Evolutionary relationships of the prolyl oligopeptidase family enzymes
Jarkko I. Vena¨la¨ inen, Risto O. Juvonen and Pekka T. Ma¨ nnisto¨
Department of Pharmacology and Toxicology, University of Kuopio, Finland
The prolyl oligopeptidase (POP) family of serine proteases
includes prolyl oligopeptidase, dipeptidyl peptidase IV,
acylaminoacyl peptidase and oligopeptidase B. The enzymes
of this family specifically hydrolyze oligopeptides with less
than 30 amino acids. Many of the POP family enzymes have
evoked pharmaceutical interest as they have roles in the
regulation of peptide hormones and are involved in a variety
of diseases such as dementia, trypanosomiasis and type 2
diabetes. In this study we have clarified the evolutionary
relationships of these four POP family enzymes and ana-
lyzed POP sequences from different sources. The phylo-
genetic trees indicate that the four enzymes were present in
the last common ancestor of all life forms and that the
b-propeller domain has been part of the family for billions
of years. There are striking differences in the mutation rates
between the enzymes and POP was found to be the most
conserved enzyme of this family. However, the localization
of this enzyme has changed throughout evolution, as three
archaeal POPs seem to be membrane bound and one third of
the bacterial as well as two eukaryotic POPs were found to be
secreted out of the cell. There are also considerable distinc-
tions between the mutation rates of the different substrate
binding subsites of POP. This information may help in the
development of species-specific POP inhibitors.
Keywords: acylaminoacyl peptidase; dipeptidyl peptidase IV;
evolution; oligopeptidase B; prolyl oligopeptidase.
The prolyl oligopeptidase family of serine proteases (clan
SC, family S9) includes a number of peptidases, from which
prolyl oligopeptidase (POP, EC 3.4.21.26), dipeptidyl pept-
idase IV (DPP IV, EC 3.4.14.5), oligopeptidase B (OB,
EC 3.4.21.83) and acylaminoacyl peptidase (ACPH,
EC 3.4.19.1) have been the enzymes under the most intense
study [1–3]. This enzyme family is different from the
classical serine protease families, trypsin and subtilisin, in
that they cleave only peptide substrates while excluding
large proteins. The mechanism of preventing the digestion
of bigger proteins was recently clarified when the 3D
structure of POP was solved [4]. The enzyme consists of a
peptidase and seven-bladed b-propeller domains. The
narrow entrance of b-propellerpreventslargerproteins
from entering into the enzyme active site. A similar b-
propeller consisting of eight instead of seven blades was
recently identified in DPP IV when its crystal structure was
solved [5].
The enzymes of the POP family have different substrate
specificities: POP hydrolyzes peptides at the carboxyl side of
the proline residue, DPP IV liberates dipeptides where the
penultimate amino acid is proline, OB cleaves peptides at
lysine and arginine residues and ACPH removes N-acetyl-
ated amino acids from blocked peptides. DPP IV is a
membrane bound enzyme, and in this way different from
the rest of the POP family members that are cytoplasmic
proteins [3]. However, a membrane bound form of POP has
also been characterized from bovine brain but the sequence
of this protein is not available at the present time [6].
Many of the POP family enzymes have become targets of
the pharmaceutical industry, e.g. POP degrades many
neuropeptides involved in learning and memory, such as
substance P, thyrotropin releasing hormone and arginine-
vasopressin. Indeed, POP inhibitors have been shown to
reverse scopolamine-induced amnesia in rats and to
improve cognition in old rats and 1-methyl-4-phenyl-
1,2,3,6-tetrahydropyridine (MPTP)-treated Parkinsonism
model monkeys [7–9]. A number of the antitrypanosomal
drugs in widespread use are OB inhibitors [10]. In addition,
inhibition of DPP IV has been proposed as a therapeutic
approach to the treatment of type 2 diabetes as this enzyme
is involved in the metabolic inactivation of a glucagon-like
peptide 1 that stimulates insulin secretion [11]. Recently,
DPP IV knockout mice were found to be protected against
obesity and insulin resistance [12].
In this study, based on public databanks and a number of
computer programs, we have clarified the evolutionary
relationships of these four POP family enzymes by gener-
ating phylogenetic trees including POP family enzymes from
different species. First, important amino acids for the
enzyme function were sought by analyzing multiple align-
ments of 72 aligned POP family sequences. Secondly, we
analyzed POP sequences from different species because POP
Correspondence to J. I. Vena
¨
la
¨
inen, Department of Pharmacology and
Toxicology, University of Kuopio, P.O. Box 1627, FIN-70211
Kuopio, Finland. Fax: + 358 17 162424, Tel.: + 358 17 163774,
E-mail: Jarkko.Venalainen@uku.fi
Abbreviations: ACPH, acylaminoacyl peptidase; DPPII, dipeptidyl
peptidase II; DPP IV, dipeptidyl peptidase IV; OB, oligopeptiadase B;
POP, prolyl oligopeptidase; GPI, glycosylphosphatidylinositol;
LUCA, last universal common ancestor.
