Deletion of AMPK minimizes graft-versus-host disease through an early impact on effector donor T cells

Allogeneic hematopoietic stem cell transplantation is a viable treatment for multiple hematologic diseases, but its application is often limited by graft-versus-host disease (GVHD), where donor T cells attack host tissues in the skin, liver, and gastrointestinal tract. Here, we examined the role of the cellular energy sensor AMP kinase (AMPK) in alloreactive T cells during GVHD development. Early posttransplant, AMPK activity increased more than 15-fold in allogeneic T cells, and transplantation of T cells deficient in both AMPKα1 and AMPKα2 decreased GVHD severity in multiple disease models. Importantly, a lack of AMPK lessened GVHD without compromising antileukemia responses or impairing lymphopenia-driven immune reconstitution. Mechanistically, absence of AMPK decreased both CD4+ and CD8+ effector T cell numbers as early as day 3 posttransplant, while simultaneously increasing regulatory T cell (Treg) percentages. Improvements in GVHD resulted from cell-intrinsic perturbations in conventional effector T cells as depletion of donor Tregs had minimal impact on AMPK-related improvements. Together, these results highlight a specific role for AMPK in allogeneic effector T cells early posttransplant and suggest that AMPK inhibition may be an innovative approach to mitigate GVHD while preserving graft-versus-leukemia responses and maintaining robust immune reconstitution.


Immunofluorescence staining
Livers from mice receiving fl/fl versus AMPK dKO T cells were recovered on day 22 post-transplant, flash frozen in optimum cutting temperature (OCT) media, and sliced into 6 µm sections. Tissue was processed and stained using a modified protocol from the University of Pittsburgh Center for Biological Imaging. Sectioned tissues were fixed with a 1:1 methanol-acetone mixture for 15 minutes at -20 ºC. Following fixation, sections were washed 3 times with 0.5% BSA in PBS, blocked with 0.5% BSA/5% goat serum in PBS, washed 5 times, and incubated with primary antibodies in 0.5% BSA/5% goat serum PBS at 4 ºC overnight. Sections were washed 5 times the next morning, followed by incubation of secondary antibody for 45 minutes at room temperature. Sections were washed 5 times with 0.5% BSA in PBS, 5 times with PBS, then counterstained with DAPI (300 nm) for 3 minutes at room temperature. Three final washes in PBS were followed by mounting in Gelvatol (PVA, glycerol, and sodium azide) prior to coverslip application and sealing of the slide. Dual fluorescence staining was performed using a primary antibody against murine CD3e followed by Alexa-Fluor 488 conjugated anti-Armenian hamster IgG and a concurrent anti-CD68 primary antibody followed by secondary staining with Alexa-Fluor 647 conjugated goat anti-rabbit. Further antibody details in Supplementary Table 4.

Image acquisition and analysis:
Following staining, slides were imaged slides using a Zeiss LSM 710 confocal microscope. For each stained section, the ten most significant peri-portal infiltrates were identified and images acquired at 20x magnification. CellProfiler open-source software was then used to develop a cell counting pipeline based on an example pipeline from CellProfiler (see supplemental Table 5 for workflow). After optimization of the cell-counting pipeline via paired manual counting of representative images, the pipeline was used to count the number of CD3-positive cells in each high-powered field, as well as calculate the percentage of CD3-positive cells (from total DAPI cells) in each image acquired. For each sample, the three images with the highest cellular count were selected for statistical analysis. Samples from 8 recipients of wild type cells and 8 recipients of AMPK dKO cells were used. Thus, a total of 24 infiltrates for each experimental condition was included in the final statistical analysis. As noted elsewhere, Student paired t-tests detected statistical differences between infiltrates characteristics in recipients of wild type versus AMPK dKO T cells.
Treg suppression assay 2x10 3 CD4 + CD25 neg Tcon from a minimum of three independent biologic replicates were labeled with CellTrace Violet and plated with 2x10 3 Mitomycin-C treated antigen presenting cells, 1 ug/ml anti-CD3 antibody, and varying ratios of fl/fl or AMPK dKO CD4 + FoxP3 + Treg from FoxP3 DTRGFP mice. Tcon "Division Index" was assessed at 72 hours, measuring the average number of divisions per cell and used to calculate percent suppression. To account for variance in Tcon division between biological replicates, the division index of Tcon at a particular Tcon/Treg ratio was divided by the baseline Tcon division for that sample SUPPLEMENTAL FIGURES Supplemental Figure 1. Antigen-driven activation of AMPK. A, When intracellular AMP/ATP ratios rise, AMPKα becomes phosphorylated on residue Thr172 by liver kinase B1 (LKB), which increases AMPK's kinase activity. Activated AMPK then phosphorylates downstream targets including Acetyl-CoA Carboxylase (ACC) and Unc51-like kinase (ULK-1). In skeletal muscle, phosphorylation of ACC inhibits its carboxylase activity, decreasing malonyl-CoA production, liberating the allosteric inhibition of CPT1a, and thereby increasing mitochondrial lipid transport as a precursor to β-oxidation. B, 1×10 6 OT-I and 1×10 6 OT-II cells were transplanted into irradiated CAG-OVA mice. On day 6 post-BMT, OT-I cells were flow-sorted and immunoblotting performed as in Figure 1A, comparing phospho/total AMPK ratios in pre-transplant (naïve) OT-I T cells to those recovered from antigen-bearing recipients on day +6 (n=10 animals per condition pooled into 3 groups).

