Graphical abstract

   graphic file with name fx1.jpg
   [73]Open in a new tab

Highlights

     * •
       Macrophage-mediated lymphoma cell phagocytosis is increased by PPP
       inhibition
     * •
       PPP inhibition is an immune-regulatory switch for macrophage
       function
     * •
       PPP inhibition is linked to the modulation of the
       UDPG-Stat1-Irg1-itaconate axis
     * •
       PPP inhibition is tolerable in vivo and boosts therapeutic B cell
       lymphoma targeting
     __________________________________________________________________

   Beielstein et al. have shown that the metabolic inhibition of the
   pentose phosphate pathway in macrophages leads to increased macrophage
   activation, less lymphoma cell support, and increased
   antibody-dependent lymphoma cell clearance by macrophages. A
   significantly improved overall survival in a therapeutic lymphoma mouse
   model is achieved.

Introduction

   The tumor microenvironment (TME) represents a hallmark of cancer, and
   interactions between its transformed and non-transformed immune
   bystander cells determine disease progression and therapeutic
   response.[74]^1 Tumor cells generate a tumor-supportive environment by
   cytokine and metabolite secretion. These mediators alter occurrence of
   bystander cells and shift the activity of the infiltrating immune cells
   from an anti- to a pro-tumoral response.[75]^2 Tumor-associated
   macrophages (TAMs) play a critical role in promoting tumor growth,
   facilitating vascularization and metastasis and suppressing other
   immune cells.[76]^3^,[77]^4

   However, we have shown that macrophages are central for tumor cell
   clearance in aggressive B cell lymphoma during immunotherapy, although
   their phagocytic capacity becomes impaired by lymphoma cells.[78]^5

   Recent decades have seen the development of numerous new treatment
   strategies for B cell malignancies, which have extended patient
   survival but struggled to substantially increase cure rates. The
   front-line strategy is chemo-immunotherapy, combining therapeutic
   antibodies like rituximab or obinutuzumab with chemotherapy. We
   demonstrated that leukemia cells in therapy-refractory niches reduce
   the engulfment of antibody-targeted tumor cells by macrophages,
   diminishing therapy efficacy and leading to relapse.[79]^5

   Macrophage function depends on their local environment, which impacts
   their differentiation and polarization. Macrophages are a heterogeneous
   population with different subtypes exerting pro-inflammatory and
   phagocytic (M1-like macrophages) and anti-inflammatory and
   tissue-regenerative (M2-like macrophages) activities. TAMs represent a
   blend of these characteristics tending toward the anti-inflammatory and
   phagocytic inactive phenotype.[80]^6^,[81]^7 The TME includes
   activation mediators such as cytokines, chemokines, and metabolites,
   which control the polarization of contained macrophages. Changes in the
   microenvironment can alter macrophage metabolism, which is crucial for
   polarization and the closely linked macrophage function.[82]^8^,[83]^9
   Changes in the cellular metabolism have the ability to repolarize the
   macrophage phenotype by which pro-tumoral TAMs could acquire
   anti-tumoral activity.[84]^10^,[85]^11^,[86]^12

   The interaction between macrophages and lymphoma cells, as well as
   macrophages metabolic sensitivity, opens up a promising strategy to
   optimize anti-cancer therapy. Modulating macrophage metabolism may
   improve their anti-tumor efficacy and diminish their tumor-supportive
   function.

   In the present study, we demonstrate that pentose phosphate pathway
   (PPP) inhibition in macrophages increases their activity and phagocytic
   capacity whereby pro-tumoral bystander function is diminished. As a
   driving mechanism, we discovered a connection between metabolism and
   immune regulation by modulation of the UDPG-Stat1-Irg1-itaconate axis.
   The effects of PPP inhibition were transmitted into human patient
   samples and also reproduced in vivo, where significantly increased
   survival in an aggressive lymphoma mouse model was achieved. These
   results open up a promising field of treatment strategy against B cell
   malignancies in clinical use.

Results

Metabolic inhibition of the PPP leads to increased phagocytic capacity of
macrophages

   To investigate how metabolic modulation of TAMs in the context of
   immunotherapy affects phagocytic capacity, we performed a
   metabolism-focused screening approach for antibody-dependent cellular
   phagocytosis (ADCP). Key metabolic pathways were blocked using
   representative inhibitors in a macrophage and humanized aggressive B
   cell lymphoma (hMB; cell line information see [87]STAR Methods)
   co-culture-assay system, and phagocytosis was assessed through specific
   antibody targeting (alemtuzumab; anti-CD52) ([88]Figure 1A). The
   antibody alemtuzumab was used as a tool compound for the first
   screening approach as several types of lymphoma downregulate CD20
   expression but not CD52 expression, also seen in hMB cells. Several
   other antibodies, currently in clinical use, were investigated in
   further analysis.

Figure 1.

   [89]Figure 1
   [90]Open in a new tab

   Metabolic inhibition of the pentose phosphate pathway leads to
   increased phagocytic rate of macrophages

   (A) Scheme of ADCP-based metabolic screening approach.

   (B–D) Summary of ADCP change compared to basal phagocytosis rate of
   J774A.1 macrophages under inhibition of respective metabolic pathways.
   (B) Inhibition of only macrophages. (C) Inhibition of only hMB cells.
   (D) Inhibition of all co-culture components. Technical replicates (B)
   n = 15–22, (C) n = 15–58, (D) n = 15–28; biological replicates (B) n =
   3–5, (C) n = 3–12, (D) n = 3–6. Data are shown as mean ± SEM. p values
   were calculated using unpaired t test. ∗p < 0.05; ∗∗p < 0.01;
   ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. See also [91]Figures S1 and [92]S2.

   Inhibition of glycolysis (via 2-deoxy-D-glucose), AMP-activated protein
   kinase (AMPK)-mediated cell energy regulation (via BML-275),
   mitochondrial ATP production (via oligomycin), and the PPP (via
   6-aminonicotinamide and oxythiamine) was screened using non-toxic
   inhibitor concentrations ([93]Figure S1). The inhibition was conducted
   in co-culture and by pre-treatment of each cell type (macrophage or hMB
   cell), to infer specific macrophage vs. lymphoma cell phagocytic
   interactions. As the basal phagocytosis rate of macrophages is
   variable, the change in phagocytosis under treatment was calculated in
   comparison to the basal phagocytosis rate (=ADCP change, [94]Figures 1A
   and [95]S2).

   Glycolysis inhibition significantly increased ADCP rates in co-culture
   (+40%, p < 0.01) and by pre-treatment of lymphoma cells (+95%,
   p < 0.0001), while macrophage pre-treatment significantly diminished
   ADCP rate (−71%, p < 0.001) ([96]Figures 1B–1D). Similarly, AMPK
   inhibition increased ADCP rate significantly by lymphoma cell
   pre-treatment (+39%, p < 0.0001) and significantly diminished ADCP rate
   by pre-treatment of macrophages (−42%, p < 0.001) ([97]Figures 1C and
   1D). Inhibition of mitochondrial ATP production also diminished ADCP
   rate significantly by macrophage pre-treatment (−21%, p < 0.0001)
   ([98]Figure 1D).

