Morphodynamics of non-canonical autophagic structures in Neurospora crassa

ABSTRACT Autophagy is a major pathway for unspooling cytoplasmic constituents and recycling them. Here, we investigate autophagy in Neurospora crassa, a close ascomycete relative to the model yeast Saccharomyces cerevisiae. High-pressure-freeze and freeze-substitution techniques proved keys to preservation of the surprising autophagic structures observed in Neurospora by transmission electron microscopy and electron tomography. Depriving Neurospora of carbon triggers two parallel processes at the plasma membrane: formation of vacuoles, initially along the plasma membrane but stuffing 70%–80% of cytoplasmic volume after several hours. The vacuoles contain material left over from digested proteins, disrupted ribosomes or glycogen clumps, plus other discrete organelle-like objects. Also, they accumulate the canonical autophagy marker Atg8, appear to be lytic from the start, and probably grow from lysosomes. Formation of diverse phagophores and autophagosomes, multi-membrane organelles manufactured de novo by the plasma membrane—a feature so far unique to Neurospora. These autophagic structures can be observed attached either to the cell membrane, to the cell wall, or free in the cytoplasm, viz. as mature organelles after detachment from the cell membrane. Although phagophores are clearly observed in carbon-replete cells, their production increases with carbon starvation—and these, too, often appear lytic. ATG1, the canonical initiator of phagophore formation in yeast, is not required for the formation of phagophores in Neurospora. This work provides clear structural and functional evidence that Neurospora’s autophagic organelles differ from those observed by others in yeast. IMPORTANCE Neurospora is a quintessential tip-growing organism, which is well known for packaging and longitudinal transport of tip-building blocks. Thus far, however, little attention has been paid to the co-essential process of reclamation, that is—taking apart of upstream, older structural elements, otherwise known as “autophagy”. We are not yet prepared to set out the chemistry of that elaborate process, but its morphological start alone is worthy of attention. Carbon starvation triggers significant autophagic changes, beginning with prolific vacuolation along the plasma membrane, and eventual filling of 70% (or more) of cytoplasmic volume. Additionally, the Neurospora plasma membrane elaborates a variety of phagophores which themselves often look lytic. These have either dual enclosing membranes, like the familiar autophagosomes, can be doubled and have four wrapping membranes, or can be compounded with multiple membrane layers. These reclamation processes must be accommodated by the mechanism of tip growth.

Multiple studies have shown that serious disruption of recycling pathways shortens lifetime and diverts or weakens particular functions in complex organisms.Although the biochemical pathways involved in these degradation and recycling processes differ considerably from species to species, and are still being worked out, the major pathways in eukaryotes are mediated by the ubiquitin-proteasome system and the vacuolar-lyso somal-autophagy mechanism.
It has been known for more than 80 years that the removal of common nutrients from the environment of free-living microorganisms leads to the upregulated production of scavenging systems-high-affinity uptake systems-for the depleted nutrients.This was elegantly demonstrated in the ascomycete Neurospora by Schneider and Wiley, in 1971, for the monosaccharides glucose, fructose, galactose, and lactose (1), and later for those and other sugars in Saccharomyces (2)(3)(4).It had also been discovered that certain nutrient depletions provoked morphological changes: proliferation of vesicles (lysosomes) and formation of frank vacuoles associated with the recycling of molecular components of cytoplasm, as first observed by de Duve, in 1963, and dubbed autophagy (5,6).By the early 1990s autophagic recycling had become a major area of investigation, pursued especially vigorously in Saccharomyces (7)(8)(9)(10)(11), in multiple animal-cell prepara tions (12)(13)(14), in plants (15)(16)(17)(18), and in mycelial fungi as well (19)(20)(21).
Working on Neurospora for other reasons-electrophysiological study of membrane transport-we noticed that carbon starvation as carried out by Schneider and Wiley (1,22) also produced numerous small vacuoles (23).These were especially prominent during the removal of the carbon/energy source, but the removal of nitrogen or phosphate gave qualitatively similar results (Slayman and Potapova, unpub).The same maneuvers also yielded hyperpolarization of the plasma membrane, from approx.−180 to −220 mV, under standard lab conditions.The voltage change was accompanied by a 7-10-fold increase in electrical resistance of the plasma membrane (23), saving energy by downregulating ion transport systems.The modest hyperpolarization implies a weaker downregulating effect on the plasma membrane's proton pump (Pma1) than upon the ensemble membrane resistance.Thus, selective upregulation of high-affinity sugar transport (1) was just a direct expression of the scavenging function.
A significant special observation with Neurospora has been that the morphological changes, but not the electrical changes, occur upon removal of glucose alone, while carbon/energy is supplied by other sugars (e.g., fructose or galactose), so implying that vacuolation is related preferentially to glucose signaling, not solely to carbon/energy starvation.However, the detailed biochemical processes which underlie these morpho logical and physiological changes remain largely to be characterized.That subject is now interesting because of recent spectacular progress in autophagy research on other systems: especially on Saccharomyces (7)(8)(9)(10)(11) and cultured mammalian cells (12)(13)(14).Starvation for phosphate, nitrogen, or specific amino acids (in auxotrophs) also triggers vacuolation in Neurospora as noted above, but we have pursued glucose starvation, because it is most conspicuous in hyphae of any diameter, and very reproducible.A direct comparison between carbon-starvation and nitrogen-starvation is shown in Fig. S1.
The purpose of this manuscript is, therefore, to set down a clear temporal description of starvation-induced morphological transformations, which are easily seen in the older (upstream) segments of hyphae, but not so easily in the younger segments, just behind the growing tips.The characteristic growth of Neurospora (and many species of mycelial fungi) is supported by longitudinal streaming, due both to pressure-driven bulk flow and to kinesin-powered sliding or walking of vesicles along microtubules, toward a Vesicle-Supply Center (i.e., spitzenkörper) just behind the tip.This means that autophagy in the older segments should be a ready and direct source of recyclable molecules for growth at hyphal tips.Historically, this idea was anticipated by vacuolation experiments on aging Neurospora mycelium (23,24), contemporaneously with the early studies of autophagy in yeast and mammalian cells (5,6).

