Identification of NoxD/Pro41 as the homologue of the
2
p22phox NADPH oxidase subunit in fungi1
Accepted Article
1
3
4
Isabelle Lacaze, Hervé Lalucque, Ulrike Siegmundǂ, Philippe Silar and Sylvain Brun*
5 6
Univ Paris Diderot, Sorbonne Paris Cité, Institut des Energies de Demain, case courrier 7040 Lamarck,
7
75205 Paris cedex 13 France.
8
Univ Paris Sud, Institut de Génétique et Microbiologie, UMR8621, 91405 Orsay cedex, France.
9
ǂInstitut
für Biologie und Biotechnologie der Pflanzen, Westfälische Wilhelms Universität,
10
Schlossplatz 8, D-48143 Münster, Germany
11
*correspondence to:
[email protected] 12 13
Running
Title:
NoxD/Pro41:
the
fungal
p22phox
homologue
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/mmi.12876
1 This article is protected by copyright. All rights reserved.
Accepted Article
14 15 16
Abstract
17
NADPH oxidases (Nox) are membrane complexes that produce O2-. Researches in mammals, plants
18
and fungi highlight the involvement of Nox-generated ROS in cell proliferation, differentiation and
19
defense. In mammals, the core enzyme gp91phox/Nox2 is associated with p22phox forming the
20
flavocytochrome b558 ready for activation by a cytosolic complex. Intriguingly, no homologue of the
21
p22phox gene has been found in fungal genomes, questioning how the flavoenzyme forms. Using
22
whole genome sequencing combined with phylogenetic analysis and structural studies, we identify
23
the fungal p22phox homologue as being mutated in the Podospora anserina mutant IDC509. Functional
24
studies show that the fungal p22phox, PaNoxD, acts along PaNox1, but not PaNox2, a second fungal
25
gp91phox homologue. Finally, cytological analysis of functional tagged versions of PaNox1, PaNoxD
26
and PaNoxR shows clear co-localization of PaNoxD and PaNox1 and unravel a dynamic assembly of
27
the complex in the endoplasmic reticulum and in the vacuolar system.
28
2 This article is protected by copyright. All rights reserved.
Accepted Article
29
30
Introduction
31
NADPH oxidases (Nox) are membrane flavoenzymes that produce extracellular or lyzosomal
32
superoxide anion (O2-) by reduction of dioxygen, concomitantly with oxidation of cytosolic NADPH.
33
The first Nox to be discovered and extensively studied was the phagocyte oxidase gp91phox/Nox2,
34
whose mutations are prevalent in patients suffering from Chronic Granulomatous Disease (CGD). This
35
therefore tended at first to restrict the roles of Nox, and the ROS (Reactive Oxygen Species) they
36
produce, to the immune response. However, subsequent identification of other Nox isoforms in
37
mammals (Nox1, Nox3, Nox4, Nox5 and DUOX1/2) shed light on their multiple and crucial roles in
38
development, cell differentiation, cell proliferation, programmed cell death and cytoskeletal
39
remodeling (reviewed in Lambeth, 2004; Sumimoto, 2008). In resting phagocytes, gp91 phox binds to
40
the p22phox transmembrane protein, together forming the inactive flavocytochrome b558.
41
Requirement for p22phox also applies to Nox1, Nox3 and Nox4, but not to Nox5 and Duox1/2. These
42
latter isoforms bear EF-hand motifs and are regulated by Ca2+. While the flavoenzyme composed of
43
p22phox-Nox4 seems constitutively active in tissues, O2- production by p22phox-Nox1/2/3 isoforms
44
requires recruitment of additional cytosolic activating subunits. Paradigm of Nox activation relies on
45
gp91phox/Nox2 in phagosomes. Upon microbial infection, a complex composed of the cytosolic
46
subunits p47phox, p40phox, p67phox and the small Rho-GTPase Rac1/2 translocates to the membrane
47
through diverse post-translational modifications in order to activate NADPH oxidation by the gp91phox
48
catalytic subunit. Within the tissues where they are expressed, the different Nox isoforms harbor
49
multiple subcellular localizations. Nox complexes are found in phagosomes, endosomes (thereby
50
called redoxosomes), mitochondria, nuclei, Endoplasmic Reticulum (ER) and at the plasma
51
membrane (reviewed in Oakley et al., 2009; Ushio-Fukai, 2009; Laurindo et al., 2014). O2- superoxide
52
anions produced by Nox are short-lived ROS unable to cross biological membranes due to their
3 This article is protected by copyright. All rights reserved.
negative electric charge. Their confinement in specialized organelles provides tightly regulated
Accepted Article
53 54
intracellular signaling and prevents thus harmful production of ROS.