Enzymes: prolyl oligopeptidase (EC 3.4.21.26); dipeptidyl peptidase IV
(EC 3.4.14.5); oligopeptidase B (EC 3.4.21.83); acylaminoacyl
peptidase (EC 3.4.19.1).
Note: The departmental website is available at http://www.uku.fi/
farmasia/fato/indexe.htm
(Received 28 March 2004, revised 28 April 2004, accepted 4 May 2004)
Eur. J. Biochem. 271, 2705–2715 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04199.x
can be considered as a model enzyme of this family, as its
crystal structure is available and many details about its
catalytic mechanism are known. In this analysis we created
a conservation profile of POP to study the mutation rates of
amino acids involved in substrate binding and to find other
essential amino acids. Finally, we pinpointed signal
sequences, and transmembrane and lipid anchor sequences
from POP enzymes of different sources to study if the
localization of the enzyme has changed during evolution.
Materials and methods
Multiple sequence alignment and construction
of phylogenetic trees of the POP family
The POP family enzymes from different sources were
identified by
BLASTP
searchesfromtheNCBInrdatabase
against human POP (NP_002717), human DPP IV
(CDHU26), human ACPH (P13798), Escherichia coli OB
(E64946) and rat DPP II (JC7668) sequences. To be
identified as a POP family member, the sequence had to
have the catalytic triad topology of Ser-Asp-His which is
different from the classical serine proteases [13]. The
iterative
PSI
-
BLAST
feature was not applied in these searches.
The aim of the searches was to obtain a large enough
number of sequences for the analysis, not to find all the
existing POP family sequences. As a result, 28 POP, 10
ACPH, 14 DPP IV, 20 OB and seven DPP II sequences
from different species were manually selected for the
analysis. The selected sequences and their accession codes
are presented in Table 1.
A multiple sequence alignment of the 79 selected
sequences was constructed by a combination of
T
-
COFFEE
and
CLUSTALX
programs [14,15]. A structure based sequence
alignment of pig POP (1QFS) and human DPP IV (IJ2E)
was created using the
T
-
COFFEE
program and other proteins
were subsequently added to this alignment using the
CLUSTALX
program until the multiple sequence alignment
of 79 sequences was obtained. The alignment was manually
edited based on the initial 3D alignment. The neighbor-
joining tree was constructed for the peptidase domains of
the enzymes (corresponding to the pig POP residues 1–72
and 428–710) and for the complete sequences using
CLUSTALX
. Bootstrap values were calculated with 1000
resamplings. DPP II sequences were used as the outgroup in
this analysis, as this enzyme is a close neighbor to the POP
family and a member of the serine protease family S28. The
NJPLOT
program was used to display the constructed
phylogenetic tree. The phylogenetic trees were also con-
structed using the maximum likelihood method with the
program
TREE
-
PUZZLE
[16]. The
TREEVIEW
program (http://
taxonomy.zoology.gla.ac.uk/rod/treeview.html) was used
to view the maximum likelihood tree.
Conservation profile of POP sequences
To study the conservation profile of POP, 28 POP sequences
alone were aligned using
T
-
COFFEE
. Multiple sequence
alignments were visualized and analyzed using
GENEDOC
program (http://www.psc.edu/biomed/genedoc/) alongside
the pig POP sequence. The conservation rates of each of the
710 amino acids were divided into four groups: 1st, £ 49%;
2nd, between 50 and 74%; 3rd, between 75 and 99% and
4th, 100% similarity at an alignment position. The similar-
ities of amino acids were based on BLOSUM62 substitution
matrix.
Prediction of transmembrane regions, lipid anchors
and signal peptides in POP sequences
All of the 28 POP sequences from different sources were
analyzed with
TMHMM
program [17] to decide whether the
enzymes contain transmembrane sequences. Lipid anchor
sites were searched with program
BIG PI
[18]. The presence of
signal sequences in the POP enzymes and their possible
cleavage sites were predicted with the
SIGNALP
V2.0 program
using hidden Markov model method [19].
Results and Discussion
Multiple sequence alignment of the POP family enzymes
As can be seen from Table 1, POP and ACPH are
distributed in archaeal, bacterial and eukaryotic species
whereas DPP IV and OB were not found from archaeal
sources. Although POP and ACPH are present in all three
forms of organisms (Bacteria, Archaea, Eucaryota), there
are several organism groups in which these enzymes were
not found. For example, POP was not found in Fungi.
Table 2 lists some identity and similarity percentages within
POP family enzymes when the whole sequences or just the
catalytic domains of the enzymes are taken into account. In
general, the sequence identity percentages between the four
enzymes are low, below 20%. The peptidase domain is
slightly more conserved, as shown by the higher identity/
similarity percentages. However, despite the low sequence
homology and distinct substrate specificities, the multiple
sequence alignment revealed 10 invariant residues between
the 72 aligned enzymes of the POP family: Arg505, Gly506,
Gly511, Asp529, Gly552, Ser554, Gly556, Gly557, Asp641
and His680 (numbering according to the pig POP sequence,
the residues are shown with downward arrows in Fig. 1).
