A hallmark of G protein signaling is alteration of cAMP levels; indeed, binding of 9(S)-HODE to the mammalian GPCR, G2A, inhibits cAMP accumulation [35]. Because the role of cAMP in development had already been described for the genetic model A. nidulans [42,43,44,45,46], the first Aspergillus species studied for oxylipin developmental effects [19], we measured cAMP levels in tissues of this fungus exposed to pure plant oxylipins. We first examined the wild type response to increasing concentrations of 13(S)-HpODE and found the cAMP burst to be released in a dose-dependent manner. Fungal tissues were treated with 12.5 μL EtOH containing increasing concentrations of 13(S)-HpODE with a resulting increasing production of cAMP (all p < 0.05; Figure 1a). Previously, individual oxylipin species were measured in homogenized A. nidulans tissues at approximately 30 to 110 nmol/g dry weight [12]. In the current study, assuming a water content of 70% in fungal tissues, 33 nanomoles oxylipin were added per gram dry weight when oxylipins were added at 100 nM and correspondingly more at higher concentrations (Figure 1a). At 10 μM, the cAMP response appeared to be at saturation (as also seen for 9(S)-HODE perceived by G2A [35]), and increasing from 10 μM to 100 μM 13(S)-HpODE did not cause a significant difference in cAMP response (p = 0.6), so oxylipins were applied at 10 μM in subsequent experiments.

The fungus was then examined for response to four oxylipins: 13(S)-HpODE, 9(S)-HpODE and their hydroxy derivatives, all of which activated the mammalian GPCR, G2A [35]. These four metabolites are generated by both plant and mammalian lipoxygenases [32,33,47,48]. As seen in Figure 1b, wild type A. nidulans produced more cAMP after a twenty second exposure to each of these oxylipins than after a twenty second exposure to the same volume of EtOH carrier (all p < 0.05). 9(S)-HpODE caused the smallest response; the other three oxylipins caused a nearly identical increase in cAMP. The act of grinding effectively spreads cAMP uniformly throughout tissues; however, local responses to oxylipins at the cellular level may have created much higher gradients of cAMP than we could measure with our methods.

As a final control to confirm accuracy of our methodology, we assessed an adenylate cyclase mutant (∆cyaA, [44]) and its parental wild type cyaA strain for their responses to 13(S)-HpODE. Adenylate cyclase is required for activation of cAMP release. As expected, addition of 13(S)-HpODE to the wild type increased cAMP production 1.8 fold over the EtOH control (p = 0.05). However, there was no significant difference between samples of the ∆cyaA mutant treated with EtOH or 13(S)-HpODE (p = 1.4; Figure S1).

Having established the cAMP response to oxylipins, we next assessed a series of GPCR mutants for alterations in oxylipin response. As a preliminary screen, we first assessed the ability of the ten GPCR deletion mutants that were constructed at the time of this study (∆gprA, ∆gprB, ∆gprC, ∆gprD, ∆gprE, ∆gprF, ∆gprG, ∆gprH, ∆gprI,and ∆gprK, Table 1) to respond to polyunsaturated fatty acid (linoleic and arachidonic acids) saturated discs by hyper-conidiation. This method provides a rapid and inexpensive screen for lack of response to oxylipins [29]. Of the ten GPCR mutants, ∆gprA and ∆gprD mutants did not exhibit hyper-conidiation when exposed to 0.5 and 1 mg fatty acid-soaked discs, with ∆gprD showing the greatest defect (data not shown).

To better assess the possibility that GprA or GprD might be involved in oxylipin perception, A. nidulans ∆gprA and ∆gprD strains, as well as ∆gprG, a randomly chosen GPCR deletion mutant exhibiting the wild type conidiation response to linoleic acid, were then assessed for their cAMP response to four pure oxylipins: 13(S)-HpODE, 9(S)-HpODE, 13(S)-HODE and 9(S)-HODE. As shown in Figure 2, both ∆gprA and ∆gprG produced significantly more cAMP when exposed to each of the four oxylipins than when treated with an equivalent volume of EtOH (all p < 0.05). In contrast, the ∆gprD mutant did not respond to any of the four oxylipins with an increase in cAMP concentrations over that of the EtOH control (all p > 0.1).

