Multimodal sensorimotor system in unicellular zoospores of a fungus

All organisms rely on sensory systems to gather information about their surroundings using external stimuli. The integration of individual sensory modalities into multisensory systems greatly increases the amount of information an organism can use to form responses and behaviors. Although multimodal sensory systems are common in multicellular, motile organisms, there are significantly fewer multisensory systems known from unicellular eukaryotes (Govorunova and Sineshchekov, 2005). The relative rarity of studies characterizing these systems in unicellular, laboratory-tractable organisms has resulted in a significant taxonomic gap between current model systems and animals.

To address this deficit, we focused on fungi characterized in part by motile, zoosporic life stages. Zoosporic fungi collectively form a clade within the ‘early-diverging lineages’ of fungi, outside the better known Ascomycota and Basidiomycota. They are largely found in freshwater ecosystems with a global distribution (James et al., 2014). Zoosporic fungi are typically characterized as saprobes, such as Allomyces, although parasitic life strategies on both plant and animal hosts also do exist (Longcore et al., 1999). Similar to all fungi, colonies of Allomyces use mycelia to absorb nutrients and ultimately grow reproductive structures. Unlike most fungi, Allomyces produce zoosporangia, terminations of mycelial branches that make, store and ultimately release a multitude of single-celled, flagellated propagules, termed zoospores (Olson, 1984). When the appropriate environmental cues are present, zoospores are produced en masse, eventually bursting from zoosporangia (James et al., 2014). Once in the water column, the zoospores rely on a single, posterior flagellum to propel themselves away from the parent colony and towards suitable substrates or hosts (Olson, 1984).

During dispersal of the zoosporic life stage, interpretation of environmental cues is critical for the survival and success of the future colony (Dukes and Apple, 1961). Allomyces zoospores have a finite amount of endogenous energy reserves, and no zoospore is known to metabolize energy from external sources (Suberkropp and Cantino, 1973). This energetic constraint places significant pressure on the zoospore to efficiently locate a favorable environment for settlement and growth. The evolution and maintenance of a sensory system within the unicellular zoospore allows it to evaluate external conditions, move towards suitable habitats and avoid hazards (James et al., 2014). Previous studies across zoosporic fungi led to the discovery of a number of sensory modalities that guide zoospore dispersal including chemotaxis, phototaxis and electrotaxis (Machlis, 1969; Morris et al., 1992; Robertson, 1972). However, these studies have neither tested a single species for multiple sensory modalities nor posited the possibility that these single senses may only be a portion of a more complex sensorimotor system guiding zoospore dispersal and settlement.

Allomyces zoospores are known to use either chemotaxis or phototaxis, depending on the species, to guide dispersal and settlement (Pommerville and Olson, 1987; Robertson, 1972). Chemotaxis towards the source of amino acid gradients allows zoospores to congregate at the site of an injury or on decaying material in the water column (Machlis, 1969). Allomyces macrogynus zoospores possess refined chemosensation, settling on substrates at varied rates in response to different amino acids (Machlis, 1969). In contrast, the zoospores of Allomyces reticulatus display positive phototaxis, potentially leading spores to swim towards the air–water interface (Robertson, 1972). In animals, positive phototaxis and subsequent ‘rafting’ on floating debris work to considerably increase the dispersal range of planktonic larvae (Epifanio et al., 1989). Similarly, spores attracted to the water surface may encounter floating debris, algal hosts or currents that aid their dispersal. Though closely related, studies to date have not tested for the presence of chemotaxis in A. reticulatus or phototaxis in A. macrogynus.

While little is known about the molecular mechanisms of chemotaxis in fungal zoospores, the underpinnings of their photosensitivity are beginning to come to light. The protein responsible for light detection in zoospores of Blastocladiella emersonii, a close relative of Allomyces, is a bacteriorhodopsin gene called CyclOps (Beme-Cycl) (Avelar et al., 2014). Unlike many bacteriorhodopsins that regulate ion channels, Beme-Cycl acts through regulating intracellular cGMP (Avelar et al., 2015). A CyclOps gene is present in the genome of A. macrogynus (Amag-Cycl), a species that has been anecdotally described as having phototactic zoospores (Olson, 1984). A recent study by Gao et al. (2015), however, contradicts claims of phototaxis in A. macrogynus by revealing the proteins encoded by Amag-Cycl are orders of magnitude less sensitive to dark/light transitions than Beme-Cycl proteins (Gao et al., 2015). This raises questions about the sensitivity of CyclOps proteins required for phototaxis, and the mutations that have occurred to precipitate the loss of function in Amag-Cycl proteins.

