Finding a place and leaving a mark in memory formation

Introduction

The ability of organisms to navigate the environment, while, integrating complex environmental information is critical for survival and reproduction. Several environmental factors including temperature, food, and water availability drive the animals to specific locations. Navigational strategies can range over large spatial scales (long distance migratory birds vs short-range homing in mice) and involve a variety of sensorimotor systems and memory processes. Studies conducted in ants, bees, wasps and other insects have shown that shared navigational strategies like landmark based guidance and path integration have built in flexibility to produce an array of behavioral outcomes and require an in-depth understanding of innate and learned mechanisms that support these strategies.

Vision plays a critical role in directed navigation and the use of visual cues and past experiences is central to navigational strategies in invertebrates and vertebrates. Spatial navigation has been studied extensively from the ethological perspectives, particularly in social insects like bees, desert ants and wasps revealing key strategies used by insects in finding ideal spots for feeding, mating and laying eggs (M. Collett, Chittka, & Collett, 2013; M. Collett, Graham, & Collett, 2017; T. Collett, 1996; T. S. Collett & Collett, 2002; T. S. Collett & Graham, 2004; T. S. Collett, Philippides, & Hempel de Ibarra, 2016; Gould, 2009; Labhart & Meyer, 2002; Srinivasan, Poteser, & Kral, 1999; Webb & Wystrach, 2016; Wolf, 2011). Most studies conducted in these insects have focused on visual cues and the role of non-visual environmental information like temperature has received far less attention. Consequently, the neural basis of these non-visual navigational strategies remains largely unexplored. Here, we will highlight key studies conducted in Drosophila melanogaster, as an experimental system to explore the genetic, neural and behavioral basis of non-visual place learning. Many of the studies described in this review article highlight the work done in the laboratory of Dr. Troy Zars at the University of Missouri where he and his research team pioneered the studies on learning and memory formation using high temperature as a negative reinforcer in a spatial operant learning paradigm. We will discuss these results in the context of navigational strategies and general mechanisms of spatial memory.

Place memory formation in Drosophila

The ability to remember and navigate requires an ability to sense and avoid danger and find places that have the resources for survival. In vertebrates, place memory is known to involve medial entorhinal cortex (MEC)-hippocampal connections but the learning-dependent changes in these circuits seem to depend on specific grid, place, boundary and other undefined spatial cells whose properties are tuned by navigational strategies (Burgess, Donnett, & O’Keefe, 1998; Hartley, Lever, Burgess, & O’Keefe, 2014; Moser, Rowland, & Moser, 2015).

While, such cells within specific regions have not yet been identified in invertebrates, circuit analysis of spatial navigation has largely relied on visual pathways. For example, navigation using polarized skylight as a visual compass cue have identified neurons (POL neurons) that are tuned to specific polarized light angles and signal via the optic lobe and central complex (cx). This neural architecture has been found in bees, butterflies, locusts and beetles.(Dacke, Nilsson, Scholtz, Byrne, & Warrant, 2003; Heinze & Reppert, 2011; Homberg, Heinze, Pfeiffer, Kinoshita, & el Jundi, 2011; Homberg et al., 2004; Merlin, Heinze, & Reppert, 2012; Stone et al., 2017; Warrant & Dacke, 2016). Evidence from Drosophila has taken these studies a step further by identifying cell types specifically, a set of columnar neurons termed ‘wedge-neurons’ that relay information from single identified domains of ellipsoid body(eb) to protocerebral bridge (pb) in flies walking on a ball in a virtual reality arena (Seelig & Jayaraman, 2015). In addition to the processing of sensory information, memory plays a central role in these navigational strategies.

