GBF/Gea mutant with a single substitution sustains fungal growth in the absence of BIG/Sec7.

Golgi Arf1-guanine nucleotide exchange factors (GEFs) belong to two subfamilies: GBF/Gea and BIG/Sec7. Both are conserved across eukaryotes, but the physiological role of each is not well understood. Aspergillus nidulans has a single member of the early Golgi GBF/Gea-subfamily, geaA, and the late Golgi BIG/Sec7-subfamily, hypB. Both geaA and hypB are essential. hypB5 conditionally blocks secretion. We sought extragenic hypB5 suppressors and obtained geaA1. geaA1 results in Tyr1022Cys within a conserved GBF/Gea-specific S(Y/W/F)(L/I) motif in GeaA. This mutation alters GeaA localization. Remarkably, geaA1 suppresses hypBΔ, indicating that a single mutant Golgi Arf1-GEF suffices for growth.

Introduction

ADP-ribosylation-factor (Arf) GTPases regulate membrane traffic by organizing vesicle budding. Their activation depends on the GEF (Guanine nucleotide Exchange Factor)-catalyzed exchange of GTP for GDP, that leads to a conformational switch stabilizing membrane insertion of their N-terminal myristoylated amphipathic helix and facilitating recruitment of specialized effectors [1], [2].

Arf1, an essential Golgi regulator, is activated by the GBF/Gea and the BIG/Sec7 GEF subfamilies [1]. These two subfamilies comprise related proteins sharing the highly conserved catalytic Sec7 domain (Sec7d) and five conserved regions denoted DCB, HUS, HDS1, -2 and -3 domains [3], [4], [5]. GBF/Gea and BIG/Sec7 are both considered essential for Golgi function in Saccharomyces cerevisiae, Drosophila melanogaster and mammalian cells [6], [7], [8], [9], [10], [11], [12], [13], [14]. Moreover, they are the only Arf-GEF subfamilies common to all eukaryotes [3]. Thus, although both GBF/Gea and BIG/Sec7 activate Arf1 at the Golgi, they have some non-overlapping essential functions.

Apart from triggering Arf1-mediated effector recruitment, Arf1-GEFs each engage specific protein interactors, resulting in variable outcomes from the activation of a single Arf. This possibly underlies non-overlapping Arf1-GEF functions. For example, Gea1p/GBF1 interacts with Sec21p/γ-COP [15].

Arf GEFs are peripheral membrane proteins: their membrane recruitment is tightly regulated to ensure precise spatiotemporal responses [16], [17], [18], [19]. However, the mechanistic bases of this regulation are incompletely understood [reviewed in [20]]. It is widely accepted that the BIG/Sec7 subfamily members act at the late/trans-Golgi, while the GBF/Gea subfamily members act at the early/cis-Golgi [1]. An HDS1-mediated interaction of S. cerevisiae Sec7p with membrane-bound Arf1-GTP contributes to Sec7p recruitment to, and activation at late Golgi compartments via a positive feedback loop [21]. The Arf-like protein Arl1 is necessary for recruitment of the Sec7-orthologues BIG1/2 at the mammalian trans-Golgi [22]. Mammalian GBF1 is a Rab1 effector [23]. Its cis-Golgi recruitment occurs in response to an increase in membrane-associated Arf-GDP [24]. The HDS1 domain of its S. cerevisiae homologue Gea1p was implicated in lipid droplet binding and Golgi recruitment [25] and in interaction with the early Golgi resident Gmh1p [26]. Coincidence detection might additionally link the above observations and determine the Arf1-GEFs localizations at the Golgi.

How different sets of interactors and/or localization determinants result in essentially distinct functions of the two Golgi Arf1-GEF subfamilies and how these functions serve cargo passage through the Golgi are unanswered questions.