I. Formation of vacuoles (vacuolation)
Within minutes of the onset of carbon starvation, the observed texture of Neurospora cytoplasm turns "sandy, " as viewed by ordinary bright-field microscopy.The progressive change is most obvious in "stem" hyphae (those ~10 µm diameter or larger) but can also be seen as serial granularity in the abundant small "seeker" hyphae.The sandy or grainy appearance is accentuated by differential interference contrast microscopy (DIC), as shown in Fig. 1, panel a.By 20-30 min of carbon starvation, small but frank "de novo" vacuoles blossom seemingly everywhere, an appearance which is sensitive to the focal plane, thus, suggesting preferential formation along the cell wall.Indeed, with longer starvations, de novo vacuoles enlarge and pack against the plasma membrane, as shown via DIC in Fig. 1b (95 min) and 1 c (195 min); via TEM (transmission electron microscopy) in Fig. 2; and via confocal microscopy in Fig. S2.
But because packing constrained at the plasma membrane defines limited (2D) space, continuing proliferation or enlargement of vacuoles must result in the literal stuffing of the cell volume, as demonstrated in Fig. 1d (270 min).The progress of such stuffing is emphasized in Fig. 3, by the pseudo-3D appearance of the phase-contrast image (Fig. 3a).
After long starvations, de novo vacuoles can come to occupy up to 70% of cytoplas mic volume (see Fig. 1d), and sharp mechanical agitation can trigger their fusion into giant vacuoles.Because overall cell volumes seem unchanged during carbon starvation, expansion of intravacuolar volume requires equivalent movement of water from the cytosol, which would be an automatic physical occurrence when the vacuoles are proteolytic.
Two possible primordia for the starvation-driven vacuolar profusion in Fig. 1c and d are the pre-existing small vacuoles, plus a native population of still smaller vesicles lying very near the plasma membrane of carbon-replete hyphae.See Fig. 4 panels a, b.The cytological appearance of these vesicles/endosomes seems to depend upon their size, somewhat on the duration of carbon starvation, and-of course-on the methods of fixation and staining applied after trapping by HPF (see Methods).After staining by osmium tetroxide, uranyl acetate and lead citrate, the smallest vesicles in carbon-replete cells (<150 nm diameter) appear empty, whereas larger ones (~300 nm and greater) are filled with lightly textured material, probably representing partially digested protein which has been fixed and stained post HPF.A similar appearance is seen in de novo vacuoles of carbon-starved cells (Fig. 5, panels a, c), but a survey of starved cells has revealed widely varied vacuolar contents: scrambled textured material and purloined organelles (Fig. 5c) (25).
In Saccharomyces, but not in Neurospora, blockade of vacuolar proteases stabilizes the microscopic picture of partially digested vacuolar contents (7).Here, we have focused instead upon ATG15 (NCU06436), whose yeast homologue is a vacuolar-resident lipase facilitating the breakdown of autophagic bodies, lipids, and organelles (26,27).After deletion of ATG15, Neurospora hyphae load up with intravacuolar (IV) fragments, as demonstrated via TEM in Fig. 6a and b.In some cases, the IV fragments are compacted The major materials stored in Neurospora vacuoles are amines, basic amino acids, polyphosphates, and calcium (28).The list goes on, but of prime interest here are several lysosomal proteins displayed in Fig. 7 by means of fluorescent tags: (i) the presumed autophagic marker Atg8, filling the interiors of both de novo and constitutive vacuoles (panels b & f ); and similarly, (ii) the aspartyl protease Pep4 (panel c).In contrast, (iii) the small GTPase Rab7-a classic late endosome-lysosome marker-appears concen trated in tonoplasts/vacuolar membranes (panel g), but clearly is not in the vacuolar bulk.Coordination of these proteins is discussed below, but two surprises arise: first, constitutive vacuoles and de novo vacuoles look the same; and second, Atg8 (Gfp-tag ged) appears throughout cytoplasm and in vacuolar bulk, but preferentially not in tonoplasts.However, removal of the Neurospora Atg1 protein (deletion of gene ATG1, NCU00188) excludes Atg8 from the vacuoles (panel j), as had been expected from the known roles of the yeast homolog in recruitment to autophagosomes (7,9).