55
Ten years ago, the first fungal isoforms of Nox were discovered in Aspergillus nidulans (Lara-Ortíz et
56
al., 2003) and Podospora anserina (Malagnac et al., 2004), opening the field for stimulating
57
researches on Nox in fungi. This group of eukaryotes has achieved great evolutionary success,
58
enabling colonization of almost all known ecological niches. As a consequence, diversity among fungi
59
is huge and lifestyle strategies are quite diverse. Remarkably, the roles of Nox and their regulation in
60
fungi are well conserved (reviewed in Takemoto et al., 2007; Scott and Eaton, 2008; Aguirre and
61
Lambeth, 2010; Tudzynski et al., 2012). Fungi possess up to three Nox isoforms, Nox1 (also called
62
NoxA), Nox2 (or NoxB) and Nox3 (or NoxC). While Nox1 and 2 are phylogenetically and structurally
63
related to mammalian Nox2, Nox3 with its EF-hand Ca2+ binding motif is closer to animals
64
Nox5/Duox and to plant Rboh (Respiratory Burst Oxidative Homologues). Nox3 has been functionally
65
studied only in P. anserina where no phenotype was found after gene deletion (Brun et al., 2009). On
66
the contrary, as their metazoan counterparts, Nox1 and Nox2 fulfill a wide range of functions in
67
fungi. They are required for cell differentiation, including cell fusion (anastomosis) (Roca et al., 2012;
68
Kayano et al., 2013; Dirschnabel et al., 2014; Chan Ho Tong et al., 2014), building of fruiting bodies
69
during sexual reproduction (Lara-Ortíz et al., 2003; Malagnac et al., 2004; Giesbert et al., 2008; Cano-
70
Domínguez et al., 2008; Dirschnabel et al., 2014), germination of melanized ascospores (Malagnac et
71
al., 2008; Cano-Domínguez et al., 2008; Lambou et al., 2008a), defense (Silar, 2005; Montero-
72
Barrientos et al., 2011; Hernández-Oñate et al., 2012), plant penetration through appressorium in
73
phytopathogens and appressorium-like development in saprophytes (Egan et al., 2007; Segmüller et
74
al., 2008; Brun et al., 2009), degradation of cellulose (Brun et al., 2009), control of hyphal growth in
75
plant-symbiotic fungi (Tanaka et al., 2006) and cytoskeletal remodeling during polarized growth
76
(Dagdas et al., 2012; Ryder et al., 2013). Regulation of Nox appears remarkably well conserved
77
among fungi. The orthologue of p67phox, called NoxR, first discovered in Epichloe festucae, was shown
78
to activate Nox1 together in a complex with the polarisome protein BemA, the GEF protein Cdc24 4
This article is protected by copyright. All rights reserved.
and the Rho-GTPAse RacA (Tanaka et al., 2006; Takemoto et al., 2006; Tanaka et al., 2008; Takemoto
Accepted Article
79 80
et al., 2011; Kayano et al., 2013). NoxR was shown to also activate Nox in P. anserina, N. crassa,
81
Botrytis cinerea and Magnaporthe grisea. In the Nox1 activation model proposed by Takemoto et al.,
82
(2011), BemA/1, which shares the crucial PX, PB1 and SH3 domains with p47phox and p40phox,
83
substitutes these two mammalian adaptators in the complex, emphasizing similarities in Nox
84
activation between mammals and fungi. Intriguingly, to date no homologue of the p22 phox subunit
85
has been described in fungi. Moreover, in all fungi, Nox1 and Nox2 have clearly distinct functions and
86
what gives specificity in functions to those two catalytic subunits remains largely unknown. The
87
Tetraspanin Pls1 could be the key factor driving Nox2/B regulation (Lambou et al., 2008a). In
88
mammals, Tetraspanins are integral membrane proteins known to organize and to regulate function
89
of multi-molecular membrane complexes (reviewed in Yáñez-Mó et al., 2009). In P. anserina, M.