Materials and methods

Deletion of hypB

hypB is AN6709 (http://www.aspgd.org/). We constructed by fusion PCR (primers 14–19, Supplementary Table II) [27], [28], a cassette (Fig. 1) for substituting the hypB ORF with Aspergillus fumigatus pyrG in wt (MAD1739-List of strains in Supplementary Table I) and geaA1 backgrounds (MAD5107). The cassette was cloned in pGEM (plasmid p2001) and checked by sequencing. A p2001 linearized (NcoI/NsiI) fragment was used in transformations for gene replacement. We assessed the lethality of hypB deletion using the heterokaryon rescue [29]. Diploids, heterokaryons and homokaryotic mini-colonies (hypBΔ) or non-sporulating colonies (hypBΔ geaA1) of transformants were genotyped by PCR (primers 14 and 19) and/or by Southern blots.

Fig. 1

(A) Deletion of the hypB gene encoding the only A. nidulans late-Golgi Arf1-GEF is virtually lethal. hypBΔ transformants, selected as pyrimidine prototrophs, were heterokaryotic (i), i.e., they carried both transformed (hypBΔ:: pyrG) and wild type (hypB + pyrG89) nuclei. Spores produced by these heterokaryons are uninucleate. On medium without pyrimidines, untransformed nuclei-containing spores cannot grow due to pyrimidine auxotrophy, which permits testing for growth of the prototroph spores that contain the deletion mutation. hypBΔ spores form aconidial microcolonies (ii). PCR genotyping with DNA from the multinuclear primary transformant mycelium (1) yielded amplification bands corresponding to both the wild type hypB and the hypBΔ::pyrG deletion cassette (iii), confirming that this was heterokaryotic. In contrast, only the band corresponding to the hypBΔ::pyrG construct was amplified from DNA from the micro-colonies on the selective medium (2), indicating that these are homokaryotic hypBΔ::pyrG colonies, which showed that HypB is virtually essential for A. nidulans. (B) GBF/Gea-subfamily mutant GeaA1 bypasses the requirement for the BIG/Sec7-subfamily HypB. Growth at the specified temperatures from spore dilutions of strains: wt (wild type, MAD2), hypB5 (MAD3574), hypBΔ (hypBΔ::pyrG89; nkuAΔ::bar pyroA4), geaA1 (MAD4062), hypB5 suA1hypB5 (MAD4041-obtained by UV mutagenesis), hypB5 geaA1 transformant (MAD4836-obtained by transformation of a hypB5 strain with a DNA fragment carrying the geaA1 mutation) and hypBΔ geaA1 (MAD5130). All photos are at the same magnification (except the one indicated as 2×, this is a 2 times magnification of the yellow rectangular region).

Ultraviolet light (UV) induced mutagenesis and molecular characterization of suppressors

40 UV-induced hypB5 suppressors were selected in MAD3574 for growth at 42 °C. In all but one case, reversion occurred within hypB (Table 1). Meiotic crosses verified that this suppressor mutation (suA1hypB5) is extragenic. Haploidization analysis [30] localized the suppressor mutation to chromosome VIII. We meiotically mapped suA1hypB5 to a position between nudA and nirA (In the process of mapping suA1hypB5, we identified the mutational lesion of sE15 as a frameshift in trxA, see Supplementary Materials and Methods). AN0112, encoding GeaA, was identified as the most likely candidate in this interval and was sequenced (primers 20–27).

Table 1

Revertants of hypB5.

Strain		HypB aminoacid sequencea
hypB5		A881P
Revertants
Revertant type	Times obtained
True revertant	1	A881
First site S	19	A881S
First site L	12	A881L
First site T	2	A881T
First site T and second site	1	D880G A881T
Second site	2	T863I A881P
Second site	1	A881P N956L
Second site	1	ΔG854ΔE855 A881P
Extragenic (hypB5 geaA1)b	1	A881P
	Total = 40

Changes in comparison with the wild type strain MAD2. Only the nucleotide sequence coding for F825 to D961 of HypB was determined (amplified with primers 1 and 2), except in the case of the “extragenic” suppressor, where the whole hypB was sequenced (primers 1 to 13 and 20) and found identical to hypB5 parent.

MAD4041.

Reconstruction of the hypB5 geaA1 strain by transformation

To confirm that the suppression phenotype in suA1hypB5 is indeed due to the mutation identified in geaA (geaA1), we PCR-amplified (from MAD4041 gDNA using primers 22 and 23) an ∼1.5 kb region of geaA1 carrying the A3065G substitution (approximately in the middle of the fragment) and used this molecule to transform hypB5 strain MAD3574. We directly selected transformants for growth at 42 °C, the restrictive temperature for hypB5, and verified by sequencing that all transformants contained geaA1. This procedure yielded MAD4836.