All of these amino acids are located at the active site of the
enzyme. This was expected, as it has been reported
previously that the greatest similarities between the amino
acid sequences of POP family members are located in the
C-terminal third of the alignment [20]. Of these conserved
residues, Ser554, Asp641 and His680 form the catalytic triad
of POP and the small residues Gly552, Gly556 and Gly557
are clustered around the catalytic serine. The three glycine
residues have been proposed to improve the binding of
substrate by preventing steric hindrance [4]. Arg505 and
Gly506 are situated in a loop between the b4-strand and the
aB¢-helix at the active site, and Gly511 is the first residue of
that a-helix. The high degree of conservation of these
residues suggests that this turn between the secondary
structure elements is crucial for the POP family enzyme
function or for its structural stability.
Figure 2 represents some amino acid similarity percent-
ages of whole sequences and catalytic domains between
human and some eukaryotic, bacterial and archaeal
sequences of POP, DPP IV and ACPH. The similarities
between human and rat sequences are very high for POP (98/
98%; whole sequences and catalytic domains, respectively)
2706 J. I. Vena
¨
la
¨
inen et al.(Eur. J. Biochem. 271) Ó FEBS 2004
Table 1. Prolyl oligopeptidase family and DPP II enzymes from different species used in this analysis.
Enzyme Species Domain of life Accession number
POP Human Eukarya NP_002717
Pig Eukarya P23687
Bovine Eukarya Q9XTA2
Mouse Eukarya NP_035286.1
Rat Eukarya NP_112614.1
Fugu rubribes Eukarya SINFRUP00000059740
Xenopus laevis Eukarya AAH47161
Arabidopsis thaliana Eukarya AAL86330.1
Dictyostelium discoideum Eukarya CAB40787.1
Drosophila melanogaster Eukarya AAF52942.1
Anophiles gambiae Eukarya EAA14977.1
Oryza sativa Eukarya BAB78619.1
Deinococcus radiodurans Bacteria NP_296223.1
Shewanella oneidensis Bacteria NP_718337.1
Trichodesmium erythraeum Bacteria ZP_00072911.1
Nostoc sp. Bacteria NP_486573.1
Nostoc punctiforme Bacteria ZP_00110050.1
Flavobacterium meningosepticum Bacteria P27028
Aeromonas punctata Bacteria AAD34991.1
Aeromonas hydrophila Bacteria Q06903
Novosphingobium capsulatum Bacteria BAA34052.1
Novosphingobium aromaticivorans Bacteria ZP_00093416.1
Myxococcus xanthus Bacteria AF127082–3
Thermobifida fusca Bacteria ZP_00058751.1
Pyrococcus abyssi Archaea NP_126828.1
Pyrococcus furiosus Archaea NP_578544.1
Pyrococcus horikoshii Archaea NP_143154.1
Sulfolobus tokodaii Archaea NP_375840
DPP IV Human Eukarya CDHU26
Bovine Eukarya P81425
Cat Eukarya Q9N2I7
Rat Eukarya A39914
Mouse Eukarya NP_034204.1
Xenopus laevis Eukarya CAA70136.1
Fugu rubribes Eukarya SINFRUP00000066299
Anopheles gambiae Eukarya EAA05700.1
Drosophila melanogaster Eukarya NP_608961.1
Aspergillus niger Eukarya CAC1019.1
Scizosaccharomyces pombe Eukarya NP_593970.1
Aspergillus fumigatus Eukarya AAC34310.1
Porphyromonas gingivalis Bacteria BAA28265.1
Flavobacterium meningosepticum Bacteria S66261
ACPH Human Eukarya P13798
Rat Eukarya NP_036632.1
Pig Eukarya JU0132
Caenorhabditis elegans Eukarya NP_500647.1
Fugu rubribes Eukarya SINFRUP00000057906
Basillus subtilis Bacteria NP_391103.1
Oceanobasillus iheyensis Bacteria NP_692002.1
Pyrococcus abyssi Archaea NP_127272.1
Pyrococcus horikoshii Archaea NP_142793.1
Deinococcus radiodurans Bacteria NP_293889.1
OB Trypanosoma brucei Eukarya AAC80459.1
Leishmania major Eukarya AAD24761.1
Escherichia coli Bacteria E64946
Shigella flexneri Bacteria NP_707707.1
Salmonella typhimurium Bacteria NP_460836.1
Yersinia pestis Bacteria NP_669832.1
Ó FEBS 2004 Evolutionary relationships of the POP family enzymes (Eur. J. Biochem. 271) 2707
and ACPH (95/96%), whereas the similarity between human
and rat DPP IV is much lower for both whole sequences and
catalytic domains (87/87%). The differences in conservation
percentages are even more striking between human/Fugu
rubribes enzymes and the same kind of conservation order
can also be found between human/Flavobacterium meningo-
septicum (55/62% of POP compared to 38/48% of DPP IV)
and human/Pyrococcus abyssi (42/50% of POP compared to
27/36% of ACPH). OB was excluded from this comparison
because it is not found in animals. However, the similarity
percentage between OB from Shewanella oneidensis and
Nostoc sp. can be compared to that of POP. Again, POP has
the higher conservation percentage: 66/73% compared to 59/
69% of OB. This analysis indicates that POP is the most
conserved peptidase of these four POP family enzymes, with
the highest similarities found between each pair of sequences
studied. The differences in conservation degrees between the
enzymes are similar when the identity percentages are
considered.