Spatial learning ability in Drosophila melanogaster has received limited attention despite the availability of genetic tools for in depth neuronal and circuit analysis. To date only a handful of behavioral paradigms have been used to investigate place and orientation memory in Drosophila. One such paradigm called, the flight simulator, uses a tethered fly to identify mechanisms underlying safe and unsafe flight directions based on visual cues displayed on a screen using high temperature as a negative reinforcer (Brembs & Heisenberg, 2000; Reiser & Dickinson, 2008). Flight simulator and its many iterations have advanced the understanding of decision making and visual pattern recognition mechanisms in flies and continue to be an area of active investigation (Strother, Nern, & Reiser, 2014; Strother et al., 2017; Tuthill, Nern, Rubin, & Reiser, 2014; Wu et al., 2016). Visual place memory has also been investigated in walking flies using a thermal-visual arena where flies have to find a safe zone (cool tile) in an environment with unsafe zones (hot tiles at 36° C) using visual cues from an electronic display of evenly spaced bars with specific orientations (Ofstad, Zuker, & Reiser, 2011). These paradigms using visual cues have led to a mechanistic dissection of visual place/orientation memory in flies, none of these paradigms or their variants has investigated non-visual components that help insects orient in their environment. A newer paradigm investigated the ability of flies to find a rewarded location using both visual and non-visual cues suggesting that the animals use multiple modalities in finding locations that maximize survival and reproductive options (Stern et al., 2019)

The current understanding of non-visual place memory in Drosophila has largely come from the work conducted by Dr. Troy Zars and his lab at the University of Wurzburg and the University of Missouri. The investigations of spatial memory were conducted using a single fly paradigm developed in the Heisenberg lab at the University of Wurzburg that uses a long chamber (26-40 mm) where flies can freely walk back and forth in the dark. Place learning in this paradigm involves training flies to avoid spatial positions associated with a high-temperature negative reinforcer and memory formation is assayed by positional preference post-training (Putz & Heisenberg, 2002; Wustmann, Rein, Wolf, & Heisenberg, 1996). In addition to identifying different behavioral and rearing parameters that alter memory formation in this paradigm, the Zars lab at the University of Missouri identified the role of key genes and biogenic amines like serotonin in place learning as described below.

There is more to white than meets the eye

One of the first experiments conducted in the Zars lab (~2002) involved the use of w¹¹¹⁸CS flies that carry a deletion on the X chromosome affecting an ABCG transporter. The transporter is required for guanine and tryptophan transport that are precursors for drosopterine and ommochromes that give the fly eye its red color (Mackenzie et al., 1999). The w¹¹¹⁸ flies are routinely used in Drosophila labs as controls as they are used for inserting transgenes including the mini-white gene (characterized by red eye phenotype) as a marker for confirming insertion. w¹¹¹⁸CS flies do not learn in the heat box and fail to avoid the unsafe zone post training. The duration and intensity (temperatures of 37 and 41° C) of the negative reinforcer fails to encode a memory as compared to wild type CS flies. The ability of these flies to sense and avoid high temperature in a themosensitivity sensory task further confirms that these flies can sense and perceive temperatures but fail to make an association between high temperature and spatial positions. This finding posed challenges as most genetic tools were in the w¹¹¹⁸ background and carried the mini-white gene which only partially rescued the learning phenotype (Diegelmann, Zars, & Zars, 2006). This required that all the transgenic flies that were to be tested in the heat box would need to be outcrossed for several generations to isogenize backgrounds and that X chromosome had to be w⁺. In spite of these challenges, a mutant allowed an inroad into the molecular mechanisms of memory formation.

Biochemical analysis of the white mutant showed a significant reduction in levels of serotonin and dopamine. Direct manipulation of the serotonin system confirmed that serotonin not dopamine is necessary for place learning (Sitaraman et al., 2008). Recent studies from the Zars lab have addressed the sufficiency of the serotonergic system in providing aversive reinforcement by pairing extrinsic activation of these neurons with a behavioral routine that leads to avoidance of spatial position associated with serotonin activation (Sitaraman, Kramer, Kahsai, Ostrowski, & Zars, 2017). Other biogenic amines like dopamine and octopamine which are also altered by the w¹¹¹⁸ deletion do not directly influence place learning (Sitaraman et al., 2008).

Navigating the unpredictable world

While Drosophila is an established experimental system for molecular and genetic analysis, the behavioral plasticity exhibited by flies gets far less attention. Around 2004, a few years after the Zars lab was established, there was an interest in understanding how rearing temperatures and pre-exposures to environmental stimuli affected learning in the heat box. Experiments with CS flies (wild type) showed that while flies preferred 24° C, memory reinforced by high-temperature reinforcers (37° C) is twice as effective as the low-temperature reinforcers (15° C). Thus, even though the temperature difference between 24°C and the low and high temperatures is constant, the effectiveness of these temperatures as reinforcers is not the same (M. Zars & Zars, 2006). Further, rearing temperatures of 18 and 24° C do not affect place memory performance reinforced with high temperatures. It is not clear as to how hardwired are temperature preferences in the fly and if changing environmental conditions can change the aversiveness of the reinforcer?