In silico analyses

Sequences used for in silico analyses that detected the GBF/Gea-specific motif were identified by Blast using as queries the S. cerevisiae Gea1p (GenBank CAA89558.1)/Sec7p (NCBI NP_010454.3), Aspergillus nidulans GeaA (AN0112)/HypB (AN6709) or Homo sapiens GBF1 (GenBank AAI17683.1)/BIG1 (NCBI NP_006412.2). Alignments used T-Coffee (http://tcoffee.crg.cat/apps/tcoffee/do:regular), while their visualization and editing were performed with GeneDoc.

GFP-tagging and microscopy

Using PCR (primers 30 through 39), we fused part of the geaA ORF (starting downstream the nucleotide that is mutated in geaA1) to gfp in frame, the 3′UTR of geaA and pyrG of A. fumigatus (Fig. 3). This fragment was cloned in pGEM generating p2198 and was sequenced to verify the absence of mutations. p2198 was linearized (BglII/SpeI) to transform MAD1739 for obtaining geaA::gfp or MAD5107 for geaA1::gfp. Transformants were analyzed by Southern blot to verify in locus integration.

Fig. 3

Intracellular distribution of GeaA-GFP and GeaA1-GFP. (A) C-terminal tagging of geaA or geaA1 with gfp was achieved by in locus integration by transformation of a linear cassette containing part of the geaA open reading frame fused to gfp, the geaA 3′UTR and the A. fumigatus pyrG as a selection marker, resulting in strains carrying a single gfp-tagged copy of the geaA or geaA1 allele. (B) Growth test showing that GFP tagged GeaA and GeaA1 are functional. Note that GeaA1-GFP suppresses hypB5 thermosensitivity, like the untagged allele. (C) Maximal intensity projections of deconvolved z-stacks. GeaA-GFP localizes at Golgi cisternae. GeaA1-GFP, although still localizing at Golgi cisternae, also labels an apical crescent and an apex-associated accumulation. The apical localization is maintained in the hypB5 background after shift to 37 °C for 40 min (n: nucleus, bar: 3 μm). (D) Fluorescence intensity profiles across the hyphae in (C) (n: position of the nucleus). Note the shift towards the apex displayed by GeaA1-GFP, compared to GeaA-GFP. (E) Hyphae treated with 200 μg/ml BFA for the indicated time, a treatment that provokes the collapse of the Golgi into large aggregates. Note that in GeaA1-GFP some labeling of the apical crescent is still visible in these conditions. (F) Staining of the plasma membrane with FM4-64 in a GeaA1-GFP-expressing strain shows that GeaA1 in the apical crescent is at or very closely associated with the plasma membrane.

In vivo microscopy used an inverted Leica DMI6000B microscope [optics/procedures as previously described in [31]]. Deconvolution was done with Huygens Professional (www.svi.nl). Further image processing was with MetaMorph (Molecular Devices). FM4-64 staining was as previously described [32].

Co-localization was studied with Li’s Intensity Correlation Analysis [33] and the Pearson’s correlation coefficient using the JACoP plugin [34] of ImageJ (http://imagej.nih.gov/ij/). Maximal intensity projections of 3-plane deconvolved z-stacks (total width = 600 nm) were used. Regions for co-localization assays were selected as described in [31]. Due to the difference in intensities of the channels, image stacks in each channel were acquired consecutively using different exposure times. Co-localization was nevertheless possible because low motility of Golgi cisternae makes the time required for shifting between the two channels irrelevant. Statistical analyses of correlation coefficients was based on [35] and used the GraphPad Prism 6 (www.graphpad.com).