The phylogenetic tree of the POP family
The multiple alignment peptidase domains of 72 POP
family sequences and seven DPP II sequences were used to
construct phylogenetic trees with distance-based (neighbor-
joining) and character-based (maximum likelihood) meth-
ods. In many cases, these two methods have been shown to
be almost equally efficient in obtaining the correct topology
[21,22]. The DPP II family was used as an outgroup for
phylogenetic constructions. The two tree-building methods
gave essentially the same tree topologies and the neigh-
bour-joining tree with bootstrap values and the maximum
likelihood tree with support values are shown in Figs 3 and
4. The phylogenetic trees clearly show that each of the four
POP family enzymes (POP, DPP IV, OB and ACPH) form
a single cluster containing all of the species included in this
analysis. Both trees show that OB is the closest relative to
POP, not ACPH as was recently stated [23] and that
DPP IV is the closest relative to ACPH. In the cases of
POP and ACPH the enzyme clusters have members from
each of the three domains of the organisms. In this
analysis, DPP IV and OB sequences were not found from
archaeal species. These four enzyme clusters are supported
by high bootstrap values in the neighbor-joining tree and
support values in the maximum likelihood tree. The clusters
are further divided in subclusters, for example, the POP
cluster forms subclusters of archaea (Pyrococcus horikoshii,
P. abyssi, Pyrococcus furiosus and Sulfolobus tokodaii)and
Table 1. Continued
Enzyme Species Domain of life Accession number
Shewanella oneidensis Bacteria NP_715786.1
Xanthomonas axonopodis Bacteria NP_640984
Nostoc sp. Bacteria NP_487951.1
Treponema denticola Bacteria AAK39550.1
Sinorhizobium meliloti Bacteria NP_385091.1
Acrobacterium tumefaciens Bacteria NP_353917.1
Brucella melitensis Bacteria NP_540282.1
Brucella suis Bacteria NP_697584.1
Mycobacterium leprae Bacteria NP_302455.1
Corynebacterium glutamicum Bacteria NP_601794.1
Rickettsia conorii Bacteria NP_360014.1
Rickettsia prowazekii Bacteria NP_220665.1
Bifidobacterium longum Bacteria NP_696390.1
Moraxella lacunata Bacteria Q59536
DPP II Rat Eukarya JC7668
Human Eukarya Q9UHL4
Mouse Eukarya Q9ET22
Arabidopsis thaliana Eukarya NP_201377.2
Anopheles gambiae Eukarya EAA04920.1
Drosophila melanogaster Eukarya AAF53897.1
Caenorhabditis elegans Eukarya NP_498718.1
Table 2. Amino acid identity/similarity percentages between POP family enzymes. The identity/similarity percentages of the peptidase domains are
shown in brackets.
POP Human ACPH Human DPP IV Human OB E. coli
POP Human – 9/24 (10/28) 15/30 (17/30) 22/41 (27/46)
ACPH Human – 10/22 (13/30) 10/24 (14/27)
DPP IV Human – 11/23 (12/25)
OB E. coli –
2708 J. I. Vena
¨
la
¨
inen et al.(Eur. J. Biochem. 271) Ó FEBS 2004
eukaryotes. It is interesting to note that according to the
POP cluster of the phylogenetic trees, Drosophila melano-
gaster and Anopheles gambiae differ more from mammals
than do the plants Oryza sativa and Arabidopsis thaliana.
The most probable reason for this apparent discrepancy is
that these two insects diverged considerably faster than
vertebrates. At the gene sequence level, these two species
that diverged 250 million years ago, differ more than even
humans and pufferfish F. rubribes – species that diverged
450 million years ago [24]. This discovery is valid also with
the POP enzyme having sequence identity of 58% between
A. gambiae and D. melanogaster and 74% between human
and F. rubribes. A similar order of sequence identities can
also be seen with DPP IV. The phylogenetic trees were also
created using the complete sequences of the enzymes (data
not shown). These analyses resulted in the same tree
topologies as seen in Figs 3 and 4, except that the branch
lengths are slightly longer due to the lower conservation of
the b-propeller domains. This shows that the b-propeller
domain has been part of this enzyme family for billions of
years.
