Inhibition of glucose metabolism selectively targets autoreactive follicular helper T cells

Abstract

Follicular helper T (T_FH) cells are expanded in systemic lupus erythematosus, where they are required to produce high affinity autoantibodies. Eliminating T_FH cells would, however compromise the production of protective antibodies against viral and bacterial pathogens. Here we show that inhibiting glucose metabolism results in a drastic reduction of the frequency and number of T_FH cells in lupus-prone mice. However, this inhibition has little effect on the production of T-cell-dependent antibodies following immunization with an exogenous antigen or on the frequency of virus-specific T_FH cells induced by infection with influenza. In contrast, glutaminolysis inhibition reduces both immunization-induced and autoimmune T_FH cells and humoral responses. Solute transporter gene signature suggests different glucose and amino acid fluxes between autoimmune T_FH cells and exogenous antigen-specific T_FH cells. Thus, blocking glucose metabolism may provide an effective therapeutic approach to treat systemic autoimmunity by eliminating autoreactive T_FH cells while preserving protective immunity against pathogens.

Fig. 1

Increased mTORC1 activity and survival in TC T_FH cells. a–c Frequency and cell number of CD69⁺ (a), CD4⁺CD44⁺CD62L⁻ T_EM cells (b), and CD4⁺CXCXR5⁺PD1⁺Bcl6⁺Foxp3⁻ T_FH (c) in CD4⁺ T cells, with the relative distribution of T_FH and Foxp3⁺ T_FR cells in CD4⁺CXCXR5⁺PD1⁺Bcl6⁺ cells. d Mean fluorescence intensity (MFI) of phospho-S6 in T_N cells (CD4⁺CD44⁻), T_Act cells (CD4⁺CD44⁺), and T_FH cells (CD4⁺CD44⁺PD-1^hiPSGL-1^lo). e Representative mTOR staining (green) in the GC (plain line) and T cell zone (dashed line) relative to non-GC B cells (IgD⁺ red) and CD4⁺ T cells (purple). Arrows point to examples of mTOR staining. Scale bars: 200 μM for 10× and 10 μM for 20×. f Frequency and number of Ki-67⁺ proliferating T_N, T_Act, and T_FH cells. g Bcl2 MFI in CD4⁺CXCXR5⁺PD1⁺ T_FH and CD4⁺CXCXR5⁻PD1⁻ non-T_FH cells, with representative FACS plots on the left. h Bcl2 mRNA expression in T_FH and non-T_FH cells as determined with a Nanostring custom panel. Spleens from 9-month-old B6 and TC mice, mean + s.e.m. of N = 3–8 mice per group compared with t tests. *P < 0.05, **P < 0.01, and ***P < 0.001

Fig. 2

Autoreactive T_FH cells are sensitive to glycolysis inhibition. Frequency (a) and cell number (b) of T_FH cells (left, CD4⁺CXCXR5⁺PD1⁺Bcl6⁺Foxp3⁻) in CD4⁺ T cells and GC B cells (middle, B220⁺GL7⁺FAS⁺) in B cells, as well as serum anti-dsDNA IgG (right) in four strains of lupus-prone mice treated with 2DG for 8 weeks starting at 5–6 months of age, as compared to untreated age-matched controls. Contemporaneous untreated B6 mice are shown as reference. Mean + s.e.m. of N = 3–28 mice per group. c–f 3–4-month-old B6 and TC mice were treated 2DG or not for 1 month. MFI of Bcl6 in T_FH cells (c) and the ratio of T_FH (CD4⁺CXCXR5⁺PD1⁺Bcl6⁺Foxp3⁻) to T_FR (CD4⁺CXCXR5⁺PD1⁺Bcl6⁺Foxp3⁺) cells (d). Phospho-S6 (e) as well as frequency and number of Ki-67⁺ (f) cells in T_N cells (CD4⁺ CD44⁻), T_Act cells (CD4⁺CD44⁺), and T_FH cells (CD4⁺CD44⁺PD-1^hiPSGL-1^lo). Mean + s.e.m. of N = 4–9 spleens per group compared with t tests. *P < 0.05, **P < 0.01, and ***P < 0.001