The uncertainty surrounding the sensory systems of Allomyces zoospores, created by these conflicting reports, demands experimental evidence to clarify the number and type of modalities used during zoospore dispersal and settlement. Addressing the current deficits in our understanding of fungal sensory systems is also motivated by the potential to discover a multimodal sensorimotor system that will further our knowledge of multisensory evolution and function in early opisthokonts. Here, we investigated the responses to chemical and light gradients in three species of Allomyces, revealing previously unknown variation in fungal sensory systems and uncovering a novel multisensory system in a zoosporic fungus.

We used Allomyces arbusculus (strain ATCC 10983), Allomyces reticulatus (strain California 70 from ATCC, cat. no. 42465) and Allomyces macrogynus (Roberson lab, Arizona State University). We kept cultures of A. macrogynus and A. arbusculus in both solid and liquid media. For solid media, we used Machlis (1953) Emerson YSS (HiMedia M773) at half strength. Colonies was transferred aseptically in a laminar flow hood every 4 weeks by moving a chunk of mycelia from the leading edge of the colony onto a new plate. For liquid media, we followed the protocol for Machlis’ medium B (1953). We inoculated liquid cultures via sterile transfer of sporangia and mycelia into a 125 ml Erlenmeyer flask containing 50 ml of liquid media and antibiotics (Machlis, 1953) for the first generation. For all subsequent generations kept in liquid culture, we used dilute salts solution to initiate sporulation of the previous generation's colonies. We then added 1 ml of this zoospore–dilute salts solution (referred to as ‘sporulation product’ from here on) into new liquid media. Both liquid and solid cultures were grown on an orbital shaker at 140 rpm and kept at room temperature (∼24°C). Cultures in liquid media were kept for a maximum of 5 days, and were considered ready for sporulation after 72 h. Because A. reticulatus did not grow well in liquid media, we cultured A. reticulatus on full-strength Emerson YSS media for no more than 6 weeks. Propagation of A. reticulatus cultures in solid media was performed identically to propagation of the other species.

Liquid cultures of A. macrogynus and A. arbusculus were considered for sporulation after 72 h. We visually inspected colonies under a microscope to confirm the absence of gametangia. Using a stainless steel sterile mesh, we strained the colonies out of growth media and rinsed them 5 times with dilute salts solution to remove the growth media. We then placed the rinsed colonies and strainer in a pyrex dish with 10 ml of dilute salts solution and allowed them to sporulate for no more than 90 min. Once either sufficient spore density had been reached [5×10⁵ spores ml⁻¹ for chemotaxis (Machlis, 1969), 1×10⁶ spores ml⁻¹ for phototaxis (Avelar et al., 2014)] or 90 min had elapsed, the mesh and colonies were lifted out of the dish (Machlis, 1969).

Because A. reticulatus was only grown on solid media, we took a surface scraping to lift sporangia from the agar and placed it into a pyrex dish with 10 ml dilute salts solution (Saranak and Foster, 1997). If no zoospores were present, we replaced the dilute salts solution every 20 min for the first hour. Sporulation typically occurred within 8 h, after which the colonies were strained from the dilute salts solution.