The phylogenetic trees show that the four POP family
enzymes were clearly set up before the archaea, prokaryota
and eucaryota diverged along their own evolutionary lines
between 2000 and 4000 million years ago. This suggests that
all POP family proteins are of ancient origin and they were
Fig. 1. Conservation profile of 28 POP sequences from different species. The conservation percentage of each amino acid along the pig POP sequence
is indicated as 0, £ 49%; 1, between 50 and 74%; 2, between 75 and 99% and 3, 100%. The secondary structure elements of pig POP are indicated by
arrows for b-sheets and by boxes for a-helices. The invariant amino acids in each of the 72 analysed POP family sequences are shown by downward
arrows and the amino acids of the catalytic triad (Ser554, Asp641 and His680) are indicated by asterisks.
Ó FEBS 2004 Evolutionary relationships of the POP family enzymes (Eur. J. Biochem. 271) 2709
present in the last universal common ancestor (LUCA) of
all life forms. Thus, the present enzyme forms are vertically
inherited from this ancestor.
The high conservation of POP family enzyme sequences
from different species and their presence in the LUCA
strongly suggest that these enzymes have important roles in
physiological processes. However, the exact roles of these
enzymes are more or less unclear at the moment. Evidently
there was a need for peptidases that cleave only small
peptides specifically after proline, lysine or arginine even
during the early days of life.
Conservation profile of POP sequences from
different species
The conservation profile of 28 aligned POP sequences is
presented in Fig. 1. It is clear that the catalytic domain
(residues 1–72 and 428–710) is a much more conserved
region than the b-propeller domain (residues 73–427). In the
b-propeller domain, only seven amino acids (2.0%) have
100% similarity compared to 65 amino acids (17.8%) in the
catalytic domain. Six of the conserved amino acids in the
b-propeller are situated in b-sheets and one (Gly369) is
located between the b-sheet structures, so that the b-sheets
seem to be more conserved than the areas between them.
The low homology in the b-propeller domain is not
unexpected, as it has been proposed that the b-propeller
of P. furiosus POP does not perform the same function as
the mammalian enzyme, i.e. the exclusion of large peptides
from the active site [25]. Clearly the role of the b-propeller
has diversified during evolution.
Table 3 lists the conservation percentages of the pig POP
active site amino acids that are involved in the substrate
binding [4]. The specificity pocket S1 has 100% similary and
almost 100% identity among the 28 studied POP sequences.
Only Val580 and Tyr599 have some variations among
different species. In addition to the amino acids of the
catalytic triad and the residues that make hydrogen bonds
with substrate, Trp595 is also invariant. This residue is
claimed to enhance substrate recognition specificity by ring
stacking between the indole ring of Trp595 and the proline
ring of the substrate, so that all of the studied POP enzymes
can be claimed to be specific for proline [4]. It is surprising
that residues Phe476, Val644, Val580 and Tyr599 also have
100% similarities and 89.3–100% identities, as their role in
substrate binding is just to provide a hydrophobic environ-
ment and appropriate lining for the proline residue [4]. Due
to this conservation, it can be predicted that the changes of
these residues would dramatically decrease the specificity
for, or binding of, the proline residue.
The specificity pocket S3 is substantially more variable
than the S1 pocket. In pig POP, the S3 pocket ensures that
there is a fairly apolar environment. However, this is not
common for all POP sequences, because in many species the
POP enzyme contains polar and even charged residues (i.e.
Asn, Gly, Ser, Asp) at this site. Hence, it seems that only the
substrate binding S1 site has remained virtually unchanged
throughout the evolution, allowing enhanced flexibility to
substrate S2 and S3 residues. There have been attempts to
develop species specific POP inhibitors, for example against
Trypanosoma cruzi [26]. According to our analysis of subsite
evolution, the specificity might be achieved by varying the
structures of P2 and P3, but not the P1 subsite of the
inhibitor.
The most interesting amino acid at the S3 subsite is
Cys255, because it is responsible for pig POP inhibition
by bulky thiol reagents. F. meningosepticum, which has a
Thr instead of Cys255, is not inhibited by thiol reagents.
In addition to accounting for the inhibition by thiol
reagents, Cys255 also improves the catalytic efficacy at
pH values above neutrality by increasing the substrate
affinity [27]. Therefore it is interesting to note that, of
the 28 studied POP sequences, only eukaryotes have
cysteine at this site. Most bacterial POP sequences have
threonine in place of Cys255 but Myxococcus xanthus has
tryptophan instead of Cys255. All of the studied archaeal
POP enzymes have tryptophan at the same location. This
variability of amino acids between the three domains of
life is important, because it clearly modifies enzyme
properties, i.e. substrate affinity and perhaps also the
regulation by oxidation state.
