the serum TRH-

245 0 00) a nd o valbumin ( 4; M
r
45 000)
was applied to t he same column.
Ó FEBS 2002 Characterization of the TRH-DE from serum and liver (Eur. J. Biochem. 269) 1281
Western blot analysis. To verify the hypothesis that the
brain T RH-DE, the serum TRH-DE and the liver TRH-DE
arederivedfromthesamegene,theenzymepreparations
were subjected to Western blotting. All three enzymes w ere
recognized by the monoclonal antibody 41H2 which was
generated by using purified TRH-DE from pig brain as
antigen (Fig. 4B). This finding indicates t hat t hese enzymes
are immunologically very similar. At this point, it is worth
noting that this antibody is specific to the enzymes of
porcine origin and does not react with t he enzymes from rat
or mouse.
Identification as glycoproteins. Asinthecaseofthebrain
TRH-DE [28], the TRH-DEs from liver and serum also
bind strongly to the Lentil-lectin Se pharose columns
which were used for the purification of these enzymes.
Thus, both enzymes could be identified as glycoproteins.
To gain more information as to the carbohydrate
structure, the three enzymes were subjected to lectin blot
analysis. As shown in T able 2, the serum enzyme and the
liver enzyme exhibit identic al properties, distinctly differ-
ent from th e brain enzyme. For example, the liver enzyme
and t he serum e nzyme are re adily recognized by the
lectin SNA (Sambucus nigra agglutinin) but not by
the lectin GNA (Galanthus nivalis agglutinin), whereas
the opposite is true for the brain enzyme. The collected

GNA (Galanthus nivalis A.) + – –
MAA (Maackia amurensis A.) – – –
DSA (Datura stramonium A.) – – –
ConA (Concanavalin A) + + +
WGA (Wheat germ A.) + + +
PHA-L (Phytohaemagglutinin-L) + – –
PHA-E (Phytohaemagglutinin-E) + – –
RCA
120
(Ricinus communis A. I) + – –
Fig. 4. SDS/PAGE and Western blot analysis of the purified porcine
TRH-DE from brain, liver and serum. As described i n Materials and
methods the solubilized and purified membrane-bound TRH-DE from
brain (lane 1) and liver (lane 3) as well a s the purified serum TRH-DE
(lane 2) were subjected to SDS/PAGE and v isualized either by silver
staining (A) or immun ologically after Western blottin g ont o a nitro-
cellulose membrane by use of the monoclonal antibody 41H2 (B). For
the identification as glycoprotein (C), the purified enzymes were either
treated (T) or not (NT) with the endoglycosidase F /N-glyco sidase F
enzyme preparation as d escrib ed in Materials and m ethods and then
subjected t o S DS/P AGE f ollow ed b y Western blotting. The proteins
were again id entified by use o f the monoclonal antibody 41H2.
1282 S. Schmitmeier et al. (Eur. J. Biochem. 269) Ó FEBS 2002
TRH-DE from pig brain were analyzed [32]. This result
clearly demonstrates that both enzymes are derived from the
same gene. Comparison of the pep tide sequences of the
serum T RH-DE w ith the cDNA deduced amino-acid
sequences of the TRH-DE from rat [32] and human [37]
brain reveals that only eight amino acids (2.8%) were
different out of the 288 amino acids i dentified, whereby at

surprising as in tissue homogenates TRH is not only
inactivated by one enzyme as in serum or plasma but is
degraded by three peptidases (reviewed in [13,40]) namely
pyroglutamate aminopeptidase and post proline cleaving
enzyme (both are cytosolic enzymes), and the membrane-
bound TRH-DE, whereby the latter peptidase exhibits
identical enzyme-chemical characteristics as the serum
TRH-DE. Using enzyme-specific conditions to determine
the activity of the TRH-specific T RH-DEs indeed we found
high enzymatic a ctivities in b rain homogenates of all three
species examined. In contrast, considerable differences in
the TRH-degrading activities were noticed in the serum of
these animals, w hereby the enzyme activity is a lmost absent
in the serum from cow. Interestingly, we also observed a
correlation between the activity of the serum TRH-DE a nd
the activity of the TRH-DE in liver homogenates, suggest-
ing that the serum enzyme m ay be of liver origin.
At present we do not have an explanation for the late
development of the serum TRH-DE or for the extremely
low activity of this enzyme in some species. Nevertheless,
these results support the notion that the serum T RH-DE,
like most serum enzymes and proteins, is derived f rom
the liver. The decrease of the activity of the TRH-DE in
rats treated with t hioacetamide, a known hepatotoxin
which induces liver c irrhosis [33,34] seemed to be in line
with this interpretation. However, a rapid decrease in the
enzymatic activity w as already observed within a few
days after thioacetamide treatment. As liver cirrhosis is
generally a long-term process and is observed histolog-
ically only after treatment with thioacetamide for several