Transmission electron microscopy
Autophagy is well studied in yeast ( 9) and the process is thought to be in large part conserved in other eukaryotic organisms (29).This assumption and massive data collected from yeast have framed yeast autophagy as "canonical, " both morphologically and biochemically (30).Although Neurospora's genome contains many atg gene homologs, the real function of Neurospora's actual autophagic organelles (phagophores and autophagosomes) is unknown.To approach this problem, we used a strain expressing eGFP-ATG8 at the native locus, and inspected mature hyphae for eGFP fluorescence ±glucose (Fig. 8, Panels a,b and c,d, respectively).Both conditions yielded intense fluorescent pinpoints throughout the cytoplasm, but especially at the cell edge.The size distribution of bright fluorescent dots spans from 0.05 μm to 1.2 µm, averaging 0.34 µm (Fig. 8g).
In yeast expressing eGFP-ATG8, bright puncta have been interpreted to reflect enhanced fluorescence intensity from lipidated eGFP-ATG8 (eGFP-ATG8-PE), decorating both membranes in nascent and mature autophagosomes (31).We have no reason to reject this view in Neurospora, but because the corresponding objects are elusive in TEM images, we do not know for certain whether bona fide autophagosomes or intermediate structures glow as the fluorescent pinpoints.Immunogold cytochemistry and correlative light electron microscopy techniques both failed to clearly answer this point, mainly because of background issues.Functionally, however, glowing eGFP-ATG8 puncta even in carbon-replete conditions implies that autophagy is a background process in Neurospora.For further testing this idea, we used the eGFP-ATG8-expressing strain to measure autophagy flux: how fast autophagosomes are degraded in C-Replete or C-Starved cells.The readout of this assay is the amount of free eGFP (fluorescence) left after attack on eGFP-ATG8 by vacuolar proteases (32,33), shown in Western blot analysis in Fig. 8e and f.The band intensity ratio free eGFP/eGFP-ATG8 fusion is 0.76 in carbon-replete versus 1.26 in starved conditions (Panel f ).Thus, Neurospora autophagy is abundantly present in carbon-replete cells, but is further induced by carbon starvation.TEM micrographs clearly demonstrate that Neurospora produces a variety of multi membrane autophagy-related structures (phagophores and autophagosomes).These structures arise predominantly from the plasma membrane, rarely-if ever-from the ER.They can be lytic in character, thus clearly differing from the yeast picture.Diameters of quasi-circular puncta were measured using ImageJ and raw data binned (326 total measurements from 10 cells, 16 bins set at 0.074 µm width, average ~0.34 µm).
Presumptive simple phagophores, with concentric dual membranes, are shown in Fig. 9, ringing a hypha, attached to the plasma membrane itself (see inset) or having been stripped from an anchor on the adjacent cell wall.In favorable hyphae, phagophores form at intervals of 1.5-2.0µm along the plasma membrane.That distribution implies 60-70 forming phagophores over the whole cylindrical surface (here, ~15 µm long and 5-6 µm diameter), viz.approximately one for every 5-6 µm 2 of hyphal surface.
More of those autophagic objects are shown in the TEM micrographs of Fig. 10a  through d: single phagophores, compound phagophores with multiple double layers of membrane; single autophagosomes, and double autophagosomes.The morphology of these objects can be appreciated by higher magnification (see below) even though full appreciation of 3D structure is limited by the 2D nature of TEM thin sections.Whether the formation of autophagic structures, or their increase by starvation, in Neurospora, uses some of the canonical machinery inferred from yeast is unresolved.To investigate this question, carbon-replete and carbon-starved mycelia have been observed via TEM and scored for the fraction of positive cells: those displaying phagophores and/or autophagosomes, out of the total cells inspected (Fig. 10a and b representative images; e left, quantification).Although autophagic structures are significantly more abundant in carbon-starved mycelium, blocking canonical autophagy by deleting the ATG1 gene yielded no useful results (Fig. 10c and d; e right).Clearly, formation of autophagic structures in Neurospora does not require ATG1 function, although Atg1 protein could have a downstream role in delivering autophagic organelles to vacuoles (34,35).
High magnification TEM micrographs show how these dynamic organelles evolve starting from the plasma membrane.Phagophores are initiated by inward growth of plasma-membrane (arrows in Fig. 11a through c), which must be sheet-like in three dimensions.The small funnel usually formed at the point of initiation appears filled with cell-wall material (white unstained in Fig. 11a) which could be part of the plug or anchor shown in Fig. 9 and inset.The growing sheets of PM subsequently recurve toward the cell surface, around a gap of tens-to-hundreds of nanometers, before fusing back to the PM (pins in Fig. 11a and b).This action entraps a small volume of cytoplasm which usually contains ribosomes, clumps of carbohydrate, and digested fragments of both.Occasionally a developing phagophore loads up, then detaches from the PM but fails to close, so resembling a canonical phagophore, but with four membranes (Fig. 11d).The obvious dissolution of ribosomes or carbohydrate clumps (Fig. 11g, j and i) seems to report the activity of lysosomal enzymes in the central cavities of the phago phores.Thus, in Neurospora, and probably in other mycelial fungi as well, autophagic recycling, downstream trafficking, and tip growth, may be able to proceed without explicit involvement of the actual defined vacuoles.Compared with the highly detailed but rather rigid picture of autophagosome structure and function in Saccharomyces, the morphology of phagophores in Neurospora can diverge widely, and is susceptible to some humorous description: "phagosacks" resembling a satchel, ~300 nm deep hung off the PM (Fig. 11e); "phagobags" ~600 nm deep, looking very floppy but pinched to the PM (Fig. 11f); "phagoballs" ~300 nm diameter, packed like a golf ball with ribosomes and either rolled snugly between the PM and an adjacent tonoplast (Fig. 11h), or welded to the PM as well as to a swollen mitochondrion (Fig. 11g).Finally, there are fully "free-floating" phagophores, resembling complete canonical autophagosomes, large (560 nm diam., Fig. 11i) or more compact (310 nm diam., Fig. 11j), and having a ladder-like structure sewn between the two membranes.
Neurospora also lays down more complicated geometries for these dual membrane phagophores, as shown in Fig. 11k, where five or more separate cytoplasmic domains are defined by the inward growth and refusion of several sheets of plasma membrane.Clusters of ribosomes are present in some of these domains, but not in others, perhaps due to a collaborating distribution of lytic enzymes.
But single phagophores and single autophagosomes (double-membrane structures) are less surprising than double phagophores and double autophagosomes (quadruplemembrane structures).Those are certainly non-canonical since they are rarely observed elsewhere.Double phagophores arise from nearly simultaneous inward growth of two adjacent patches of plasma membranes, shown in Fig. 12a and b, which then curve back to the PM forming quadruple-membrane spherical bodies (Fig. 12c through j).The usual clusters of ribosomes and clumps of carbohydrates fill the central core of such structures as the inward-growing membranes corral underlying cytoplasm.
To simplify the description of the layered structures, we call the core C0, and count the enclosing shells outward as S1, S2, & S3, labeled as on Fig. 12c.S1 and S3 resemble the outer layer for most single phagophores (Fig. 11f and j) in being generally clear.S2 appears puffy or gritty but is devoid of ribosomes or obvious carbohydrate clumps.
Independent maturation of each phagophore, whether single or doubled, requires several steps away from the PM: (i) actual detachment from the membrane (and anchor); (ii) submersion into the bulk cytoplasm; and (iii) closure of the outer membrane sheath (Fig. 12c-inset, 12d-inset).Phagophores thus separated have been seen squeezed against vacuoles, but we have not caught any actually fusing to vacuolar membranes in Neurospora.Although multilayer structures of this sort have a morphological resem blance to the transient so-called "cradles" which support autophagosome formation in cultured mammalian cells (36), single or double non-canonical phagophores usually appear as completely closed structures by 2D imaging, having two, three, or four internal spaces.The inner-most of these can contain intact fragments of cytoplasm, including ribosomal clusters and carbohydrate clumps.
Notably, the inner cavities of phagophores and autophagosomes often appear devoid of well-defined cytoplasmic components, such as ribosomes and glycogen clumps.This is true in about 40% of the observed autophagic structures in Neurospora and strongly suggests a self-sufficient lytic function of these organelles.Fig. 13 shows a few examples of lytic phagophores/autophagosomes (panels a, b, c, e, f ), and a non-lytic autophagosome as a control (panel d).Although PEP4-mCherry and eGFP-ATG8 did not clearly colocalize (data not shown), other hydrolytic enzymes could work within these organelles.