We conducted phototaxis trials in a custom-made 1×3×5 cm (W×H×L) Plexiglas chamber. We added 10 ml of sporulation product, diluted to 1×10⁶ spores ml⁻¹ (Avelar et al., 2014), to the test chamber and left the solution for 15 min in total darkness to dark adapt and randomize spore distribution. Preliminary trials showed that an adjustment period of 15 min, rather than 30 min (Avelar et al., 2014), was sufficient for zoospore randomization and dark adaptation. After the adjustment period, spores were exposed to a white light (USHIO halogen bulb) through a 5 mm diameter fiber optic cable positioned 5 cm from the leading edge of the test chamber. To calibrate the light intensity, we used a JAZ Oceanoptics light sensor with Spectrasuite v2.0.162. We adjusted the intensity of the light to 1.8–1.0×10¹³ mol of photons cm⁻² on the edge closest to the light source. We measured temperature with a thermometer placed in the center of the test chamber with the light source on for 15 min. The chamber was designed with thick plastic to limit heat transfer to the zoospore suspension; no appreciable thermal turbulence or change in temperature was observed after 15 min of light exposure, at which point we divided the test chamber into four sequential sub-chambers (1×3×1.25 cm) using sterile glass slides. This resulted in four sub-chambers (1–4) arranged linearly so that sub-chamber 1 was closest to the light source, while sub-chamber 4 was the farthest away (Fig. S1A). We gently agitated the liquid in each sub-chamber to homogenize swimming spore distribution and counted swimming spore density in four, 10 μl samples from each sub-chamber using a hemocytometer. To account for the variation in spore counts between each trial in both control and light treatments, we quantified the number of swimming spores in each chamber as a proportion of the total number of spores counted in that trial. This resulted in less inter-trial variation. A total of 10 control treatments (no light exposure) and 18 experimental treatments were conducted for each species.

Chemotaxis trials followed the protocol established by Machlis (1969). The amino acids and combinations thereof we tested were lysine (K), leucine (L), proline (P), L+K, L+P, K+P, L+K+P and buffer (5×10⁻³ mol l⁻¹ KH₂PO₄) solution (referred to as ‘treatment solutions’ from here on). All amino acid concentrations were 5×10⁻⁴ mol l⁻¹ for each amino acid in all treatments. We created a chemical dispersal apparatus by drilling a hole through the lid of a 60×15 mm Petri dish and inserting a 5 mm inner diameter glass pipette. We secured dialysis membrane (3500 MWCO) to the tip of the pipette and positioned it 3 mm above the bottom of the Petri dish (Fig. S1B). This creates a gradient in the Petri dish of whatever solution is placed behind the dialysis membrane, allowing zoospores to navigate to the membrane, where they settle and can later be counted. The dialysis membrane was soaked and rinsed with deionized water for 24 h to remove potential contaminants and bubbles that would affect results (Carlile and Machlis, 1965). Turbulence in the Petri dish would greatly affect the strength of the gradient. To avoid this, we placed the modified Petri dishes in a larger, sealed container to avoid drafts and the treatment solution was added last. This resulted in no movement of any container between the addition of treatment solution and the end of the trial.

To test chemotactic and settlement responses to these treatments, we added 10 ml of sporulation product (diluted to 5×10⁵ spores ml⁻¹) to the Petri dish and 300 μl of treatment solution into the pipette. As a control, we used 300 μl of buffer alone. We allowed the spores to react to the gradient in total darkness for 90 min. At the end of the trial time, we removed the pipette and dialysis membrane from the dish and gently shook it to remove excess liquid (Machlis, 1969). We counted spores settled on the membrane under an Olympus szx7 microscope at 400× or greater magnification.

Because genomic data existed for A. macrogynus but expression data did not, we used PCR to attempt to identify whether CyclOps genes are expressed in A. macrogynus zoospores. mRNA was extracted from zoospores using a NucleoSpin RNA XS kit. We synthesized cDNA using the Clontech cDNA synthesis kit. Primers were designed from putative rhodopsin/guanalyl cyclase fusion proteins identified from the BROAD institute's Allomyces macrogynus genome, using the CyclOps protein from Blastocladiella emersonii as the bait sequence (Avelar et al., 2014). Sets of primers were designed in IDT PrimerQuest. All PCR products were visualized using a 1% agarose gel with 100 bp ladder. PCR for CyclOps was only done on A. macrogynus as transcriptomes for A. reticulatus and A. arbusculus would yield expression data.