Peptide 4+ *543-GHSVFQRQ LQDYLTIHKY GNAARNDLWNT LSEA
Peptide 5+ 598-GYP VITIFGNTTA ENRII
Peptide 6à 677-GSWL LGNI
Peptide 7à 751-DFLPWHAASK
Peptide 8+ *958-NSK LISGVTEFLN TEGELKELKN
Peptide 9à 985-SYDGVA AASFSRAVET VEANVRW
Peptide 10à *1009-M LYQDELFQWL GKALRH
Ó FEBS 2002 Characterization of the TRH-DE from serum and liver (Eur. J. Biochem. 269) 1283
the activity of the TRH-DE released into the culture
medium, suggesting that the increase in the enzymatic
activity is due to the increased synthetic a ctivity of
hepatocytes and not due to cell leakage.
For direct analysis we purified the membrane-bound liver
TRH-DE after solubilization by trypsin and the serum
TRH-DE to elec trophoretic homogeneity b y following the
procedure described for the isolation of the membrane-
bound TRH-DE from pig brain [28]. By gel filtration a
molecular mass of 250 000 Da could b e estimated for the
truncated liver enzyme, a value which corresponds well with
the molecular mass of the p apain-solubilized liver enzyme
[15] and the molecular mass of the serum enzyme [9]
reported before (260 000 Da) but differs from the molecular
mass of 230 000 Da determined for the trypsin-solubilized
brain enzyme [28]. After SDS/PAGE under reducing
conditions a molecular mass of 125 000 Da was estimated
for the liver enzyme and the serum enzyme and a molecular
mass of 116 000 Da for the brain enzyme, indicating that all
these enzymes exist as homodimers, like many surface
peptidases [42]. Interestingly, TRH-DEs from brain, liver
and s erum were all r ecognized by the monoclonal antibody

immunoglobulin l [51,52]) or by proteolytic cleavage of the
membrane-bound liver enzym e as demonstrated for various
membrane-bound proteins with soluble c ounterparts
(reviewed in [53]). By fragmentation a nalysis of t he purified
serum TRH-DE, two peptide sequences (peptide 1 and 2)
(Table 1) could be identified which correspond to the
sequences 160–173 and 192–221 of the cDNA deduced
amino-acid sequence of the membrane-bound brain
TRH-DE. As both peptides are encoded b y e xon 1 which
ends at the position of amino-acid 260 [37,54], we can
conclude that the serum enzyme is not a product of
alternative mRNA s plicing but must b e generated by
proteolysis. Whe ther the s erum enzyme is released f rom the
plasma membrane of hepatocytes by proteases acting as
sheddases o r secretases (also designated as membrane
protein-solubilizing proteases, MPSPs) [53,55–57] remains
to be elucidated. Preliminary experiments indicate that the
release of the serum enzyme is not affected by inhibitors
directed against well characterized sheddases [name ly
b-secretase, c-secretase and TNFa protease (TACE)]. The
present results indicate furthermore that the serum e nyzme
might be generated intracellularly b ecause after homogeni-
zation of isolated hepatocytes and high speed centrifuga-
tion, 40% of the TRH-degrading activity could be found in
the cytosolic fraction and 60% of the e nzyme activity was
recovered from the particulate fraction ( Schmitmeier, S. &
Bauer, K., unpublished observation). This asp ect warrants
further investigation.
ACKNOWLEDGEMENTS
We would like to thank Prof Dr P. W. Jungblut for his interest and

9. Bauer, K. & Nowak, P. (1979) Characterization of a thy-
roliberin-degrading serum enzyme catalyzing the hydrolysis of
thyroliberin at the pyroglutamyl-histidine bond. Eur. J. Biochem.
99, 239–246.
10. Bauer, K., Nowak, P. & Kleinkauf, H. (1981) Specificity of a
serum peptidase hydrolyzing thyroliberin at pyroglutamyl-histi-
dine bone. Eur. J. Biochem. 11 8, 173–176.
11. O’Connor, B . & O’Cuinn, G. (1984) L ocalization of a narrow-
specificity thyroliberin hydrolyzing pyroglutamate aminopepti-
dase in synaptosomal membranes of guinea-pig brain. Eur.
J. Bioc hem. 144, 2 71–278.
12. Wilk, S. & Wilk, E.K. (1987) Pyroglutamyl peptidase II, a t hy-
rotropin releasing hormone degrading enzyme: purification and
1284 S. Schmitmeier et al. (Eur. J. Biochem. 269) Ó FEBS 2002
specificity studies of the rabbit b rain enzyme. Neurochem. Int. 15 ,
81–89.
13. O’Cuinn, G., O’Connor, B. & Elmore, M. (1990) Degradation of
thyrotropin-releasing hormone and luteinising hormone-releasing
hormone b y enzymes of bra in tissue. J. Neurochem. 54, 1–13.
14. O’Leary, R.M. & O’Connor, B. (1995) A study of a synaptosomal
thyrotropin releasing hormone-inactivating pyroglutamate ami-
nopeptidase from bovine brain. Int. J. Biochem. Cell Biol. 27,
881–890.
15. Scharfmann, R., Morgat, J.L. & Aratan-Spire, S. (1989) Presence
of a particulate thyrotropin-releasing hormone-degrading pyro-
glutamate aminopeptidase activity in rat liver. Neuroendocrinology
49, 442 –448.
16. Bauer, K. (1976) Regulation of degradation of thyrotropin
releasing h ormone by thyroid hormones. Na tu re 259, 5 91–593.
17. Dupont, A., Labrie, F., Levasseur, L., D ussault, J.H. & S chally,