Results and discussion

The A. nidulans genome codes for a single member of each of the two Arf-GEF subfamilies. AN0112 encodes the GBF/Gea homologue, GeaA. HypB (encoded by AN6709, www.aspgd.org) is the BIG/Sec7 homologue [36], co-localizing with the Arf1- and PtIns4P-binding, late-Golgi marker mRFP-PHOSBP [37], [38]. HypB/mRFP-PHOSBP-containing cisternae are resolvable from Sed5-containing early Golgi cisternae in the unstacked A. nidulans Golgi [39], [40]. Late Golgi cisternae are transient and mature to RabERab11 post-Golgi carriers [41].

Fungal life is crucially dependent on exocytosis, required to maintain the cell wall. The hypBΔ null mutation severely impairs hyphal growth [36]. Using heterokaryon rescue [29] (Fig. 1A), we found that hypBΔ spores give rise to aconidial microcolonies on solid medium at 30 and 37 °C, but they do not grow at 42 °C (Fig. 1A and B). We conclude that hypBΔ is virtually lethal. Spores carrying the geaAΔ null mutation are also inviable (Supplementary Fig. 1) and, thus, both HypB and GeaA are essential for fungal growth.

hypB5 is a ts mutation [42], [43], resulting in Ala881Pro in the catalytic Sec7d [36] (Fig. 2A). hypB5 strains grow well at 30 °C but form aconidial microcolonies at 42 °C (Fig. 1B). Ala881 lying in a conserved region of the Sec7d α-helix F contributes to a hydrophobic network situated across from the catalytic hydrophobic groove that contacts Arf1 [44]. By interfering (putatively) with the folding of α-helix F, Ala881Pro might decrease HypB stability at elevated temperatures. Shifting hypB5 cells from 28 °C to 37 °C results in cessation of apical extension and mislocalization of exocytic-carriers from an apical crescent to intracellular structures [31], [41], [45], showing that exocytosis is prevented.

Fig. 2

GeaA1-Y1022C substitution lies in a region between HDS1 and HDS2 and alters a conserved GBF/Gea-specific motif. (A) Upper panel: scheme of HypB sequence homology domains [named DCB, HUS, Sec7d, HDS1-HDS4 [3], [4], identified by T-coffee alignment with yeast and mammalian homologue sequences]. HypB5 (A881P) affects a conserved amino acid residue within the catalytic Sec7d. Lower panel: scheme of GeaA sequence homology domains. GeaA-Y1022C lies between HDS1 and HDS2. (B) Summary of multiple alignments of GBF/Gea homologues from Opisthokonta and Arabidopsis thaliana in the region between HDS1 and HDS2, where GeaA1-Y1022C lies (sequences displayed are from: A. nidulans XP_657716, Saccharomyces cerevisiae CAA89558, Pichia pastoris XP_002490563, Ustilago maydis XP_757309, Rhizopus delemar EIE78027, Mus musculus BAD32197, Homo sapiens NP_004184, Danio rerio XP_694714, Strongylocentrotus purpuratus XP_003728128, Caenorhabditis elegans NP_001255140, Drosophila melanogaster NP_610761, A. thaliana NP_198766). We looked for robust alignment of HDS1 and 2. Then we inspected and manually adjusted the less conserved region between HDS1 and HDS2. 11 sequences of Aspergilli GBF/Gea invariably contain serine, tyrosine (Y1022 in GeaA), leucine (SYL) in the region between HDS1 and 2. In 27 further fungal sequences, including basidiomycetes, zygomycetes, ascomycetes, the motif is SY (L/I/V), with few exceptions (P. pastoris: SFM, Candida albicans: SFL). >30 sequences of vertebrates (including mammals, zebrafish, xenopus) contain a SFVSWL sequence, corresponding to a duplication of the motif. Echinodermata have SYF. In ecdysozoa, Drosophila has SFI, Ceratitis capitata and Bombyx mori have SY L/I, C. elegans has SWF. We conclude that a serine-aromatic-hydrophobic aminoacid motif, SΩΦ (yellow box), is conserved in GBF/Gea proteins in Opisthokonta. The A. thaliana GBF-subfamily member GLN1, functioning at the Golgi [53], possesses SFI between HDS1 and HDS2. On the contrary, among 30 BIG/Sec7 subfamily fungal members tested, only S. cerevisiae Sec7p and the Ashbya gossypii and Candida albicans Sec7 homologues contain SFF, a GBF/Gea-motif related sequence, in the region between HDS1 and HDS2 (no SFF motif variant is found in any GBF/Gea member), while no motif was found in BIG members of vertebrates/ecdysozoa.