Electron tomography
All of the EM structures shown so far have emerged from TEM thin sections (~60 nm), which reveals the majority phagophores to be circular-or en route to becoming circular-and consisting of two or four separate quasi-concentric membranes.Fig. 11k, however, suggests that Neurospora's phagophoric structures in general may be more complicated than can be revealed by the two-dimensional technique.We therefore moved to tomography (3D) on thick sections (~250 nm), then reconstructed images from tilt series, as detailed in Fig. 14, with a simple two-membrane non-canonical phagophore.Although the structures we could access this way were incomplete, they nevertheless did reveal important complexities.This phagophore was formed by the inward growth of two parallel PM sheets, similar to those displayed in Fig. 11a through c, e and f.In this case, however, the in-growing sheets did not simply close back to adjacent PM, but branched and partitioned off a second, parallel, internal space before RE-fusing to the PM.(This is a distinction in space, not in time.)The data set do not include the top or bottom portions of the whole cylindrical structure, but the midportion can be built up progressively by superposing individual slice views, from 8 to 140, as shown in the right-hand column.Sections 20 and 48 (panels a and b) demonstrate that the large right-hand cavity (still connected to the cytoplasm) is bounded by two layers of plasma membrane.For the smaller left-hand cavity, that fact is not obvious, but the small intrusion at the base (slice views 20 and 48) does become continuous with the left-hand cavity.The two "central" cavities here are nearly devoid of ribosomal clusters from the immediately surrounding cytoplasm, suggesting that lytic enzymes were active throughout formation of this complicated phagophore.
Tomographic results for a non-canonical doubled phagophore, similar to those depicted in Fig. 12, are displayed in Fig. 15.The resulting structure proved not quasispherical, but cylindrical, having two almost complete double-membrane rings around the lower half (near slice-view c), but opening from a discrete focus (plug-like object) against the cell wall in slice-view a.The top of this double cylinder had been truncated by the microtome, but the bottom was left nearly intact, closed (above slice-view c) for the inner cylinder, and imaged (poorly) for the outer cylinder, as if it lay between the two membranes.The 3D projection in Fig. 15d displays the whole object almost edge-on, seen outward from the PM.The projection in Fig. 15e shows the object tilted about 60 o , depicting the upper surfaces of both bottom membrane wraps as background, and the top edges of both membranes highlighted in blue/green.The projection in Fig. 15f shows the inner surface (back wall) of the cavity formed in this double phagophore by the central ring/wrap of dual membranes.It also displays the discontinuities of both double rings pointing toward the actual plasma membrane, as well as the local merger of the two double rings.

