Glucocorticoids regulate lipid mediator networks by reciprocal modulation of 15-lipoxygenase isoforms affecting inflammation resolution

Significance

The specialized proresolving mediators (SPM) are potent immunoresolvents, formed from omega-3-fatty acids by lipoxygenases, which foster resolution of inflammation. Currently, no pharmacological treatments are approved to stimulate SPM biosynthesis for promoting inflammation resolution. Glucocorticoids are widely used anti-inflammatory drugs that dampen immune responses and excessive inflammation, but they also display inflammation-resolving features. However, by which mechanism glucocorticoids resolve inflammation and how they impact SPM biosynthesis remained elusive. We show that in human monocyte/macrophages, glucocorticoids strongly up-regulate 15-lipoxygenase-2/ALOX15B expression via the glucocorticoid receptor that binds to ALOX15B intron 3 but suppress 15-lipoxygenase-1/ALOX15 isoform induction with striking consequences for SPM biosynthesis. Conclusively, glucocorticoids mediate their inflammation-resolving effects via elevated SPM formation, and thus advance the understanding of how glucocorticoids manipulate immune responses.

Abstract

Glucocorticoids (GC) are potent anti-inflammatory agents, broadly used to treat acute and chronic inflammatory diseases, e.g., critically ill COVID-19 patients or patients with chronic inflammatory bowel diseases. GC not only limit inflammation but also promote its resolution although the underlying mechanisms are obscure. Here, we reveal reciprocal regulation of 15-lipoxygenase (LOX) isoform expression in human monocyte/macrophage lineages by GC with respective consequences for the biosynthesis of specialized proresolving mediators (SPM) and their 15-LOX-derived monohydroxylated precursors (mono-15-OH). Dexamethasone robustly up-regulated pre-mRNA, mRNA, and protein levels of ALOX15B/15-LOX-2 in blood monocyte–derived macrophage (MDM) phenotypes, causing elevated SPM and mono-15-OH production in inflammatory cell types. In sharp contrast, dexamethasone blocked ALOX15/15-LOX-1 expression and impaired SPM formation in proresolving M2-MDM. These dexamethasone actions were mimicked by prednisolone and hydrocortisone but not by progesterone, and they were counteracted by the GC receptor (GR) antagonist RU486. Chromatin immunoprecipitation (ChIP) assays revealed robust GR recruitment to a putative enhancer region within intron 3 of the ALOX15B gene but not to the transcription start site. Knockdown of 15-LOX-2 in M1-MDM abolished GC-induced SPM formation and mono-15-OH production. Finally, ALOX15B/15-LOX-2 upregulation was evident in human monocytes from patients with GC-treated COVID-19 or patients with IBD. Our findings may explain the proresolving GC actions and offer opportunities for optimizing GC pharmacotherapy and proresolving mediator production.

Glucocorticoids (GC) crucially impact various physiological processes, including metabolism, development, and inflammation (1). Due to their marked immunomodulatory features, they are frequently used to treat inflammatory and allergic diseases including rheumatoid arthritis (RA), lupus erythematosus, inflammatory bowel disease (IBD), and asthma (1, 2). The potent anti-inflammatory GC dexamethasone (Dex) was recently reported as lifesaving drug for critically ill COVID-19 patients in the RECOVERY trial (3). Interestingly, GC not only block inflammatory responses but also can promote inflammation resolution (2, 4), but the cellular and molecular basis of GC actions underlying resolution of inflammation are still elusive. Furthermore, the use of GC is associated with severe side effects and the occurrence of GC resistance (1), which demands for deciphering the cellular and molecular mechanisms of GC actions in order to develop more specific and safer anti-inflammatory treatment options (5).

The inflammatory response is initiated and maintained by the proinflammatory lipid mediators (LM) prostaglandins (PG) and leukotrienes (LT), produced via enzymatic oxygenation of arachidonic acid (AA) by COX and 5-lipoxygenase (LOX), respectively (6) (Fig. 1A). In contrast, the specialized proresolving mediators (SPM) are distinct LM, also formed from AA but mainly from the ω-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which promote resolution of inflammation, tissue regeneration and the return to homeostasis (7, 8). SPM are grouped into lipoxins (LXs), resolvins (RVs), protectins (PDs), and maresins (MaRs) that are potent immunoresolvents and provide a basis for new therapeutic approaches for treating inflammatory diseases and for stimulating tissue regeneration (8). For production of AA- and DHA-derived SPM (Fig. 1A), two sequential oxygenation steps are necessary, involving either 12-/15-LOXs alone to form dihydroxylated PDs and/or MaRs, or together with 5-LOX to form dihydroxylated/trihydroxylated RVs and LXs (9). Notably, in humans, the 15-LOX-1 is a dual C12/15- (for AA and EPA) or C14/17-lipoxygenating (for DHA) enzyme with a ratio of approx. 1:9 to 1:1, whereas the 15-LOX-2 isoform solely oxygenates C15 in AA and EPA and C17 in DHA (10, 11).

Fig. 1.