Transmembrane regions and signal peptides
in POP sequences
Twenty eight POP sequences were analyzed with
TMHMM
program to detect transmembraneous regions in the
enzyme, because POP has also been characterized in a
membrane bound form from bovine brain [6]. Unfortu-
nately, the sequence of this apparently membrane bound
POP has not been published. Therefore, it is impossible to
conclude whether the enzyme is another form of cytosolic
POP or some other enzyme possessing similar properties to
POP. The program used in this analysis was recently
evaluated to have the best overall performance of the
currently available and most widely used transmembrane
prediction tools [28]. According to our analysis conducted
using the
TMHMM
program, none of the sequences were
predicted to contain transmembrane regions. However,
Novosphingobium capsulatum POP had a weak possibility
(0.45) of a transmebrane region. To decide whether this
protein is membrane bound or not, we analyzed this
sequence with another transmembrane prediction program,
SOSUI
[29]. This program also predicted the sequence to be of
Fig. 2. Amino acid similarity percentages between human–rat, human–
F. rubribes, human–F. meningosepticum and human–P. abyssi se-
quences of POP, ACPH and DPP IV. The whole bar and the lower
part of the bar represent the similarity percentages of the catalytic
domains and the complete sequences, respectively.
2710 J. I. Vena
¨
la
¨
inen et al.(Eur. J. Biochem. 271) Ó FEBS 2004
Fig. 3. The neighbor-joining tree of POP family enzymes. Protein sequences were aligned with
T
-
COFFEE
and
CLUSTALX
programs and the tree with
bootstrap values was then constructed with
CLUSTALX
program. DPP II sequences were used as outgroups and numbers represent the percentages
of 1000 bootsraps. The tree was then visualized with
NJPLOT
program.
Ó FEBS 2004 Evolutionary relationships of the POP family enzymes (Eur. J. Biochem. 271) 2711
Fig. 4. The maximum likelihood tree of POP family enzymes. Protein sequences were aligned with
T
-
COFFEE
and
CLUSTALX
programs and the
maximum likelihood tree with support values was calculated using
TREE
-
PUZZLE
version 5.0. DPP II sequences were used as outgroups and the tree
was visualized with
TREEVIEW
program.
2712 J. I. Vena
¨
la
¨
inen et al.(Eur. J. Biochem. 271) Ó FEBS 2004
a soluble protein so we believe that this enzyme is not
membrane bound.
Proteins can also be membrane bound even if they do not
possess a transmembrane sequence, if they contain a lipid
anchor. In that case the protein is post-translationally
modified with a glycosylphosphatidylinositol (GPI) moiety
and anchored on the extracellular side of the plasma
membrane [18]. The entry to the GPI-modification route is
directed by a C-terminal sequence signal, consisting of
about 20 amino acids. These signal sequences were searched
with the
BIG PI
program. None of the eukaryotic and
bacterial sequences possessed lipid anchor sequences, but
archaeal POP enzymes P. horikoshi, P. abyssi and P. furio-
sus seemed to contain the signal sequence with false positive
probabilities of 0.0147, 0.0172 and 0.0173, respectively. The
predicted attachment sites of the GPI moiety were Ala594,
Ala596 and Ala595 which all correspond to the Gly683 of
pig POP. The search was carried out using the metazoa
prediction function of the program and it is unclear whether
the result is valid for archaeal sequences. However, GPI-
linked proteins closely related to eukaryotes have also been
found from archaeal sources [30], suggesting that the
prediction may be correct. Naturally, this result will need
to be verified experimentally, but to our knowledge, this is
the first hint of a possible mechanism by which POP could
be attached to the cell membrane.
Sequence analysis with the
SIGNALP
program resulted in
the identification of four bacterial POP sequences that
contain signal peptide sequences, i.e. the enzymes are
secreted through the cell membrane. The POP forms are
secreted from Gram negative bacterias F. meningosepti-
cum, N. capsulatum, Novosphingibium aromaticivorans and
Shewanella oneidensis. The calculated signal peptide
probabilities of these enzymes varied from 0.971 to
1.000. The
SIGNALP
output of N. capsulatum POP is
presented in Fig. 5A. The output contains n-, h- and
c-region probabilities and the most likely cleavage site,
which is between residues 22 (alanine) and 23 (glutamine).
The cleavage sites of F. meningosepticum, N. aromaticivo-
rans and S. oneidensis signal peptides were predicted to be
between residues 20–21 (alanine-glutamine), 30–31 (serine-
glutamic acid) and 33–34 (alanine-alanine), respectively.
The signal sequences and their potential cleavage sites are
presented in Fig. 5B.
SIGNALP
predicted correctly the F. meninosepticum POP
signal peptide, as this enzyme has been shown experiment-
ally to be periplasmic, the cleavage site of the signal peptide
being between residues 20 (alanine) and 21 (glutamine) [31].