   Sole inhibition of the PPP induced significantly increased ADCP rates
   by co-culture treatment and macrophage pre-treatment. The increase was
   induced by both inhibition of the oxidative part of the PPP via
   6-phosphogluconate dehydrogenase inhibition (6Pgd; inhibitor
   6-aminonicotinamide) (co-culture +40% p < 0.01; macrophage
   pre-treatment +15%, p < 0.05) and inhibition of the non-oxidative part
   via transketolase inhibition (Tkt; inhibitor oxythiamine)
   (co-culture +51% p < 0.001; macrophage pre-treatment +28%, p < 0.01)
   ([99]Figures 1B and 1D). Moreover, lymphoma cell pre-treatment with
   oxythiamine increased phagocytic rate significantly (+19%, p < 0.0001)
   ([100]Figure 1C).

   Of note, Tkt inhibition induced the highest increase in phagocytic
   capacity in the co-culture and by pre-treatment of macrophages in the
   screening approach.

   Thus, inhibition of glycolysis, AMPK, and mitochondrial ATP production
   negatively affected macrophages’ phagocytic capacity, while blocking
   PPP favored lymphoma cell clearance by macrophages.

Cross validation of PPP inhibition in macrophages confirms increased ADCP
rates

   To further investigate the PPP in the context of macrophages’ function
   and as a target for improving immunotherapy, we applied alternative
   inhibitors (6Pgd: physcion, Tkt:
   p-hydroxyphenylpyruvate)[101]^13^,[102]^14 and confirmed significant
   increases in ADCP rates (physcion +26%, p < 0.0001;
   p-hydroxyphenylpyruvate +25%, p < 0.001) ([103]Figure 2A).
   Additionally, we recapitulated the phagocytosis assays with the human
   monocyte cell line THP1 using an alternative antibody (obinutuzumab;
   anti-CD20 type II), also identifying significant induction of ADCP
   (6-aminonicotinamide p < 0.05; oxythiamine p < 0.01) ([104]Figure 2B).

Figure 2.

   [105]Figure 2
   [106]Open in a new tab

   Cross validation of PPP inhibition in macrophages confirms increased
   ADCP rates

   (A–C) ADCP change compared to basal phagocytosis rate of J774A.1
   macrophages under inhibition of PPP. (A) Alternative inhibitor physcion
   of 6-phosphogluconate dehydrogenase (6Pgd) in oxidative part of PPP
   (red) and p-hydroxyphenylpyruvate for inhibition of transketolase (Tkt)
   in non-oxidative part of PPP (blue). (B) Using human monocyte cell line
   THP1 and CD20 antibody obinutuzumab under inhibition of 6Pgd by
   6-aminonicotinamide (red) and inhibition of Tkt by oxythiamine (blue).
   (C) ADCP assay performed under hypoxic conditions and inhibition of
   6Pgd by physcion (red) or inhibition of Tkt by oxythiamine (blue).

   (D) Antibody-independent cellular phagocytosis of hMB cells by J774A.1
   macrophages compared to control under inhibition of 6Pgd by
   6-aminonicotinamide (left) and physcion (right).

   (E) ADCP change compared to basal phagocytosis rate of empty vector
   control J774A.1 macrophages under shRNA-mediated knockdown of 6Pgd
   (red) and Tkt (blue).

   (F) ADCP change compared to basal phagocytosis rate of J774A.1
   macrophages under supplementation of metabolites of the PPP. Enzyme
   reactions in focus colored in violet (6Pgd) and blue (Tkt). E4P,
   erythrose-4-phosphate; F6P, fructose-6-phosphate; G3P,
   glyceraldehyde-3-phosphate; Glc6P, glucose-6-phosphate; R5P,
   ribose-5-phosphate; Ru5P, ribulose-5-phosphate; S7P,
   sedoheptulose-7-phosphate; X5P, xylulose-5-phosphate. Data are shown as
   mean ± SEM. Technical replicates (A) n = 30, (B) n = 17–25, (C) n =
   25–28, (D) n = 30, (E) n = 20–23, (F) n = 13–20; biological replicates
   (A) n = 6, (B) n = 4–5, (C) n = 5–6, (D) n = 6, (E) n = 4–5, (F) n =
   3–4. p values were calculated using one-way ANOVA. ∗p < 0.05;
   ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. See also [107]Figures S1 and
   [108]S3.

   Since hypoxia is a functional aspect of the TME in vivo, we
   also conducted ADCP assays under hypoxic conditions (O[2] 1.5%) and
   observed significantly increased ADCP rates (physcion +12%, p < 0.01;
   oxythiamine +20%, p < 0.01) ([109]Figure 2C).

   To evaluate if PPP inhibition also increases phagocytic capacity of
   macrophages without the targeting function of antibodies, we assessed
   antibody-independent cellular phagocytosis (AICP) ([110]Figure 2D) and
   observed significantly increased AICP rates only by inhibition of the
   oxidative part of the PPP (6-aminonicotinamide p < 0.01, physcion
   p < 0.01) ([111]Figure S3D).

   To abrogate off-target effects of the PPP inhibitors, we generated
   short hairpin RNA (shRNA) knockdown for the respective enzymes in
   macrophages ([112]Figures S3P–S3S). Silencing of 6Pgd and Tkt
   significantly increased macrophages’ ADCP rates (6Pgd +87%, p < 0.0001;
   Tkt +45%, p < 0.001) ([113]Figure 2E).

   Altogether, we demonstrate that PPP enzyme inhibition by metabolic
   inhibitors as well as 6Pgd and Tkt knockdown in macrophages promotes
   phagocytosis of lymphoma cells and warrants further investigation of
   the molecular function.

Increased phagocytosis is driven by PPP enzyme inhibition and not by PPP
metabolite shifting

   To identify which specific components and metabolites of the PPP
   directly affect phagocytic function, we performed ADCP assays with
   supplementation of single educts and products of the
   PPP ([114]Figure 2F). We observed unaltered ADCP rates using the
   non-exclusive PPP metabolites glucose-6-phosphate (Glc6P),
   ribose-5-phosphate (R5P), xylulose-5-phosphate (X5P),
   glyceraldehyde-3-phosphate (G3P), and fructose-6-phosphate (F6P). In
   contrast, supplementation of the PPP-exclusive glucose-6-phosphate
   dehydrogenase (G6pd) product 6-phosphogluconolactone significantly
   increased ADCP rate (+13%, p < 0.05), as well as the products of 6pgd
   (ribulose-5-phosphate [Ru5P]; +17%, p < 0.0001) and of Tkt
   (sedoheptulose-7-phosphate [S7P]; +20%, p < 0.05) ([115]Figure 2F;
   right panel). In contrast, supplementing the Tkt educt
   erythrose-4-phosphate (E4P) significantly reduced ADCP rate (−12%,
   p < 0.01) ([116]Figure 2F; right panel).

   In conclusion, products of Tkt and 6Pgd promote macrophages’ phagocytic
   activity while enzyme educts diminish it, indicating that inhibition of
   the enzymes itself and not a decrease in their products causes
   increased phagocytic capacity.