Dexamethasone (Dex) causes reciprocal modulation of 15-LOX isoform expression and dictates LM signature profiles. (A) Schematic illustration of the LM-biosynthetic pathways with relevance for this study. (B) Human M0_GM-CSF and M0_M-CSF (2 × 10⁶ cells/mL, each) were preincubated with 100 nM Dex or vehicle (0.1% DMSO) for 15 min and then polarized for 48 h to M1/M1_Dex or M2/M2_Dex, respectively. Protein expression of 15-LOX-1 and 15-LOX-2 normalized to β-actin was determined by Western blot; n = 3 separate donors, ratio paired t-test vs. vehicle. (C) MDM (2 × 10⁶ cells/mL; differentiated with 10 ng/mL GM-SCF and 10 ng/mL M-CSF for 6 d) or (D) monocytes (2 × 10⁶ cells/mL) were incubated with Dex (100 nM) or vehicle (0.1% DMSO) for 48 h. Protein expression of 15-LOX-2 normalized to β-actin was determined by Western blot; n = 3 separate donors, ratio paired t-test vs. vehicle. (E) M0GM-CSF and M0M-CSF (2 × 106 cells/ml, each) were preincubated with 100 nM Dex or vehicle (0.1% DMSO) for 15 min and then polarized for 48 h to M1/M1Dex or M2/M2Dex, respectively. Protein expression of COX-2, COX-1, 5-LOX, and FLAP normalized to β-actin was determined by Western blot; n = 3 separate donors, ratio paired t-test vs. vehicle. (F) Secreted cytokines were quantified by ELISA; n = 3-5 separate donors, ratio paired t-test vs. vehicle. (G) Expression of CD54 and CD80 (M1) as well as CD163 and CD206 (M2) among viable CD14+ cells was analyzed, the mean fluorescence intensity (MFI) is shown as mean + S.E.M, n = 3-4, ratio paired t-test vs. vehicle control. (H) M0GM-CSF and M0M-CSF (2 × 106 cells/ml, each) were preincubated with 100 nM Dex or vehicle (0.1% DMSO) for 15 min and then polarized for 48 h to M1/M1Dex or M2/M2Dex, respectively. MDM (2 × 106 cells/ml; differentiated with 10 ng/ml GM-CSF and 10 ng/ml M-CSF for six days) or monocytes (2 × 106 cells/ml) were incubated with Dex (100 nM) or vehicle (0.1% DMSO) for 48 h. Cells were incubated with E. coli (O6:K2:H1; ratio 1:50) for another 180 min. Formed LM were analyzed and normalized to total protein amounts given as pg/0.15 mg protein. LOD for LM analysis was 3 pg/0.15 mg protein. Data in the radar blots are means of n = 3-4 separate donors for different LM. Statistics: ratio paired t-test Dex-group vs. vehicle group. (B-H) Level of significances: *P < 0.05, **P < 0.01, and ***P < 0.001.

Human macrophages and monocytes can express both, 5-LOX and 15-LOX-1/2 isoforms to various extents that depends on distinct environmental signals (12, 13) with specific LM signature profiles in relation to their different phenotypes (14, 15). Macrophages adapt distinct inflammatory (M1-like) and proresolving (M2-like) phenotypes with specific tasks in the immune response and tissue homeostasis, where M1 produce proinflammatory LTB₄ while M2 macrophages generate various SPM (14–16). GC act effectively on innate immune cells, suppressing inflammatory reactions along with induction of anti-inflammatory monocyte and macrophage populations by down-regulating proinflammatory cytokines and PGs (4, 17, 18). But how GC impact SPM and their key biosynthetic enzymes (i.e., 15-LOXs) in monocyte/macrophage populations is presently unknown. Therefore, we studied how GC affect LM biosynthetic pathways and connected LM signature profiles in human monocytes and various macrophage subsets with focus on inflammation-resolving SPM. GC cause consistent and robust induction of 15-LOX-2 in all investigated human monocyte/monocyte-derived macrophage (MDM) subsets, that is, peripheral blood monocytes as well as unpolarized MDM and M1- and M2-MDM but block 15-LOX-1 expression during M2 polarization. These alterations result in tremendous changes in the biosynthetic spectrum and capacity of the cells to produce LM, especially SPM and their mono-15-OH precursors.

Results

Dexamethasone Causes Reciprocal Modulation of 15-LOX Isoform Expression and Dictates LM Signature Profiles.

We first studied how Dex affects phenotype-specific LM pathways (Fig. 1A) in human monocytes and MDM subsets by assessing related key enzymes on the protein level. Monocytes and unpolarized MDM (M0) as well as MDM during polarization into M1 or M2 were exposed to 100 nM Dex for 48 h. M0_GM-CSF were polarized to classically (LPS/IFNγ-)activated M1 and M0_M-CSF to alternatively (IL-4-)activated M2, or kept as unpolarized M0. Since GC affect macrophage polarization yielding specific phenotypes (19), we designated Dex-treated cells as Mono_Dex, M0_Dex, M1_Dex or M2_Dex. Besides monocytic cells, also neutrophils express LOXs and produce abundant LM (10, 20), but they are short-lived apoptotic cells after isolation and thus not suitable for studying protein expression modulation in vitro.

15-LOX-2 protein levels strikingly increased during polarization to M1_Dex or M2_Dex versus cells polarized without Dex (Fig. 1B). In contrast, Dex completely prevented 15-LOX-1 protein induction during polarization to M2_Dex, while the protein levels of 5-LOX, 5-LOX-activating protein (FLAP), and COX-1 were hardly altered (Fig. 1 B and E). Dex blocked induction of COX-2 protein in M1_Dex (Fig. 1E) and reduced the levels of cytosolic phospholipase A₂α (cPLA₂α) in M1_Dex and M2_Dex (Fig. 2A). All these modulatory Dex effects were concentration-dependent starting at 10 nM with maximal effects at 100 nM (SI Appendix, Fig. S1A).

Fig. 2.

