Converging on a core cognitive deficit: the impact of various neurodevelopmental insults on cognitive control

Abstract

Despite substantial effort and immense need, the treatment options for major neuropsychiatric illnesses like schizophrenia are limited and largely ineffective at improving the most debilitating cognitive symptoms that are central to mental illness. These symptoms include cognitive control deficits, the inability to selectively use information that is currently relevant and ignore what is currently irrelevant. Contemporary attempts to accelerate progress are in part founded on an effort to reconceptualize neuropsychiatric illness as a disorder of neural development. This neuro-developmental framework emphasizes abnormal neural circuits on the one hand, and on the other, it suggests there are therapeutic opportunities to exploit the developmental processes of excitatory neuron pruning, inhibitory neuron proliferation, elaboration of myelination, and other circuit refinements that extend through adolescence and into early adulthood. We have crafted a preclinical research program aimed at cognition failures that may be relevant to mental illness. By working with a variety of neurodevelopmental rodent models, we strive to identify a common pathophysiology that underlies cognitive control failure as well as a common strategy for improving cognition in the face of neural circuit abnormalities. Here we review our work to characterize cognitive control deficits in rats with a neonatal ventral hippocampus lesion and rats that were exposed to Methylazoxymethanol acetate (MAM) in utero. We review our findings as they pertain to early developmental processes, including neurogenesis, as well as the power of cognitive experience to refine neural circuit function within the mature and maturing brain's cognitive circuitry.

Keywords: cognitive control; hippocampus; mental illness; neural coordination; neurodevelopmental insults; schizophrenia models.

Figure 1

Cognitive control depends on neural coordination, the coordinated activity of neural networks within or among brain regions. (A) During the Stroop test, the test subject is presented with a word that means a color. The word meaning and color match for color-meaning congruent words, whereas there is a mismatch for color incongruent words. (B) Neural representations are a coactive ensemble of neurons that coalesce into a temporally defined pattern to represent a concept. Neural coordination to express the appropriate representation is required for appropriate responds during the Stroop test. Coactivity implies that the neurons are functionally coupled to drive the emergence of the ensemble. The cognitive challenge is escalated for presentations of color incongruent words because the appropriate response requires suppression of the inappropriate neural representation, despite the presence of functional coupling to drive the neural representation for the word name. The appropriate response contrasts with the case of an inappropriate response to the color-incongruent word. Instead of the neural representation of the color, the inappropriate representation of the meaning emerges, suppressing the appropriate representation of the color. The neural equivalent of confusion results when both the color and meaning representations are simultaneously active, which is altogether a different inappropriate representation. (C) The hippocampus is a complex circuit in which multiple streams of information converge on a single area. For example, information from CA3 and entorhinal cortex must be coordinated in order to activate the appropriate response in CA1, a point of convergence for CA3 and EC activity. The dominant contemporary model asserts that the dentate gyrus mediates pattern separation because a few, sparsely organized neurons are activated to relay incoming cortical information to CA3. CA3 mediates pattern completion because the recurrent collaterals provides for relatively high autoassociative connectivity, and a less sparse pattern of activation. This means that the CA3 output will minimize differences between different representations once they are small and the differences will be exaggerated if they are large. The CA1 subfield is thought to relay the result of the “processed” combined dentate gyrus and CA3 computations to the neocortex via the subiculum. Given that CA1 also receives direct neocortical input, this subfield may function to integrate and compare the two direct and processed inputs. CA1, cornu ammonis 1 subregion; CA3, cornu ammonis 3 subregion; DG, dentate gyrus; SUB, subiculum; EC, entorhinal cortex.