While, the genetic and neural basis of place learning were an important focus of the lab, the Zars lab also explored how prior experiences shaped learning. As part of a series of such experiments a unique pre-exposure effect was identified. Here flies that were pre-exposed to aversive high temperatures of 41° C, showed higher conditioned avoidance of 30° C in a subsequent post-test phase as compared to flies pre-exposed to 30° C (Sitaraman, Zars, & Zars, 2007). Since, the flies were subjected to pre-exposure and had no control over it, it was difficult to know if the high temperature or the unexpected value of pre-exposure changed conditioned preferences. To distinguish between these, the pre-exposure was controlled in one set of animals by behavioral choice. The second set of yoked flies, by contrast, received high temperature pre-exposure dependent on the animal that is in control. Thus, although both sets of animals received the same number, intensity, duration of heat pulses during pre-exposure, one set of animals could control the onset and offset of pulses while the other set (yoked) had no control. Results clearly showed that it was not the high temperature exposure per se that induces the change in conditioned place preference but rather the unexpected, unpredictable and uncertain nature of the pre-exposure that biases later learning (Sitaraman & Zars, 2010). The unpredicted aversive temperatures in the heat-box also showed increases in escape latencies. Escape latencies or time delay in escaping an environmental stressor have been shown in models of learned helplessness in rodents and involve serotonin signaling (Maier & Watkins, 2005). After months and years of discussions during Tuesday afternoon meetings the lab settled on calling this phenomenon in flies, uncertainty bias. The reviewers did not take an instant liking to this terminology and after some argument over the effect of pre-exposure vs the unexpected nature of this pre-exposure, the term was allowed to be put in the discussion but never the title.

Manipulation of the serotonergic system using broad drivers (Trh-, Ddc-Gal4; ThGal80 and Tph-Gal4) showed that activation of these neurons during pre-exposure was sufficient to induce the increase in later learning using low temperature reinforcers. Taken together, it appears that serotonergic system underlies aversive reinforcement, escape latencies and memory levels in the presence and absence of unexpected aversive events (Sitaraman et al., 2017). With Gal4 transgenic lines targeting specific cis-regulatory domains of Trh and SERT gene, it became evident that with manipulation of as few as 5 serotonergic neurons, aversive reinforcement could be altered but the same neurons did not produce the pre-exposure effect. Indeed, only with a large portion of the serotonergic system, a place memory enhancement through unexpected activation could be induced (Sitaraman et al., 2017). These studies suggest that while the same biogenic amine is necessary and sufficient for both behaviors, the subsets used are distinct and likely act via different neural systems.

Beyond aversive learning and uncertainty bias

Behavioral decision making is a dynamic process and action selection pathways are updated by experience as evidenced by behavioral outcomes. To study this process in the context of place learning air, light, and gravity sources were presented to individual flies in naïve and conditioning situations. Wild type flies showed an increase in memory performance in the presence of gravitaxic cue while light and air had no effect. Given that organisms encounter multiple sensory cues in their environment, these studies were the first attempts to examine potential interactions of overt sensory cues and place learning and memory in behavioral action selection (Baggett et al., 2018). The interaction between innate response to sensory stimuli and conditioned behavior in place learning paradigm appears to be cue specific and is largely unexplored in the field of learning and memory.

The heat box has more recently been used to develop thermogenetic tools to manipulate neurons. The Drosophila gustatory receptor gene family, has been largely unexplored with respect to temperature responses and provided a good target for development of new tools. Drosophila Gr28bD when expressed in Xenopus oocytes produced a large inward cation current and showed incapacitation when expressed pan-neuronally at temperatures between 33 and 36°C. These channels provide a novel thermogenetic tool with tremendous potential for change in temperature sensitivity, ionic selectivity and localization within the cell (Mishra et al., 2018).