Fig. 3

Glycolysis inhibition does not affect the TD-humoral response. After pre-treatment with 2DG for 2 weeks, 8–10-week-old B6 and TC mice were immunized with NP-KLH in alum and maintained under 2DG treatment until sacrifice. Mice were analyzed 10 d after immunization (primary), or 7 weeks after the first immunization following 2 boosts with same antigen 2 and 6 weeks after the primary immunization (memory). Frequency and number of total (a, b) or NP-specific (d, e) GC B cells and plasma cells (gating shown in supplementary Fig. 3). Serum anti-dsDNA IgG (c) and NP-specific IgG antibody-forming cells (f) in TC mice in the memory response. g Serum levels of high-affinity anti-NP₄ and low-affinity anti-NP₂₅ IgG1 and IgG2a in the primary (left) and memory (right) responses. Frequency and number of splenic T_FH cells (h) and T_FH/T_FR cell ratio (i) in the primary and memory responses. All cell analyses were performed with splenocytes. Mean + s.e.m. of N = 5 mice per group compared with t tests. *P < 0.05, **P < 0.01, and ***P < 0.001. For simplification, statistical differences between strains are not shown

Fig. 4

Glycolysis inhibition does not impair the humoral response to influenza immunization. 8–10-week-old B6 and TC mice were infected with PR8 influenza virus. 2DG treatment was initiated 2 weeks before infection and maintained until termination. CD4⁺ and CD8⁺ T cells were analyzed 10 d after infection. a Representative FACS plot (left) and percentage (right) of splenic influenza virus-specific CD4⁺ T cells (CD4⁺CD44⁺NP-Tet^pos). b, c Frequency and number of CD4⁺CD44⁺PD-1^hiPSGL-1^lo T_FH cells in PR8 virus-specific NP-Tet^posCD4⁺ T cells (b) and in total CD4⁺ T cells (c). Serum levels of PR8 NP-specific IgG1, IgG2b and IgG2c (d), and anti-dsDNA IgG (e) 30 d after low dose infection. f, g Frequency and number of IFNγ⁺ in total CD4⁺ T cells (f) and PR8 virus-specific NP-Tet^posCD4⁺ T cells (g) in the lung. h, i Analysis of NP-Tet^posCD8⁺ T cells in the lung: representative FACS plots (h), total number (i), frequency and number of SLEC (j) and MPEC (k) among these NP-Tet^posCD8⁺ T cells, and SLEC/MPEC ratio (l). Mean + s.e.m. of N = 3–8 mice per group compared with t tests. *P < 0.05 and **P < 0.01

Fig. 5

Spontaneous TC T_FH cells are glycolytic and show an altered expression of solute transporters. a Hif1a and Mct4 gene expression was compared between total CD4⁺ T cells and T_FH cells (CD4⁺CXCR5⁺PD-1⁺) from TC mice using qRT-PCR and normalized to Ppia. b Hif1α protein expression (as determined by MFI using flow cytometry) in total CD4⁺ T cells and T_FH cells from B6 and TC mice. c–d ECAR during a mitochondrial stress test conducted on T_FH (CD4⁺CD44⁺PD-1^hiPSGL-1^lo) and CD44⁻ CD4⁺ T cells from B6 and TC mice. c Time-course with the arrows indicating the addition of oligomycin (Olig), trifluoromethoxy carbonylcyanide phenylhydrazone (FCCP), and actimycin A and rotenone (Act & Rot). The horizontal gray bar indicates the three time-point measurements of basal ECAR, and the white bar indicates maximum ECAR. The TC and B6 T_FH cells plots were significantly different (two-way ANOVA, P < 0.001). d Basal and max ECAR averages in B6 and TC T_FH cells. e 2NDBG uptake by B6 and TC T_N cells (CD4⁺CD44⁻), T_Act cells (CD4⁺CD44⁺), and T_FH cells (CD4⁺CD44⁺PD-1^hiPSGL-1^lo). f Microarray heat-map and scatter plot of solute transporter genes differentially expressed between B6 and TC T_FH cells. Heatmap coloring is mean centered (white), standard deviation normalized to 1 with red indicating above average expression and blue for below average expression. Dotplot coloring is based on signal intensity RMA normalized values, blue is lowest expression and red is highest expression. Selected genes are labeled and only the range 3–9 is shown on dotplot. g Expression of solute transporter genes in T_FH cells was analyzed using qRT-PCR, and normalized to Hmbs. h Representative FACS plot (left) and MFI (right) of CD98 expression on TC and B6 T_FH cells. All T cells were isolated from spleens from 7–8 month-old mice. Mean + s.e.m. of N = 4–7 per strain, compared with t tests (a, b, d, e, g, h), *P < 0.05, **P < 0.01, and ***P < 0.001