RNA was isolated from zoospores of A. reticulatus and A. arbusculus using a Nucleospin xsRNA kit, and cDNA was synthesized using the NEB Next RNA First and Second Strand Synthesis modules. cDNA was sequenced using a multiplexed Illumina HiSeq lane at approximately 50× coverage.

We trimmed Illumina data using Trimmomatic (Bolger et al., 2014), assembled using Trinity 2.0, and analyzed on the UCSB Osiris bioinformatics platform (Giardine et al., 2005; Oakley et al., 2014). Putative CyclOps proteins were identified using Beme-Cycl as a bait sequence in BLASTn searches against the A. macrogynus genome (NCBI bioproject 20563) and the new A. reticulatus and A. arbusculus zoospore transcriptomes. Any sequence with an e-score lower than 1e−40 was considered as a candidate. We then used the ‘get orf’ feature from Trinity to produce predicted proteins from candidate genes, selecting the longest open reading frames (ORFs) with the highest similarity to Beme-Cycl protein when reciprocally BLASTed using blastp. We identified the putative CyclOps gene in both A. reticulatus and A. arbusculus (Aarb-Cycl). Reads from the A. arbusculus transcriptome were mapped back to the putative Aarb-Cycl using Bowtie2 (Langmead and Salzberg, 2012) and visualized using IGV viewer (Thorvaldsdóttir et al., 2013). Because the bacteriorhodopsin and guanylate cyclase domains appeared in different ORFs on the same strand, each base was manually examined for uncertainty and low support. A guanine at site 378 was manually removed because of low coverage and low support in the reads, implying that the addition of guanine at position 378 was most likely an assembly artifact. The manually edited Aarb-Cycl gene produced a single predicted ORF with the appropriate bacteriorhodopsin–guanylyl cyclase domains (Fig. S2).

Candidate proteins were aligned using MAFFT under the L-INS-i strategy. Outgroups were selected based on a previous analysis (Avelar et al., 2014) (Fig. S2). The alignment was used to create a phylogeny of candidate genes with RAxML 8 (Stamatakis, 2014) and 100 bootstrap replicates using the GTR+Γ model. Trees were visualized in Evolview (He et al., 2016) and annotations were added in Adobe Illustrator.

Comparisons of zoospore phototaxis behavior were analyzed using JMP v13.0. Average spore counts per sub-chamber per trial were analyzed using a one-way ANOVA. Although sample size was low, each sample represents an average of four replicates for each treatment. Zoospore chemotaxis was analyzed using Wilcoxon each pair test because of the non-parametric distribution of results and small sample size (n=10).

Zoospores of A. reticulatus showed no significant deviation from the control when exposed to any amino acid treatment (P>0.05 for all treatments) (Fig. 1A). In accordance with existing literature, A. reticulatus showed a significant response to a directional light source (Robertson, 1972). The number of zoospores swimming in the sub-chamber closest to the light source (Fig. 2A) was significantly higher than that when no light source was present (P=0.0001).

As seen in previous experiments, A. macrogynus zoospores displayed a significant response to all amino acid treatments compared with the control (Machlis, 1969) (P, P=0.0294; L, P=0.0275; K, P=0.0294; and any combination of two or three amino acids when compared to a control P<0.001) (Fig. 1C). The zoospore response to increasing treatment complexity was non-linear though roughly equal for all unique combinations of equal complexity. Allomyces macrogynus zoospores showed no response when exposed to a directional light source (Fig. 2C). The number of zoospores in all sub-chambers was the same for both light and dark trials (P=0.8801).

As expected from existing literature, A. arbusculus zoospores responded similarly to the zoospores of A. macrogynus when exposed to amino acid treatments. The number of spores settled increased in a non-linear fashion as the complexity of the treatment increased, though at a lower average number of settled spores when compared with A. macrogynus (K+P, P=0.0014; K+L, P=0.0014; L+P, P=0.0008; K+L+P, P=0.0004) (Fig. 1B). As opposed to A. macrogynus, A. arbusculus did not respond to, or could not detect, gradients of single amino acids (P>0.05 for all single amino acid treatments) with the possible exception of proline. We found that no zoospores settled when A. arbusculus was exposed to trials of proline alone, potentially indicating negative chemotaxis or inhibition of settlement in response to gradients of proline by itself. However, we cannot definitively resolve this reaction with our sample size (P versus control: P=0.072, N=10). When exposed to a directional light source, A. arbusculus displayed positive phototaxis (Fig. 2B). The number of zoospores in the sub-chamber closest to the light was significantly higher (P<0.0001) in light versus dark trials and was comparable to the response of A. reticulatus zoospores.

