Structural and functional analysis of the interaction
of the AAA-peroxins Pex1p and Pex6p
Ingvild Birschmann
1,
*
,
†, Katja Rosenkranz
2,
†, Ralf Erdmann
2
and Wolf-H Kunau
1
1 Abteilung fu
¨r Zellbiochemie, Medizinische Fakulta
¨t der Ruhr-Universita
¨t Bochum, Germany
2 Abteilung fu
¨r Systembiochemie, Medizinische Fakulta
¨t der Ruhr-Universita
¨t Bochum, Germany
Peroxisomes are ubiquitous, single membrane-bound
organelles involved in many metabolic pathways [1].
Thirty-two proteins required for peroxisome biogenesis
have been described [2–6]. They are encoded by PEX
genes and are collectively called peroxins. Most of the
peroxins are directly involved in the import process for
peroxisomal matrix and membrane proteins while oth-
ers are required for proliferation and inheritance of the
organelles [2]. The current model of peroxisome bio-
genesis suggests that peroxisomal proteins are synthes-
ized on free ribosomes and post-translationally
targeted to the organelle [7–9]. Targeting and trans-
location of newly synthesized peroxisomal matrix pro-
teins requires ATP and depends on peroxisomal
targeting signals [10], PTS1 and PTS2, and their
cognate receptors Pex5p and Pex7p, respectively
[11,12]. These two peroxins bind their cargo in the
cytoplasm, transport them to (and possibly across) the
peroxisomal membrane and return to the cytoplasm
for the next round of import. The membrane-bound
steps of this receptor cycle are currently under inten-
sive investigation [13,14].
PEX1 and PEX6 genes encode members of the
AAA family of ATPases, a large superfamily of pro-
teins involved in the ATP-dependent rearrangement of
protein complexes [15–17]. AAA-proteins are found in
all organisms and are essential for many activities, e.g.
cell cycle function, vesicular transport, mitochondrial
Keywords
AAA-proteins, peroxisomal biogenesis,
Pex1p, Pex6p, peroxin
Correspondence
Dr Ralf Erdmann, Institut fu
¨r
Physiologische Chemie, Abteilung fu
¨r
Systembiochemie, Medizinische Fakulta
¨t der
Ruhr-Universita
¨t Bochum, D-44780 Bochum,
Germany. Tel: +49 234322 4943
Fax: +49 234321 4266
E-mail: Ralf.Erdmann@ruhr-uni-bochum.de
*Present address
Institut fu
¨r Klinische Biochemie und Patho-
biochemie, Medizinische Universita
¨tsklinik,
D-97078 Wu
¨rzburg, Germany
†Both authors contributed equally to this
manuscript
(Received 21 July 2004, accepted 25 August
2004)
doi:10.1111/j.1432-1033.2004.04393.x
The AAA-peroxins Pex1p and Pex6p play a critical role in peroxisome bio-
genesis but their precise function remains to be established. These two
peroxins consist of three distinct regions (N, D1, D2), two of which (D1,
D2) contain a conserved 230 amino acid cassette, which is common to
all ATPases associated with various cellular activities (AAA). Here we
show that Pex1p and Pex6p from Saccharomyces cerevisiae do interact
in vivo. We assigned their corresponding binding sites and elucidated the
importance of ATP-binding and -hydrolysis of Pex1p and Pex6p for their
interaction. We show that the interaction of Pex1p and Pex6p involves
their first AAA-cassettes and demonstrate that ATP-binding but not
ATP-hydrolysis in the second AAA-cassette (D2) of Pex1p is required for
the Pex1p–Pex6p interaction. Furthermore, we could prove that the second
AAA-cassettes (D2) of both Pex1p and Pex6p were essential for peroxi-
somal biogenesis and thus probably comprise the overall activity of the
proteins.
Abbreviations
AAA, ATPases associated with various cellular activities; NSF, N-ethylmaleimide-sensitive fusion protein.
FEBS Journal 272 (2005) 47–58 ª2004 FEBS 47
functions and proteolysis [17,18]. AAA-proteins share
the presence of one or two AAA-cassettes, comprising
about 230 amino acids and are characterized by Wal-
kerA and B motifs for ATP-binding and ATP-hydro-
lysis [19,20]. Pex1p and Pex6p are functionally
nonredundant AAA-proteins required for the biogen-
esis of peroxisomes. A direct interaction of Pex1p and
Pex6p has been demonstrated in some organisms [21–
24] indicating that the two peroxins cooperate in per-
oxisome assembly. Sequence comparison of Pex1p and
Pex6p with other AAA members indicates that the two
AAA-peroxins belong to an AAA-protein subfamily
characterized by a tripartite structure. These proteins
possess large N-terminal regions followed by two AAA-
cassettes. In N-ethylmaleimide-sensitive fusion protein
(NSF), the best characterized AAA-protein, distinct
functions could be assigned to its three parts [25,26].