26. Peterson, G.L. (1977) A simplification of the protein assay method
of Lowry et al. which is more generally applicable. Anal. Biochem.
83, 346 –356.
27. Bauer, K., Carmeliet, P., Schulz, M., Baes, M. & Denef, C. (1990)
Regulation and cellular localization of the membrane-bound
thyrotropin-releasing h ormone-degrading enzyme in primary
cultures of neur onal, glial and aden ohypophyseal cells.
Endocrinology 127 , 1224–1233.
28. Bauer, K. (1994) Purification and chara cterization of the thyro-
tropin-releasing-hormone-degrading ectoenzyme. Eur. J. Biochem.
224, 387–396.
29. Laemmli, U .K. (1970) Cle avage of structural proteins during the
assembly of the head o f bacteriophage T4 . Nature 227 , 680–685.
30. Towbin, H ., Staehelin, T. & Gordon, J. ( 1979) Electrop horetic
transfer of proteins from polyacrylamide gels to nitrocellulose
sheets: procedure and s ome applications. Proc. Natl Acad. Sci.
USA 76, 4350–4354.
31. Hase lbeck, A., Schickaneder, E., von der Eltz , H. & Hos el, W.
(1990) Structural characterization of glycoprotein carbohydrate
chains by using digoxigenin-labeled lectins on blots. Anal.
Biochem. 191, 25–30.
32. Schauder, B., Schomburg, L., Ko
¨
hrle, J. & Bauer, K. (1994)
Cloning of a cDNA encoding an ectoenzyme that degrades
thyrotropin-releasing hormone. Proc. Natl Acad. Sci. USA 91,
9534–9538.
33. Muller, A., Machnik, F., Zimmermann, T. & Schubert, H. (1988)
Thioacetamide-induced cirrhosis-like liver lesions in rats – use-
fulness and reliability of this animal model. Exp. Pathol. 34,

signalling sys tem. Results Probl. C ell Differ. 26, 1 3–42.
41. Nuber, R., T eu tsch, H.F. & Sasse, D. ( 1980) Metabolic zonation
in thioacetamide-induced liver cirrhosis. Histoc hemistry 69,
277–288.
42. Maraux, S. (1987) Structural and topological aspects. In Mam-
malian Ectoenzymes (Kenny, A.J. & Turner, A.J., eds), p p. 15–45.
Elsevier, A msterdam, the Nether lands.
43. Krusius, T. & Finne, J. (1977) Structural features of tissue
glycoproteins. Fraction atio n a nd methylation analysis of glyco-
peptides derived from rat brain, kidney and liver. Eur. J. Biochem.
78, 369 –379.
44. Hoffmann, A., Nimtz, M., Wurster, U. & Conradt, H.S. (1994)
Carbohydrate structures of beta-trace protein from human
cerebrospinal fluid: evidence for Ôbrain-typeÕ N-glycosylation.
J. Neurochem. 63 , 2185–2196.
45. Finne, J. & Krusius, T. (1979) Structural features of the
carbohydrate units of plasma glycoproteins. Eur. J. Biochem. 102,
583–588.
46. Ashwell, G. & Harford, J . (1982) Carbohydrate-specific receptors
of the liver. Ann. Re v. Biochem. 51, 5 31–554.
47. Rademacher, T.W., Parekh, R.B. & Dwek, R.A. (1988)
Glycobiology. Ann. Rev. Biochem. 57, 785–838.
48. Paulson, J .C. (1989) Glycoproteins: what are the sugar chains for?
Trends Bioc hem. Sci. 14, 272–276.
49. Lis, H. & Sharon, N. (1993) Protein glycosylation. Structural and
functional aspects. Eur. J. Bio c hem. 218, 1–27.
50. Parekh, R.B. (1994) Site-specific protein glycosylation. Adv. Drug
Del. Rev. 13, 2 51–266.
51. Early, P., Rogers, J., Davis, M., Calame, K., Bond, M., Wall, R. &
Hood, L. (1980) Two mRNAs can be produced from a single