Phylogenetic analysis of putative CyclOps genes revealed the presence of CyclOps in all three samples (A. arbusculus and A. reticulatus: transcriptome data; A. macrogynus: previously available genome data). The single copy of CyclOps recovered from the A. arbusculus transcriptome showed a possible truncation of the guanylyl cyclase domain (Fig. 3). Aarb-Cycl and Amag-Cycl share a mutation at the putative functional residue F313 to I313. Despite the success of positive controls indicating successful PCR amplification, no primers successfully amplified Amag-Cycl from A. macrogynus zoospores.

How sensory modalities evolve and integrate with other behavioral circuits remains an open question in both neurobiology and evolutionary biology. The Allomyces genus, with new variation in sensory systems first discovered in this study, will aid in answering these questions. Previous studies showed Allomyces spores use either chemotaxis or phototaxis to guide dispersal. Here, we reveal that the sensorimotor system in A. arbusculus is multimodal – able to process both chemical and light cues. Additionally, our results reveal the previously unknown complexity and variation of sensorimotor systems in Allomyces.

We report previously unknown variation in the distribution of sensory modalities across the genus. This variation manifests in two ways: the types of sensory modalities used by each species of Allomyces and the responses of A. arbusculus and A. macrogynus zoospores to the same amino acids. The lack of phototaxis, coupled with the inability to amplify CyclOps from zoospore mRNA in A. macrogynus is quite interesting because of conflicting reports between previous studies. These studies state either that A. macrogynus zoospores are phototactic (through anecdotal evidence) (Olson, 1984) or that the photosensitive protein, Amag-Cycl, does not respond effectively to light (Gao et al., 2015). Our findings support the hypothesis that CyclOps in A. macrogynus no longer effectively differentiates between light and dark, and suggest that CyclOps expression has been lost in A. macrogynus zoospores. The shared distribution of CyclOps expression and photosensitivity in A. reticulatus, A. arbusculus and Blastocladiella make it likely that phototaxis was an ancestral character state in Allomyces (Fig. S3). It is currently unclear whether this loss occurred in a natural ecosystem or as a byproduct of being kept in culture. Anecdotal evidence exists for the loss/suppression of phototaxis in lab-cultivated fungus (Olson, 1984), and particular long-term algal cultures show degraded eyespot organization (Moldrup et al., 2013). Unfortunately, because all evidence of phototaxis in A. macrogynus is anecdotal, there is no way to know the state of the original culture without re-isolation of a wild strain. If phototaxis was lost in a natural setting, the dispersal and settlement patterns of A. macrogynus zoospores would likely deviate greatly from those of both A. reticulatus and A. arbusculus in the wild. These differences would present a mechanism for potential speciation driven by mutation in sensorimotor systems. However, if the relatively recent cultivation of A. macrogynus led to the degradation of photosensitivity in zoospores, it would present an excellent opportunity to examine how multisensory systems in simple organisms remodel after the loss of an input.

The variation in Allomyces sensory systems discovered in this study coupled with the convergent function of CyclOps and animal opsins make it critical to our understanding of the evolutionary history of light sensing. Animal photoreception, mediated through Type II opsins, operates through intracellular regulation of cyclic nucleotides. Non-animal photoreception, mediated through Type I opsins, operates through channel and sensory rhodopsins (SRII). However, unlike Type II opsins, SRIIs do not regulate cyclic nucleotide concentrations. Instead, SRIIs often indirectly regulate CheY, a protein that controls flagellar motion (Klare et al., 2004). Type I and II opsins are considered a spectacular example of convergent evolution (Larusso et al., 2008). CyclOps represents a functional evolutionary convergence, where the protein may modulate intracellular cGMP levels like some Type II phototransduction cascades, yet the sequence of CyclOps indicates its origin as a Type I opsin (Avelar et al., 2015). Studying the evolution of CyclOps sequence function through the variation in Allomyces will yield further insight into the evolution of this novel photosensory mechanism.