Because of the growth and sporulation defects of the A. nidulans ∆gprD strain [50] and it not being a pathogen of note, we followed up with physiology studies with mutants of the homologous gprD genes in the agriculturally relevant species, A. flavus. As A. nidulans is a model organism for other filamentous fungi, including A. flavus, we reasoned that if GprD was involved in oxylipin perception, then it could play a role in A. flavus biology, as A. flavus oxylipin mutants are aberrant in spore and sclerotial development [16,17]. A. nidulans encodes at least sixteen putative GPCRs [38]. BLAST searches querying each of these GPCRs identified fifteen putative GPCRs in A. flavus. Of these, two appear to be highly similar to A. nidulans GprD. The first of these is A. flavus GprD (AFLA_135680), which shares 67% amino acid identity (e-value 2e-164) with A. nidulans GprD. The second we call GprC (AFLA_074150), and it shares 47% identity (e-value 6e-87) with A. nidulans GprD and 47% identity with A. flavus GprD (e-value 1e-119). An alignment of all three GPCRs is provided in Figure S2.

Individual deletion mutants of both gprC and gprD were created by replacing each gene with pyrG (Table 1, Figure S3). A third strain, KD::gprCD (for “knock-down of gprC and gprD”) with both genes down-regulated by RNAi [52], was also created with the thought that the two proteins may have overlapping function due to their high identity (Table 1, Figure S3). All mutants were confirmed with Southern blots, and in the case of KD::gprCD, Northern blots as well (Figure S3).

Considering the critical role of oxylipins in A. flavus quorum sensing development and the possibility that GprC/D could be involved in oxylipin perception, the A. flavus GprC and GprD mutants were examined for density dependent development. The fungi were grown at low (10³ spores per plate) and high (10⁷ spores per plate) densities, and production of conidia, sclerotia, and AF were measured after seven days (Figure 3). At low density, there were subtle differences between the mutants and wild type (NRRL3357), but the most striking differences were seen at high density. Here, production of conidia was high in the wild type, but it was reduced 2.3 fold in the ∆gprC strain and was almost completely absent in the KD::gprCD mutant (both p < 0.001, Figure 3a,d). Conversely, sclerotia production was minimal in the wild type at high density. As seen in Figure 3b,d, all three mutants exhibited profound increases in sclerotia production at high density. The ∆gprC and ∆gprD strains produced 35.8 and 30.9 fold more sclerotia, respectively, than the wild type, while the KD::gprCD strain produced 60.7 fold more sclerotia than the wild type, approximately doubling what the two single mutants produced (all p < 0.001). AF biosynthesis was also regulated in response to population density changes, being up-regulated at low density and down-regulated at high density as previously reported [16]. The ∆gprD and KD::gprCD mutants showed modest increases in AF compared to the wild type at low density. Most remarkably, the KD::gprCD mutant produced more AF at high density than wild type AF production at the favorable low density conditions (Figure 3e).

Because it appeared that depletion of both gprC and gprD caused the fungus to stay locked in a low-density development pattern, we hypothesized that these receptors may be required for transmitting a signal produced at high density. The extract from the spent medium of a high density culture induced a significant increase in conidiation in the wild type [16], so this extract and an extract from un-inoculated plates (media control) were applied to cultures of the gprC and gprD mutants (Figure 4). As previously reported, the high-density extract stimulated a significant increase in conidiation when applied to the wild type (p = 0.002). In contrast to wild type, conidiation for all three mutants appeared to be slightly decreased in response to the high-density extract, though this was only minimally significant for the ∆gprC strain (p = 0.049).

Next the three mutants were assessed for their ability to respond to linoleic acid (LA), a known inducer of conidiation in A. flavus [29]. Conidiation was measured in response to disks containing 1 mg LA or an ethanol control (Figure 5). As previously reported, the wild type exhibited a significant increase in conidiation in response to LA (p = 0.003). In a similar fashion, all three mutants also produced significantly more conidia when exposed to LA (p = 0.002 for ∆gprC, p = 0.001 for ∆gprD, and p = 0.0008 for KD::gprCD). Although the relative increase in conidiation for the KD::gprCD strain from the ethanol control, 5.5-fold, was greater than that of any other strain tested, it still produced fewer conidia than any of the other strains (data not shown), as seen in the previous experiments.

10⁶ spores/mL were washed in GMM and suspended in 400 mL GMM in 1 L flasks, and grown at 37 °C at 300 r.p.m. for 18 h. At this time, mycelia were harvested by vacuum filtration, washed once with sterile GMM, and divided into 200 mg pieces which were inoculated into 20 mL fresh, sterile GMM in 50 mL conical tubes. These subcultures were equilibrated for 1 h at 37 °C at 300 r.p.m.