DISCUSSION
The first surprising effect of carbon starvation in Neurospora is the enormous vacuolar expansion.That is not a conservative process.Filling 70% of a stem hypha with ~3 µm membrane-bounded spheres means manufacturing or collecting about 4-to 5-fold as much membrane as would be present in the starting plasma membrane.Also, if the de novo vacuoles in Fig. 1 in fact do arise from the small vesicles shown in Fig. 4, then the needed expansion of surface area can be estimated from the relative mean diameters: less than 300 nm in Fig. 4, and ~3000 nm in Fig. 1d, implying a ~100-fold increase of vesicular surface area.At least two forces would be driving this membrane expansion: osmosis and biochemical regulation.Cytosol and vacuolar fluid must remain in osmotic equilibrium, so cytosolic water would osmose into the vacuoles as organelles and proteins are degraded into small osmotically active molecules, unless and until these small molecules are extruded.For such lysis-induced expansion really to work, the smallest starting vesicles should contain lytic enzymes, meaning that the visible vacuoles grow from pre-existing lysosomes.So, even without stains for cytochemical proof, the smallest vesicles in Fig. 4 should be lysosomes.These might operate auton omously, except for the energy required to build or recruit vacuolar membrane; and for the energy needed to absorb or extrude cytoplasmic components.In fact, small vesicles labeling with vacuolar markers have been widely observed in filamentous fungi and considered to be part of the complicated "vacuolar compartment, " which includes textbook-classic spherical vacuoles as well as interconnected tubular-vesicular structures (37)(38)(39).Given the magnitude and speed of vacuolar membrane expansion (Fig. 1), sources for phospholipids feeding sustained tonoplast growth are an interesting and important mystery.Two possibly significant sources are (i) classic vesicular trafficking, which mediates lipid transfer between organelles via vesicle fusion (40) and (ii) bulk lipid transfer at overt contact sites between organelles (14,41,42).
The canonical hallmark of autophagy is the formation of so-called autophago somes, drone-like organelles which engulf cytoplasmic components either randomly (non-selective/bulk autophagy), or specifically via receptor-mediated cargo recognition factors.In yeast and mammalian cells, autophagosome biogenesis involves about 20 autophagy-specific genes/proteins, which make the canonical core ATG machinery (43,44).This machinery carries out nucleation, expansion, and closure of specialized cupshaped cisternae called phagophores-a single-membrane element formed by fusion of diverse vesicles (29,30).The process is assisted by scaffolding proteins, and occurs at the vacuolar membrane in yeast (29), or in mammalian cells, at a discrete region of the ER-termed the omegasome (36).
In Neurospora, by contrast, we found that autophagosome biogenesis results in a surprising variety of structures, having shapes far from the conventional idea of quasi-spherical organelles: for example, double phagophores (Fig. 10, 12 and 15), compound phagophores (Fig. 10 and 11), and twinned and branched phagophores (Fig. 14).The origin of these structures is at the plasma membrane, clearly shown by the continuity between the limiting membrane of most phagophores and the plasma membrane (Fig. 11 to 14).Otherwise documented sources include the ER (36,45), ER-Golgi intermediate compartments (ERGIC) (46), the mitochondrial outer membrane (47), the nuclear membrane (48), and the plasma membrane (49).The nuclear membrane as a source of autophagosomes is particularly relevant because of the morphological similarity between double phagophores/autophagosomes made by macrophage nuclear membranes in response to herpes simplex viral infections (48) and those made by Neurospora's plasma membrane in response to carbon starvation (Fig. 12c, d, e, g and  h).Whereas plasma membrane contribution to autophagosome formation is often seen as a mechanism for extra-membrane during autophagic surges (49), Neurospora's plasma membrane builds whole phagophores and autophagosomes in situ (Fig. 11 to 15).
The peculiar feature that forming phagophores can have a firm plug on (or anchor to) the cell wall through the plasma membrane (Fig. 9 and 15) has not been repor ted elsewhere.That plug/anchor has an unknown function for single phagophores in Neurospora.
In canonical autophagosomes, the space between outer and inner membranes is described as a tight sub-compartment which can be expanded by fusion with lyso somes to allow degradation of inner membrane and resident cargo in a newly formed autolysosome (13,29).In Neurospora double and compound phagophores, the picture is more complicated: for example, the intermediate space (#2) between the two dual membranes is devoid of cytoplasmic objects (Fig. 12c through h), implying that proteases must be activated within the space captured by advancing pair folds of PM (Fig. 12a and  b); also, central cavities of compound phagophores lack cytoplasmic material (e.g., Fig. 13  and 14) strongly suggesting degradative functions.
We do not know whether these organelles carry out their lytic process instead of the expected cargo-transport function, or both.However, the conspicuous filling of vacuoles with green fluorescence (Fig. 8d) strongly suggests that cargo-competent phagophores at the plasma membrane must undergo maturation and detachment from the plasma membrane before fusing to de novo vacuoles during starvation.Thus, robust autophagy flux even in carbon-replete cells (Fig. 8e and f) could be supported by lytic-competent phagophores and autophagosomes, as well as by other components of Neurospora's elaborate vacuolar system (37)(38)(39).
The direct involvement of plasma membrane in the manufacture of phagophores and autophagosomes, in addition to the apparent lytic autonomy of those organelles in Neurospora, could frame our understanding of how the familiar autophagosomes may have originated and evolved from organisms not yet equipped with a system of endomembranes (50,51).

