The Aspergillus nidulans Peripheral ER: Disorganization by ER Stress and Persistence during Mitosis

The genetically amenable fungus Aspergillus nidulans is well suited for cell biology studies involving the secretory pathway and its relationship with hyphal tip growth by apical extension. We exploited live-cell epifluorescence microscopy of the ER labeled with the translocon component Sec63, endogenously tagged with GFP, to study the organization of ‘secretory’ ER domains. The Sec63 A. nidulans ER network includes brightly fluorescent peripheral strands and more faintly labeled nuclear envelopes. In hyphae, the most abundant peripheral ER structures correspond to plasma membrane-associated strands that are polarized, but do not invade the hyphal tip dome, at least in part because the subapical collar of endocytic actin patches constrict the cortical strands in this region. Thus the subapical endocytic ring might provide an attachment for ER strands, thereby ensuring that the growing tip remains ‘loaded’ with secretory ER. Acute disruption of secretory ER function by reductive stress-mediated induction of the unfolded protein response results in the reversible aggregation of ER strands, cessation of exocytosis and swelling of the hyphal tips. The secretory ER is insensitive to brefeldin A treatment and does not undergo changes during mitosis, in agreement with the reports that apical extension continues at normal rates during this period.


Introduction
A remarkable aspect of the biology of filamentous fungi is their ability to grow exclusively by apical extension. Aspergillus nidulans is a useful experimental system to study cell biological processes due to its amenability to classical and reverse genetic analyses and to its suitability to live microscopy studies [1]. A fundamental question is how growth of hyphal tip cells is maintained sustainably. Such polarity maintenance [2] is critically dependent on exocytosis. Exocytic carriers fueling tip growth are mainly, if not exclusively, targeted to the apex [3,4,5]. Indeed hyphal tip cells contain, underneath the apical plasma membrane, an idiosyncratic structure denoted the Spitzenkörper, where secretory carriers accumulate before their fusion with the plasma membrane [6,7,8].
The spatial organization of the classical secretory pathway reflects the needs imposed by the polarization of exocytosis. Both the late and the early Golgi cisternae show strong polarization, predominating ahead of the leading nucleus [9,10,11]. In agreement with the accepted view that early Golgi cisternae form by coalescence of ER-derived COPII carriers [12], ER exit sites are also polarized, as it is for the ER in another Aspergillus species [13], although to a lesser extent than the Golgi [9]. Thus, assuming that the distribution of the ER ultimately contributes to the overall polarization of the secretory pathway, the key issue is what determines the ER polarization. In Saccharomyces cerevisiae, the ER includes two different 'domains' [14], namely nuclear envelopes (NEs) and 'peripheral' ER, the latter including all ER strands excepting the NEs. The largest 'subdomain' of the peripheral ER is composed of plasma membrane (pm)-associated cortical ER sheets [15]. Hyphal tip growth is maintained during mitosis [16], indicating that the exocytic compartments remain functionally active during this period. Indeed the Golgi is not disorganized during mitosis [9], indicating that ER function is sufficiently preserved. As the A. nidulans nuclear envelopes (NEs) remain intact during mitosis [17,18], mitotic ER function is attributable, at least in part, to the NE domains; however, the potential changes undergone by peripheral ER domains have not yet been analyzed. In their seminal studies of the ER in filamentous fungi, carried out with the basidiomycete Ustilago maydis, Wedlich-Soldner et al. used an ER-targeted GFP construct carrying a HDEL ER retention signal. This luminal fluorescent protein probe labeled the ER as well as vesicles mediating early-Golgi-to-ER recycling [19], and thus is not ideally suited to study secretory ER domains.
Therefore, to study the A. nidulans ER secretory domains we chose a component of the ER translocon. The term 'translocon', coined by Walter and Lingappa [20], denotes the 'pores' and the molecular engines mediating the delivery of eukaryotic secretory proteins synthesized in the cytosol into the ER lumen. Studies in S. cerevisiae have been instrumental to understand the mechanistic and genetic basis of translocation. These studies showed that the composition of the translocon varies depending on whether translocation occurs co-translationally or post-translationally. A core component of all translocons is the Sec61 heterotrimer, composed of Sss1, Sbh1 and the pore-forming subunit Sec61 [21]. Post-translational translocation, the pathway that is seemingly followed by a larger fraction of proteins in fungi [21], is mediated by the so denoted 'SEC complex' containing, in addition to the Sec61 heterotrimer, the Sec62/63 complex including Sec62, Sec63, Sec71 and Sec72 [22,23]. Co-translational translocation requires, in vitro, the Sec61 complex and the signal recognition particle (SRP) receptor [24]. However Sec63 appears to be involved in all translocon complexes in vivo, a conclusion strongly supported by the finding that a SEC' complex containing all SEC components excepting Sec62 is required for co-translational translocation [25]. In view of this and of the fact that Sec63 is an integral membrane protein localizing to membrane domains, not to the ER lumen, we set out to study secretory ER domains using Sec63 as marker.
Here we show that the ER structure is dramatically altered by reductive stress, correlating with cessation of exocytosis. We show that the secretory ER, like NEs, remains intact through mitosis. Lastly we show that F-actin plays a key role in maintaining the polarization of the ER, which is characteristically associated with the subapical ring of endocytic patches.
Microscopy. Germlings and hyphae were cultured at 28uC in Lab-Tek chambers (Thermo Fischer Scientific; 0.3 ml of medium per well) at 25-28uC in pH 6.5 'watch minimal medium' (WMM) [32] containing 0.1% glucose and 5 mM ammonium tartrate as sole carbon and nitrogen source, respectively. When indicated, dithiothreitol (DTT) was added at 8 mM to induce UPR. Latrunculin B was added at 100 mM to depolymerize F-actin [9]. Brefeldin A was used at 300 mg/ml [9]. Benomyl was used at 4.8 mg/ml, a concentration that led to depolymerization of MTs within 10 min, as determined with a control strain expressing GFP-a-tubulin (TubA) [16].
Images were acquired with a Leica DMI6000B inverted fluorescence microscope driven by MetamorphH software (Molecular Dynamics), coupled to an ORCA ER-II (Hamamatsu) camera. Excitation of fluorescent proteins was achieved with an external light source equipped with a metal halide lamp for epifluorescence excitation (Leica, EL6000). The microscope was also equipped with GFP (Semrock GFP-3035B) and mRFP/ mCherry (Semrock TXRED-4040) filter sets, and a HCX 6361.4 NA objective (Leica 11506187). To minimize photodamage, the excitation light attenuator was set to the minimal position giving a reasonable intensity of fluorescence. Maximal intensity projections were obtained from z-stacks of images that had been contrasted with the 'unsharp mask' filter of Metamorph (settings: filter width, 9 pixels; scaling factor 0.75; result scale 2) or deconvolved using Huygens Software (Hilversum, The Netherlands). Time-lapse sequences (5 dimensions: x, y, z, t, red and green channels) were obtained from maximal intensity projections of deconvolved single time point z-stacks. Images were converted to 8-bit greyscale (and usually shown in inverted contrast) or to 24-bit RGB, and annotated with Corel Draw (Corel). Time-lapse sequences were converted to QuickTime using ImageJ 1.37 (http://rsb.info.nih. gov/ij/).