90
grisea and B. cinerea, deletion mutants of Pls1 and Nox2/B exhibit exactly the same phenotypes and
91
both BcNoxB and BcPls1 co-localize in B. cinerea (Clergeot et al., 2001; Lambou et al., 2008a;
92
Siegmund et al., 2013). Moreover, in all fungal genomes analyzed to date, Nox2/B and Pls1 are
93
always either both present or both absent (Brun et al., 2009), suggesting that they cooperate in the
94
same signaling complex. In fungi, Pls1 may thus substitute to p22phox during the assembly of the
95
flavoenzyme containing Nox2/B. What could specifically drive Nox1/A signaling remains unknown
96
and whether Nox1/A associates with a membrane protein as other gp91phox catalytic subunits
97
remains to be addressed.
98
The ascomycete fungus P. anserina is a powerful genetics model that is well suited to help
99
deciphering Nox signaling. The phenotypes of the PaNox1 mutants are clearly different from those of
100
the PaNox2 ones (Malagnac et al., 2004; Lambou et al., 2008b; Brun et al., 2009). The PaNox1
101
mutant IDC343 was identified in a screen designed to uncover genes regulating the development of
102
Crippled Growth (CG), an epigenetically-regulated mycelium degeneration phenomenon (Haedens et
103
al., 2005). The IDC mutants (Impaired in Development of Crippled Growth) share many phenotypic
104
alterations with IDC343, but not with PaNox2. The PaNoxR mutant is an exception because this 5
This article is protected by copyright. All rights reserved.
factor activates both PaNox1 and PaNox2 (Brun et al., 2009). This led us hypothesize that the genes
Accepted Article
105 106
inactivated in the other IDC mutants could specifically be involved in PaNox1 signaling. Interestingly,
107
IDC mutants may display slightly different phenotypes related to cellulose degradation. Notably, they
108
may or may not recover female fertility on cellulose and may or may not present defects in
109
appressorium-like development on cellophane.
110
In this study, we started to identify by genome sequencing the gene affected in a mutant
111
called IDC509, which was identified in the same screen as that of PaNox1 (Haedens et al., 2005).
112
Careful phenotypic analysis of the IDC509 mutant revealed perfect identity with those of the PaNox1
113
IDC343 mutant, suggesting that both genes work in concert. Phylogenetic and cytological analyses
114
supported the hypothesis that PaNoxD, the gene affected in IDC509 was the fungal p22phox.
115 116
Results
117
The IDC509 mutant has phenotypes identical to those of the IDC343 mutant affected in PaNox1
118
IDC mutants may display different phenotypes related to sexual development and cellulose
119
degradation. Careful phenotype analysis of the IDC509 mutant revealed perfect identity with those of
120
IDC343 mutant. As other IDC mutants, both IDC509 and IDC343 grew flat on standard medium (M2),
121
were not pigmented and were female sterile (Fig. 1A). However, IDC509 x IDC509 crosses were fertile
122
on media containing cellophane or paper, as IDC343 homozygous crosses (data not shown). Mosaic
123
and grafting analyses of the IDC509 mutant showed that the gene mutated in IDC509 is specifically
124
required in the peridium of perithecium during fruiting body development, like PaNox1 (Fig. S1)
125
(Silar, 2011). The cellulolytic capacity of the IDC509 mutant, as measured by paper pad weight loss
126
during mycelium growth, was enhanced as that of IDC343, correlatively with an increased O2-
127
production that can be visualized by an NBT staining assay on whole mycelia (Fig. S2). IDC509 was as
128
IDC343 defective in needle-like hyphae that were differentiated during penetration of cellophane (Fig.
6 This article is protected by copyright. All rights reserved.
1B), since IDC509 hyphae were able to reorient their growth towards cellophane and to efficiently
Accepted Article
129 130
establish contacts; however needle-like hyphae never emerged from the contact cushion. It has
131
recently been shown that Nox1 and NoxR mutants are impaired in cell fusion (Roca et al., 2012;
132
Kayano et al., 2013; Dirschnabel et al., 2014; Chan Ho Tong et al., 2014). As those of the PaNox1
133
mutant IDC343, IDC509 hyphae failed to fuse (Fig. 1C).