To determine whether the essential late Golgi function of HypB for secretion can be bypassed, we sought suppressors of hypB5 at 42 °C. Out of 40 strains able to grow at 42 °C, one extragenic mutation, suA1hypB5, was obtained (Table 1). Suppression of hypB5 is partial (Fig. 1B) and suA1hypB5 itself results in constricted growth (Fig. 1B) and is dominant in diploids. Genetic analyses localized it between AN0118 (nudA) and AN0098 (nirA). The nudA to nirA 75 kb-interval contains 24 autocalled genes including AN0112 encoding the early Golgi Arf1-GEF homologue, GeaA. Sequencing of AN0112 in suA1hypB5 revealed the presence of a missense A3065G mutation (geaA1) resulting in Tyr1022Cys (GeaA1). Reconstruction by transformation of the geaA1 mutation in a hypB5 strain showed that geaA1 suppresses hypB5ts (i.e., geaA1 is indeed suA1hypB5) (Fig. 1B).

To determine whether geaA1 also suppresses hypBΔ, we deleted hypB in a geaA1 strain and, in contrast to the heterokaryotic transformants obtained when hypB was deleted in the wild type background (Fig. 1A), we recovered homokaryotic hypBΔ geaA1 double mutants, showing that HypB is dispensable in the geaA1 genetic background (Fig. 1B).

GeaA-Tyr1022 lies between the HDS1 and HDS2 domains (Fig. 2A and B) [3], [4]. Using fungal GBF/Gea and BIG/Sec7 homologues in multiple alignments, we found that Tyr1022 lies in a previously unidentified Gea-specific motif Ser-Ω-Φ, where Ω indicates any aromatic and Φ any hydrophobic residue [fungal consensus: SYL]. The Ser-Ω-Φ motif is conserved in GBF/Gea of both vertebrates and ecdysozoa (Fig. 2B).

The finding that the substitution alters a conserved GBF/Gea-specific motif suggests that it is the (partial or total) loss of one GEA-specific function in GeaA1 (GeaAY1022C) that makes SEC7-specific functions dispensable. Members of the GBF/Gea subfamily, although predominating at the early Golgi [4], have been recently reported to localize also at the trans-Golgi and TGN [46], [47]. An attractive hypothesis was that Tyr1022Cys shifts GeaA1 localization towards the late Golgi, perhaps by reducing its affinity for an early Golgi receptor, and that this suffices to bypass HypB partially.

Therefore, we investigated GeaA localization, replacing the endogenous geaA coding region by the geaA::gfp fusion allele (Fig. 3A). GeaA-GFP is functional, as shown by almost wild type growth of a strain carrying solely the geaA::gfp allele (Fig. 3B). GeaA-GFP labels punctate cytosolic structures resembling in shape and polarization towards the tip the A. nidulans Golgi cisternae (Fig. 3C and D) [38]. In growing hyphae, GeaA-GFP structures are found in the proximity of the apex, like early Golgi cisternae (late Golgi are excluded from the most proximal sub-apical area) [38], [39]. GeaA-GFP is also partly cytosolic, as suggested by the fact that nuclei are visible as “empty” hollows against the fluorescent cytosolic background (Fig. 3C). Upon brefeldin A (BFA) treatment triggering the collapse of the A. nidulans Golgi network into large aggregates [38], GeaA-GFP structures collapse into large aggregates (Fig. 3E). We constructed strains simultaneously expressing GeaA-GFP and the early Golgi marker mCherry-SedVSed5 or the late Golgi marker mRFP-PHOSBP [31], [38], [39] for co-localization analyses (Section 2 and Fig. 4 legend). These showed that GeaA-GFP substantially co-localizes with mCherry-SedVSed5 at the early Golgi, while it largely segregates from mRFP-PHOSBP at the late Golgi (Fig. 4A, C and D). Moreover, GeaA-GFP largely colocalizes with mCherry-SedV in BFA bodies (not shown). These data are consistent with localization at the early-/cis-Golgi reported for members of the GBF/Gea subfamily of Arf1 GEFs (see for example [48]).

