Early life cancer and chemotherapy lead to cognitive deficits related to alterations in microglial-associated gene expression in prefrontal cortex

Abstract

Children that survive leukemia are at an increased risk for cognitive difficulties. A better understanding of the neurobiological changes in response to early life chemotherapy will help develop therapeutic strategies to improve quality of life for leukemia survivors. To that end, we used a translationally-relevant mouse model consisting of leukemic cell line (L1210) injection into postnatal day (P)19 mice followed by methotrexate, vincristine, and leucovorin chemotherapy. Beginning one week after the end of chemotherapy, social behavior, recognition memory and executive function (using the 5 choice serial reaction time task (5CSRTT)) were tested in male and female mice. Prefrontal cortex (PFC) and hippocampus (HPC) were collected at the conclusion of behavioral assays for gene expression analysis. Mice exposed to early life cancer + chemotherapy were slower to progress through increasingly difficult stages of the 5CSRTT and showed an increase in premature errors, indicating impulsive action. A cluster of microglial-related genes in the PFC were found to be associated with performance in the 5CSRTT and acquisition of the operant response, and long-term changes in gene expression were evident in both PFC and HPC. This work identifies gene expression changes in PFC and HPC that may underlie cognitive deficits in survivors of early life exposure to cancer + chemotherapy.

Figure 1.. Experimental timeline.

10³ L1210 cells were injected intrascapularly at P19. Chemotherapy was initiated two days later with i.p. vincristine, methotrexate and leucovorin given every four days for four cycles. Vincristine was injected at 1600 followed by MTX 30 minutes later. Leucovorin was injected the following day 16 hours post chemotherapy. Social behavior testing was started on ~d42, open field/novel object recognition on ~d56, and 5CSRTT training beginning on d136. Brain tissue was collected on PN 182 (females) or 193 (males), one day after conclusion of behavioral testing. Social= social preference and social recognition, OF=open field, NOR= novel object recognition.

Figure 2.. Recognition memory deficits in animals exposed to cancer+chemotherapy.

Exposed males and females showed an impairment in recognition memory for both a familiar mouse (A) and a familiar object (B), as evidenced by a recognition index not different from 0.5. *p<.05 (one sample t-test if different from 0.5). Open circles- saline, black triangle-cancer+chemotherapy. Mean ± SEM (n=12, except male saline social recognition where n=10)

Figure 3.. Executive function deficits in animals exposed to cancer+chemotherapy.

(A) Exposed mice and females reached criterion in FR1 (70 trials within a 30 min) in significantly fewer days. (B) Exposed mice were significantly slower to progress through increasingly more challenging stages of the 5CSRTT, while female mice were faster than males. With an 8 sec stimulus light, male exposed animals showed an increase in premature responses (C) and a decrease in accuracy (D). With a 4 sec stimulus, both males and females showed an increase in premature responses (E) and during titration, males had an increase in incorrect responses (F). Open circles- saline, black triangle-cancer+chemotherapy. *p<.05, **p<.01, ***p<.001, ****p<.0001 Mean ± SEM (n=12)

Figure 4.. Cancer+chemotherapy increased gene expression in the PFC.

Dnmt1, Dnmt3a and Ptgs2 were increased in the PFC by cancer+chemotherapy exposure. Open circles- saline, black triangle-cancer+chemotherapy. *p<.05. Mean ± SEM Gene expression (male control n=9, male cancer/chemo = 9, female control = 10, female cancer/chemo = 9)

Figure 5.. Cancer+chemotherapy decreased gene expression in the HPC.

Pdgfra (A), Cldn1 (C) and P2ry12 (E) expression levels were decreased in female exposed animals only. Expression of Tlr4 (B), Mylk (D) and Dnmt3a (F) were reduced in all exposed animals. Open circles- saline, black triangle-cancer+chemotherapy. *p<.05, **p<.01, ****p<.0001 Mean ± SEM. Gene expression (male control n=10, male cancer/chemo = 11, female control = 9, female cancer/chemo = 9)

Figure 6.. Heatmap of Pearson correlations between gene expression and behavioral endpoints in PFC and HPC.

Cluster analysis of Pearson correlations identified a cluster of microglial-associated genes that were related to cognitive performance (boxed in red). No such relationships were identified in the HPC. 5C endpoints are the average across all three sessions, NOR is recognition index expressed as a z-score, titr=titration.

Figure 7.. Significant gene × behavior correlations.

Significant correlations between performance in the 5CSRTT and microglial-associated genes (log fold change) in the PFC were identified. Tlr4 and Tgfbr1 expression were positively related to % correct (r²=.26, p=.0014; r²=.24, p=.0025, respectively) and incorrect errors (r²=.26, p=.0014; r²=.22, p=.0043, respectively) P-values in red were significant using a p_corr=0.0045.

Figure 8.. Significant correlation of microglial genes and synaptophysin in the PFC with FR1.

A cluster of microglial-associated genes (Tlr4, Tgfbr1, Grm5, Itgam and P2ry12; log fold change) were all positively related to the time required to acquire the FR1 response, while syp was negatively related to FR1 days to criterion. Tlr4: r²=.36, p=.0001; Itgam: r²=.27, p=.0013; Tgfbr1; r²=.34, p=.0002; P2ry12: r²=.29, p=.0007; Grm5: r²=.35, p<.0001, Syp; r²=.20, p.0057.

Figure 9.

Summary figure.