Self-Organizing Feature Maps Identify Proteins Critical to Learning in a Mouse Model of Down Syndrome

Abstract

Down syndrome (DS) is a chromosomal abnormality (trisomy of human chromosome 21) associated with intellectual disability and affecting approximately one in 1000 live births worldwide. The overexpression of genes encoded by the extra copy of a normal chromosome in DS is believed to be sufficient to perturb normal pathways and normal responses to stimulation, causing learning and memory deficits. In this work, we have designed a strategy based on the unsupervised clustering method, Self Organizing Maps (SOM), to identify biologically important differences in protein levels in mice exposed to context fear conditioning (CFC). We analyzed expression levels of 77 proteins obtained from normal genotype control mice and from their trisomic littermates (Ts65Dn) both with and without treatment with the drug memantine. Control mice learn successfully while the trisomic mice fail, unless they are first treated with the drug, which rescues their learning ability. The SOM approach identified reduced subsets of proteins predicted to make the most critical contributions to normal learning, to failed learning and rescued learning, and provides a visual representation of the data that allows the user to extract patterns that may underlie novel biological responses to the different kinds of learning and the response to memantine. Results suggest that the application of SOM to new experimental data sets of complex protein profiles can be used to identify common critical protein responses, which in turn may aid in identifying potentially more effective drug targets.

Fig 1. Classes of mice.

(A) There are eight classes of mice based on genotype (control, c, and trisomy, t), stimulation to learn (Context-Shock, CS, and Shock-Context, SC) and treatment (saline, s, and memantine, m). Learning outcome indicates the response to learning for each class. (B) Number of mice in each class. (C) Format of protein expression data: rows are individual mice, and columns, P _{1 …} P_77, are the measured levels of the 77 proteins. The color coding in the last column identifies each class; color coding was used in visualization of the SOMs but not in clustering.

Fig 2

(A) SOM clustering of four classes of control mice. Node color indicates the majority class of the measurements clustered within it: brown: c-SC-s; orange: c-SC-m; green: c-CS-s, yelow: c-CS-m. Nodes are labeled with the name(s) of the majority and minority class(es) and the total number of measurements contained within them is also shown. (B) Strong class-specific clusters are outlined as black: c-SC-s; brown: c-SC-m; green: c-CS-s; dark red: c-CS-m.

Fig 3. Venn diagrams of discriminating proteins.

(A) Intersection of proteins discriminating learning in control mice. Colors indicate the proteins that respond in the four comparisons that reflect successful learning. (B) Intersection of proteins discriminating between normal, failed and rescued learning. Red circle: number of discriminant proteins between CS-s and SC-s in control (Normal learning; repeated from panel A). Blue circle: number of proteins that respond when learning is rescued in trisomic mice with memantine (CS-m vs. SC-m). Green circle: number of discriminant proteins in comparison CS-s vs. SC-s in trisomy (Failed learning).

Fig 4. SOM clustering with subsets of protein expressions data from control mice.

(A) SOM obtained using only 11 proteins (out of 77) that discriminate between the four classes of c-CS and c-SC. The dashed line indicates the border between the two main classes. (B) SOM obtained using the remaining 66 proteins. Outlined in green are nodes that contain a majority of CS mice surrounded by nodes with SC mice. Outlined in brown are nodes with SC mice surrounded by nodes of CS mice.

Fig 5. SOM clustering with data from the 11 proteins that discriminate between context-shock and shock-context (c-CS and c-SC) plus.

(A) the 12 proteins that discriminate between context-shock with and without memantine (c-CS-m and c-CS-s). Dashed black line, clusters of CS-m; green outline, clusters of CS-s. (B) the 13 proteins that discriminate between c-SC-m and c-SC-s. Brown outline: c-SC-s clusters; orange dashed outline: c-SC-m cluster.

Fig 6. SOM clustering of trisomic mice data using 77 proteins.

Light blue: t-SC-s; dark blue: t-SC-m; light pink: t-CS-s; dark pink: t-CS-m. Nodes forming each cluster are outlined: blue: t-SC-s; black: t-SC-m; yellow: t-CS-s; green: t-CS-m.

Fig 7. SOM clustering using subsets of proteins for trisomic mice.

(A) Clustering of trisomic mice with proteins common to the two comparisons that reflected rescued learning: t-CS-m vs. t-SC-m and t-CS-m vs. t-SC-s (15 proteins), plus the initial effects of memantine: t-SC-m vs. t-SC-s (12 proteins). The t-SC-m cluster is outlined in solid blue and the t-SC-s cluster is outlined in dashed blue. (B) Clustering with the former 15 proteins (rescued learning) plus the 9 discriminant between t-CS-m and t-CS-s (S2 Table, c5).

Fig 8. SOM clustering of CS classes of control and trisomic mice.

(A) t-CS-s (failed learning, light pink nodes), c-CS-s (green) and c-CS-m (yellow), using as input the levels of all 77 proteins. Pink: clusters of t-CS-s. (B) t-CS-s, c-CS-s and c-CS-m using as input only the 10 proteins that discriminate t-CS-s from both c-CS-s and c-CS-m. Pink: clusters of t-CS-s. (C) t-CS-m (rescued learning, dark pink nodes), c-CS-s and c-CS-m using as input only the set of 10 proteins that discriminate t-CS-s from both c-CS-s and c-CS-m. Dashed black line: clusters of t-CS-m mice. Black squares: nodes with mixed classes of t-CS-m and controls.

Fig 9. SOM clustering of shock-context classes of control and trisomic mice.

(A) Clustering of classes: t-SC-s (light blue), t-SC-m (dark blue), c-SC-m (orange), c-SC-s (brown) with 77 proteins. (B) Clustering of the three classes with the 21 discriminant proteins found between t-SC-s and the two other classes.