Temporal modulation of LM pathways by dexamethasone (Dex) during M1- and M2-polarization. (A) Human M0_GM-CSF and M0_M-CSF (2 × 10⁶ cells/mL, each) were preincubated with Dex (100 nM) or vehicle (0.1% DMSO) for 15 min and then polarized for the indicated times to M1/M1_Dex or M2/M2_Dex, respectively. Protein expression of 15-LOX-1, 15-LOX-2 and cPLA₂α, normalized to β-actin, was determined by western blot; n = 3-4 separate donors, two-way ANOVA–Bonferroni multiple comparison test. (B) Human M0_M-CSF (2 × 10⁶ cells/mL) were polarized to M2 for 48 h and then treated with Dex (100 nM) or vehicle (0.1% DMSO) for another 48 h. 15-LOX-1 and 15-LOX-2 protein expression, normalized to β-actin, was determined by western blot; n = 3 separate donors, ratio paired t-test vs. vehicle. (C–E) M0GM-CSF and M0M-CSF (2 × 106 cells/ml, each) were preincubated with Dex (100 nM) or vehicle (0.1% DMSO) for 15 min and then polarized for the indicated times to M1/M1Dex or M2/M2Dex, respectively. (C and D) Total RNA, including mRNA (C) and pre-mRNA (D), was isolated, transcribed into cDNA and quantified by qPCR; fold increase vs. control (vehicle, t=0 h) is given; n = 3 separate donors, two-way ANOVA-Bonferroni multiple comparison test. (E) Cells were incubated with E. coli (O6:K2:H1; ratio 1:50) for 180 min and formed LM were analyzed. LM were normalized to total protein amounts; n = 4 separate donors, two-way ANOVA-Bonferroni multiple comparison test. (A–E) Level of significances: *P < 0.05, **P < 0.01, and ***P < 0.001.

Like in M1, the 15-LOX-1 and -2 are hardly expressed in monocytes and M0, but Dex treatment markedly elevated 15-LOX-2 protein in M0_Dex and Mono_Dex, too (Fig. 1 C and D). We also studied how Dex affects cytokine formation and macrophage phenotype markers during polarization. As expected, Dex significantly decreased TNF-α and IL-6 release in M1_Dex and M2_Dex (Fig. 1F). Dex slightly impaired the M1 markers CD54 and CD80 but markedly increased the M2 markers CD163 and CD206 in M1_Dex and in M2_Dex (Fig. 1G). Together, our data reveal reciprocal modulation of 15-LOX isoform expression in monocytes and MDM subsets by Dex, along with the well-known impact of Dex on COX-2/cPLA₂ suppression, cytokine release, and macrophage phenotype switch.

We next investigated the impact of Dex on the LM signature profiles using targeted LM metabololipidomics employing UPLC–MS/MS (15, 21). Monocytes as well as MDM subtypes cultured for 48 h with or without 100 nM Dex were challenged with pathogenic E. coli (O6:K2:H1, ratio 1:50) for 3 h to elicit LM biosynthesis (15, 21). In agreement with strong 15-LOX-2 expression, all 15-LOX-derived monohydroxylated products (“mono-15-OH” = 15-HETE, 15-HEPE, 17-HDHA) were massively up-regulated in Mono_Dex, M0_Dex and M1_Dex and to a lesser extent in M2_Dex (Fig. 1H and SI Appendix, Table S1). Like for di-hydroxylated and tri-hydroxylated SPM, also for mono-15-OH proresolving activities were reported, e.g., 15-HETE increased efferocytotic capacity (22) and 17-HDHA promoted phagocytosis and impaired the development of a proinflammatory macrophage phenotype (23). Note that in M1_Dex also the formation of the 17-HDHA-derived RvD5 and PDs and in M0_Dex RvD5 was considerably elevated. Interestingly “mono-12-OH”, i.e., 12-HETE, 12-HEPE, and 14-HDHA as well as 14-HDHA-derived MaR1 formed by 15-LOX-1 with dual 12- and 15-lipoxygenating specificity (11), were hardly increased. SPM levels in monocyte incubations were below the LOD (Fig. 1H and SI Appendix, Table S1). In line with the impaired cPLA₂ and COX-2 expression, formation of proinflammatory PGs and LTs as well as 5-LOX-derived monohydroxylated products (“mono-5-OH”, i.e., 5-HETE, 5-HEPE and 7-HDHA) was strongly decreased in M1_Dex and Mono_Dex and to a minor extent also in M2_Dex and M0_Dex (Fig. 1H and SI Appendix, Table S1). Suppression of PGs in M1 as well as RVs, MaR1 and mono-12-OH in M2 occurred already at 1 nM Dex being most pronounced at 10 nM (SI Appendix, Fig. S1B). Notably, the release of AA, EPA, and DHA was hardly altered by Dex in either cell type (SI Appendix, Table S1), regardless of the concentration (1 to 1,000 nM, SI Appendix, Fig. S1B). Short-term treatment of polarized M1- and M2 with 100 nM Dex for 3 h did not affect E. coli-induced LM profiles (SI Appendix, Table S2), excluding direct effects of Dex on the LM-biosynthetic process itself.

In M2 that usually generate abundant SPM due to strong 15-LOX-1 expression (15, 21), formation of RvD5, MaR1, and mono-12-OH was suppressed, while mono-15-OH were elevated and PDs were unaltered (Fig. 1H and SI Appendix, Table S1). Thus, reduced 15-LOX-1 expression due to Dex impairs the SPM-biosynthetic capacity, while Dex-induced 15-LOX-2 upregulation may compensate PD formation, implying an overall superordinated role of 15-LOX-1 for SPM formation in M2.