Morphology of the A. nidulans ER Visualized with Endogenously Tagged Sec63
To visualize the A. nidulans ER we GFP-tagged endogenously Sec63, a key component of the protein complex mediating translocation of newly synthesized proteins into the ER [33]. Sec63 corresponds to AN0834 (www. aspgd.org). Strains carrying sec63-gfp displayed wild-type growth, showing that the fusion protein is functional. z-stacks of Sec63-GFP hyphal tip cell images, processed with a deconvolution algorithm to remove out-of-focus signal, provided a very detailed view of the ER. Sec63-GFP localizes to a network of 'strands' corresponding to the different ER 'domains' reported in yeast [14], namely nuclear envelopes (NEs) and 'peripheral' ER, including pm-associated ER strands and a faint network of interconnecting tubules (Figure 1). NEs were only faintly labeled with Sec63, which contrasts with the strong labeling of NEs by Erg24 reported previously [34]. Like in S. cerevisiae [15], cortical strands associated to the plasma membrane represented the largest peripheral ER subdomain ( Figure 1). These cortical strands were polarized, as described for A. oryzae [13] and for dikaryotic hyphae of U. maydis [19]. Peripheral ER strands were excluded from the apical dome ( Figure 1). In the tip region the ER network was frequently capped by a finger-like protrusion (arrowed in Figure 1) directed towards, but not reaching the apex (apex indicated by an asterisk).
The relative importance of the different domains varied depending on whether short germlings or long hyphae were examined ( Figure 2). In conidiospores growing isotropically, and in germlings photographed shortly after establishing polarity, the ER appeared as a faint reticulate network in which NEs were hardly noticeable without the aid of a nuclear marker (Figure 2, A and B). NEs became more visible after nuclei moved into the germtube ( Figure 2B). At approximately this stage, strands associated to the plasma membrane became prominent in the proximity of the tip; more so as apical extension proceeded ( Figure 2C, C1 through C6), until reaching the stage of long hyphae. In tip cells of long hyphae, polarized pm-associated strands became the most abundant Sec63-containing ER structures (Figures 1 and 2D). These strands do not overlap with the plasma membrane labeled with FM4-64 ( Figure 2E). Given that long hyphae grow ,5 times faster than germlings [16], and that cortical strands generally correspond to ER sheets enriched for poly-ribosomes and translocation complexes [14], the prominence and polarization of pm-associated ER strands in hyphal tip cells possibly reflects the dependence that these have on secretion for rapid apical extension. Video S1 shows the dynamics of the ER network over a 30 min period (time resolution, 2 frames/min), illustrating the prominence of the polarized pm-associated strands as growth by apical extension proceeds. Video S2 shows a high time-resolution series (2.5 frames/sec over a 30 sec period) displaying the shortterm dynamics of the ER strands.