PPP inhibition induces pro-inflammatory polarization and activation in
macrophages

   To test whether PPP inhibition alters macrophage differentiation and
   activation, we assessed expression of markers delineating polarization
   by flow cytometry ([117]Figures 3A and 3B; [118]Table S1). We observed
   a trend of increased M1-like marker expression and decreased M2-like
   and TAM marker expression under PPP inhibition and knockdown.

Figure 3.

   [119]Figure 3
   [120]Open in a new tab

   PPP inhibition induces pro-inflammatory polarization and activation in
   macrophages

   (A and B) Radar plot of surface marker expression of J774A.1
   macrophages. Expression of characteristic surface marker for different
   macrophage subtypes measured by immunofluorescent staining. Mean
   fluorescence intensity (MFI) is depicted. To improve readability, high
   MFI has been downscaled (factor named in brackets next to marker).
   (A) Compound mediated PPP inhibition. (B) shRNA-mediated PPP knockdown.

   (C) Immunofluorescent microscopy of J774A.1 macrophages under
   compound-mediated PPP inhibition and shRNA-mediated PPP knockdown.
   Blue, phalloidin staining of nucleus; green, actin staining of
   cytoskeleton.

   (D–G) Measurement of metabolic activity of J774A.1 macrophages under
   compound-mediated PPP inhibition by Seahorse analysis. Inhibition of
   non-oxidative part of PPP by oxythiamine, inhibition of oxidative part
   of PPP by physcion. (D) One representative example of XF Mito Stress
   test measurement of ECAR und OCR. (E) Respiratory basal rate and
   capacity. (F) Glycolytic basal rate and capacity. (G) ATP production.
   Data are shown in (A and B) as mean of four replicates, in (D) as mean
   of six replicates in one experiment ±SD, and in (E–G) as mean ± 5–95
   percentile. Technical replicates (A and B) n = 4, (D) n = 6, (E–G) n =
   18–27; biological replicates (A and B) n = 4, (D) n = 1, (E–G) n = 6–9.
   p values were calculated using one-way ANOVA, (D) using two-way ANOVA.
   ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. See also
   [121]Figure S4 and [122]Table S1.

   To evaluate macrophage morphology, we performed fluorescent microscopy
   ([123]Figure 3C). Under PPP inhibition the macrophages underwent a
   profound change in morphology from a round, centered appearance to a
   spread and outlaying phenotype with filopodia surrounding the cell
   body.

   As macrophages’ metabolic status greatly influences their activity and
   polarization, we assessed glycolytic and mitochondrial activity with
   the Seahorse XF Mito Stress test ([124]Figures 3D–3G, [125]Figure S4).
   We observed a significant increase of the oxygen consumption rate (OCR)
   (p < 0.0001) and extracellular acidification rate (ECAR) (p < 0.0001)
   ([126]Figure 3D) indicating an increased mitochondrial respiration and
   glycolytic activity and thus an increased metabolic activity of
   macrophages. Further analysis identified significant increase of
   mitochondrial basal activity (physcion p < 0.01; oxythiamine
   p < 0.001), mitochondrial maximal capacity (physcion p < 0.05;
   oxythiamine p < 0.01), glycolytic maximal capacity (physcion p < 0.05;
   oxythiamine p < 0.001), and ATP production (physcion p < 0.05;
   oxythiamine p < 0.0001) of macrophages ([127]Figures 3E–3G).

   Taken together, these data show an activation of macrophages by shifted
   polarization, cytoskeletal reorganization, and increased metabolic
   activity under PPP inhibition as possible functional basis of increased
   phagocytosis.

PPP inhibition changes the proteomic profile of macrophages toward
pro-inflammatory activity

   To investigate the mediators of increased phagocytic capacity in
   macrophages, we performed a multi-omics (proteomics, phosphoproteomics,
   and metabolomics) screening under chemical or shRNA-mediated PPP
   inhibition.

   A uniform regulation pattern of proteins involved in macrophage
   polarization and activation was observed by the use of independent
   inhibitors and PPP enzyme knockdowns ([128]Figures 4A and 4B,
   [129]Table S2). Under compound-mediated PPP inhibition, the
   anti-inflammatory proteins Ptgs1, Sqstm1, and
   Ybx3[130]^15^,[131]^16^,[132]^17 and the TAM- and M2-typical proteins
   Hagh and Ezr[133]^18^,[134]^19 were significantly downregulated, while
   the pro-inflammatory proteins Pam16 and Gosr1 were significantly
   upregulated ([135]Figure 4A). By using shRNA, an even more pronounced
   regulation was seen. Negative regulators of cytokine expression and
   pro-inflammatory signaling were significantly downregulated while
   promoters of pro-inflammatory activation were significantly upregulated
   (Atg16l1, Cast, Csf1r, Cybb, Inppl1, Oas3, Parp14, Fkbp5, Ilf2,
   Tlr7).[136]^20^,[137]^21 Moreover, there was a significant increase in
   protein expression needed for phagocytosis (Actn1, Actr1a, Iqgap3,
   Itgav, Lrp1, Myl12a, Necap1, Sh3bp1) ([138]Figure 4B; for
   phosphoproteomic analysis see [139]Table S3).

Figure 4.

   [140]Figure 4
   [141]Open in a new tab

   PPP inhibition changes the proteomic profile of macrophages towards
   pro-inflammatory activity

   (A and B) Volcano plots showing mean change of proteomic transcription
   under (A) compound-mediated PPP inhibition by 6-aminonicotinamide and
   oxythiamine compared to untreated J774A.1 macrophages and
   (B) shRNA-mediated PPP knockdown of 6Pgd and Tkt compared to empty
   vector control J774A.1 macrophages. Circle size represents number of
   significantly changed conditions. Red circles: significantly
   downregulated abundance; green circles: significantly upregulated
   abundance. Proteins known to participate in immune system are annotated
   in significant groups.

   (C and D) Pathway enrichment analysis of (C) proteomics and
   (D) phosphoproteomics of J774A.1 macrophages under compound-mediated
   PPP inhibition and shRNA-mediated PPP knockdown. Protein count in
   listed pathways represented in circle size, mean −log10 p value
   represented in heatmap analysis.

   (E and F) Analysis of significantly negative changed protein activity
   in normalized upstream kinase score (NUKS). (E) Top five most
   downregulated enzymes in NUKS analysis under shRNA-mediated PPP
   knockdown of 6Pgd and Tkt and (F) integrative analysis of
   compound-mediated PPP inhibition by physcion and oxythiamine.

   (G) Western blot analysis of Ptk2b expression in J774A.1 macrophages
   under shRNA-mediated PPP knockdown of 6Pgd and Tkt compared to empty
   vector control.

   (H) Scheme of hypothesized mechanism leading to pro-inflammatory
   phenotype of macrophages. In (G) data are shown as mean ± SEM.
   Technical replicates (A–F) n = 1, (G) n = 5; biological replicates
   (A–F) n = 3, (G) n = 5. p values in (G) were calculated using one-way
   ANOVA. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. See also
   [142]Tables S2–S5.