This correct prediction increases the reliability of
SIGNALP
results. The biological relevance of the periplasmic POP
activity is not clear. However, secretion of POP in bacterial
sources seems to be quite common, as four of the studied 12
bacterial sequences (33%) contained the signal sequence.
In addition to bacteria, secretion signal sequences were
also found from eukaryotes A. gambiae and Xenopus laevis
with probabilities of 0.905 and 0.808, respectively. The
cleavage sites were predicted to be between residues 24–25
(glycine-lysine) and 34–35 (alanine-serine). To our know-
ledge, these are the first eukaryotic POP enzymes that are
thought to be secreted out of the cell. It is interesting to note
the difference of POP localization between the fruit fly
D. melanogaster and the malaria transmitting mosquito
A. gambiae. Despite the different localization and rather
low sequence identity (58%), the POP proteins of A. gamb-
iae and D. melanogaster are likely to have similar catalytic
Table 3. Conservation percentages of the pig POP amino acids involved
in substrate binding.
Location Amino acid Role
Identity/
similarity (%)
S1-Pocket Ser554 Catalysis 100/100
Asp641 Catalysis 100/100
His680 Catalysis 100/100
Trp595 Ring stacking 100/100
Asn555 H-bond with S 100/100
Tyr473 H-bond with S 100/100
Phe476 Lining 100/100
Val644 Lining 100/100
Val580 Lining 92.9/100
Tyr599 Lining 89.3/100
S2-Pocket Arg643 H-bond with S 100/100
S3-Pocket Trp595 H-bond with S 100/100
Phe173 Lining 75.0/82.1
Met235 Lining 28.6/32.1
Cys255 Lining 42.9/42.9
Ile591 Lining 71.4/78.6
Ala594 Lining 57.1/57.1
Fig. 5. The secreted POP sequences. (A) The
SIGNALP
output of
Novosphingobium capsulatum POP. Predicted n-, h- and c-regions are
shown and the predicted cleavage site between residues 22 and 23 is
shown with a downward arrow. (B) The amino acid sequences of
secreted POP forms, the predicted cleavage sites are shown with
underlined letters.
Ó FEBS 2004 Evolutionary relationships of the POP family enzymes (Eur. J. Biochem. 271) 2713
properties because their amino acids involved in substrate
binding (Table 3) are identical. A. gambiae has only one
POP gene but D. melanogaster has an extra POP-like gene
(NP_610129) in addition to the POP sequence used in this
study (AAF52942). These proteins have sequence identity
and similarity percentages of 60% and 73% and their
substrate binding residues are identical with one important
exception: the C-terminal part starting from Val660 has
been deleted from NP_610129 and hence the third member
of the catalytic triad (His680) is missing. It is probable that
this protein is inactive or has a different function than POP
and that the extra POP-like gene is a product of gene
duplication in D. melanogaster.
A. gambiae and D. melanogaster belong to the same
taxonomic order, but have different lifestyles. Due to blood
feeding, A. gambiae is exposed to parasites such as
Plasmodium falciparum, the human malaria parasite.
A. gambiae efficiently combats the P. falciparum infection
and therefore an understanding of the immune system of
A. gambiae could be a very useful way to obtain clues to
controlling malaria. This has been done by comparing the
differences between immune-related genes of A. gambiae
and D. melanogaster [32]. Interestingly, POP has been
claimed to play a role in immunopathological processes
associated with lupus erythematosus and rheumatoid arth-
ritis [33]. Furthermore, several serine proteases have been
shown to regulate invertebrate defense responses such as
antimicrobial peptide synthesis [34]. Therefore, it is possible
that the secreted POP might play a role in the immune
responses of A. gambiae.
In summary, POP family enzymes were found to be of
ancient origin, as they were already present in the last
universal common ancestor of life. With respect to the
studied enzymes of the POP family, POP seems to be the
most conserved enzyme. Ten conserved amino acids were
found at the active site of the enzyme of each of the studied
POP family enzymes, indicating that those residues are
probably critical to the enzyme function. In POP, the S1
specificity pocket was found to be highly conserved,
compared to the more variable S3 specificity pocket. This
finding may help to develop species-specific POP-inhibitors.
Signal sequences were found in one third of bacterial POP
sequences and also in two eukaryotic species. Lipid anchor
sequences were found from three archaeal sources, indica-
ting that the POP enzyme in these species is membrane
bound.
Acknowledgements
This work was supported by National Technology Agency of Finland
and Ministry of Education of Finland (to J. I. V.). We wish to thank
Prof. Dan Larhammar, University of Uppsala, for his advice and
extremely helpful comments on the manuscript and Dr Ewen
MacDonald for linguistic advice.
References
1. Kanatani, A., Masuda, T., Shimoda, T., Misoka, F., Lin, X.S.,
Yoshimoto, T. & Tsuru, D. (1991) Protease II from Escherichia
coli: sequencing and expression of the enzyme gene and char-
acterization of the expressed enzyme. J. Biochem. (Tokyo) 110,
315–320.