134
Pa_1_7250 is the gene mutated in IDC509
135
To identify the gene mutated in IDC509, we performed Illumina high throughput sequencing on the
136
IDC509 genome and compared the reads to the isogenic reference genome (Grognet et al., 2014) (see
137
Experimental procedures for IDC509 strain genome sequencing). Three mutations were detected. Only
138
one occurred in the coding sequence of a predicted gene (Pa_1_7250), leading to a premature stop
139
codon. The Pa_1_7250 coding sequence is 435 nucleotides long, interrupted by two introns and
140
encodes a 144 amino acids protein. In IDC509, a CAG to TAG transition replaced the Gln110 codon by a
141
stop (Fig. 2B). ESTs encompassing this gene are available (Espagne et al., 2008) showing that
142
Pa_1_7250 is expressed and confirming intron/exon predictions. Microarray analysis showed that
143
Pa_1_7250 was not differentially expressed during mycelium growth (Bidard et al., 2012) and during
144
fruiting body development (V. Berteaux-Lecellier and F. Bidard, personal communication). In the
145
following parts, Pa_1_7250 was renamed PaNoxD.
146
In order to assess if lack of functional PaNoxD was actually responsible for the phenotypes of
147
IDC509, the mutant was transformed with the pPaNoxD plasmid carrying the wild-type PaNoxD CDS
148
(Table S3). The pPaNoxD plasmid complemented the defects of the IDC509 mutant (See Experimental
149
procedures). Notably, female fertility was restored in 5 arbitrarily selected wild-type transformants
150
and, in one transformant selected for further phenotypic analysis, all the phenotypes tested were as
151
in the wild type (Fig. 1 & S2). This led us to conclude that the PaNoxD mutation was indeed
152
responsible for all the phenotypes observed in IDC509.
7 This article is protected by copyright. All rights reserved.
We then deleted PaNoxD by replacement of its CDS with a Hygromycin B-resistance cassette
Accepted Article
153 154
(see Experimental procedures). All the tested phenotypes of the PaNoxD∆ strain were identical to
155
those of IDC509 (Fig. 1 & S2). Therefore, IDC509 was a null mutant of the PaNoxD gene. We then
156
constructed the PaNoxD∆ IDC343 double mutant and determined that its phenotypes were identical to
157
those of each single mutant, i.e defects in vegetative mycelium and cellulolytic capacities (Fig S2 &
158
S6), anastomosis, fruiting body formation and appressorium-like development) (data not shown).
159
PaNoxD and SmPro41 are fungal homologues of mammalian p22phox
160
PaNoxD has been annotated as being the orthologue of SmPro41 of Sordaria macrospora, another
161
fungus belonging to the order Sordariales (Nowrousian et al., 2007). Interestingly, like PaNoxD,
162
SmPro41 is required for fruiting body development in S. macrospora. SmPro41 encodes a
163
transmembrane protein with no known function located at the ER, a location also found for Nox1 in
164
several fungi (Rinnerthaler et al., 2012; Siegmund et al., 2013). This intriguing coincidence led us to
165
consider that PaNoxD/Pro41 and PaNox1 proteins might act in a same complex. Strikingly, in silico
166
topology analysis of the PaNoxD protein using TMHMM yielded the same transmembrane domain
167
organization as p22phox (Fig. 2), the protein that binds p91phox/Nox2, forming the active
168
flavocytochrome b558 in mammals (Meijles et al., 2012). Indeed, despite divergent primary
169
sequences, both p22phox and PaNoxD have three transmembrane domains (Fig. 2), the first one
170
overlapping a potential secretion-signal (although in both cases contradictory predictions were
171
obtained with SignalP and PrediSi, as only the latter predicted a signal peptide). Moreover, a
172
conserved domain of about 30 amino acids was present downstream the third transmembrane
173
segment. Note that the C-terminus of p22phox that binds the p47phox protein was absent in
174
PaNoxD/SmPro41. In line with this, we could not detect any p47phox homologue in fungal genomes.
175
Because all the data shown in this paper point towards the homology between PaNoxD/SmPro41 and
176
p22phox (see above and below) and because evidence that BcNoxD directly binds to Nox1/A catalytic
177
subunit are brought by studies in the phytopathogenic fungus Botrytis cinerea (see companion paper
8 This article is protected by copyright. All rights reserved.
by Siegmund et al., n.d.), we called this new fungal homologue of p22phox, PaNoxD for PaNox1-
Accepted Article
178 179
Docking.