Temporal Modulation of LM Pathways by Dex during M1- and M2-polarization.

Next, we studied the temporal expression of LM-biosynthetic enzymes and corresponding LM signature profiles in more detail. Dex strongly increased 15-LOX-2 protein levels during M1- and M2-polarization at 24 h, peaking at 48 (M2_Dex) and 72 (M1_Dex) h (Fig. 2A). Expression of 15-LOX-1 in M2_Dex was suppressed over the entire polarization process (Fig. 2A). However, the high 15-LOX-1 protein levels in polarized M2 were unaffected after subsequent Dex treatment for 48 h (Fig. 2B), suggesting that Dex does not cause 15-LOX-1 protein degradation. In parallel, 15-LOX-2 was up-regulated by Dex in these polarized M2, corroborating that Dex-induced 15-LOX-2 upregulation is also accomplished in already polarized MDMs (Fig. 2B). We also detected impaired expression of COX-2 protein and of cPLA₂α at 48 h in M1_Dex, while in M2_Dex the cPLA₂α protein levels were consistently suppressed (Fig. 2A). In parallel to elevation of 15-LOX-2 protein, expression of ALOX15B mRNA was rapidly induced by Dex in M2 starting at 3 h with a peak (50-fold increase) at 6 h, remaining elevated up to 72 h. In M1_Dex, ALOX15B mRNA was also rapidly enhanced (3 to 6 h) but continuously increased up to 72 h up to >1,500-fold (Fig. 2C). As expected from the 15-LOX-1 protein analysis, ALOX15 mRNA in M2_Dex was not up-regulated at any of the time points up to 72 h (Fig. 2C). Analysis of pre-mRNA of ALOX15 in M2/M2_Dex and ALOX15B in M1/M1_Dex and M2/M2_Dex as surrogates for newly transcribed mRNA (24) revealed comparable modulation by Dex treatment (Fig. 2D) as for mRNA, supporting that Dex regulates the expression of these ALOX genes at the transcriptional level.

LM formation in E. coli-stimulated M1_Dex and M2_Dex essentially concurred with the respective temporal enzyme expression pattern. In M1, Dex increased levels of mono-15-OH and RvD5 as early as 24 h, with marked effects at 48 and 72 h (Fig. 2E and SI Appendix, Table S3). In M2_Dex, the formation of mono-12-OH, RvD5, PDs, and MaR1 was steadily blocked, while mono-15-OH were increased at 24 h and maintained at 48 and 72 h (Fig. 2E). These temporal patterns of 12/15-LOX product formation concur with the reciprocal 15-LOX isoform expression with rapid increases of 15-LOX-2 in M1_Dex and M2_Dex, but suppression of 15-LOX-1 in M2_Dex. The 5-LOX product levels were hardly affected by Dex but as expected, PG formation was suppressed in M1_Dex and M2_Dex.

Modulation of LM Pathways by Dex is Mediated by the Glucocorticoid Receptor.

In analogy to Dex, the potent GC prednisolone (100 nM) strongly increased 15-LOX-2 protein levels during 48 h MDM polarization to M1_Pred and M2_Pred and impaired the levels of COX-2 and of 15-LOX-1 (SI Appendix, Fig. S2A). Accordingly, in M1_Pred, formation of 15-LOX-2-derived mono-15-OH, RvD5, and PDs increased along with impaired PG levels while in M2_Pred the levels of mono-12-OH, PGs, and SPM were lowered with moderate elevation of mono-15-OH (SI Appendix, Fig. S2B and Table S4). Similarly, the endogenous GC hydrocortisone (0.1 or 1 µM) mimicked the effects of Dex on 15-LOX-1 and -2 expression in M2 (SI Appendix, Fig. S2C). To corroborate that Dex mediates LM pathway modulation via the GR, we employed the GR antagonist RU486 (25). Cotreatment with 1 µM RU486 prevented Dex-induced modulation of 15-LOX-1 and 15-LOX-2 expression on the mRNA (Fig. 3A) and on the protein (Fig. 3B) level of M1_Dex and M2_Dex, and reversed the altered LM profiles of these cells upon stimulation with E. coli (Fig. 3C and SI Appendix, Table S5). Since RU486 acts also as progesterone receptor (PR) antagonist, we assessed whether 15-LOX isoform expression is affected by progesterone. In contrast to Dex, progesterone (0.1 or 1 µM) failed to elevate 15-LOX-2 protein levels and to suppress those of 15-LOX-1 in M2_Prog (SI Appendix, Fig. S2C), neglecting a role of the PR. Together, these data suggest that the reciprocal modulation of the 15-LOX-isoforms by Dex is mediated via the GR.

Fig. 3.