Media, strains, genetics, DNA constructs
Neurospora strain RL21a, brought from the Tatum collection at Rockefeller University, was maintained in silica-gel cultures at 4°C.Similarly, the Oak Ridge wild-type strains OR-74A and 74-OR8-1a, along with several single-gene KO strains, were obtained from the Fungal Genetics Stock Center (fgsc.net;Department of Plant Pathology, Kansas State University, Manhattan, KS 66506) and stored as silica-gels.All strains could be retrieved from the frozen cultures as needed, and were normally refreshed at yearly intervals, thence maintained on Vogel's minimal medium (52) +2% sucrose.New constructs usually were started with FGSC strain 9718 a (mus51Δ:: bar + ), in which Neurospora's robust heterologous recombination is suppressed (53).
Strain eGFP-ATG8 was constructed by adding a non-dimerizing variant of eGFP) to the N-terminus of Neurospora's ATG8 gene (NCU01545).We then constructed four principal strains (Table 1, strains 7-10) all bearing eGFP-ATG8 as a supposed marker for conven tional autophagy.This construct was paired with four other constructs.
The homokaryotic double-mutants for Strains 7-9 were each obtained by standard crossing of parents bearing the singly tagged genes (Table 1, strains 1-6).However, Strain 10 required more work and was obtained by transformation of an eGFP-ATG8 strain carrying the nourseothricin selection marker with an ATG15-deletion cassette bearing the hygromycin B selection marker.