180
The next evidence that PaNoxD functions along with PaNox1 and could be the p22phox
181
component of the Nox complex came from phylogenetic analyses of p22phox and PaNoxD proteins in
182
fungi. While BLAST analysis did not allow retrieving p22phox homologues in the genome of P. anserina
183
and did not allow retrieving PaNoxD homologues in mammalian genomes, query of basal fungi
184
(Chytridiomycota) genomes clearly identified the same new potential homologues (Table S1). Query
185
with either p22phox, NoxD and the new proteins from Chytridiomycetes of additional fungal genomes
186
as well as that of unicellular organisms related to both animals and fungi, uncovered potential
187
homologues in many genomes (Table S1). Potential p22phox orthologues were present in the genomes
188
of unicellular organisms related to fungi, including in Rozella allomycis, a Cryptomycota, indicating
189
that presence of NoxD could be an ancestral character in fungi. In most fungal species, analysis
190
identified only a single NoxD homologue. Strikingly, apart from two exceptions in Wallemiales and
191
some Sporidobolales, both Nox1 and NoxD were either both present or both absent from fungal
192
genomes, including that of most Ascomycota and Basidiomycota (Table S1). Based on the known
193
phylogeny of fungi, we could estimate that both genes were lost together more than 10 times, which
194
leaves very little doubt that losses were independent. Interestingly, in some fungi without Nox1, the
195
Nox2 and/or Nox3 isoforms may remain, yet NoxD was lost from these genomes. Multiple alignment
196
(Fig. S3) showed that all these proteins were related, but showed a high sequence divergence,
197
explaining why orthology was not previously detected. Phylogenetic tree defined several groups of
198
p22phox/NoxD in relation with the known phylogeny of species (Fig. 2). All potential homologues had a
199
topology similar to that of P22phox/NoxD and in many of them a potential secretion signal could be
200
predicted by Predisi (but not SignalP).
201
PaNoxD locates both at the ER and the VS in P. anserina
9 This article is protected by copyright. All rights reserved.
With the aim to determine the sub-cellular localization of PaNoxD in P. anserina, we constructed two
Accepted Article
202 203
fusion proteins with GFP or mCherry fused in frame to the C-terminus of PaNoxD. Both chimaeric
204
proteins were put under the control of the upstream regulatory sequences of PaNoxD or of those of
205
the strong and constitutive promoter of the AS4 gene. All constructs were fully functional (data not
206
shown) and, except fluorescence intensity which was stronger in the strains overexpressing PaNoxD-
207
GFP and PaNoxD-mCherry, localization of the four chimaeric proteins was identical (see Experimental
208
procedures and Fig S5 for non-overexpressed conditions). For the sake of simplicity, only
209
observations made with the overexpressed strains are reported here. PaNoxD tagging displayed
210
different patterns presumably depending on maturation of hyphae. It was essentially punctuate and
211
more rarely reticulate and perinuclear (Fig. 3A & 3B). In the cortical area of hyphae articles (fungal
212
cells), the punctuate tagging pattern likely corresponded to vesicles either separated or fused in a
213
network running beneath the plasma membrane (Fig. 3A, 3C, 6C and 6D).
214
To get insights into PaNoxD localization as regards to the ER, we constructed a transgenic P.
215
anserina strain with its ER tagged with mCherry. To this end, we added both the predicted secretion-
216
signal sequence of PaNoxD at the N-terminus of mCherry and the “KDEL” ER-retention motif at the C-
217
terminus as in Nowrousian et al. (2007). The fluorescence pattern observed in mCherry-ER strains
218
was the one expected for an ER protein (Fig. 3A and 4A). Indeed, tagging appeared reticulate all
219
along hyphae, around nuclei (perinuclear) and punctuate especially beneath the plasma membrane
220
(cortical area), suggesting that mCherry-ER efficiently marked the ER from the nuclear envelope to
221
small vesicles possibly dedicated to intra cellular trafficking. This ER-tagging pattern validated the
222
prediction of the signal secretion peptide of PaNoxD. We also noted that the ER morphology changed
223
with maturation of hyphae. Indeed, while tagging was essentially reticulate and perinuclear in
224
hyphae of the leading edge (