Role of activating transcription factor 4 in the hepatic response to amino acid depletion by asparaginase

Abstract

The anti-leukemic agent asparaginase activates the integrated stress response (ISR) kinase GCN2 and inhibits signaling via mechanistic target of rapamycin complex 1 (mTORC1). The study objective was to investigate the protective role of activating transcription factor 4 (ATF4) in controlling the hepatic transcriptome and mediating GCN2-mTORC1 signaling during asparaginase. We compared global gene expression patterns in livers from wildtype, Gcn2 ^-/-, and Atf4 ^-/- mice treated with asparaginase or excipient and further explored selected responses in livers from Atf4 ^+/- mice. Here, we show that ATF4 controls a hepatic gene expression profile that overlaps with GCN2 but is not required for downregulation of mTORC1 during asparaginase. Ingenuity pathway analysis indicates GCN2 independently influences inflammation-mediated hepatic processes whereas ATF4 uniquely associates with cholesterol metabolism and endoplasmic reticulum (ER) stress. Livers from Atf4 ^-/- or Atf4 ^+/- mice displayed an amplification of the amino acid response and ER stress response transcriptional signatures. In contrast, reduction in hepatic mTORC1 signaling was retained in Atf4 ^-/- mice treated with asparaginase.

Conclusions: GCN2 and ATF4 serve complementary roles in the hepatic response to asparaginase. GCN2 functions to limit inflammation and mTORC1 signaling whereas ATF4 serves to limit the amino acid response and prevent ER stress during amino acid depletion by asparaginase.

Figure 1

Percent weight change of liver, pancreas, and spleen relative to body weight following 8 daily injections of asparaginase (3 IU per gram body weight, ASNase) or phosphate buffered saline excipient (PBS) in wild type mice (WT) or mice deleted for Gcn2 (Gcn2 ^−/−) or Atf4 (Atf4 ^−/−). Data are represented as the average value ± standard error of the mean, n = 4–6 animals per group. Means not sharing a letter are different according to Tukey post hoc analysis following ANOVA, P < 0.05.

Figure 2

Transcriptional profiling of livers following 8 daily injections of asparaginase (3 IU per gram body weight, ASNase) or phosphate buffered saline excipient (PBS) in wild type mice (WT) or mice deleted for Gcn2 (Gcn2 ^−/−) or Atf4 (Atf4 ^−/−). (a) Venn diagram shows the number of hepatic genes altered basally, categorized as unique to Gcn2 ^−/− [A], unique to Atf4 ^−/− [C] or common to deletion of either [B]. (b) Overview of the top overrepresented canonical pathways based on gene expression changes in panel a. (c) Venn diagram shows ASNase alters the transcriptome differently in WT, Gcn2 ^−/− and Atf4 ^−/− mice, with gene changes clustered as unique to WT [D], unique to Gcn2 ^−/− [E]; unique to Atf4 ^−/− [G]; common to Gcn2 ^−/− and Atf4 ^−/− [F]; common to WT and Gcn2 ^−/− [H]; common to WT and Atf4 ^−/− [J]; and common to all strains [I]. (d) Overview of the top overrepresented canonical pathways based on gene expression changes in panel c. Data represent n = 3 per group. All differentially expressed genes were statistically significant at a false discovery rate (q value) < 0.1.

Figure 3

Changes in the amino acid response following 8 daily injections of asparaginase (3 IU per gram body weight, ASNase) or phosphate buffered saline excipient (PBS) in wild type mice (WT) or mice deleted for Gcn2 (Gcn2 ^−/−) or Atf4 (Atf4 ^−/−). (a) Phosphorylation of eIF2 alpha at serine 51 in liver by immunoblot analysis. A full immunoblot is presented in Supplementary Fig. S7. (b) Clustered RNA-Seq expression data (n = 3 per group) presented in a heat map illustrates relative mean change in amino acid stress response genes to ASNase in WT, Gcn2 ^−/− and Atf4 ^−/− mice. Color scale reflects increased (yellow) and decreased (blue) expression of each gene relative across treatment groups. (c) Gene expression of Fgf21 (d) Asns and (e) Atf3 in liver as measured by RT-qPCR. Data in (a,c,d,e) are represented as the average value ± standard error of the mean, n = 3–4 per group. Means without a common letter are different according to Tukey post hoc analysis following ANOVA, P < 0.05.