For yeast and human Pex6p it has been reported that
the N-terminal region of this protein functionally inter-
acts with the yeast peroxin Pex15p or human Pex26p,
respectively [3,27]. However, despite these findings and
the fact that both AAA-peroxins are conserved from
yeast to humans their function in peroxisome biogenesis
remains unknown.
In this study we mapped the mutual binding sites
and assayed the effects of deletion and point mutations
in Pex1p and Pex6p for the interaction of the proteins
and for their overall function in peroxisome biogenesis.
Our results demonstrate a different role of the two
AAA-cassettes for the Pex1p–Pex6p interaction and
their functional role in peroxisome biogenesis.
Results
The contribution of Pex1p and Pex6p to peroxisomal
biogenesis is well established on the basis of the pheno-
types of the corresponding null mutants. Both pex1D
[24,28] and pex6D[29,30] mutants show a characteristic
pex phenotype with only residual, ghost-like peroxi-
somal structures and mislocalization of peroxisomal
matrix proteins to the cytosol. Here we confirm and
extend earlier studies of these AAA-peroxins and give
a further detailed functional analysis of their cassette
structure and interaction.
The interaction of Pex1p and Pex6p involves
their first AAA-cassettes (D1)
Pex1p and Pex6p have been shown to interact in Pichia
pastoris,Hansenula polymorpha and human [21–24]. To
further limit the corresponding binding regions, we first
confirmed the interaction of the two proteins in bakers
yeast. For this purpose, we constructed plasmids carry-
ing PEX1 fused to the coding sequence of the activation
domain of Gal4p (PEX1-GAL4-AD) or PEX6 linked to
the coding region of the DNA binding domain of Gal4p
(PEX6-GAL4-BD). The two-hybrid reporter strain
PCY2 was cotransformed with the two plasmids. Acti-
vation of the reporter gene lacZ, indicated by blue
colonies on X-Gal medium was observed, when PEX6-
GAL4-BD was coexpressed with PEX1-GAL4-AD
(Fig. 1A,B). Transformation of either of these two plas-
mids alone did not lead to an activation of the reporter
gene. The same results were obtained when the reporter
strain HF7c was used to assay for histidine prototrophy
(data not shown). These data demonstrate that ScPex1p
interacts with ScPex6p. The two-hybrid studies did not
provide an indication for homo-oligomerization of the
two AAA-peroxins when PEX1-GAL4-BD was coex-
pressed with PEX1-GAL4-AD or PEX6-GAL4-BD was
coexpressed with PEX6-GAL4-AD (data not shown).
This finding in turn is in agreement with the assumption
that these two AAA-peroxins fulfill their function in
peroxisome biogenesis as a transient or stable hetero-
meric protein complex. To investigate the influence of
the recent published interaction between Pex6p and
Pex15p [27], we carried out the two-hybrid assay in the
absence of Pex15p (PCY2 pex15D). Pex1p and Pex6p
still showed interaction in pex15Dindicating that the
association of the two proteins does not depend on
Pex15p (data not shown).
To identify the regions in Pex1p and Pex6p that are
responsible for their interaction, truncated versions of
the proteins were tested for interaction in the yeast
two-hybrid system. According to the predicted domain
structure, the Pex1p sequence was divided into three
parts: the N-terminal region (N, aa1–400), the first
AAA-cassette (D1, aa394–681) and the second AAA-
cassette (D2, aa669–1043). DNA fragments encoding
these parts of Pex1p were fused to the GAL4-AD and
coexpressed with PEX6-GAL4-BD in PCY2.As
judged by the lack of lacZ gene expression, none of
these cassettes alone can mediate binding to Pex6p
(Fig. 1A). However, a Pex1p fragment consisting of
the N-terminal part and the first AAA-cassette
(N + D1, aa1–681) gave rise to a low but significant
lacZ gene activation (Fig. 1A). The fragment compri-
sing both AAA-cassettes (D1 + D2), however, led to
ab-galactosidase activity comparable to that of the
full-length Pex1p (Fig. 1A). These findings indicate
that binding of Pex6p is best in the presence of both
AAA-cassettes of Pex1p. However, the data also indi-
cate that the first AAA-cassette together with the
N-terminal region or the second cassette of Pex1p is
already sufficient for interaction of the two AAA-
peroxins. Consequently, these results demonstrate the
Domain function of Pex1p and Pex6p I. Birschmann et al.