Modulation of LM pathways by dexamethasone (Dex) is mediated by the GC receptor. Human M0_GM-CSF and M0_M-CSF (2 × 10⁶ cells/mL, each) were preincubated with (A–C) RU486 (1 µM) for 15 min before treatment with 100 nM Dex or vehicle (0.1% DMSO) for another 15 min, and then cells were polarized for 48 h to M1 or M2, respectively. (A) Total RNA was isolated, transcribed into cDNA, and quantified via qPCR; data are given as the -fold increase vs. vehicle; n = 3 separate donors, RM one-way ANOVA-Bonferroni multiple comparison test (vehicle vs. Dex, vehicle vs. RU486, Dex vs. RU486 + Dex). (B) Protein expression of 15-LOX isoforms, normalized to β-actin, was determined by western blot; n = 3 to 4 separate donors, RM one-way ANOVA–Bonferroni multiple comparison test (vehicle vs. Dex, vehicle vs. RU486, Dex vs. RU486 + Dex). (C) MDM were incubated with E. coli (O6:K2:H1; ratio 1:50) for another 180 min. Formed LM were extracted and analyzed. LM were normalized to total protein amounts. Data are mean ± SEM from n = 4 separate donors, RM one-way ANOVA-Bonferroni multiple comparison test (vehicle vs. Dex, vehicle vs. RU486, Dex vs. RU486 + Dex). SPM include PDX, PD1, MaR1 and RvD5. (D, Upper) GR occupancy on ALOX15B and ALOX15 genes from representative ChIP-seq data from Wang et al. (26) in vehicle (veh) and GC (triamcinolone; TA)-treated human macrophages. Lower: ChIP-qPCR data showing occupancy of GR at ALOX15B TSS and intron 3, and at ALOX15 TSS (amplicons indicated in orange, blue and purple, respectively) in MDM treated with IL-4 (M2) or Dex (100 nM)/IL-4 (M2_Dex) for 6 h. Data are presented as percentage of input DNA. IgG IP served as negative control. Bars represent mean ± SEM (n = 3 donors), independent ratio paired t test Dex vs. vehicle for IgG and GR IP. Level of significances: *P < 0.05, **P < 0.01 and ***P < 0.001.

Next, we studied whether modulation of 15-LOX-1/2 expression results from direct interaction of the GR with ALOX15 and ALOX15B genes using chromatin immunoprecipitation (ChIP) assays. In accord with published ChIP-seq data comparing GR occupancy in vehicle- and GC triamcinolone (TA)–treated macrophages (26), ChIP-qPCR in M2_Dex showed significant GR enrichment at intron 3 of the ALOX15B gene but not at ALOX15B transcription start site (TSS) or at ALOX15 TSS (Fig. 3D). Inspection of the ALOX15B intron 3 spanning genomic region indicates the GR-binding site to overlap with a putative enhancer marked by histone 3 lysine 27 acetylation (H3K27ac) and histone H3 lysine 4 mono-methylation (H3K4me1) obtained from ENCODE data (SI Appendix, Fig. S3A) (27). To support direct effects of Dex/GR on ALOX15B mRNA expression without the requirement of another de novo induced protein, we employed the protein biosynthesis inhibitor cycloheximide (CHX). While 1 µM CHX repressed 15-LOX-2 protein expression during 48 h polarization to M2_Dex as expected (SI Appendix, Fig. S3B), ALOX15B mRNA levels were not reduced (SI Appendix, Fig. S3C). Mechanistically, the GR binds to the ALOX15B gene and activates its transcription possibly via the putative intronic enhancer.

Impact of Genetic Interference with ALOX15B mRNA on Dex-modulated LM Formation.

To confirm that the increased formation of RvD5, PDs, and mono-15-OH due to Dex not only correlates but indeed is caused by elevated 15-LOX-2 levels, we silenced ALOX15B during MDM polarization in the presence of Dex using siRNA. Because Dex abolished 15-LOX-1 expression, ALOX15 knockdown experiments appear not expedient and were not performed. 15-LOX-2 protein expression in M1 and M2 was strongly reduced by ALOX15B silencing regardless of Dex treatment (Fig. 4A), while 5-LOX, COX-2, and 15-LOX-1 were not affected (Fig. 4A and SI Appendix, Fig. S4), and nontarget siRNA still enabled marked 15-LOX-2 protein expression upon Dex treatment (Fig. 4A).

Fig. 4.

Impact of genetic interference with ALOX15B mRNA on Dex-modulated LM formation. (A and B) Human M0_GM-CSF and M0_M-CSF were subjected to electroporation with sequence-specific siRNA against ALOX15B or nontarget siRNA (200 nM, each) and then preincubated with Dex (100 nM) for 15 min before polarization for 48 h to M1/M1_Dex or M2/M2_Dex, respectively. (A) Protein expression of 15-LOX-1, 15-LOX-2, and COX-2, normalized to β-actin, was determined by western blot; n = 3 separate donors, RM one-way ANOVA-Bonferroni multiple comparison test (nontarget vs. ALOX15B, nontarget vs. nontarget + Dex, ALOX15B vs. ALOX15B + Dex, nontarget + Dex vs. ALOX15B + Dex). (B) Cells were incubated with E. coli (O6:K2:H1; ratio 1:50) for another 180 min. Formed LM were analyzed and normalized to total protein amount. In the heatmap, the amounts of each LM in nontarget control group are shown in pg/0.15 mg protein given as means, and additionally the -fold change of each LM after different treatments vs. treatment with nontarget siRNA is given. For values with LOD ≤ 3 pg/0.15 mg protein, the -fold change was calculated on the basis of 3 pg/0.15 mg protein.