Fig. 4

Intra-Golgi distribution of GeaA-GFP and GeaA1-GFP. Maximal intensity projections of deconvolved z-stacks, using strains co-expressing GeaA-GFP and the early Golgi marker mCherry-SedV (left panel) or the late Golgi marker mRFP-PHOSBP (right). Note the high degree of overlap (yellow color in the “merge” image) between GeaA-GFP and mCherry-SedV, contrasting the low overlap of GeaA-GFP and mRFP-PHOSBP. Coincidence in shape of fluorescent structures indicated true co-localization (examples included in the green and red boxes). Bar = 2 μm. (B) As in (A), but with strains expressing mutant GeaA1-GFP. (C) Li’s green channel intensity correlation analysis of the indicated pair of markers. The analysis was done for a region of the cells displayed in (A and B). The y-value reflects the intensity distribution in the GFP channel. The x-value depends on co-variance of the two channels. The cloud of points towards the right of the y axis (positive values) reflects positive correlation of the two channels (co-localization). The horizontal dotted green line intersects the y axis at the mean intensity value. Note that, in the GeaA/SedV graph, the majority of the “above the average intensity” pixels form a cloud on the right of the y axis, indicating localization of GeaA to the early Golgi. The opposite is true for GeaA/PHOSBP, where high intensity pixels have negative x-values, indicating inverted correlation (ICQ: the intensity correlation quotient. ICQ approaches 0.5 in complete co-localization or 0.0 in random staining) (Ai: gfp pixel intensity, a: gfp average pixel intensity, Bi: rfp pixel intensity, b: rfp average pixel intensity). (D) Graph showing the dispersion and the mean value of the Pearson’s correlation coefficient (PCC) calculated from co-localization analysis in (n) examples for each pair of markers. Blue bars indicate the standard error of the mean. The difference between the mean PCC of the pair GeaA/SedV versus the mean PCC of GeaA/PHOSBP is large and very significant, showing that GeaA-GFP mainly localizes at the early Golgi. The difference in PCC between GeaA1/SedV versus GeaA1/PHOSBP is also large, suggesting that the mutant GeaA1-GFP also predominates at the early Golgi. However, a drop in the mean PCC of GeaA1/SedV compared to GeaA/SedV suggests that localization of mutant GeaA1 at the early Golgi is reduced. This correlates with GeaA1 localization shift to the apical crescent in (B). A small increase in PCC of GeaA1/PHOSBP versus GeaA/PHOSBP is also observed, although the correlation between the two markers is still weak in the mutant GeaA1.

We next tagged the geaA1 mutant allele with gfp (Fig. 3A). GeaA1-GFP is functional as shown by typical geaA1 growth of the GeaA1-GFP-expressing strain (Fig. 3B). Moreover, geaA1::gfp suppresses hypB5 thermosensitivity similarly to geaA1 (Fig. 3B), establishing that GFP labeling has no detectable effect on GeaA1 function. Like GeaA-GFP, GeaA1-GFP localizes to Golgi-resembling cytosolic structures and the cytosol (Fig. 3C). Strikingly, it also labels, strongly, an apical crescent and an apex-associated material, resembling the localization of the secretory v-snare SynA [49] at the apical plasma membrane crescent, where exocytosis predominates, and at the Spitzenkörper, where post-Golgi carriers accumulate [41]. Plots of fluorescence intensity across the hyphae confirm this GeaA1-GFP redistribution towards the apex (Fig. 3D). Indeed, staining the plasma membrane of GeaA1-GFP-expressing hyphae with the lipophilic dye FM4-64 [50], showed that GeaA1-GFP at the apical crescent is at or very closely associated with the plasma membrane (Fig. 3F). Upon BFA treatment the majority of GeaA1-GFP collapsed in Golgi-like aggregates, but some plasma membrane labeling persisted (Fig. 3E). Using strains co-expressing Golgi markers, we determined that GeaA1-GFP partially co-localizes with the early Golgi marker SedV, although to a lesser extent than GeaA-GFP (Fig. 4B and D). Decrease in co-localization with the early Golgi marker was accompanied by a minor increase in the co-localization with the late Golgi marker (Fig. 4B and D). GeaA1 extensively co-localizes with SedV in BFA bodies (not shown). However, an increase in the proportion of BFA mRFP-PHOSBP late Golgi aggregates containing detectable levels of GeaA1-GFP (87% in geaA1 versus 44% in geaA) was observed, suggesting some change in Golgi dynamics.