Supplemental Figure 2. Intact thymic and peripheral T cell development in the absence of AMPK. A-C,
AMPKα1 fl/fl AMPKα2 fl/fl mice were crossed to CD4-Cre animals and thymi recovered from naïve mice at eight weeks of age. Total thymocyte number (A), percentage of double positive (DP) thymocytes (B), and number of DP thymocytes (C) was evaluated by flow cytometry. D-F, Splenocytes were recovered from donors in A-C and the percentage of B220 versus TCR-β cells (D), total TCR-β + cells (E), and CD4 + versus CD8 + percentages within the TCR-β + gate (F) were quantitated. N= 4 animals/group and data represent two independent experiments. **p<0.01 by unpaired Student T test.

Supplemental Figure 3. AMPK dKO T cells mediate equivalent leukemia clearance.
A-C, 1×10 6 fl/fl or AMPK dKO donor T cells were transplanted into irradiated B6D2F1 recipients with 5×10 6 TCD BM cells and 10 5 GFP + p815 leukemia cells (n= 15 mice/group). Recipients transplanted with only BM and p815 cells served as a positive control. Leukemia-bearing mice were weighed twice weekly and changes from baseline graphed as a function of time (A). On day 13 post-transplant, the total number of GFP + cells was quantitated in the liver, spleen, and BM of recipient mice (B). C, In a separate cohort, the total number of GFP + cells was quantitated in the liver and spleen of recipient animals on day 28, around the time of median overall survival )n= 6-7 mice/group). D, 1×10 6 or 0.25×10 6 fl/fl or AMPK dKO donor T cells were transplanted as above with 5×10 6 TCD BM cells and 10 5 GFP + p815 leukemia cells and recipient mice were monitored for survival out to 8 weeks post-transplant. P values in (D) represent comparisons to survival of no T cell control mice. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. P value for survival curves was determined by log-rank (Mantel Cox) test. All other statistics were by Student T test. . Livers were recovered from the same animals, stained with H&E, and scored for portal inflammation, bile duct injury, and central perivenulitis on a scale of 0 to 3 by a trained pathologist in a blinded fashion. These components were then totaled to generate an overall liver pathology score, with a max score of 9 (B). n= 10 mice/group in A/B. C, Computer-assisted confocal image analysis quantitated the percentage of CD3 + cells (of all nucleated (DAPI + ) cells) in each high-powered field of freshly frozen livers collected from a separate cohort of B6D2F1 transplanted mice (n= 24 hpf/group). *p<0.05, ***p<0.001 by unpaired Student T test.

Supplemental Figure 7. Autophagy remains unchanged in allogeneic AMPK dKO T cells.
A-D, 2x10 6 CD45.1 + B6 T cells were left unmanipulated (naïve) or labeled with CellTrace violet and transplanted with 5×10 6 B6 BM cells into irradiated B6 (Syn) or B6D2F1 (Allo) recipients. On day 7 post-transplant, the percentage of donor T cells positive for Cyto-ID, a marker of autophagolysosome (APL) formation, was evaluated by flow cytometry (A) and the median fluorescence intensity (MFI) for Cyto-ID staining averaged in donor T cells from multiple recipients (B). fl/fl or AMPK dKO CD8 T cells were flow sorted from recipients on day 7 and cell lysates blotted for LC3-II formation (n= 10 mice pooled into 3 sets/group). D, fl/fl or AMPK dKO donor CD8 + T cells were flow-sorted on day 7 and cell lysates immunoblotted for phospho-and total S6 protein levels (n=3-5 mice/group). ****p<0.0001 by Student T test. Supplemental Figure 13. Transplantation of wildtype Treg with fl/fl or AMPK dKO Tcon. Wildtype or AMPK dKO T cells (both CD90.1 + /CD90.2 + ) were depleted of CD25 + cells, added to wildtype CD90.2 + Treg (which totaled 15% of all CD4 + cells), and transplanted into irradiated B6D2F1 recipients. On day 7, the percentage of FoxP3 + cells, from within the total donor CD4 + population, or specifically the original CD4 + CD90.2 + Treg population, was assessed by flow cytometry. Data are displayed in Figure 6C . B-C, B6D2F1 recipients were transplanted with AMPK dKO_FoxP3 DTR donor T cells and administered 50 ng/kg diphtheria toxin (DT) on day 0 and +1 post-transplant. The percentage of FoxP3 + cells in spleen/mesenteric LN (B) or the total number of FoxP3 + cells in the spleen (C) was quantitated on day 7 (n= 4-6 mice/group, data represent 2 independent experiments). D-E, FoxP3 DTR fl/fl versus AMPK dKO donor T cells were transplanted into B6D2F1 recipients, treated on day 0 and +1 with DT, and followed for survival (D) and weight loss (E) to 10 weeks post-transplant (n=13-16 mice/group). Mice receiving TCD BM only served as a no-GVHD control (n=8).Asterisks in (E) refer to statistically significant differences between fl/fl and AMPK dKO groups. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by Student T test. P values for survival curves were determined by log-rank (Mantel Cox) test.