ALOX15B silencing in M1_Dex reversed the up-regulatory effect of Dex on PD1, RvD5, and mono-15-OH levels in E. coli-stimulated M1_Dex (Fig. 4B and SI Appendix, Table S6). In M2_Dex, ALOX15B silencing only reversed upregulation of mono-15-OH without alteration of SPM and mono-12-OH (Fig. 4B and SI Appendix, Table S6). In M1 without Dex subjected to ALOX15B silencing, the levels of RvD5 and mono-15-OH were lower compared to M1 controls, but in M2 the formation of mono-15-OH and SPM was not altered (Fig. 4B and SI Appendix, Table S5). Therefore, without Dex, mono-15-OH, and some SPM like RvD5 derive from 15-LOX-2 in M1, while in M2 the 15-LOX-1 plays a superordinated role in this respect. ALOX15B deficiency may have functional consequences for MDMs. Although Dex modulated several MDM responses (see above, Fig. 1 F and G), ALOX15B silencing during 48 h polarization to M1/M1_Dex or M2/M2_Dex neither affected IL-6 and TNFα release nor expression of surface markers for M1 (CD54, CD80) or M2 (CD163, CD206) phenotypes (SI Appendix, Fig. S5 A and B). Moreover, ALOX15B silencing did not alter the phagocytotic capacity of M0-MDM regardless of Dex-treatment (SI Appendix, Fig. S5 C and D).

Upregulation of 15-LOX-2 in Monocytes during Clinical GC Therapy in Acute (COVID-19) and Chronic (IBD) Inflammation.

We next studied if GC therapy causes modulation of 15-LOX expression also in vivo in patients in the context of clinical therapy of COVID-19 and IBD, representing acute hyperinflammatory and chronic inflammatory diseases, respectively. GC therapy with Dex (6 mg/day, i.v.) of severely diseased COVID-19 patients requiring additional oxygen support (SI Appendix, Fig. S6A and Table S7) impaired hyperinflammation, as indicated by a significant drop of C-reactive protein (CRP) plasma levels 4 d post Dex-treatment (Fig. 5A). Similarly, IL-6 serum levels were lowered by tendency, whereas TNF-α slightly increased. Differential blood counts showed significantly elevated total numbers of leukocytes, neutrophils, and lymphocyte subtypes (from defined lymphopenia to normal range), while monocyte numbers remained essentially unchanged after Dex. No significant changes in these parameters were found in the group of COVID-19 patients without GC therapy. We assessed mRNA and protein expression levels of 15-LOX isoforms and other LM-biosynthetic enzymes in peripheral blood monocytes and neutrophils from these patients. GC treatment increased ALOX15B mRNA in both, monocytes and neutrophils (Fig. 5B). The increase of 15-LOX-2 was confirmed on the protein level in monocytes but not in neutrophils (Fig. 5C). In contrast, ALOX15 mRNA levels in monocytes and neutrophils were only moderately elevated by Dex with overall very low absolute mRNA amounts (for some donors not detectable at all) and below detection limit for 15-LOX-1 protein. Similarly, COX-2 protein was undetectable in monocytes and neutrophils, and 5-LOX protein expression remained unchanged in either cell type. In the plasma of Dex-treated patients none of the analyzed LM (mono-15-OH, mono-12-OH, mono-5-OH, LT, and PG) were significantly changed versus untreated controls (Fig. 5D).

Fig. 5.

Upregulation of 15-LOX-2 in monocytes during clinical GC therapy in acute (COVID-19) and chronic (IBD) inflammation. (A–D) COVID-19 study. (A) Blood samples were tested for clinical parameters before and after a median of 4 d (IQR 3/6). Median absolute values and median -fold changes are shown, Wilcoxon matched-pairs signed rank test vs. control (day 0). (B and C) Monocytes (CD14+) and neutrophils (CD16+) were isolated by magnetic beads cell sorting. (B) mRNA expression of ALOX15 and ALOX15B, given as -fold change vs. control (day 0, n = 5 to 7). (C) Protein expression of 15-LOX-2 and 5-LOX, normalized to β-actin (n = 6 to 7 separate donors). Statistics: Friedman test with Dunn multiple comparison test vs. control (day 0). (D) Plasma (1 mL) was prepared followed by LM analysis (n = 6 to 7); Friedman test with Dunn multiple comparison test vs. control (day 0). (E–H) IBD study. (E) Blood was tested for clinical parameters. Median absolute values and median -fold changes are shown from n = 6 to 11 patients, Wilcoxon matched-pairs signed rank test day 4 vs. control (day 0). (F) Plasma was prepared from blood at day 0 and day 4 and analyzed for LM. LM are given in pg/2 mL of plasma; n = 11 separate donors, Wilcoxon matched-pairs signed rank test day 4 vs. control (day 0). (G and H) Monocytes were isolated from blood (18 mL) at day 0 and day 4, cultivated for 24 h at 37 °C, and (G) used for determination of protein expression of 15-LOX-2 (n = 11) and 5-LOX (n = 7), normalized to β-actin, using western blot, or (H) stimulated with E. coli-conditioned medium (0.5%) for 90 min, followed by LM analysis. LM are given as mean ± SEM in pg/mg total protein (n = 8); Wilcoxon matched-pairs signed rank test day 4 vs. control (day 0). Level of significances: *P < 0.05, **P < 0.01 and ***P < 0.001.

In patients with acute flares of IBD (SI Appendix, Table S8), prednisolone (1 mg/kg, p.o.) for 4 d (SI Appendix, Fig. S6B) significantly reduced CRP plasma levels, and the changes in the differential blood count resembled those after Dex-treatment in COVID-19 patients, while IL-6 and TNF-α remained unchanged (Fig. 5E). In monocytes isolated from these IBD patients, prednisolone therapy significantly increased 15-LOX-2 protein amounts about fourfold, whereas 5-LOX protein levels were rather impaired, and 15-LOX-1 was not detectable (Fig. 5G). Note that double bands for 15-LOX-2 protein were apparent, possibly due to posttranslational modifications such as phosphorylation or acetylation, or due to splicing variants of ALOX15B, found in macrophages, differing by 3.3 kDa (28); additional studies are necessary to confirm the identity of these bands and to determine the modifications involved. When these monocytes were stimulated with E. coli-derived exotoxins to induce LM formation ex vivo, production of mono-15-OH was significantly elevated due to prednisolone, while formation of mono-12-OH, mono-5-OH, and in particular PGs was impaired (Fig. 5H). We tested if prednisolone-induced upregulation of 15-LOX-2 might impact LM levels in plasma but we observed no increase in mono-15-OH under GC therapy, nor of any other LM (mono-12-OH, mono-5-OH and PGs) (Fig. 5F).