To assess the localization of GeaA1-GFP in a hypB5 background, we co-cultivated a geaA1::gfp hypB5 double mutant and a hypB5 single mutant. Double mutants are easily identifiable because of their green fluorescence, but also because at 28 °C geaA1::gfp hypB5 hyphae are thicker than hypB5 hyphae. After a temperature shift to 37 °C, growth of hypB5 cells ceased as previously described [31], whereas geaA1::gfp hypB5 strains continued to grow, confirming that geaA1::gfp suppresses hypB5 in these conditions. Importantly, GeaA1::GFP apical localization is maintained in the double mutant background at 37 °C (Fig. 3C and D), consistent with the possibility that this localization is involved in suppression.

It is largely unknown why two Arf1-GEF subfamilies, asymmetrically located in the Golgi, are conserved among eukaryotes. Our finding that an Arf1-GEF mutation bypasses this asymmetry, rendering a single Arf1-GEF capable of maintaining growth, is unprecedented. GeaA1 partially bypasses HypB. Moreover, geaA1 is hypomorphic (or GeaA1 is somewhat deleterious), as geaA1 single mutants grow less than wild type. This implies that acquisition of HypB-bypassing capability by GeaA1 occurs at the expense of its physiological role. We have demonstrated that Tyr1022Cys induces a shift in the localization of GeaA1 at the expense of its early Golgi localization. This change in localization appears to be a forward displacement within the secretory pathway, as GeaA1 is, in part, redistributed towards the apical plasma membrane, where exocytic carriers are preferentially delivered. This localization shift might reflect the mutationally altered Golgi dynamics and/or Golgi exit, resulting in re-routing a portion of GeaA1 towards the plasma membrane. Alternatively, the GeaA1 localization shift might reflect a direct role of the Ser-Ω-Φ motif on GeaA localization, in which case the change in localization per se would effect the suppression, perhaps by GeaA1 acquiring function at its new apical locale. In this case, one highly speculative possibility is that GeaA1 reorganizes the exocytic pathway in such a way that the Golgi is partially bypassed. Reorganization of the secretory pathway by manipulation of the early Golgi Arf1-GEF has precedents in intracellular membrane remodeling via GBF1 function alteration by RNA viruses [51]. However, despite its minor localization at the late Golgi, we cannot discard the possibility that GeaA1 would have gained function at this compartment. According to this hypothetical possibility, GeaA1 would be capable of performing HypB essential functions at the late Golgi and Arf-GEFs would be, to some extent, interchangeable.

At present we can only speculate on the mechanism by which Tyr1022Cys shifts GeaA localization. The substitution might impair a Ser-Ω-Φ motif-mediated interaction of GeaA with a ‘recycling’ factor/adaptor hypothetically required to restore GeaA localization to the early Golgi during cisternal maturation. This implies that if GeaA1 is inefficiently recycled retrogradely, it must be rapidly sorted into exocytic carriers, as we have only detected a small increase in GeaA1 residence at the late Golgi. A slightly modified alternative is that the motif might bind one key component of a coincidence detection module restricting GeaA localization to the early Golgi. Loss of interaction with this component could make GeaA1 localization largely dependent on a second component with broader distribution. It is notable that a GBF1 HDS1 and HDS2 module containing the Ser-Ω-Φ motif has been shown to redirect GBF1 to the leading cell edge through phosphoinositide binding [52]. Similarly, the GeaA1 mutation might modify a lipid-recognizing module, such that loss of affinity for a Golgi-specific lipid would be accompanied by gain of affinity for an apical crescent-specific lipid, resulting in the shift in GeaA1 localization. In any case, future studies addressing how the GeaA1 substitution has such an impact on its localization constitute a promising tool to understand the mechanisms by which GeaA specifically localizes to the Golgi.