48 FEBS Journal 272 (2005) 47–58 ª2004 FEBS
critical role of the first AAA-cassette of Pex1p for the
binding of Pex6p. Our data suggest that the binding
site for Pex6p is comprised within the first AAA-
cassette (D1) of Pex1p. Binding efficiency can be
increased by the additional presence of either the
N-terminal fragment or more drastically by the pres-
ence of the second AAA-cassette. This is consistent
with the observation that a mutation in the beginning
of Pex1pD2 (G843D in human, corresponding to
G700 in yeast) attenuates the interaction of the two
AAA-peroxins [23].
To identify the Pex6p binding site for Pex1p, we tes-
ted the N-terminal region (N, aa1–428), the first AAA-
cassette (D1, aa421–716) and the second AAA-cassette
(D2, aa704–1030) and all possible combinations for
interaction with Pex1p in the yeast two-hybrid system.
The corresponding PEX6 fragments were fused to
GAL4-BD and coexpressed with PEX1–GAL4-AD.
The results shown in Fig. 1B demonstrate that an acti-
vation of the lacZ reporter gene occurred only in dou-
ble transformants carrying PEX1–GAL4-AD and the
construct coding for the Pex6p region comprising the
N-terminal fragment and the first AAA-cassette of
PEX6 (PEX6N + D1–GAL4-BD). Neither of the cas-
settes alone nor the two AAA-cassettes together or the
N-terminal part together with the second cassette led
to an activation of the reporter gene and thus is suffi-
cient to maintain the interaction with Pex1p. These
findings demonstrate the importance of the first AAA-
cassette together with the N-terminal region of Pex6p
for the described interaction.
To confirm these results, we carried out coimmuno-
precipitation with different genomic integrated con-
structs: Pex1pN–ProteinA, Pex1pN + D1–ProteinA,
Pex1p–ProteinA, Pex6pN–ProteinA, Pex6pN + D1–
ProteinA, Pex6p–ProteinA. For this purpose, we
generated strains that express the proteins of interest,
C-terminally fused to two IgG-binding domains derived
from Staphylococcus aureus protein A (ProtA). An
advantage of these fusion proteins was the possibility
to detect even the truncated constructs via the ProtA-
tag. Moreover, as expression of the constructs was
under the control of the endogenous promoters,
overexpession was avoided, which has been shown to
A
BD
C
Fig. 1. Coimmunoprecipitation and two-hybrid interaction of Pex1p and Pex6p. A schematic representation of the Pex1p and Pex6p con-
structs which were analyzed in the two-hybrid assay (A,B) or by coimmunoprecipitation (C,D) is shown on the left. Analysis of PCY2 trans-
formants expressing the indicated fusion proteins of (A) Pex1p truncations and full-length Pex6p or (B) Pex6p truncations and full-length
Pex1p. The interactions were analyzed for b-galactosidase activity by filter assays with X-gal as the substrate. Three independent double-
transformations are shown. Extracts of oleic acid-induced wild-type cells expressing (C) Pex1p and Pex1p-truncations or (D) Pex6p and
Pex6p-truncations fused to ProteinA were immunoprecipitated with anti-IgG and immunoblotted with the same antisera (C,D), or antibodies
to Pex1p (D) or Pex6p (C). As a control, wild-type cells expressing no ProteinA fusion protein were treated equally (C,D; lane 4).
I. Birschmann et al.Domain function of Pex1p and Pex6p
FEBS Journal 272 (2005) 47–58 ª2004 FEBS 49
affect peroxisome biogenesis [31–33]. To ascertain
whether the ProtA fusion proteins were functional or
not, we investigated the ability of the strains to grow
on medium containing oleic acid as the sole carbon
source and monitored for the correct proliferation of
peroxisomes via the analysis of cell morphology by
electron microscopy. We observed that both full
length proteins Pex1p and Pex6p fused to ProtA
were functional in vivo, whereas the four deleted
strains (Pex1pN–ProtA, Pex1pN + D1–ProtA, Pex6pN–
ProtA, Pex6pN + D1–ProtA) showed neither growth
on oleic acid nor morphologically detectable peroxi-
somes (data not shown).