Discussion

GC potently dampen excessive inflammation (1), but also promote inflammation resolution, especially by repressing inflammatory macrophage activation and by induction of anti-inflammatory monocyte/macrophage populations, connected to induction of anti-inflammatory molecules like annexin A1, IL-10, and CD163 (2, 4). Monocytes and macrophages possess high capacities to produce diverse LM that regulate all stages of inflammation, especially the SPM that are potent immunoresolvents and key for promoting resolution (14, 29). Therefore, GC could mediate their proresolving actions in these cells, at least in part, by favoring SPM biosynthesis (30). We here uncover bioactions of GC in monocyte/macrophage subsets to impact the inflammatory process, that is, robust induction of 15-LOX-2 and blockade of 15-LOX-1 expression with tremendous consequences for the LM spectrum and the LM-forming capacity, especially for SPM. Thus, in inflammatory M1 and monocytes Dex robustly increased formation of SPM and mono-15-OH due to elevated 15-LOX-2 at the expense of proinflammatory PG and LT. We propose that GC promote inflammation resolution via 15-LOX-2/SPM induction, particularly in proinflammatory monocyte and (M1-like) macrophage populations (4). Of interest, annexin A1 was reported to stimulate SPM production (2). However, GC could also impede resolution by blocking 15-LOX-1 expression and related SPM formation in M2-like macrophages within resolving tissues.

We confirm typical anti-inflammatory GC actions in monocytes/macrophages, namely, the block of TNF-α and IL-6 expression (31), but also uncover proresolving events such as upregulation of the M2 markers CD163 and CD206 in Dex-treated MDM and promotion of SPM biosynthesis. Manipulation of LM biosynthesis by GC was shown before but essentially focused on suppression of proinflammatory eicosanoids, due to interference with inducible COX-2 and cPLA₂α (18, 32, 33). Our data confirm suppression of COX-2 induction in M1_Dex and downregulation of cPLA₂α in M1_Dex and more consistently in M2_Dex. GC can also affect expression of human 5-LOX [decreased in synovial biopsies (34), increased in monocytes (35)] and of its helper protein FLAP (increased in leukocytes) with consequences for LT formation (35) but regulation of 15-LOX isoforms in monocytes/macrophages by GC remained unexplored. Interestingly, two recent studies showed that GC treatment elevated SPM levels in mice and in men. Thus, in lungs of ovalbumin-sensitized and challenged mice, robust activation of the DHA/17-HDHA pathway by Dex was evident within 6 h, with >3-fold higher PD1 and PDX levels (36). In humans, SPM levels were elevated in plasma of COVID-19 patients after treatment with 6 mg/day Dex (37). However, the underlying mechanisms of the elevated SPM levels were not the focus of these contributions.

15-LOX-1 is the key enzyme for all SPM that require oxygenation at C12 and C15 in AA and EPA and at C14 and C17 in DHA. These include LXs, D-series RV, MaRs, PDs, and RvE4, while RvE1-E3 require C18-oxygenation by COX-2 or CYP450 enzymes (9). 15-LOX-1, but not 15-LOX-2, displays dual reaction specificity forming both 15-HETE and 12-HETE in a 9:1 ratio, 15-HEPE and 12-HEPE in an 8.5:1.5 ratio, and 17-HDHA and 14-HDHA in nearly equal amounts. In contrast, 15-LOX-2 exclusively forms 15-HETE, 15-HEPE and 17-HDHA (10, 11). Hence, cells expressing solely the 15-LOX-2 isoform may accomplish formation of PDs, D-RVs, and LXs but not MaRs, while cells expressing 15-LOX-1 can generate all these (38–40), which is reflected by our present results.

In humans 15-LOX-1 is expressed in immature red blood cells, eosinophils, and airway epithelial cells, and at lower levels also in neutrophils, alveolar macrophages, vascular cells, and atherosclerotic lesions (10). 15-LOX-2 is mainly present in epithelial cells of prostate and lung but also in macrophages, neutrophils, skin, and hair roots (41) and also in atherosclerotic plaques (13). M0 and M1 that hardly express 15-LOXs per se (12, 15, 42) acquired abundant 15-LOX-2 protein upon Dex treatment, and accordingly generated substantial amounts of mono-15-OH and related SPM, namely PDX, PD1, and RvD5. 15-LOX-2 expression was also up-regulated in human monocytes/macrophages by IL-4 and under hypoxia (13) as well as in M2 upon TLR-2 and -4 stimulation along with elevated 5-LOX levels and increase in IL-10 (42). ALOX15B silencing reversed Dex-induced elevation of mono-15-OH in M1_Dex and M2_Dex confirming that the Dex-mediated changes in LM formation are indeed caused by higher 15-LOX-2 expression, which excludes other mechanisms like enhanced autoxidation or changes in posttranslational modification of LM-biosynthesizing enzymes. Interestingly, ALOX15B silencing did not decrease SPM biosynthesis in M2, as compared to ALOX15 knockdown that strongly lowered SPM formation (43), confirming that SPM biosynthesis in M2 strongly depends on 15-LOX-1. Although ALOX15B silencing in unpolarized human MDM decreased cytokine secretion (44), silencing of ALOX15B during MDM polarization neither affected cytokine release, expression of M1 or M2 phenotype markers, nor phagocytosis in M0-MDM. Polarization tightly impacts macrophage responses and thus preponderate the influence of 15-LOX-2 as compared to unpolarized MDM (44); furthermore, the 15-LOX-2-derived LM may act not only in an autocrine but also in a paracrine fashion on other LM-susceptible cell types, which requires more detailed investigations. Together, Dex confers inflammatory M1_Dex a proresolving phenotype by shifting from COX-2 expression and PGE₂ and LTB₄ formation to 15-LOX-2 expression and elevated SPM/mono-15-OH production. This is in line with GC repressing inflammatory macrophage activation but induction of anti-inflammatory macrophage populations to foster resolution (4).

