Discovery of genes involved with learning and memory: an experimental synthesis of Hirschian and Benzerian perspectives

Abstract

The biological bases of learning and memory are being revealed today with a wide array of molecular approaches, most of which entail the analysis of dysfunction produced by gene disruptions. This perspective derives both from early "genetic dissections" of learning in mutant Drosophila by Seymour Benzer and colleagues and from earlier behavior-genetic analyses of learning and in Diptera by Jerry Hirsh and coworkers. Three quantitative-genetic insights derived from these latter studies serve as guiding principles for the former. First, interacting polygenes underlie complex traits. Consequently, learning/memory defects associated with single-gene mutants can be quantified accurately only in equilibrated, heterogeneous genetic backgrounds. Second, complex behavioral responses will be composed of genetically distinct functional components. Thus, genetic dissection of complex traits into specific biobehavioral properties is likely. Finally, disruptions of genes involved with learning/memory are likely to have pleiotropic effects. As a result, task-relevant sensorimotor responses required for normal learning must be assessed carefully to interpret performance in learning/memory experiments. In addition, more specific conclusions will be obtained from reverse-genetic experiments, in which gene disruptions are restricted in time and/or space.

Figure 1

Bidirectional selection for learning in blowflies. (A) Food-deprived individual blowflies were subjected to 15 trials of a classical conditioning procedure that paired one of two tarsal chemosensory stimuli (CSs; either water or saline) with a sucrose (US; reward) stimulus applied to the proboscis. Normally, the US produced a robust PER. After a few paired CS-US trials, the CS also began to elicit a conditioned PER. Learning scores were based on the number of CS-induced PERs during the last eight training trials. Eight pairs of the highest or lowest scoring flies were mated together each generation in the bright or dull strains, respectively. The response to selection required several generations to reach an asymptote, suggesting a polygenic basis. After 12 generations, mean scores for the bright and dull strains differed significantly from each other and from that of a free-mated control strain. (Data replotted from ref. .) (B) Food-deprived but water-satiated individual blowflies were subjected to a water pretest delivered to the tarsi, followed immediately by tarsal stimulation with sucrose. Flies then were subjected to a tarsal water posttest either 15, 30, 45, or 60 s later. Proboscis extensions to the 15-, 30-, 45-, or 60-s water posttests were given scores of 1, 2, 3, or 4, respectively. Each fly received three trials with each of the four posttest periods. Sensitization scores thus ranged from 0 to 30. Eight pairs of the highest or lowest scoring flies then were mated together in the high or low strains, respectively. The response to selection was nearly complete in one generation, suggesting a one-gene mode of inheritance. After one generation, mean scores for the high and low strains differed significantly from each other and from that of a free-mated control strain. (Data replotted from ref. .)

Figure 2

Frequency distribution of individual sensitization scores from selected high and low strains and F₁ hybrid progeny (low × high). In spite of a one-gene mode of inheritance, individual scores for each genotype overlap considerably. (Data from ref. .)

Figure 3

Memory retention after Pavlovian olfactory learning in fruit flies. During one training session (1×), about 100 flies were exposed sequentially to two odors (CSs) piped through the training chamber on air currents. During the first odor presentation, but not during the second, flies received 12 1-s pulses of footshock (US). Massed training (10× massed) consisted of 10 of these training sessions one after the other. Spaced training (10× spaced) consisted of 10 training sessions with a 15-min rest interval between each session. Conditioned odor avoidance responses were tested at various times after training by transferring the flies to the choice-point of a T-maze, where they were exposed to convergent currents of air carrying the odors used during training. Odor concentrations were adjusted for these conditioning experiments so that untrained flies distributed themselves 50:50. At most retention times after training, however, a majority of flies avoided the shock-paired odor. For a complete experiment, a second group of 100 flies was trained to the reciprocal odor combination. The performance index was an average of reciprocally trained groups and was calculated to be zero if flies distributed themselves 50:50 or 100 if all flies avoided the shock-paired odor. (Data from ref. ; also see ref. for more procedural details.)

Figure 4

Information processing during memory formation after Pavlovian olfactory learning. Newly acquired information (LRN) first is processed sequentially through short-term memory (STM) and middle-term memory (MTM) phases. Then information processing branches into two functionally independent phases: anesthesia-resistant memory (ARM) and long-term memory (LTM). Massed training induces LRN, STM, MTM, and ARM, whereas spaced training induces LTM along with similar amounts of LRN, STM, MTM, and ARM. Different single-gene mutants (latheo, linotte, dunce, rutabaga, amnesiac, and radish) disrupt this process at different steps. Notably, the radish mutation disrupts ARM without affecting LTM, and the protein synthesis inhibitor cycloheximide (CXM) or inducible dominant-negative CREB transgenes (dCreb2-b) disrupt LTM without affecting any other aspect of learning or memory. Thus, ARM and LTM appear to be functionally independent, parallel memory phases. (Data from ref. .)

Figure 5

Olfactory learning and memory in normal and mutant flies. (A) Memory retention in wild-type (Can-S) flies or dunce^M11 mutants. Each point represents the average avoidance responses of about 1200 individuals. The genetic backgrounds of each strain were heterogeneous and equilibrated, so the difference between mean scores represents the average effect of the dunce^M11 mutation in homozygotes (see text). (Data from ref. .) (B) Learning in wild-type (Can-S), dunce^M11 (dnc^M11) or rutabaga¹ (rut¹) single-gene mutants and dunce^M11 rutabaga¹ (dnc, rut) double-mutants. The learning defect in the double mutant is more severe than either single mutant, revealing a quantitative, polygenic basis for olfactory learning. (Data from ref. .)

Figure 6

LTM formation in normal flies or transgenic flies carrying inducible CREB transgenes. (A) CREB repressor blocks LTM. Seven-day retention after 10× spaced training in wild-type (Can-S) flies or transgenic flies carrying a heat-shock inducible CREB repressor (hs-dCREB2-b). LTM formation is normal in transgenic flies in the absence of heat shock (−hs), indicating that no developmental defects impinge on LTM. In contrast, LTM formation is completely blocked in transgenic flies when trained 3 hr after a 30-min heat shock induction (+hs) of CREB repressor. (B) CREB activator enhances LTM. Seven-day retention in wild-type (Can-S) flies or transgenic flies carrying a heat-shock inducible CREB activator (hs-dCREB2-a). LTM formation is normal after 10× spaced training, or normally absent after 1× training, in transgenic flies in the absence of heat shock (−hs), indicating that no developmental defects impinge on LTM. When subjected to 10× spaced training 3 hr after a 30-min heat shock induction of CREB activator (+hs), LTM formation is neither enhanced or suppressed in transgenic flies. LTM forms after only one training session, however, in transgenic flies after such heat shock. Thus, induced expression of CREB activator enhances LTM by promoting its formation after less training. [Data from refs. and (Copyright 1994 and 1995, Cell Press).]