   We performed pathway enrichment analysis for functional annotation
   ([143]Figures 4C and 4D), showing similar enrichment clusters for both
   compound-mediated and shRNA-mediated inhibition of the oxidative and
   the non-oxidative part of the PPP. A strong enrichment was seen for
   immune activity ([144]Figure 4C) including cytokine signaling, antigen
   processing, and antigen presentation with up to 124 involved proteins
   and significantly changed phosphorylation patterns ([145]Figure 4D).
   Moreover, enrichment in proteins relevant to phagocytosis and
   cytoskeletal organization was observed. In line with our metabolic flux
   analysis ([146]Figures 3D–3G), we observed great enrichment for
   proteins influencing mitochondrial and glycolytic activity
   ([147]Figure 4C). Particularly analysis of phosphoproteomics uncovered
   a significant enrichment in signaling pathways important for immune
   signaling (mitogen-activated protein kinase [Mapk]-Erk, Egf/Egfr, p53,
   Pi3k-Akt) and metabolic regulation (Pi3k-Akt, Hif1a) ([148]Figure 4D).

   To further analyze the impact of altered protein phosphorylation, we
   performed an adapted upstream kinase analysis on the basis of
   integrative inferred kinase activity (INKA) analysis
   ([149]Table S4).[150]^22 The five most inactivated kinases are
   displayed ([151]Figure 4E), highlighting the decrease of Hck in the
   normalized upstream kinase score (NUKS). Hck supports M2-like
   macrophage polarization, TAM activity, tumor growth, and tumor cell
   evasion[152]^23 and activates the Csf1 receptor (Csf1r).[153]^24 Csf1r
   signaling likewise induces M2-like macrophage polarization.[154]^25
   Csf1r and its downstream kinase Mapk1 were also one of the five most
   inactivated kinases ([155]Figure 4E). In combined analysis of PPP
   inhibitors, the Csf1r downstream kinase Ptk2b (Pyk2) was the most
   negatively regulated kinase ([156]Figure 4F). Accordingly, a
   significant downregulation of Ptk2b in PPP knockdown macrophages was
   observed ([157]Figures 4G and [158]S5A). Furthermore, the most
   downregulated protein in both knockdown macrophages was Sema4d
   ([159]Figure 4B), which is an activator of the Ptk2b pathway.[160]^26

   Following the Csf1r pathway further downstream ([161]Figure 4H),
   decreased immune-regulatory gene 1 (Irg1 = Acod1) expression, a major
   node in immunosuppressive regulation of macrophages, was seen in
   proteomic analysis ([162]Table S2). Changed Irg1 expression is one
   possible mechanism leading to altered macrophage activity and
   phagocytosis.[163]^27 With the exception of Hmox-1, all included signal
   molecules of the regarded Csf1r pathway were significantly
   downregulated in proteomic analysis ([164]Figure 4H) ([165]Table S2).

PPP inhibition modulates glycogen metabolism and the immune response
signaling axis UDPG-Stat1-Irg1-itaconate of macrophages

   Regarding the critical role of Irg1 on macrophage polarization, we
   aimed to explore the connection between metabolic modulation, Irg1
   regulation, and the resulting macrophage phenotype.

   PPP and glycogenolysis activity are coupled causing suppression of both
   pathways if one is inhibited.[166]^28 We quantified glycogen levels
   identifying significant decreased glycogen amount under all conditions
   (p < 0.0001, [167]Figure 5A). Glycogen metabolism influences signaling
   regulating Stat1 activity.[168]^28 Thus, we hypothesized that
   inhibition of PPP would lead to suppression of glycogenolysis with
   subsequent decreased uridine diphosphate glucose (UDPG) production and
   thereby to an inhibition of P2y14 expression with following decreased
   Stat1 activity. Decreased Stat1 activity leads to less Irf1 and thereby
   to a decreased Irg1 expression, which possibly leads to functional
   increasing macrophage activity and phagocytosis.[169]^29^,[170]^30

Figure 5.

   [171]Figure 5
   [172]Open in a new tab

   PPP inhibition modulates glycogen metabolism and the immune response
   signaling axis UDPG-Stat1-Irg1-itaconate of macrophages

   (A) Total glycogen amount in J774A.1 macrophages under
   compound-mediated PPP inhibition and shRNA-mediated knockdown of 6Pgd
   and Tkt.

   (B) Western blot analysis of protein expression of hypothesized
   connecting pathway in J774A.1 macrophages under shRNA-mediated
   knockdown of 6Pgd and Tkt compared to empty vector control. Mean
   expression displayed in bar graph analysis and one representative
   western blot example.

   (C) Scheme of working hypothesis of PPP metabolism modulating immune
   response.

   (D) Amount of 6pgd product ribulose-5-phosphate and Tkt product
   sedoheptulose-7-phosphate under shRNA-mediated inhibition of 6pgd and
   Tkt in J774A1 macrophages.

   (E) Metabolomic analysis of tricarboxylic acid cycle and citrate
   metabolism with display of enzyme expression of key enzymes under
   shRNA-mediated PPP knockdown of 6Pgd and Tkt compared to empty vector
   control J774A.1 macrophages. Amount of metabolites displayed in box and
   whiskers. Change in enzyme expression displayed in bar graphs. Genes:
   succinate dehydrogenase (Sdh), ATP citrate lyase (Acly),
   immunoregulatory gene 1 (Irg1 = Acod1).

   (F and G) Cytokine expression under 6Pgd inhibition by
   6-aminonicotinamide or physcion and Tkt inhibition by oxythiamine in
   J774A.1 macrophages. (F) IL-6 expression compared to untreated control.
   (G) IL-10 expression compared to untreated control.

   (H) ADCP assay of bone marrow-derived macrophages of Irg1^+/+ wild-type
   mice and Irg1^−/− knockout mice. Macrophages differentiated out of
   femoral bone marrow with M-CSF. In (A), (B), and (F–H), bar plots are
   shown as mean ± SEM; in (D and E) metabolite amount is shown as minimum
   to maximum and protein expression is shown as calculated −Log2 fold
   change of control and knockdown macrophages. Technical replicates (A)
   n = 15, (B) n = 3–6, (D) n = 3, (E) n = 3, (F and G) n = 9–18, (H) n =
   35; biological replicates (A) n = 3, (B) n = 3–6, (D) n = 3, (E) n = 3,
   (F and G) n = 3–6, (H) n = 7. p values were calculated in (A, B, D, and
   E) using one-way ANOVA, protein expression in (E) using student’s t
   test, and in (F–H) using unpaired t test. ∗p < 0.05; ∗∗p < 0.01;
   ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. See also [173]Figures S4 and [174]S5 and
   [175]Tables S6 and [176]S7.

   To validate this hypothesis, we performed western blot analysis of the
   hypothesized pathway-associated proteins and identified significant
   reductions in expression (p < 0.0001, [177]Figure 5B) with the highest
   decline of Irg1 amount (>80%, p < 0.0001, [178]Figure S5). The
   hypothesized pathway linking PPP activity and Irg1 expression is
   displayed in [179]Figure 5C.

   As Irg1 acts as a metabolic enzyme and changes in immune activity are
   driven by its product itaconate, we investigated the connection between
   metabolism and enzyme expression by metabolomic assessment
   ([180]Tables S5 and [181]S6) and compared it to changes in enzyme
   expression of interest detected in proteomics ([182]Table S2)
   ([183]Figures 5D and [184]S4C).