Preparations for microscopy
In most experiments, mycelium was grown from an agar-button inoculum on cellophane, underlain by Vogel's minimal medium plus 2% sucrose and 2% agar.After 18 h at 25°C, a square of cellophane with mycelium attached was cut out and transferred to ~8 mL of Vogel's medium in a large (6-7") watch crystal and agitated to float mycelium away from the cellophane.(We call this very thin, free-floating mycelium "gossamer.") Gossamer mycelium, submerged by repeated droplets of Vogel's medium, could then be main tained in the growth medium, or transferred (via a coverslip) to glucose-free medium for any appropriate interval.
For light microscopy, the gossamer was picked up on a large coverslip, drained of most fluid, and inverted into a microscope chamber bearing small fluid ports.There after, the chamber was perfused with the appropriate medium and could be viewed eGFP-ATG8, mCherry-RAB7 eGFP-ATG8-hph, mCherry-RAB7-hph Cross [1] x [3] eGFP-ATG8, PEP4-mCherry eGFP-ATG8-hph, PEP4-mCherry-hph Cross [1] x [4] eGFP-ATG8, atg1Δ eGFP-ATG8-hph, atg1::hph Cross [1] x [5] eGFP-ATG8 by bright-field, phase-contrast, or DIC optics, without staining or fixation, but with continuing solution flow, to prevent anoxia.For electron microscopy after high-pressure freezing (HPF), the appropriately conditioned gossamer (i.e., carbon-starved or carbon-replete) was plucked "by the tail" with jeweler's forceps, blotted lightly, then dunked in a drop of hexadecene, removed, blotted lightly again, loaded into a 3 mm brass carrier, and high-pressure frozen using a Leica EM HPM100 at 2100 psi in liquid nitrogen.Frozen samples were freeze-substituted via a Leica Freeze-Substitution unit (AFS2) using the following protocol: 3 h at −95°C in 1% osmium tetroxide/0.1% uranyl acetate in pure acetone; 46 h at −90°C; 20 h at −50°C; then rinsed in acetone over 4 h, to −20°C.Samples were then infiltrated with graded LX112 epoxy resin (Ladd) in acetone over 4 h, to 4°C.After two changes into pure LX112, samples were placed in gelatin capsules and infiltrated under vacuum in pure resin over 2 hr, then hardened overnight in a 60°C oven.
Sections (60 nm thick) were cut using a Leica EM UC7 ultramicrotome, collected onto formvar/carbon coated grids, and stained with 2% uranyl acetate plus lead citrate.Grids were examined on a FEI Tecnai Biotwin TEM at an accelerating voltage of 80 kV.Digital images were recorded with a Morada CCD camera and iTEM (Olympus) imaging software.
For electron tomography, 250 nm sections from the same blocks were cut with the same ultramicrotome, and 15 nm fiducial gold particles were added to the sections before imaging.The dualaxis tilt series was collected via a FEI Tecnai TF20 at 200 kV TEM equipped with a field emission gun.Images were recorded using SerialEM software (UC, Boulder) and an FEI Eagle 4K x 4K CCD camera.Dual-axis tilting angles were from −60° to 60° at 1° increments.Tomogram reconstruction, segmentation, and modeling were performed using IMOD software (54).
Technical note concerning terminology for EM tomography: In Fig. 14 and 15, the TEM-like images (left-hand columns) are computed and represent surfaces only, not objects with depth.They are properly termed "slice views." However, for the sake of simplicity, we sometimes refer to the slice views displayed as "sections."

FIG 1
FIG 1 Profuse production and growth of vacuoles within Neurospora hyphae, during carbon starvation.Panel a: Cytoplasm perceptibly granulated (630x), representing the onset of vesicle enlargement, after 15 min of carbon starvation (CS).Panels b-d: Lengthening starvation times, with visualization by the differential interference contrast (DIC) microscopy.Prior to carbon starvation, Neurospora stem hyphae typically possess a single spherical constitutive vacuole (CV) in every second cell.Gossamer mycelium was grown and prepared as described in Methods.In separate experiments, control cells in carbonreplete conditions (1% glucose) did not vacuolate up to 6 h incubation if the gossamer was handled gently.(XW) cross wall.(Neurospora's crosswalls are perforated, allowing continuous cytoplasmic flow for several centimeters in a mature colony.The flow here was rightward.)Strain: RL21a, Rockefeller-Tatum wild type.

FIG 3 5 FIG 4 6 FIG 5
FIG 3 Radial display of vacuoles elicited by C-starvation, with phase-contrast image to capture the 3-D effect.Panel a: phase-contrast highlights 3D by revealing interspaces that are hidden by fringes in Panel b: the differential interference contrast (DIC) image.Same experiment as Fig. 1.

FIG 7
FIG 7 Localization of a few "key" enzymes in Neurospora's pathway for autolysis.Panels a-d: strain expressing ATG8 tagged with eGFP (N-terminus) and PEP4 (vacuolar peptidase; mCherry tag to C-terminus).ATG8 & PEP4 both enter the whole vacuolar space.Panels e-h: strain expressing e-GFP-ATG8 and RAB7 (small GTPase, late endosome marker; mCherry tag to N-terminus).ATG8 fills the vacuoles while RAB7 distributes to vacuolar membrane.Panels i-l: strain knocked out for ATG1 gene (atg1Δ), expressing eGFP-ATG8, and stained with FM4-64 (a dye for vacuolar membrane).ATG8 is blocked out of the vacuoles by deletion of ATG1.Panels m, n, o emphasize fluorescence intensity distribution along digital sections drawn across a single vacuole (see white lines in merged images).Mycelia of all strains were carbon-starved for 2 h.Scale bars in DIC images (10 µm) apply to the corresponding horizontal image set.Negative controls (non-tagged strains) show negligible autofluorescence.