Figure 4

Atf4 deletion selectively engages the UPR following 8 daily injections of asparaginase (3 IU per gram body weight, ASNase) or phosphate buffered saline excipient (PBS) in wild type mice (WT) or mice deleted for Gcn2 (Gcn2 ^−/−) or Atf4 (Atf4 ^−/−). (a) Phosphorylation of PERK in liver by immunoblot analysis. (b) CHOP protein expression by immunoblot analysis. (c) Gene expression of spliced (sXbp1) and unspliced (uXbp1) Xbp1 mRNA measured by RT-qPCR. (d) Atf6 mRNA levels measured by RT-qPCR. Data are represented as the average value ± standard error of the mean, n = 3–4 per group. Means without a common letter are different according to Tukey post hoc analysis following ANOVA, P < 0.05. PC, liver from a tunicamycin-treated mouse as positive control. Full immunoblots are presented in Supplementary Figure S7.

Figure 5

The mTORC1 pathway is amplified in livers from Gcn2 ^−/− but not Atf4 ^−/− mice following 8 daily injections of asparaginase (3 IU per gram body weight, ASNase). (a) Representative immunoblots for panels (b–e). (b) The ratio of phospho-S6K1 at threonine 389 to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). (c) Phosphorylation of 4E-BP1 expressed as the ratio of the γ-form to the α + β + γ sum. (d) The ratio of phospho Akt at threonine 308 to total Akt. (e) The ratio of Sestrin2 phospho-forms to the sum of all Sesn2 resolved forms. Data are represented as the average value ± standard error of the mean, n = 4–6 per group. Means without a common letter are different according to Tukey post hoc analysis following ANOVA, P < 0.05. Full immunoblots are presented in Supplementary Fig. S7.

Figure 6

Atf4 haploinsufficiency amplifies the hepatic AAR and liver damage in response to 8 daily injections of asparaginase (3 IU per gram body weight, ASNase). (a) Phospho-eIF2 alpha was measure by immunoblot analysis and quantified relative to total eIF2 alpha. (b) Gene expression of Atf4 in liver as measured by RT-qPCR. n.d. = not detectible. (c) Hepatic expression of ISR genes Asns, Atf3, Atf5, Fgf21, Eif4ebp1, and Ppp1r15a in WT, Atf4 ^+/− and Atf4 ^−/− mice. (d) Apoptosis was ascertained in de-identified histological sections by manual counting of TUNEL-positive nuclei using Image J software. (e) Fragmented DNA visualization by TUNEL method in representative frozen liver sections (10 μm thick). Images taken at 40X magnification show visual features determined in WT, Atf4 ^+/− and Atf4 ^−/− mice. Scale bar represents 50 μm. Data are represented as the average value ± standard error of the mean, n = 3–6 per group. Means without a common letter are different according to Tukey post hoc analysis following ANOVA, P < 0.05. Full immunoblots are presented in Supplementary Fig. S7.

Figure 7

Atf4 haploinsufficiency predisposes mice to amplified ER stress when treated with 8 daily injections of asparaginase (3 IU per gram body weight, ASNase). Gene expression levels of: (a) Ddit3 (CHOP), (b) spliced Xbp1 and unspliced Xbp1, (c) Hspa5 (BiP/Grp78), and (d) Atf6 and oxidative stress marker Nrf2 were measured in liver by RT-qPCR. Data are represented as the average value ± standard error of the mean, n = 4–6 per group. Means without a common letter are different according to Tukey post hoc analysis following ANOVA, P < 0.05.