   The metabolomic screening[185]^31 confirmed a significantly decreased
   amount of exclusive 6Pgd and Tkt products ribulose-5-phosphate and
   sedoheptulose-7-phosphate (p < 0.0001, [186]Figure 5D) in 6pgd and Tkt
   knockdown macrophages, while drug-mediated inhibition did not show a
   significantly decreased amount ([187]Figure S4C).

   In line with the hypothesized pathway, a significant downregulation of
   Irg1 was observed under PPP inhibition (p < 0.01) with subsequent
   significantly decreased amount of itaconate (p < 0.0001)
   ([188]Figure 5E). Itaconate is an inhibitor of succinate dehydrogenase
   (Sdh). Accordingly, there was a significant decrease of the Sdh educt
   succinate (p < 0.001) and a significant increase of the Sdh product
   malate (p < 0.0001) ([189]Figure 5E). This indicates less suppression
   of Sdh due to decreased itaconate production by which mitochondrial
   oxidative activity is increased as observed in the metabolic flux
   analysis ([190]Figures 3F and 3G). Besides the use of citrate for
   itaconate production, citrate is also an educt of the ATP citrate lyase
   (Acly) for acetyl-coenzyme A production. It has been shown that Acly
   activity is an inducer of macrophage activation and supports
   pro-inflammatory cytokine production—for example interleukin (IL)-6—in
   macrophages.[191]^32 In proteomic analysis a significant increase of
   Acly expression under PPP inhibition was observed (p < 0.01,
   [192]Figure 5E).

   Itaconate is widely known as a regulatory immunosuppressive metabolite
   in macrophages, which promotes anti-inflammatory IL-10 secretion and
   inhibits pro-inflammatory IL-6 secretion.[193]^27 Moreover, we observed
   significantly increased nuclear factor κB1 expression under PPP
   inhibition and knockdown ([194]Table S2), an activator of IL-6 and
   inhibitor of IL-10 production. In line with these findings, we observed
   significantly increased IL-6 secretion (p < 0.05, [195]Figure 5F) and
   significantly decreased IL-10 secretion (p < 0.0001, [196]Figure 5G) by
   PPP inhibition.

   To further prove the functional role of the Irg1-itaconate pathway, we
   evaluated the phagocytic activity of macrophages of Irg1^−/− knockout
   mice. A significantly increased phagocytic activity of bone
   marrow-derived macrophages in ADCP assay ex vivo was observed in
   comparison to Irg1^+/+ wild-type mice (+34%, p < 0.05, [197]Figure 5H).

   In conclusion, we connected metabolic activity and immune regulation in
   macrophages via the UDPG-Stat1-Irg1-itaconate signaling axis provoked
   by PPP activity. Irg1 downregulation increases macrophage activation
   via itaconate reduction with subsequent metabolic activation and a
   pro-inflammatory shift in cytokine secretion. This also leads in vivo
   to an increased phagocytic capacity of macrophages.

PPP inhibition in primary human cells increases phagocytic capacity of
macrophages and decreases their bystander function

   To translate our findings into human context, we isolated primary human
   monocytes from healthy donors and differentiated them into macrophages
   by macrophage colony stimulating factor (M-CSF) under PPP inhibition.
   After testing cytotoxicity of the PPP inhibitors to primary human
   macrophages, ADCP assays with non-toxic concentrations of inhibitors
   were performed. Inhibition of both parts of the PPP significantly
   increased ADCP rates (physcion +64%; oxythiamine +92%, p < 0.0001,
   [198]Figures 6A and 6B). The human macrophages showed a similar switch
   toward pro-inflammatory cytokine secretion with significantly increased
   IL-6 (p < 0.05) and significantly decreased IL-10 secretion
   (p < 0.0001) ([199]Figures 6C and 6D).

Figure 6.

   [200]Figure 6
   [201]Open in a new tab

   PPP inhibition in primary human cells increases phagocytic capacity of
   macrophages and decreases their tumor-supportive bystander function

   (A and B) ADCP change compared to basal phagocytosis rate of human
   monocyte-derived macrophages. (A) ADCP change of monocyte-derived
   macrophages differentiated in the presence of physcion and M-CSF, (B)
   ADCP change of monocyte-derived macrophages differentiated in the
   presence of oxythiamine and M-CSF.

   (C and D) Cytokine expression of monocyte-derived macrophages
   differentiated in the presence of oxythiamine and M-CSF. (C) IL-6
   expression, (D) IL-10 expression.

   (E and F) ADCP change of J774A.1 macrophages phagocyting primary CLL
   patient cells compared to basal phagocytosis rate. (E) ADCP change
   under compound-mediated PPP inhibition. (F) ADCP change under
   shRNA-mediated PPP knockdown.

   (G) ADCP change of monocyte-derived macrophages differentiated in the
   presence of oxythiamine and M-CSF phagocyting primary CLL patient cells
   compared to basal phagocytosis rate.

   (H) Viability of primary CLL patient cells after incubation with PPP
   inhibitors physcion or oxythiamine in mono-culture and in co-culture
   with J774A.1 macrophages. In co-culture setting, cells were treated in
   parallel or macrophages were pre-treated before onset of co-culture.

   (I–L) Half maximal inhibitory concentration (IC[50]) for individual
   primary CLL patient cell samples to bendamustine treatment compared to
   control. Cells were incubated with bendamustine after protective
   macrophage co-culture with untreated J774A.1 macrophages vs. PPP
   inhibition. (I and J) Inhibition of 6Pgd in oxidative part of PPP by
   physcion, (I) co-culture treatment, (J) macrophage pre-treatment. (K
   and L) Inhibition of Tkt in non-oxidative part of PPP by oxythiamine,
   (K) co-culture treatment, (L) macrophage pre-treatment. In (A–H) data
   are shown as mean ± SEM, in (I–L) as minimum to maximum. Technical
   replicates (A) n = 28, (B) n = 20, (C and D) n = 18, (E and F) n = 20,
   (G) n = 65, (H) n = 30, (I–L) n = 30; biological replicates (A) n = 6,
   (B) n = 4, (C and D) n = 6, (E and F) n = 5, (G) n = 12, (H) n = 10,
   (I–L) n = 10. p values were calculated in (A–G) using one-way ANOVA, in
   (H) using repeated measures (RM) one-way ANOVA, and in (I–L) using
   paired t test. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. See
   also [202]Figure S6.

   To address effector function of macrophages in the context of primary
   human leukemia cells, primary chronic lymphocytic leukemia (CLL) cells
   of five individual patients were used for ADCP assays. A significant
   increase of phagocytosis was observed under PPP inhibition (+22%,
   p < 0.001, [203]Figure 6E) and by using knockdown macrophages
   (Tkt +60%; 6Pgd +92%, p < 0.0001, [204]Figure 6F) ([205]Figures S6C and
   S6D).

   To evaluate phagocytosis in a fully human setting, we performed ADCP
   assay with primary human monocyte-derived macrophages differentiated in
   the presence of PPP inhibitors and primary CLL patient cells (12
   individual patients). A significantly increased phagocytic capacity was
   observed (+24%, p < 0.0001, [206]Figures 6G and [207]S6E). Thereby, we
   have demonstrated that primary indolent lymphoma and primary human
   macrophages are also affected by PPP modulation.