ProtA fusion proteins in the eluates were detected
by immunoblot analysis with anti-IgG (Fig. 1C,D).
The different migration behaviour reflects the different
sizes of the fusion proteins. Immunological analysis
revealed the presence of Pex6p in the precipitate of
full-length Pex1p–ProtA and also the presence of
Pex1p in the full-length Pex6p–ProtA-precipitate
(Fig. 1C,D). These data confirm the in vivo interaction
of the two proteins. We tested two truncations for
their interaction behaviour, the N-terminal region
and the N-terminal region together with the first AAA-
cassette. Attempts to express ProtA fusion constructs
comprising the first and the second cassette were not
successful. Neither Pex1p nor Pex6p was present in the
precipitates of the N-terminal region of Pex6, or
Pex1p, respectively (Fig. 1C,D). These data indicate
that the N-terminal fragment alone does not interact
with the corresponding binding partner. However, a
significant amount of Pex1p and Pex6p was detected in
the precipitates of the N + D1 fusions (Fig. 1C,D).
These data support the two-hybrid results and indicate
that D1 of both Pex1p and Pex6p contributes to the
binding of the two proteins. Truncation of either pro-
tein, however, did result in a significant decrease in the
coimmunoprecipitation of the binding partner. These
data might indicate that the contact sites between the
two proteins are not limited to the first cassette but
also comprise regions of the N-terminal fragment or
the second cassette, or that the binding site is com-
prised by D1 but its binding capability is significantly
enhanced in the presence of the N-terminal fragment
or D2.
ATP-binding but not ATP-hydrolysis in the
second AAA-cassette (D2) of Pex1p is required
for the Pex1p–Pex6p interaction
Pex1p and Pex6p both contain two AAA-cassettes and
thus two consensus ATP-binding sites. Typically,
Walker-type nucleotide-binding sites consist of two
conserved motifs. The WalkerA motif is essential for
nucleotide-binding, while the WalkerB motif is
required for hydrolysis of the nucleotide. To investi-
gate the influence of the binding and ⁄or hydrolysis of
ATP on the interaction of the AAA-peroxins, Pex1p
and Pex6p carrying mutated ATP-binding sites were
tested for interaction with the binding partner in the
yeast two-hybrid system. In the first set of mutants,
the lysine residue in the GXXGXGKT sequence of
either one of the two WalkerA motifs was replaced by
a glutamate or alanine residue which led to Pex1pA1-
(K467E), Pex1pA2(K744E), Pex6pA1(K489A) and
Pex6pA2(K778A). These invariant lysine residues have
been shown to be essential for the biological activity of
a number of ATP- and GTP-binding proteins and their
replacement yields proteins with significantly reduced
ATP-binding capacity [34]. Such an impairment of the
biological activity has also been reported for Pex1p
and NSF [19,35,36]. Similarly, we also introduced
point mutations of amino acids in the ATP-hydrolysis
sites (conserved sequence is four hydrophobic amino
acid D ⁄E) to investigate their influence on the des-
cribed interaction. We changed the conserved aspartate
of the WalkerB motifs of Pex1p and Pex6p into gluta-
mine leading to Pex1pB1(D525Q), Pex1pB2(D797Q)
and Pex6pB2(D831Q) [37]. It was not necessary to cre-
ate a B1 point mutation because wild-type Pex6p
already contains an alanine instead of the critical
aspartate (aa548), strongly suggesting that D1 of
Pex6p can bind but not hydrolyse ATP.
DNA fragments encoding the entire Pex1p or Pex6p
harboring the different mutations in the AAA-cassettes
were fused to GAL4-AD ⁄GAL4-BD and coexpressed
with PEX6–GAL4-BD or PEX1–GAL4-AD, respect-
ively (Fig. 2). Compared to wild-type Pex1p or Pex6p,
only the Pex1pA2 mutation resulted in a significantly
less efficient interaction with Pex6p as judged by the
decreased activation of the lacZ gene (Fig. 2A). Neither
a mutation in Pex1pA1 nor in Pex6pA1 or Pex6pA2 had
any influence on the Pex1p–Pex6p interaction (Fig. 2).
Moreover, the mutations of the WalkerB motif did not
affect the interaction of the AAA-peroxins (Fig. 2). All
constructs used for two-hybrid analyses were over-
expressed and showed the same protein level, demon-
strating that the described effects were not a result of
different expression levels (data not shown).