Figure 2

The active place avoidance paradigm is used to study cognitive control. The apparatus consists of an 82-cm diameter disk-shaped arena with transparent walls and extra-arena visual cues placed within the room. (A) Experimental protocol. Pretraining (Pre 1–2), the animal explores the stationary arena, during which time no shock is delivered. To evaluate active place avoidance, the animals are placed on the rotating arena (1 rpm) and a mild foot-shock (~0.3 mA) is delivered for 500 ms whenever the rat enters the shock zone, which is a 60° sector that is defined by the computer in room coordinates. The shock is repeated every 1500 ms until the subject leaves the shock zone. Avoidance is measured by counting the number of errors, defined as occasions that the animal enters the shock zone. Animals are trained over the course of several days to study learning, memory retention, and cognitive flexibility. Each trial [training 1–16 (T1–16), retention (RT), and conflict learning (C1–8)] is 10 min and the interval between trials is at least 10 min. After training for multiple days, cognitive flexibility is assessed during conflict learning after shifting the shock zone 180° from the original location. (B) Behavior of a single rat. The pretraining sessions are open field tests and can be used to assess overall locomotor activity and behavioral habituation in response to novelty. Testing over multiple days allows examination of within-session learning across trials and between-session learning across days. We also assess within-trial learning (inset of graph) over the 10 min trial by assessing the number of errors made during two min periods. For the animal presented in this figure, the learning curve over the first trial shows that the rat received more shocks in minutes 4–6 than in the first 2 min. This pattern is atypical but not uncommon, as avoidance can occur by chance for a few minutes if the animal runs away in response to shock. Avoidance by chance is unlikely for prolonged periods of time, which is why end point measures are taken over a 10 min period or longer. Consolidation can also be examined by comparing the performance at the end of day one to the performance at the beginning of day two. Long-term memory (LTM) is tested in a single retention trial (RT) 1 day after training to the initial shock zone. Reversal learning is examined over eight conflict trials (C1–8) with the shock zone shifted 180°. Rodents quickly learn both the original and reversed shock location. (C,D). The two-frame active place avoidance task resembles the Stroop test, where the room frame represents relevant information (similar to the word color) while the arena frame represents irrelevant information (similar to the word meaning). Place avoidance was measured as the number of entries into the shock zone (errors). Initially (T1), a control animal makes errors. By the sixteenth trail (T16), this animal makes no errors. The data presented in B-D are from the same animal. Red circles indicate shocks. The gray lines are the path of the animal throughout the trial with respect to the Room or Arena frame. The 60° area defined by the red lines is the shock zone, stationary within the room frame.

Figure 3

Neurodevelopmental insults differently affect cognitive ability. Three animal models of schizophrenia are generated in the Long Evans strain of rats and are being used to test the neural discoordination hypothesis. Cognitive control was tested using the two-frame active place avoidance task. Generation of the ttxNVHL and iboNVHL is as follows: on postnatal day 7, male pups were anesthetized by hypothermia and bilateral injections of TTX (30ng/μL; 0.3 μL/hemisphere), ibotenic acid (10 μg/μL; 0.3 μL/hemisphere), or an equal volume of saline was injected into each ventral hippocampus. To generate MAM animals, timed pregnant females were given intraperitoneal injection of MAM (26 mg/kg) or an equal volume of saline at gestational day 17 (GD17). (A) ttxNVHL animals do not have cognitive deficits in the two-frame active place avoidance task and are predicted to have normal neural coordination. (B) Adult iboNVHL animals have deficits in the two-frame active place avoidance task and also have altered interhippocampal neural coordination (Lee et al., 2012). (C) GD17-MAM animals have difficulty performing the two-frame active place avoidance task and are predicted to have poor neural coordination. Data are presented as average ± s.e.m.

Figure 4

Cognitive flexibility is differently altered by neurodevelopmental insults. Active place avoidance can be used to test cognitive flexibility. (A) One day after training to the initial shock zone, the animals are tested in a retention trial and eight trials in which the shock zone is shifted 180° from the initial location. The ability to adapt to the conflict between the shifted shock zone location and the memory of the initial shock zone location was measured as the number of entries into the shifted shock zone location (errors). (B) Cognitive flexibility is normal in ttxNVHL animals. (C,D) As indicated by the retention trial, both the iboNVHL and GD17-MAM rats are capable of performing the active place avoidance task after 2 or 4 days of training, respectively. However, cognitive flexibility during the 180° conflict trials is disrupted in iboNVHL and GD17-MAM animals. Data are presented as average ± s.e.m.

Figure 5

Adolescent experience alters adult behavior in iboNVHL animals. (A) At postnatal day 35 (P35), iboNVHL animals were habituated to the room on the stationary arena for two 10 min sessions with a 10 min inter trial interval. Adolescence training consisted of eight trials per day for 2 days (P36 and P37). To control for noncognitive experience, a group of adolescent animals was received the same experience with the shock off. (B) During adolescence, iboNVHL animals are able to perform the task as well as saline controls. (C) Adolescence training had a positive impact on adult behavior in the active place avoidance task. iboNVHL animals that received cognitive training during adolescence performed as well as saline control animals when tested as adults. The animals that received noncognitive experience during adolescence remained impaired. Data are presented as average ± s.e.m.

Figure 6

Potential circuit modifications to attenuate cognitive deficits. One hypothesized cause of cognitive deficits is an imbalance between the entorhinal cortex and CA3 inputs to CA1 that may carry perceptual and memory related information, respectively. Cognitive training may induce synaptic restructuring, for example by increasing the CA3 input and decreasing the entorhinal input to normalize CA1 responses. A second hypothesized cause of cognitive deficits is the abnormal pattern separation computation that might underlie cognitive flexibility deficits. Cognitive training may increase the survival:death ratio of newly born neurons in the dentate gyrus to promote better pattern separation. The adolescent brain may be more receptive to these changes than the adult brain because juvenile brains are actively undergoing similar modifications. CA1, cornu ammonis 1 subregion; CA3, cornu ammonis 3 subregion; DG, dentate gyrus; SUB, subiculum; EC, entorhinal cortex.