   Beyond the inefficient phagocytic function, TAMs exert
   direct supportive effects on tumor cells. CLL cells depend on
   macrophages as “nurse-like” bystander cells to survive.[208]^33
   Macrophages in the microenvironment of CLL are polarized toward
   tumor-promoting TAMs and support CLL cells by chemokine secretion and
   immunosuppressive signaling. We therefore evaluated the effect of
   PPP-inhibited macrophages on primary CLL cells. Interestingly, PPP
   inhibition in mono-cultured primary CLL cells decreased their viability
   significantly (p < 0.0001, [209]Figure 6H, left), as well as inhibition
   of the non-oxidative part of the PPP in co-culture (p < 0.01,
   [210]Figure 6H, right).

   As TAMs are also important mediators in chemotherapy resistance, we
   evaluated if the co-cultivation under PPP inhibition affects the
   susceptibility of primary CLL cells toward apoptosis by chemotherapy.
   We observed significantly increased bendamustine-induced apoptosis
   among primary CLL cells under inhibition of both parts of the PPP
   (p < 0.001, [211]Figures 6I–6L) (for individual patient data see
   [212]Figures S6F–S6J). This boost in apoptosis was achieved by PPP
   inhibition in co-culture (p < 0.001, [213]Figures 6I and 6K) and by
   macrophages pre-treatment before exposing them to primary CLL cells
   (p < 0.01, [214]Figures 6J and 6L).

   These observations underline the role of altered macrophage support
   under PPP inhibition in the TME such as direct leukemia cell support or
   resistance to chemotherapy.

PPP inhibition increases macrophages’ maturation and pro-inflammatory
polarization in vivo

   To evaluate if PPP inhibition preserves its effect on macrophages
   in vivo, we treated C57BL/6J mice with the PPP inhibitor S3
   (1-hydroxy-8-methoxy-anthraquinone). S3 is a more stable derivate of
   the 6Pgd inhibitor physcion.[215]^13

   Investigating the myelopoiesis under PPP inhibition, a significant
   increase of cells in the LSK compartment (Lin^−, Sca-1^+, c-Kit^+) was
   seen (p < 0.0001, [216]Figure 7A). PPP inhibition significantly
   increased the amount of hematopoietic stem cells (HSCs; p < 0.05,
   [217]Figure 7B) and multipotent progenitor (MPP) cells
   ([218]Figure 7B), including MPP pools known to fuel the myeloid
   compartment (MPP2 p < 0.001; MPP3 p < 0.01; MPP5 p < 0.01,
   [219]Figure 7B).[220]^34^,[221]^35 This coincided with a significantly
   increased frequency of myeloid progenitor cells (p < 0.01,
   [222]Figure 7C) and macrophages (p < 0.001, [223]Figure 7C) in the bone
   marrow, indicating propelled myelopoiesis.

Figure 7.

   [224]Figure 7
   [225]Open in a new tab

   PPP inhibition increases myelopoiesis, macrophages’ maturation, and
   pro-inflammatory polarization in vivo and boosts anti-leukemic
   treatment response in an aggressive humanized lymphoma mouse model

   (A) Progenitor cell compartment LSK (Lin^−, Sca-1^+, c-Kit^+) in bone
   marrow of C57BL/6J mice treated with vehicle (control) or PPP inhibitor
   S3 i.p. for 7 days.

   (B) Multipotent progenitor (MPP) subsets in bone marrow of C57BL/6J
   mice treated with vehicle (control) or PPP inhibitor S3
   intraperitoneally (i.p.) for 7 days. HSC (CD34^−, CD48^−, CD150^+,
   CD135^-), MPP1 (CD34^+, CD48^−, CD150^+, CD135^-), MPP2 (CD34^+,
   CD48^+, CD150^+, CD135^-), MPP3 (CD34^+, CD48^+, CD150^-, CD135^-),
   MPP4 (CD34^+, CD48^+, CD150^-, CD135^+), MPP5 (CD34^+, CD48^−, CD150^-,
   CD135^-).

   (C) Percentage of myeloid lineage cells of whole cell amount in bone
   marrow of NSG mice transfected with hMB cells, treated with vehicle or
   PPP inhibitor S3 i.p. for 12 days, and euthanized on day 15. Common
   myeloid progenitor cells (CD41^+, CD34^+), monocytes (Ly6c^+,
   CX3CR1^+), macrophages (F4/80^+, CD64^+).

   (D) Expression of characteristic surface marker for different
   macrophage subtypes on peritoneal macrophages measured by
   immunofluorescent staining. Mean fluorescence intensity (MFI) is
   depicted. To improve readability, high MFI has been downscaled (factor
   named in brackets next to marker). C57BL/6J mice treated with vehicle
   (control) or PPP inhibitor S3 i.p. for 7 days.

   (E) ADCP assay of bone marrow-derived macrophages. C57BL/6J mice
   treated with vehicle (control) or PPP inhibitor S3 i.p. for 7 days,
   macrophages differentiated out of femoral bone marrow with M-CSF.

   (F) Survival curve of aggressive lymphoma (hMB) bearing mice treated
   with PPP inhibitor S3 +/− therapeutic antibody alemtuzumab.

   (G) One representative example of immunohistochemical staining of hMB
   cells (CD19^+) and macrophages (CD68^+) in spleen of aggressive
   lymphoma (hMB) bearing mice treated with vehicle or alemtuzumab + S3.
   In (A–C and E) data are shown as mean ± SEM, in (D) data are shown as
   mean of ten replicates. Technical replicates (A and B) n = 12, (C) n =
   3, (D) n = 9–10, (E) n = 70–75, (F) n = 21–25, (G) n = 4; biological
   replicates (A and B) n = 12, (C) n = 3, (D) n = 9–10, (E) n = 14–15,
   (F) n = 21–25, (G) n = 4. p values were calculated in (A–E) using
   unpaired t test and in (F) using Benjamini-Hochberg test. ∗p < 0.05;
   ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. See also [226]Figure S7 and
   [227]Table S8.

   We further investigated the polarization of in vivo macrophages
   ([228]Table S8) and detected a shift away from M2-like subtype with
   significantly decreased Arg1 expression (p < 0.05) and toward
   pro-phagocytic M1-like subtype with significantly increased CD38
   expression (p < 0.05) in unstimulated peritoneal macrophages
   ([229]Figure 7D) (for other compartments see [230]Figure S7A). These
   findings are in line with our in vitro observations.

   Moreover, in vivo PPP inhibition significantly increased phagocytic
   activity of bone marrow-derived macrophages in ADCP assay ex vivo
   (+74%, p < 0.001, [231]Figure 7E).

   Altogether, we have shown that PPP inhibition in vivo activates
   macrophages’ inflammatory polarization and maturation as well as their
   phagocytic capacity, which increases their anti-tumor function in vivo.

PPP inhibition boosts anti-leukemic treatment and thereby prolongs survival
in an aggressive lymphoma mouse model

   To focus on the therapeutic effect of PPP inhibition in vivo, we
   evaluated treatment effects in an aggressive lymphoma mouse model.