2. Rawlings, N.D., Polga
´
r, L. & Barrett, A.J. (1991) A new family of
serine-type peptidases related to prolyl oligopeptidase. Biochem. J.
279, 907–908.
3. Polga
´
r, L. (2002) The prolyl oligopeptidase family. Cell. Mol. Life
Sci. 59, 349–362.
4. Fu
¨
lo
¨
p, V., Bo
¨
cskei, Z. & Polga
´
r, L. (1998) Prolyl oligopeptidase:
an unusual beta-propeller domain regulates proteolysis. Cell 94,
161–170.
5. Hiramatsu, H., Kyono, K., Higashiyama, Y., Fukushima, C.,
Shima,H.,Sugiyama,S.,Inaka,K.,Yamamoto,A.&Shimizu,R.
(2003) The structure and function of human dipeptidyl peptidase
IV, possessing a unique eight-bladed beta-propeller fold. Biochem.
Biophys. Res. Commun. 302, 849–854.
6. O’Leary, R.M., Gallagher, S.P. & O’Connor, B. (1996) Purifica-
tion and characterization of a novel membrane-bound form of
prolyl endopeptidase from bovine brain. Int. J. Biochem. Cell.
Biol. 28, 441–449.
7. Yoshimoto, T., Kado, K., Matsubara, F., Koriyama, N.,
Kaneto, H. & Tsura, D. (1987) Specific inhibitors for prolyl
endopeptidase and their anti-amnesic effect. J. Pharmacobio-dyn.
10, 730–735.
8. Atack, J.R., Suman-Chauhan, N., Dawson, G. & Kulagowski, J.J.
(1991) In vitro and in vivo inhibition of prolyl endopeptidase. Eur.
J. Pharmacol. 205, 157–163.
9. Marighetto, A., Touzani, K., Etchamendy, N., Torrea, C.C., De
Nanteuil, G., Guez, D., Jaffard, R. & Morain, P. (2000) Further
evidence for a dissociation between different forms of mnemonic
expressions in a mouse model of age-related cognitive decline:
effects of tacrine and S 17092, a novel prolyl endopeptidase
inhibitor. Learn. Mem. 7, 159–169.
10. Morty, R.E., Troeberg, L., Powers, J.C., Ono, S., Lonsdale-
Eccles, J.D. & Coetzer, T.H. (2000) Characterisation of the
antitrypanosomal activity of peptidyl alpha-aminoalkyl phos-
phonate diphenyl esters. Biochem. Pharmacol. 60, 1497–1504.
11. Hughes, T.E., Mone, M.D., Russell, M.E., Weldon, S.C. &
Villhauer, E.B. (1999) NVP-DPP728 (1-[[[2-[(5-cyanopyridin-2-yl)
amino]ethyl]amino]acetyl]-2-cyano-(S)-pyrrolidine), a slow-binding
inhibitor of dipeptidyl peptidase IV. Biochemistry 38, 11597–11603.
12. Conarello, S.L., Li, Z., Ronan, J., Roy, R.S., Zhu, L., Jiang, G.,
Liu, F., Woods, J., Zycband, E., Moller, D.E., Thornberry, N.A.
& Zhang, B.B. (2003) Mice lacking dipeptidyl peptidase IV are
protected against obesity and insulin resistance. Proc. Natl Acad.
Sci. USA 100, 6825–6830.
13. Polga
´
r, L. (1992) Structural relationship between lipases and
peptidases of the prolyl oligopeptidase family. FEBS Lett. 311,
281–284.
14. Notredame, C., Higgins, D.G. & Heringa, J. (2000) T-Coffee: a
novel method for fast and accurate multiple sequence alignment.
J. Mol. Biol. 302, 205–217.
15. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. &
Higgins, D.G. (1997) The CLUSTALX windows interface: flexible
strategies for multiple sequence alignment aided by quality ana-
lysis tools. Nucleic Acids Res. 25, 4876–4882.
16. Schmidt, H.A., Strimmer, K., Vingron, M. & von Haeseler, A.
(2002) TREE-PUZZLE: maximum likelihood phylogenetic ana-
lysis using quartets and parallel computing. Bioinformatics 18,
502–504.
17. Krogh, A., Larsson, B., von Heijne, G. & Sonnhammer, E.L.
(2001) Predicting transmembrane protein topology with a hidden
Markov model: application to complete genomes. J. Mol. Biol.
305, 567–580.
18. Eisenhaber, F., Eisenhaber, B., Kubina, W., Maurer-Stroh, S.,
Neuberger, G., Schneider, G. & Wildpaner, M. (2003) Prediction
of lipid posttranslational modifications and localization signals
from protein sequences: big-Pi, NMT and PTS1. Nucleic Acids
Res. 31, 3631–3634.
2714 J. I. Vena
¨
la
¨
inen et al.(Eur. J. Biochem. 271) Ó FEBS 2004