The Subapical Actin Collar Prevents the ER from Invading the Apical Dome
The organization of the ER in the tip region was consistent with a physical barrier, perhaps elements of the cytoskeleton, impeding strands to move forward. We tested this idea with anti-microtubule (MT) and anti-actin treatments. Benomyl (used at 4.8 mg/ml) led to the rapid (within 10 min) depolymerization of MTs, as determined with a control strain expressing TubA-GFP ( Figure 3A). Under equivalent conditions, benomyl affected the structure of the ER network, which was less organized. However, ER strands maintained their polarization, accumulating in the tip region, ahead of the leading nucleus ( Figure 3B; note the characteristic loss of directionality resulting from MT depolymerization in the benomyl-treated hyphal tip cell [16]).
In contrast, the effect of anti-actin treatment was striking. Upon incubation with latrunculin B, the ER underwent major reorganization, adopting a more regular reticulate structure connecting internal and cortical strands (Figure 4, panels A and B). This result suggested strongly that F-actin scaffolds the ER structure. More importantly, the ER strands invaded the swollen latrunculin Btreated tips, and the apical ER protrusion was no longer seen, suggesting that the organization of the ER in the tip region involves a 'barrier' dependent on F-actin (Figure 4, B2). An obvious candidate to form this barrier was the subapical endocytic collar of actin patches [3,30,35]. Thus, we co-imaged Sec63-GFP with the actin patch marker AbpA-mRFP. These images showed that the position of the endocytic collar is consistent with this interpretation ( Figure 4C). Moreover, the ER protrusion and the endocytic collar showed complementary shapes in maximal intensity projections, strongly indicating that the endocytic collar actually constricts the ER ( Figure 4C). Therefore, the cluster of F- actin-containing patches forming the endocytic collar, or structures associated to it, likely act(s) as a physical barrier excluding the ER from the apical dome.

Secretion Arrest by DTT-induced UPR
DTT has been shown to result in ER stress and unfolded protein response (UPR) in yeast and filamentous fungi, including A. nidulans [36,37,38,39]. We asked whether we could detect ER stress morphologically at the subcellular level, exploiting that activation of the UPR should block ER exit, thus leading to an arrest in secretion. The v-SNARE SynA Snc1 is a useful marker for secretion because SynA Snc1 traveling with secretory carriers critically contributes to the tetrahelical SNARE bundle mediating docking and fusion of these carriers with the plasma membrane. Like its Saccharomyces cerevisiae orthologues Snc1p, SynA Snc1 is strongly polarized by endocytic recycling, localizing to the apical dome and to the Spitzenkörper [3,10,31,40]. Impairment of the secretory function of the ER would be expected to result in tip swelling and mislocalization of SynA Snc1 , resembling the effects of the Golgi-disorganizing drug brefeldin A [9]. Figure 5 demonstrates that this is indeed the case. DTT treatment resulted in tip swelling and detectable SynA Snc1 mislocalization within 15 min, both effects being even more marked after 45 min. Tip growth was restored, as were the normal morphology of the tip and the normal localization of SynA Snc1 (to an apical crescent and the Spitzenkörper), after washing out DTT ( Figure 5). Collectively these data show that induction of UPR results in a sharp yet reversible arrest in secretion.