Fig. 6

Spontaneous but not virus-specific TC T_FH cells expressed lower levels of glutamine transporters. a Gene signature of spontaneous CD4⁺CXCR5⁺PD-1⁺ T_FH cells using microarray (left) and Nanostring (right) on two independent cohorts of TC and B6 mice. b, c Slc1a5 expression was analyzed with the Nanostring panel in spontaneous T_FH cells (b), as well as in influenza virus-specific NP⁺ T_FH cells (CD4⁺CXCR5⁺PD-1⁺NP-Tet^pos, right) versus non-specific T_FH cells (CD4⁺CXCR5⁺PD-1⁺NP-Tet⁻, left) (c). d Representative FACS plot (left) and MFI (right) of CD98 expression on influenza virus-specific NP-Tet^pos T_FH cells (CD4⁺CXCR5⁺PD-1⁺NP-Tet^pos) and non-specific T_FH cells (CD4⁺CXCR5⁺PD-1⁺NP-Tet^neg), as well as non-T_FH (CD4⁺CXCXR5⁻PD1⁻) cells. All T cells were isolated from spleens. Mean + s.e.m. of N = 3–9 mice per group compared with t tests, *P < 0.05, **P < 0.01, and ***P < 0.001

Fig. 7

Glutamine metabolism is required for both induced and spontaneous antibody responses. a Experimental design for immunization with NP-KLH in alum in 8–10-week-old B6 and TC mice treated or not with glutamine antagonist DON. Frequency and cell number of T_FH cells (b), and ratio of T_FH to T_FR cells (CD4⁺CXCXR5⁺PD1⁺Bcl6⁺FOXP³⁻ and CD4⁺CXCXR5⁺PD1⁺Bcl6⁺FOXP3⁺) (c). d MFI of Bcl6, CD40L and ICOS in T_FH cells. e Representative spleen sections with GCs shown in the dotted lines. The first rows show composite images and the second rows show GL7-staining only. All images are 10× amplification (scale bar: 200 μM). f, g Frequency and number of NP-specific GC B cells (f) and number of NP-specific plasma cells (g). Serum levels of high-affinity anti-NP₄, low-affinity anti-NP₂₅ IgG1, and anti-NP₂₅ IgM NP-specific antibodies (h), and total IgG and IgM (i). All cell analyses were performed with splenocytes. Mean + s.e.m. of N = 4–12 mice per group compared with t tests, *P < 0.05, **P < 0.01, and ***P < 0.001

Fig. 8

Effect of DON treatment on spontaneous GC formation. a–c B6 and TC mice were treated with DON 3 times a week for 2 weeks starting at 7–8 months of age. Frequency and number of total B cells (a), GC B cells (b) and plasma cells (c). d Representative spleen sections stained with B220-PE (red), CD4-APC (purple) and GL7-FITC (green). The first rows show composite images and the second rows show GL7-staining only. GCs are shown in the dotted lines. All images are 10× amplification (scale bar: 200 μM). e, f GC (e) and follicle (f) surface area as measured by GL7⁺ pixels (e, left) and GL7 intensity in these areas (f, right), and by B220⁺ pixels in 3–5 high power fields from 2 mice per group. g–i Frequency and number of total CD4⁺ T cells (g), T_FH cells (h), and MFI of Bcl6 in T_FH cells (i). j Serum anti-dsDNA IgG represented as percent change from pre-treatment value. Statistical comparisons were made between terminal samples. All cell analyses were performed with splenocytes. Mean + s.e.m. N = 6 mice per group compared with t tests, *P < 0.05, **P < 0.01, and ***P < 0.001