Molecular mechanisms of memory formation

In addition to the neural basis of memory formation, the amenability of Drosophila to genetic manipulations makes it a particularly appealing model to conduct behavioral studies. Prior to work done in the Zars lab, there were only a handful of genes identified in operant place memory formation. These included amnesiac, dunce, and rutabaga, all believed to be components of the cAMP signaling cascade. Mutations in each of these genes lead to a decrease in memory formation following operant conditioning in the heat-box compared to wild-type (Diegelmann et al., 2006; Feany & Quinn, 1995; Waddell, Armstrong, Kitamoto, Kaiser, & Quinn, 2000; Wustmann et al., 1996; T. Zars, Wolf, Davis, & Heisenberg, 2000).

The Zars lab identified and/or characterized several genes underlying operant place memory formation in addition to the white gene. These genes included rutabaga, tribbles, arouser EPS8L3, radish, latheo, pastrel, and rac (LaFerriere et al., 2008; LaFerriere, Ostrowski, Guarnieri, & Zars, 2011; LaFerriere, Speichinger, Stromhaug, & Zars, 2011; Ostrowski, Kahsai, Kramer, Knutson, & Zars, 2015). We will focus on three of the most well-characterized genes: tribbles, arouser, and radish. Two of these, tribbles and arouser, were identified during a screen containing eleven P-element insertion fly lines. Five of these P-element insertion lines had an altered sensitivity to ethanol and six had normal ethanol sensitivity as compared to wild type (LaFerriere et al., 2008; LaFerriere, Ostrowski, et al., 2011; LaFerriere, Speichinger, et al., 2011; LaFerriere & Zars, 2017). These fly lines were utilized to determine whether there is a correlation between the genes important for memory formation in multiple learning paradigms and response to ethanol.

As most of the identified genes performed housekeeping function, the Zars lab saw this as an opportunity to identify the cell biological mechanisms involved in learning and memory formation. In pursuit of this, genes were explored in the context of several behavioral paradigms. Here we focus on the role of these genes in themosensitivity, place learning and memory formation.

tribbles:

Flies mutant for the tribbles gene have reduced place memory performance as compared to wild type in the same genetic background using the 20 minute asymptotic memory protocol (LaFerriere et al., 2008; LaFerriere & Zars, 2017). Multiple alleles with trbl disruption, confirm this learning phenotype and immunostaining approaches to visualize the gene expression (using anti-Trbl) shows widespread distribution with increased signal in the cellular rind as compared to neuropils. Further, the memory phenotype can be rescued with transgenic expression of the trbl gene in otherwise mutant flies using the trbl[3-54] Gal4.

However, unlike antibody staining, trbl [3-54] Gal4 driven GFP expression is strong in the antennal lobe, ellipsoid body, and median bundle, regions of the fly brain shown to be especially important for place memory using selective rescue of the rutabaga (type I adenylyl cyclase mutant). All flies tested (with altered trbl expression) were able to sense and avoid the reinforcement temperature ruling out sensory deficits but questions as to the role of trbl in place memory including where in the organism and when in development does this gene function need further investigation.

The trbl gene encodes a pseudokinase. When examining its sequence, the conserved serine/threonine kinase domain is missing residues necessary for normal kinase activity and there is evidence that trbl functions in targeting specific proteins to a ubiquitin ligase and degradation in Drosophila and mammals (Eyers, Keeshan, & Kannan, 2017; Grosshans & Wieschaus, 2000; Mata, Curado, Ephrussi, & Rorth, 2000; Seher & Leptin, 2000). In Drosophila, trbl regulates both string/CDC26 and slow boarders by targeting them for turnover via a proteasome (Masoner et al., 2013; Mata et al., 2000).

arouser EPS8L3:

The arouser (aru) EPS8L3 gene was also identified in the same P-element screen of ethanol sensitive mutants as trbl and was found to be important for memory formation and normal response to ethanol (Eddison et al., 2011; LaFerriere et al., 2008). Flies mutant for the aru gene have different effects on memory formation depending on their genetic background. Specifically, flies with an insertion in the aru gene perform significantly worse than wild-type Canton S flies for place memory, but perform normally in the wild-type Berlin genetic background. Examination of aru expression levels using RT-PCR indicate that the insertion in a CS genetic background causes an increase in aru expression and a decrease in a Berlin genetic background. It is believed the interaction of the Berlin and CS alleles at some number of genes with the insertion allele gives rise to the differential memory deficits. An imprecise excision of the insert reverts the phenotype but the expression level of the gene is reduced, while a precise excision of the P-element reverts the phenotype and expression levels are comparable to wild-type. Like the tribbles mutant, these flies were normal in their ability to sense and avoid the high temperature reinforcement. These results taken together suggests that the appropriate level of the aru gene product is critical for normal place memory formation and that genetic background can have specific effects on a mutation leading to highly variable behavioral phenotypes. The molecular role of aru EPS8L3 in memory formation is also yet to be determined. However, aru interacts with the EGF-receptor/ Erk and PI3K / AKT signaling cascades to regulate ethanol sensitivity (Eddison et al., 2011).

radish:

While, the radish (rsh) gene was identified as important for anesthesia resistant memory using aversive olfactory conditioning (Folkers, Drain, & Quinn, 1993; Folkers, Waddell, & Quinn, 2006; LaFerriere, Speichinger, et al., 2011), we found that this gene plays a critical role in place memory formation. Mutant flies with a premature stop codon, which reduces Rsh protein levels to below detection limits were used and first described in (Folkers et al., 1993). These flies performed similar to wild-type directly following training with a high temperature reinforcement for both 6 minutes and 20 minutes, but after several increasing time points following spaced training, they perform significantly worse. Flies were trained with intermittent spaced training consisting of three 6-minute sessions with 3-minute intervals using 24/41°C temperatures and then held for a retention interval of varying times (1, 10, 20, 30 or 40 minutes) before being tested for memory with a short 1-minute reminder training. The rsh flies had memory performance similar to wild-type CS levels with a 1-minute delay between training and the memory test, but at 10, 20, and 40 minutes rsh flies performed significantly worse than wild-type CS flies. The post-training time period and its dependence on radish gene reveals a potentially interesting dynamic occurring during this time point that requires further investigation. This was the first mutation identified that reduces place memory levels without altering conditioned behavior during training. Based on these results, it looks like rsh has a more general function in memory formation with specific temporal dynamics making it distinct from other memory mutants.

In a switch from the single gene approach to identify the genes important for memory formation, the most recent publication of results examining the genetic factors that contribute to operant place memory formation by members of the Zars lab in collaboration with Dr. Elizabeth G King examined 741 recombinant inbred fly lines that naturally vary in learning and memory (Williams-Simon et al., 2019). 16 different loci were identified as affecting place learning and/or memory performance. 5 of those loci affect both learning and memory. Using RNA-Seq data along with the identified loci, 9 candidate genes were identified as potentially important in learning and memory performance. As opposed to the single gene approach, the methods utilized in this work may better lead to the identification of genetic components that more subtly influence learning and memory in natural populations.

A place in our memory

In writing this review and reflecting on the breadth of work conducted in spatial learning, it’s hard not to think about how Troy conceptualized and executed such a wide range of projects in his lab. As the first set of graduate students in his lab we had an opportunity to observe some of his process and see the lab in its early days. Troy read a lot and liked asking many questions. He expected the same of his students. Designing experiments, developing protocols and analysis with him always felt like an unsatisfactory process in the short term and only after years of leaving the lab we can appreciate the value in his process. We would go into his office to discuss some specific experiments/questions and after listening patiently he would hand you a copy of a book (e.g. Matching law by Richard Herrnstein) or engage you in a discussion of a different project. He would always be quick to follow up on the initial discussion and while there were no quick or sure answers, there were always more questions and ambiguities to resolve. He would often urge us to go into unrelated territories and develop our questions and approaches from a much deeper and broader perspective. He was equally at ease when discussing theoretical models in psychology or what GAL4 to use to target some specific neurons. He taught us to tolerate and accept ambiguity and keep asking questions. His ability to look at and break down extremely complicated problems into simple ideas and clear experiments is evident in all of his work.

Working with Troy was a collaborative experience and his unique approach to science has left an indelible mark in memory formation. Be it an impromptu middle of the week lab trip to Jefferson City, MO to see the flooding zone or “toe Jenga” after dissertation defense, Troy had an original approach to everything from science to life. His originality will be missed.