FIG 8
FIG 8 Autophagy indicators in Neurospora under carbon-replete and carbon-starved conditions.Mycelial gossamers were incubated for 2 h in the presence of 1% glucose (C-Replete, panels a, b) or zero glucose (C-Starved, panels c, d), mounted on a slide, and immediately imaged using confocal microscopy.Homokaryotic strain expressing ATG8 tagged with eGFP at the N-terminus.Fluorescent puncta are interpreted as autophagosomes/phagophores carrying lipidated eGFP-ATG8.These puncta are present in carbon-replete as well as in carbon-starved Neurospora cells.Arrows indicate fluorescent puncta close to the plasma membrane.Arrowheads indicate puncta close to the cross walls (xw).Constitutive vacuole (cv).Scale bar in panel a applies to all four images.Panel e: autophagy flux measured as degradation of eGFP-ATG8 to free eGFP by vacuolar proteases of replete vs. starved mycelial gossamers; Western Blot of crude total protein extracts (2.5 µg protein/lane) probed with a GFP antibody.Minor crossreacting bands most likely represent intermediate degradation or aggregation products.Western blot is representative of four independent experiments that gave similar results.Band intensities from Western blot in panel e were quantified in panel f using ImageJ software.Panel g: size distribution of eGFP-ATG8 fluorescent puncta (Gaussian fit).

FIG 9 11 FIG 10 12 FIG 11
FIG 9 Phagophores proliferated around a single hypha.Cytoplasm fortuitously shrunken either during HPF or during fixing & staining.PM thus torn from the cell wall revealed multiple forming phagophores: designated by A & white arrows.A robust plug (or anchor) remained attached to the wall opposite each phagophore (P & black arrows), while a corresponding "mouth" gapped the cytoplasm.Slender fibers (F) along the gap: part of the severed membrane-to-wall attachments.Labels (V, Sv, & M) as in Fig. 4 and 5. Wild-type strain RL21a.Mycelium carbon starved for 4 h.Inset: Detail of one forming phagophore (A).The enlargement of phagophore-to-membrane-wall boundary reveals a faintly bilobed structure in the contact mouth.Scale bar: 2 µm for main, 0.2 µm for inset.

FIG 12 15 FIG 13
FIG 12 Formation and character of non-canonical double phagophores.Panels a, b: Simultaneous in-growth of four sheets of plasma membrane at each panel, fused in pairs before re-connecting to the plasma membrane (PM).(w) Cell wall.Panels c, d: double phagophores detaching from the plasma membrane, but with the outer membrane still attached to PM, as shown by both insets.Panel c: Numbered spaces: 0 = main core, 1 = first wrap, 2 = 2 nd core, 3 = 2 nd wrap.Panels e, h, two mature double autophagosomes, cores still loaded with ribosomes, presumably because lysosomal enzymes are missing or inactive.Panel f: Similar to panel c, except that the core space is branched.Panel g: An anemic-looking double autophagosome is paired with a robust single phagophore, the core of which is filled with ribosomes and glycogen clumps.WT strain carbon-starved for 2 h.The scale bar for all panels is 400 nm.

FIG 14 3 -
FIG 14 3-D tomographic reconstruction of a branched and twinned single phagophore, growing from the plasma membrane.Left-hand column: Computer-generated plane sections: a = "near bottom" to d = "top." Thickness of section images, 1.29 nm.Right-hand column, e to h: gradual construction of 170 nm tall cylindrical structure by stacking of membranes in the top 132 sections, represented by the four calculated sections in panels a to d (left-hand column).Data obtained by multiply tilted imaging of a thick block, via the same software as in Fig. 15.Stacks 86 to 140 show a clear vertical-space extension between the two membranes surrounding the left-hand cavity.See Fig. S3 for fixed 3D prospectives of these data, for a montage of TEM slice views 8 →140, and ) for full reconstruction of stacks, made through multiple angles.Same gossamer and HPF material as in Fig. 12. Scale bar for all panels is 100 nm.(PM) plasma membrane, (W) cell wall.

FIG 15
FIG 15 Electron tomographic reconstruction/segmentation of a doubled phagophore in Neurospora.Panels a, b, c: Representative slice-views of the tomogram.Panels d, e, f: Segmentation of the same tomogram at different viewing angles (see Methods).Numbers in panel d are the distances, in nm, from the top edge of the outer wrap of membranes to the slice-views a, b, & c, respectively.Distance between adjacent slices is 1.108 nm, and the entire tomogram is 153 nm thick (tall).Top contour lines are painted blue/green to emphasize the presence of dual membrane sheets (each contour line represents one membrane bilayer).P the non-descript plug (or anchor) against the cell wall, visible in sections a & b, recalls Fig. 9, above.The plasma membrane (PM), which is clear in b & c, has been added to projection f.The same gossamer culture and HPF material as in Fig. 12. Horizontal scale bar in all panels is 100 nm.Reconstruction/segmentation carried out via IMOD software (see Methods).(W) cell wall.

TABLE 1
Strains used in this study