   We used the humanized double-hit lymphoma mouse model (hMB),[232]^36
   which is amenable for modeling human-specific antibody therapy. We
   treated the mice with the therapeutic antibody alemtuzumab and the PPP
   inhibitor S3. As the lymphoma reflects aggressive disease, untreated
   mice died rapidly after tumor cell injection (median overall survival
   [mOS] 22 days, [233]Figure 7F). By treatment with S3 only, this rapid
   tumor progression persisted. As shown in our previous work, treatment
   with alemtuzumab increases survival significantly in this aggressive
   lymphoma mouse model[234]^5 (mOS 25 days, p = 1.4e−05, [235]Figures 7F
   and [236]S7B). By adding the PPP inhibitor S3 to alemtuzumab, an
   additional significant prolongation of mouse survival was achieved in
   comparison to antibody treatment only (mOS 27 days, p = 0.0059,
   [237]Figures 7F and [238]S7B) with a stable increased number at risk up
   to day 25 (survival of 88%, [239]Figure 7F). Immunohistochemical
   analysis of spleens showed a marked reduction of CD19^+ lymphoma cell
   infiltration with concomitant increase of CD68^+ macrophage
   infiltration after treatment with alemtuzumab and S3 in comparison to
   vehicle control ([240]Figures 7G and [241]S7C).

   We finally demonstrated in vivo that PPP inhibition in the context of a
   highly aggressive lymphoma model increases the efficacy of antibody
   therapy to prolong overall survival significantly.

Discussion

   TAMs are key drivers in various cancers associated with poor outcome
   and diminished efficacy of immunotherapies.[242]^1^,[243]^3^,[244]^5
   The influence of glucose and mitochondrial metabolism on macrophages’
   polarization and activity has been
   established.[245]^8^,[246]^9^,[247]^37 Activation of the PPP in
   macrophages has been implicated in immune tolerance and granuloma
   formation.[248]^38^,[249]^39 However, no functional implications of the
   PPP in TAM regulation have been established nor the effects of PPP
   inhibition on immune regulation have been identified. Here, we show
   that modulation of the PPP in TAMs serves as a robust regulator of
   phagocytic function and macrophage activity and prolongs survival in
   aggressive B cell lymphoma therapy.

   Metabolic inhibition screening in lymphoma-macrophage co-cultures
   emphasized detrimental effects on macrophage function for the majority
   of investigated pathways. Only inhibition of the PPP showed a
   significant increase of phagocytosis with a synergistic effect on
   effector and target cells. This was true across various in vitro and
   in vivo macrophage model systems of both murine and human descent by
   compound and genetic targeting. Even though the time-dependent effects
   of PPP inhibition differ between chemical compounds and genetic
   targeting, we observed similar phenotype under compound-mediated
   as well as shRNA-mediated PPP inhibition. The increased phagocytosis
   rate appeared also under hypoxic conditions as an approximation of
   physiological status of therapy-refractory niches of lymphoma—the lymph
   nodes and bone marrow[250]^40^,[251]^41—indicating therapeutic efficacy
   by metabolic inhibition in contrast to other therapy modalities in
   these niches.

   Previous reports demonstrated reduced cancer and leukemia cell growth
   in mice upon PPP inhibition.[252]^13^,[253]^42 Especially in CLL,
   macrophages play a pivotal role as supportive bystander cells in the
   TME, without which CLL cells would undergo spontaneous
   apoptosis.[254]^33 We have shown that PPP inhibition diminishes this
   pro-survival bystander function of macrophages and acts as a sensitizer
   to genotoxic regimens.

   Alterations of the PPP enzymes 6PGD and TKT have been previously
   described in many cancer
   types.[255]^43^,[256]^44^,[257]^45^,[258]^46^,[259]^47 Overexpression
   of TKT was closely associated with aggressive hepatocellular carcinoma
   features,[260]^48 and 6PGD was shown to promote metastasis,[261]^49
   while suppression of 6Pgd attenuates cell proliferation and tumor
   growth[262]^42 and overcomes cisplatin resistance[263]^50 and Tkt
   inhibition sensitizes cancer cells to targeted therapy and reduces
   growth of metastatic lesions.[264]^48 PPP inhibition by physcion, S3,
   or 6-aminonicotinamide has demonstrated anti-tumorigenic effects in
   several solid tumor types and chemotherapeutic-resistant acute myeloid
   leukemia cells,[265]^47^,[266]^51^,[267]^52 without affecting
   non-malignant cells.[268]^42 Moreover, 6Pgd inhibition in CD8^+ T cells
   led to an increased effector function with higher tumoricidal
   activity.[269]^53

   We have previously shown that macrophage effector polarization is
   crucial in therapeutic antibody-based regimens of B cell lymphoma and
   can be modulated.[270]^54^,[271]^55^,[272]^56 We now identified
   macrophage metabolism as an essential switch of macrophage effector
   function in lymphoma.

   We demonstrated that increase of phagocytosis is driven by PPP enzyme
   inhibition, rather than metabolite accumulation. Non-exclusive PPP
   metabolites did not influence phagocytosis, possibly due to degradation
   via glycolysis (G3P, F6P) or nucleotide synthesis (R5P) before entering
   PPP flux. In contrast, exclusive PPP metabolites altered phagocytosis
   activity: supplementation of the G6pd product 6-phosphogluconate, the
   6Pgd product ribulose-5-phosphate, and the Tkt product
   sedoheptulose-7-phosphate increased phagocytosis, while supplementation
   of the Tkt educt erythrose-4-phosphate decreased phagocytic activity in
   macrophages. This points to a feedback inhibition of the respective PPP
   enzymes and emphasizes the enzyme inhibition as driving force for
   increased phagocytosis.

   PPP is a central linker between glucose metabolism, amino acid
   biosynthesis, fatty acid metabolism, and redox homeostasis.[273]^43 A
   gain in metabolic activity was observed under PPP inhibition with
   increased glycolytic and mitochondrial capacity causing enhanced ATP
   production, fueling macrophages’ activation. An increase of glycolysis
   is well described within the phenotypical switch to pro-inflammatory
   macrophages.[274]^57 We observed profound alteration of morphology and
   macrophage polarity, demonstrated by a decrease of markers associated
   with M2-like macrophages and TAMs, which represent immunosuppressive
   and tumor-promoting macrophage subtypes,[275]^3^,[276]^58^,[277]^59
   while exclusive M1 marker, expressed on pro-inflammatory macrophages,
   was increased.[278]^60^,[279]^61^,[280]^62^,[281]^63

   In total, the restriction of one metabolic pathway—the PPP—gives rise
   to numerous paths of activation, which renders a profound alteration of
   phenotype and particular phagocytic activity in macrophages. Thereby
   the anti-tumor function could be improved from independent directions.

   Our detailed multi-omics and functional analysis provides evidence that
   these effects are directly related to 6Pgd and Tkt enzyme activity
   loss, which polarizes macrophages to a pro-inflammatory phenotype
   through downregulation of Stat1 and Irg1. The functional switch between
   PPP enzyme activity and subsequent polarization program is Csf1r
   expression and activity of glycogen metabolism.