UPR Induces ER Aggregation
Next, we examined the effects of DTT treatment on the ER structure. Marked disorganization of the pm-associated ER at the tip was noticeable after 15 min of treatment, a time at which tip swelling was already conspicuous ( Figure 6A). Following a 30-min incubation with DTT, abnormal aggregates of ER membranes became noticeable in the hyphal tips (Fig. 6A). After 45 min, aggregates also formed in regions distant to the tip (Fig. 6A,  arrowheads), indicating that the extent of ER aggregation depends on the time of incubation in reductive stress conditions. In some examples, examination of individual planes of Z-stacks revealed that ER strands in these aggregates formed donought-like structures ( Figure 6B). All these changes were reversible upon washing out the reducing agent, demonstrating that this treatment does not result in cell death. Of note, the NE sheets appeared unaffected by DTT/UPR (red arrows; Figure 6A). Thus, DTTinduced UPR results in a subcellular response leading to the    The Aspergillus nidulans Peripheral ER PLOS ONE | www.plosone.org collapsing and aggregation of the cortical ER but not of NEs. This is consistent with the view that cortical ER strands in which Sec63 (i.e., the translocon complexes) predominates correspond to the most active ER subdomain in terms of secretion.
In contrast to the marked effects of DTT, we demonstrated that the Golgi-targeting drug brefeldin A (BFA) does not affect the ER. BFA treatment leads to aggregation of Golgi cisternae into larger 'brefeldin bodies' (Fig. 7; note the aggregation of late Golgi cisternae labeled with mRFP-PH OSBP ) [9,10,41]. However, the ER, both NE and peripheral, was unaffected (Fig. 7), even though the hyphal tips swelled due to secretion impairment. Thus, the A. nidulans ER is BFA-resistant and the effects of BFA are restricted to the Golgi.

The Peripheral ER does not Undergo Changes during Mitosis
The fate of NE sheets during A. nidulans mitosis has been extensively studied by Ukil et al., who used fluorescent Erg24 (a 'general' ER marker) to show that a double NE constriction generates, in addition to the two daughter nuclei, a nucleolar remnant that is later disassembled [34]. To examine the fate of peripheral ER subdomains during mitosis, we collected time-lapse series of cells coexpressing Sec63-GFP with HhoA HH1 -mCherry, to label chromatin. As noted above, due to their relatively faint staining with Sec63 the NEs of nuclei undergoing mitosis were hardly traceable against the background of peripheral strands ( Figure 8A; NEs arrowed when possible). In contrast, pmassociated strands enriched in translocon (i.e., Sec63) complexes were very conspicuous all throughout mitosis. Figure 8B (Video S3) shows a hyphal tip cell undergoing mitosis, growing at 0.78 mm/min. The kymograph shows that the rate of apical extension does not change even though the three nuclei shown undergo mitosis. The sequence clearly shows that the peripheral ER and, in particular, the polarized pm-associated strands, are not affected to any detectable extent by mitosis ( Figure 8B).
Polarity maintenance requires that the growing regions of hyphal tip cells be populated by secretory organelles. Our data showing that polarized pm-associated ER strands are intimately connected with the F-actin-containing subapical endocytic collars would be consistent with the 'tip' of the ER strands being carried along with the hyphal tip as apical extension occurs. Whether this occurs simply by 'constriction' of the ER by the collar of actin patches and/or by proteins associated to it, or whether there is involvement of proteinaceous apical polarity landmarks [42], of septin-dependent diffusion barriers/structural elements [43,44,45], or even of lipid domains associated with the tip establishing direct contact with the ER strands, are important questions for the future.
In mammalian cells the peripheral ER, unlike NEs, does not disassemble during mitosis but undergoes major changes in structure correlating with detectable inhibition of ER export [46,47]. Yeasts of the basidiomycete U. maydis display a particular way of 'open mitosis' in which the nuclear envelope is brokendown [48], which leads to distortion of the ER network [19]. The biology of A. nidulans, like that of other filamentous ascomycetes, is conditioned by the compulsory need of its hyphae to grow by apical extension. Apical extension is crucially dependent on exocytosis, which implies that, during vegetative growth, the secretory pathway cannot be compromised under any circumstances. Apical extension is not slowed down by mitosis [16,49]. Thus, the demonstration that impairment of ER function results in cessation of growth and tip swelling (this work) leads to the prediction that the peripheral ER should not undergo major reorganizations during this period. We have thoroughly documented that this is the case. Thus, mitosis does not result in NEs disassembly [17,34]; it does not affect the peripheral ER either (this work) and, consequently, is does not affect the Golgi [9]. In summary, the full capacity of the A. nidulans classical secretory pathway is maintained during mitosis. This, together with the persistence during mitosis of a tip-associated subpopulation of cytoplasmic MTs that possibly contribute to the targeting of secretory carriers to the tip, provide the fungus with the basic tools to maintain apical extension under mitotic circumstances [16,49].

Supporting Information
Video S1 Dynamics of the Sec63-labeled ER in growing hyphae. Sequences were assembled from deconvolved z-stacks of Sec63-GFP images covering growth of several hyphal tip cells over a 30 min period. Time is in min:sec.