These results give rise to the notion that ATP-bind-
ing to D1 of Pex1p and to D1 and D2 of Pex6p as
well as the capability for ATP-hydrolysis in general is
dispensible for the Pex1p–Pex6p interaction. However,
the results also clearly demonstrate that the ability of
the second AAA-cassette of Pex1p to bind ATP is
required for the interaction of Pex1p and Pex6p.
Domain function of Pex1p and Pex6p I. Birschmann et al.
50 FEBS Journal 272 (2005) 47–58 ª2004 FEBS
The second AAA-cassettes (D2) of Pex1p and
Pex6p are essential for peroxisomal biogenesis
To investigate the effects of the described WalkerA
and WalkerB mutants of the ATP-binding sites for the
function of Pex1p and Pex6p in peroxisomal biogen-
esis, we analyzed the functional and morphological
phenotypes of the transformed mutants in further
detail.
Cells deficient in Pex1p or Pex6p are characterized
by the inability to grow on oleic acid as the single car-
bon source, mislocalization of matrix proteins to the
cytosol and the absence of morphologically detectable
matrix-filled peroxisomes [21,24,28,30,35,38]. First, we
tested different mutant constructs for their ability to
complement the oleic acid-growth phenotype of the
pex1 or pex6 null mutant. As demonstrated in Fig. 3,
null mutants expressing Pex1p mutated at the first
ATP-binding site (Pex1pA1, Pex1pB1) showed the
wild-type phenotype with respect to growth on oleic
acid medium. The same has been shown for Pex6p
(Pex6pA1) [27]. Complementation of the mutant
strains is indicated by growth on oleic acid medium
(YNO)-agar plates, which gives rise to a typical halo
reflecting the consumption of oleic acid. Interestingly,
the null mutants expressing Pex1p or Pex6p mutated
at WalkerA or WalkerB of the second ATP-binding
site were not able to grow on oleic acid as sole carbon
source (Fig. 3; [27,35]), indicating that both ATP-bind-
ing and ATP-hydrolysis at the conserved D2 is
required for Pex1p and Pex6p function in peroxisomal
biogenesis. This is also supported by the ultrastructural
appearance of the corresponding mutants (Figs 4 and
5). Oleic acid-induced pex1D(Fig. 4) or pex6D(Fig. 5)
mutant cells transformed with plasmids encoding
Pex1pA2 (Fig. 4E), Pex1pB2 (Fig. 4F), Pex6pA2
(Fig. 5E), or Pex6pB2 (Fig. 5F) are characterized by
the absence of morphologically detectable peroxisomes
and thus exhibit the same phenotype of the corres-
ponding pex1 (Fig. 4B) and pex6 (Fig. 5B) null
mutants, indicative of no complementation. Peroxi-
somes reappear, however, upon complementation of
pex1Dor pex6Dstrains with the wild-type Pex1p
(Fig. 4A) or Pex6p (Fig. 5C) proteins or proteins har-
boring mutations in the first AAA-cassette [Pex1pA1
(Fig. 4C), Pex1pB1 (Fig. 4D), Pex6pA1 (Fig. 5D)].
These results are consistent with an essential role of
the conserved AAA-cassette for Pex1p and Pex6p func-
tion in peroxsiome biogenesis. The results also demon-
strate that ATP-binding and ATP-hydrolysis at the
conserved AAA-cassette is required for the biological
function of Pex1p and Pex6p.
A
B
Fig. 2. Two-hybrid interaction of Pex1p and Pex6p harboring point
mutations of the WalkerA and B motifs of their ATP-binding sites.
A schematic representation of the Pex1p and Pex6p mutants which
were tested for two-hybrid interaction is shown on the left. PCY2
transformants expressed the indicated fusion protein combinations
of (A) Pex1p mutations and full-length Pex6p or (B) Pex6p muta-
tions and full-length Pex1p. The interactions were analyzed for b-ga-
lactosidase activity by filter assays with X-gal as substrate and
three independent double-transformations are shown.
Fig. 3. Effects of point mutation of the WalkerA and B motifs of
the ATP-binding sites on the complementation activity of Pex1p.
Growth behaviour on oleic acid medium (YNO) was analyzed for
wild-type,pex1Dand pex1Dexpressing genes encoding wild-type
or indicated point mutated Pex1p. Complementation is indicated by
growth on YNO-agar plates, which gives rise to a typical halo
reflecting the consumption of oleic acid.
I. Birschmann et al.Domain function of Pex1p and Pex6p
FEBS Journal 272 (2005) 47–58 ª2004 FEBS 51