Initially, different phenological stages of A. nidulans were compared for their response to 13(S)-HpODE: (i) 12-h and (ii) 18-h liquid shake cultures (cultured as described above), and (iii) tissues harvested from 18-h liquid shake cultures by filtration and asexually induced by spreading onto GMM agar and incubating for 4 h at 37 °C in constant light. Eighteen hours liquid shake cultures were the only ones in which 13(S)-HpODE treatment produced a significant change in cAMP concentrations (data not shown), so these conditions were used for all subsequent experiments. Depending on the experiment, three or four biological replicates were performed for each treatment. Each experiment was performed twice with similar results.

After equilibration, cultures were treated with 12.5 μL of 5 mg/mL pure oxylipin (in EtOH) or with the same volume of pure EtOH as a control. Thus, oxylipins were at a final concentration of 10 μM. When other concentrations of oxylipins were needed, concentrated or diluted stocks were used so that the same volume, 12.5 μL (in EtOH), was always added to each sample. Samples were shaken by hand for precisely twenty seconds. Next, mycelia was separated from growth medium by filtration using Grade 1 Whatman filter paper (11 μM), scraped into microcentrifuge tubes, and flash-frozen in liquid nitrogen. Samples were stored for a maximum of 24 h at −80 °C until extraction and cAMP quantification.

Frozen mycelia were ground to a fine powder with a clean, pre-chilled mortar and pestle in liquid nitrogen and weighed on a microbalance. Tissue was resuspended by vortexing in 50 mM Tris.HCl/4 mM EDTA (pH 7.5) followed by boiling for five minutes and subsequent acidification by addition of concentrated HCl to a final concentration of 0.1 M. Samples were then incubated for ten minutes on ice and centrifuged for five minutes at 13,000 r.p.m. at 4 °C. The supernatant was transferred to a fresh microcentrifuge tube and flash-frozen in liquid nitrogen. Acidified supernatants were split into two subsamples, one for quantification of cAMP and the other for quantification of protein.

cAMP was measured using the Direct cAMP colorimetric (EIA) kit (Enzo Life Sciences, Exeter, UK) according to the manufacturer’s directions. Reactions were performed in 96 well plates and optical density (OD₄₀₅–OD₅₉₅) was measured using a Victor³ V multiwell plate reader (Perkin Elmer, Waltham, MA, USA). Corrected absorbance readings were fitted to a four-parameter sigmoidal model, and cAMP concentrations of unknowns from each plate were interpolated from that individual plate’s standard curve. The R² for all plate-specific standard curves was always ≥0.99 except for the ∆cyaA test plate, for which the R² was 0.94. Two technical replications were performed on each sample.

Protein content was measured using the Bradford reagent supplied by BioRad (Hercules, CA, USA) according to the manufacturer’s directions. To neutralize the 0.1 M HCl in the samples to be used for protein quantification, an equal volume of 0.1 N NaOH was added to each subsample.

Extracts from high-density and media-only plates were collected as described [16]. Briefly, thirty 100-mm diameter plates containing 25 mL GMM + 2% sorbitol (1.5% agar) were overlaid with 10 mL GMM + 2% sorbitol (0.75% agar) containing 10⁷A. flavus wild type (NRRL3357) spores. The same set-up was repeated with no spores added for a media-only control. All plates were incubated in the dark at 29 °C for five days. The plate contents were homogenized in a blender (ten plates plus 250 mL sterile distilled H₂O a time) and transferred to a 4-L beaker. One volume of ethyl acetate was mixed in, and the macerate was allowed to settle overnight. The organic layer was removed, filtered through Whatman™ #1 filter paper and evaporated using a rotary evaporator. The final extract was resuspended in 2.5 mL ethanol, topped with nitrogen gas, and stored at −80 °C.

The extracts were diluted 1:10 in ethanol, and 100 μL of diluted extract was spread onto 60-mm diameter plates containing 10 mL GMM + 2% sorbitol (1.5% agar). Plates were allowed to dry for fifteen minutes and were then inoculated with 10⁶ spores in 3 mL GMM + 2% sorbitol (0.75% agar). Four replicates per strain per treatment were prepared and placed in the dark at 29 °C for three days. Enumeration of conidia was carried out as described for density-dependent physiology tests.