   Csf1r activation induces Hmox-1 expression,[282]^64^,[283]^65 which
   induces Irg1 expression,[284]^66 a central inhibitory regulator of
   macrophage activation.[285]^27 We demonstrated significant
   downregulation of Csf1r pathway proteins. Via Csf1r signaling,
   macrophages are polarized toward an M2-like or TAM phenotype by
   directly activating Erk1/2 (Mapk1/2) and Hck
   signaling.[286]^24^,[287]^25^,[288]^67^,[289]^68^,[290]^69 Both, Mapk1
   and Hck activity was shown to be decreased under PPP inhibition in
   upstream kinase analysis.

   Considering the relevant role of Csf1r in macrophage ontogeny,
   activation, and polarization, reduced Csf1r expression might be
   responsible for macrophage activation under PPP inhibition. Therefore,
   Csf1r blockade might be a promising strategy to increase macrophage
   activity in the context of tumor therapy. Several CSF1R inhibitors are
   currently under clinical investigation.[291]^70 Nevertheless, as CSF1R
   is a macrophage-exclusive receptor, only a macrophage-exclusive effect
   could be achieved by using CSF1R inhibition, in contrast to the
   previously described multi-cellular effects of 6PGD and TKT inhibition.

   PPP inhibition has been functionally linked to inhibition of
   glycogenolysis causing decreased UDPG production.[292]^28 UDPG as a
   signaling molecule activates the P2y14 receptor (P2y14r). P2y14r
   activation in turn increases Stat1 expression and Mapk1
   phosphorylation.[293]^71 Considering that Stat1 is a major regulator of
   Irf1 expression,[294]^29 which is the transcription factor of
   Irg1,[295]^72 the signaling cascade induced by PPP inhibition in
   macrophages is causing an altered Irg1 expression. Irg1 is almost
   exclusively expressed in activated immune cells and a key driver of
   immune inhibition via itaconate production.[296]^27 Itaconate inhibits
   glycolysis and mitochondrial activity by Sdh inhibition[297]^73 and
   promotes anti-inflammatory macrophage phenotype[298]^74 and tumor
   growth by increased reactive oxygen species secretion by TAMs.[299]^75
   In contrast to the publication of Ma et al.,[300]^28 we observed a
   decreased amount of glycogen under PPP inhibition, but simultaneously
   increased glycolysis as a possible indicator for a shift of glucose
   processing causing decreased glycogen synthesis and glycogenolysis.
   Subsequently, we observed decreased expression of all
   UDPG-Stat1-Irg1-itaconate pathway proteins under PPP inhibition,
   particularly lower amounts of itaconate and an increased Sdh and Acly
   activity. Acly activity is known to induce macrophage activation and
   pro-inflammatory cytokine production.[301]^32 These changes
   subsequently resulted in a pro-inflammatory cytokine switch, phenotypic
   shift toward M1-like macrophages, and diminished primary leukemia cell
   support. The metabolic alterations observed here support the hypothesis
   of the connection between PPP inhibition, itaconate abundance, and
   immune regulation.

   Furthermore, the transcription factor Irf1 induces inducible nitric
   oxide synthase (iNos) expression,[302]^72 consistent with the observed
   decreased iNos expression under PPP inhibition. High iNos expression
   and activity have been correlated with malignancy and poor survival in
   several solid tumors and leukemia.[303]^76

   Despite the pro-inflammatory transcriptional function of Stat1 in
   LPS-stimulated macrophages, we show an alternative mechanism of
   macrophage activation by metabolic depression of the anti-inflammatory
   properties of Stat1 via itaconate regulation leading to a
   pro-phagocytic phenotype of macrophages. In TAMs, Stat1 has been shown
   to be the generator of the blended M1/M2 phenotype and supporter of the
   anti-inflammatory and pro-tumorigenic properties.[304]^77^,[305]^78

   As reduced expression of Irf1 also influences Stat1 and Csf1r
   expression,[306]^79 Irf1 appears as the junction between all
   recapitulated pathways, which are leading to the changed macrophage
   activity and polarization.

   Taken together, these signaling connections show a narrow network,
   which links PPP inhibition and immune regulation in macrophages and
   cancer cells.

   We demonstrated a highly significant increase of phagocytosis in
   primary human cells, indicating efficacy for potential clinical use.

   To model lymphoma in patients, we performed in vivo experiments with
   the PPP inhibitor S3, whose low toxicity and high effectiveness in
   treatment of other tumor entities were demonstrated
   before.[307]^13^,[308]^42 In the aggressive lymphoma mouse
   model,[309]^36 PPP inhibition has an amplification effect on antibody
   therapy leading to significant prolonged overall survival in comparison
   to antibody treatment only with associated increased macrophage
   lymphoma infiltration. PPP inhibition in vivo increased myelopoiesis
   and gave rise to progenitor cell expansion, indicating increased
   provision of a variety of immune cells. Macrophages displayed
   pro-inflammatory polarization and significantly increased phagocytic
   capacity after PPP inhibition in vivo.

   We have proven in vivo the efficacy of PPP inhibition leading to
   macrophage activation and improving therapy response by
   antibody-mediated phagocytic clearance of lymphoma with prolongation of
   overall survival.

   In conclusion, PPP inhibition may serve as immune-modulatory therapy
   repolarizing macrophages. We demonstrated metabolic modulation as a key
   mechanism of macrophage regulation. PPP inhibition causes diminished
   Irg1 expression leading to reduced anti-inflammatory itaconate
   production. Our work indicates PPP inhibition as a dual-principle
   therapy targeting cancer cells and their immune-microenvironment
   simultaneously, with implications for cancer treatment, especially in
   the context of antibody-based regimens.

Limitations of the study

   This study has several important limitations to consider. The screening
   process was conducted primarily from the perspective of ADCP and
   emphasized a therapeutic angle, which may have limited the scope of the
   findings. While the PPP was a focus, its role in modulating lymphoma
   progression requires further investigation to fully understand its
   impact on the TME. The study concentrated on macrophages as targets for
   PPP inhibition. The effects of this inhibition on other
   immunotherapeutic approaches, such as bispecific antibodies and CAR-T
   cells, remain to be demonstrated. Crucially, the role of PPP inhibition
   in human patients has yet to be established through clinical trials.
   These limitations highlight areas for future research to build upon
   this study’s results and provide a more comprehensive understanding of
   PPP inhibition in lymphoma treatment.

Resource availability

Lead contact

   Further requests for information should be directed to and will be
   fulfilled by the lead contact, Christian P. Pallasch
   (christian.pallasch@uk-koeln.de).

Materials availability

   This study did not generate new unique reagents.

Data and code availability

     * •
       Proteomic and phosphoproteomic data have been deposited at
       PRIDE:PXD042428. Metabolomic data have been deposited at UCSD
       Metabolomics Workbench: ST003516.
     * •
       The data are publicly available as of the date of publication.
       Accession numbers are listed in the [310]key resources table.
     * •
       This paper does not report original code.
     * •
       Any additional information required to reanalyze the data reported
       in this paper is available from the [311]lead contact upon request.

Acknowledgments