The failure of A. reticulatus to move and settle on the cellulose membrane does not necessarily indicate a complete lack of chemotaxis, but does highlight variation in Allomyces sensory system evolution. The amino acids tested here are frequently associated with wounded tissue and decaying organic matter, and were isolated from crushed, natural substrate in previous studies, making them appropriate for guiding the distribution and settlement of sessile saprobes (Carlile and Machlis, 1965; Machlis, 1969). While previous studies uncovered differences in the combinations of amino acids that A. arbusculus and A. macrogynus zoospores move toward (Machlis, 1969), no one tested the possible loss of this chemotactic response in closely related species. The absence of chemotaxis to these amino acids in A. reticulatus may reveal a turnover in sensory capabilities responsible for controlling a vital behavior across the genus. In future studies, we will use the variation in both chemotaxis and phototaxis across Allomyces as a platform to understand behavioral integration, multisensory evolution and sensory remodeling in an organism closely resembling ancestral opisthokonts.

Allomyces arbusculus zoospores present an easily culturable, laboratory tractable system for investigating multimodal sensation in unicellular systems and its underlying mechanisms. Previous studies have independently confirmed that zoospores use a diversity of senses, but fungal zoospores have only been tested for a single sense per species (Avelar et al., 2014; Machlis, 1969; Morris et al., 1992; Robertson, 1972). Our findings represent the first study of zoospore multimodal sensing, and a concrete example of multimodal sensorimotor control in a unicellular opisthokont (Figs 1,2). Though choanoflagellates fall within Opisthokonta (Cavalier-Smith et al., 2014) and potentially exhibit both aerotaxis and pH-taxis, it remains unknown whether the sensory modalities in colonies (aerotaxis) are also used to direct the dispersal stage (pH-taxis) (Kirkegaard et al., 2016; Miño et al., 2017).

Zoospores, as unicellular, flagellated cells, might closely represent the ancestral opisthokont phenotype (Cavalier-Smith et al., 2014). Specialized cell types and functions may often evolve through subfunctionalization followed by elaboration of the ancestral cell's functions (Arendt et al., 2016). This implies that as multicellular opisthokonts evolved, the foundation for specialized sensory modalities/cells already existed in ‘pluripotent’, generalized pathways. Under the subfunctionalization hypothesis, multimodal systems in unicellular organisms, such as we report in A. arbusculus, must have evolved prior to subfunctionalization in ancestral, multicellular opisthokonts. Future studies of the multimodal sensorimotor system in A. arbusculus zoospores and the variation in modalities across Allomyces may uncover how multiple senses became integrated into behavioral responses in ancestral opisthokonts. Understanding these mechanisms in the context of the cellular subfunctionalization hypothesis will further our understanding of sensorimotor evolution, elaboration and individuation through cellular specialization.

We present a multimodal sensorimotor system in a unicellular life stage of a fungus. The multisensory system of A. arbusculus zoospores is an excellent system to study how sensory modalities integrate into existing behavioral regimes. Together with existing models of unicellular sensory mechanisms, the variation in sensory modalities in Allomyces and other early diverging fungi will allow us to formulate more accurate conclusions about the evolution of complex sensory systems and multisensory systems in ancestral eukaryotes. Lastly, the relatively narrow taxonomic breadth associated with the multiple transitions in sensory systems during Allomyces evolution will allow testing of broad questions in evolution; such as the role multimodal cell types play in the origin and evolution of specialized sensory systems, and how emergent behaviors evolve during sensory remodeling.

The authors would like to thank Robert Roberson for supplying A. macrogynus and Jason Stajich for his advice and guidance on molecular work in fungus systems. Additionally, the authors would like to thank Oren Malvey for his help in maintaining cultures.