SfN14 blogging: Parvalbumin interneurons and cognitive inflexibility

Figure 1: Parvalbumin interneurons are also 
categorized as "basket" cells because of their 
shape and they fire with a high frequency. 
Cartoon by Dr. Immy Smith, 
@Cartoon_Neuron

Schizophrenia is a multi-faceted disease. Many people are aware of the positive (hallucinations) and negative (low emotionality) symptoms but unaware of the cognitive symptoms like memory deficits. Samuel Paskewitz works at Columbia University doing research with Dr. Holly Moore. He investigates cognition in a mouse model of schizophrenia.

Parvalbumin interneurons are a type of cell that is important for learning and memory. In people with schizophrenia, the number of parvalbumin interneurons is reduced. People with schizophrenia also tend to have deficits in working memory and task shifting, a type of cognitive flexibility. Sam hypothesized that cognitive flexibility is related to a decrease in parvalbumin interneuron number.

To test this hypothesis he used mice with a gene, Ccnd2, deleted. Mice with missing or altered genes are called transgenic mice and, in this case, they're named Ccnd2 knock-out mice (Ccnd2-KO). Deleting the Ccnd2 gene results in mice with 40% fewer parvalbumin interneurons in adulthood. In order to assess cognitive flexibility he used a strategy switching cross maze, which is a rodent equivalent to the Wisconsin Card Sorting Test used in humans.

Figure 2: The egocentric rule. 

In the Wisconsin Card Sorting Test (WCST), subjects are trained to sort cards based on a rule, for example, by color. Then the rule changes, for example, to sort by shape. Subjects with schizophrenia tend to make more mistakes when learning the new rule. In the rodent version, the strategy switching cross maze, mice are trained to choose a specific arm of the maze (Figure 2). Once they've mastered the rule, a new rule is introduced and they must learn to solve the maze based on that.

The two rules used in the task are egocentric (ego = self, centric = centered) and allocentric (allo = other). For an egocentric rule, the mouse is trained to always "turn right", for example (Figure 2). So no matter if the mouse is placed in the west or east side of the maze, they are rewarded to always turn right.

An example of the allocentric rule, is when mice are trained to always "go north" (Figure 3). When the mouse is placed in the west arm of the maze it is rewarded to turn north, which is its left. When placed in the east arm of the maze it is rewarded to turn north, which is now its right.

Figure 3: The allocentric rule.

Some mice learned the egocentric version first and some learned the allocentric version first. When the rule was switched, there were two types of errors that could be made. If the mouse continued to turn the same way they were trained rather than the new way it was a "perseverative error", because they persevered in trying to solve the maze the old way. People with schizophrenia tend to make preservative errors when learning a new rule in the WCST. In the strategy switching maze, if mouse turned in a new direction that they had never been rewarded for, it was a "never reinforced" error.

Sam trained the mice for 24 trials/ day until they chose the correct arm 8 out of 10 times (80% efficiency). During initial learning, the number of errors made by non-transgenic and Ccnd2-KO mice was not different. When the rule was switched, Ccnd2-KO mice had difficulty learning the new rule. They made more errors than the non-transgenic mice, especially if they had first learned the allocentric rule. The type of error was surprising. Rather than making “perseverative” errors, as people with schizophrenia tend to do in the WCST, the mice made more "never reinforced" errors. It seemed they weren't learning the new rule effectively but it wasn’t because they were still stuck on the old rule.

The non-transgenic animals were particularly good at switching to the new rule if they learned the egocentric version first. Sam thinks this is because while they were learning the egocentric version they passively made a spatial map of the maze even though they didn't need to in order to solve the maze. This type of learning is called latent learning. It seems to have given the non-transgenic mice an advantage when learning the allocentric version of the task second.

Sam says the results mean that the Ccnd2-KO mice aren't necessarily bad at rule switching but that the non-transgenic mice are inherently better at it. The Ccnd2-KO mice may just have difficulty forming a spatial map of their surroundings, which other recent studies have also suggested. In order to test this further he could take a new group of mice and let them passively explore the maze, to gain a spatial map through latent learning, and then train them to solve the maze using an allocentric rule. Sam plans to continue his education and research career in graduate school.