In accordance with recent studies (13, 15, 42), M2 express 15-LOX-1, but not monocytes or M1. Dex treatment completely abolished IL-4-induced 15-LOX-1 upregulation in M2_Dex. Notably, Dex failed to lower existing 15-LOX-1 protein levels in polarized M2, suggesting that Dex interferes with IL-4 signaling that induces this enzyme (45) rather than by supporting 15-LOX-1 protein degradation. We showed before that blocking vacuolar-ATPase abrogated 15-LOX-1 expression and SPM formation during IL-4-induced M2 polarization due to compromised ERK-1/2 signaling (46). Indeed, GC inhibited ERKs in monocytes (47) which may underly the blockade of IL-4-induced 15-LOX-1 in M2_Dex. Our results from ChIP studies exclude binding of the GR to the ALOX15 gene, in line with the lack of GR ChIP-seq peaks in GC-treated human macrophages (26). Finally, the complex changes of the LM profile in M2_Dex are due to the reciprocal 15-LOX isoform modulation, where mono-12-OH and the majority of SPM (i.e., PD1, MaR1 and RvD5) were impaired due to 15-LOX-1 suppression. However, mono-15-OH, now formed by Dex-induced 15-LOX-2, were still produced with ~threefold higher levels. This suggests a major role for 15-LOX-1 in SPM biosynthesis in M2.

We propose that Dex mediates the reciprocal regulation of the 15-LOX-1 and -2 via the GR, reflected by comparable modulation using prednisolone and hydrocortisone and reversal of the Dex effects by the GR antagonist RU486. The results from ChIP studies in Dex-treated M2 show that the GR is strongly recruited to intron 3, which may serve as a putative enhancer of the ALOX15B gene, and pre-mRNA analyses reveal upregulation of ALOX15B expression at the transcriptional level. Despite protein biosynthesis blockage by CHX, Dex elevated ALOX15B mRNA levels underscoring that Dex-stimulated GR directly activates ALOX15B gene transcription, precluding induction of another protein.

The RECOVERY trial revealed Dex as efficient treatment for critically ill COVID-19 patients (3), and Dex became standard of care for patients with severe hyperinflammatory conditions (48). Our studies with monocytes and neutrophils from Dex-treated COVID-19 patients corroborate the GC-induced increase of 15-LOX-2. Why neutrophils do not translate elevated mRNA into corresponding higher protein levels requires further investigation. Yet, increased 15-LOX-2 expression in monocytes from Dex-treated patients was not reflected by increased mono-15-OH or SPM plasma levels, possibly due to the fact that monocytes constitute only a small fraction of total peripheral blood cells, and to the generally minute mono-15-OH/SPM levels in plasma compared to other body fluids or tissues (9). Interestingly, 15-LOX-2 was also elevated in monocytes from prednisolone-treated IBD patients. Again, LM levels in plasma were not altered by GC, but mono-15-OH formation in monocytes ex vivo was elevated. SPM act as tissue mediators, where local concentrations are often orders of magnitude higher than in blood (49). Local elevation of SPM in GC-treated humans in vivo is reasonable where tissue macrophages reside and are impacted by GC; assessment of tissue from the patients in these two studies could not be performed for several reasons. Nevertheless, in a comparable study (37), Dex-treatment of COVID-19 patients increased plasma SPM, in line with our findings.

Conclusively, reciprocal modulation of 15-LOX isoforms by GC in different monocyte/macrophage subtypes has a striking impact on LM biosynthesis. Hypothetically, GC may promote their proresolving effects via elevated SPM formation (30, 36, 37), strongly supported by our findings. SPM promote inflammation resolution by limiting excessive neutrophil tissue infiltration and stimulating local macrophage–mediated clearance of apoptotic neutrophils, cellular debris, and microbes, as well as by counter-regulating proinflammatory eicosanoid/cytokine production (8). Elevations of SPM and mono-15-OH by GC via 15-LOX-2 induction in M1 may be beneficial in the pharmacotherapy at early stages of inflammation, where the M1 phenotype determines the acute (hyper-)inflammatory response. In contrast, GC may have detrimental impact by downregulation of 15-LOX-1 and thus SPM biosynthesis in M2, a phenotype that acts at later stages to resolve inflammation and to repair and regenerate tissues for restoring homeostasis (14, 50). Multiple clinical trials in tendinopathies indeed reveal effectiveness of GC injections in the short but not in the long term (51). Thus, benefits of GC injections until 6 wk in patients with tennis elbow are paradoxically reversed with higher recurrence rates thereafter (52), supporting that GC may impede inflammation resolution. Our findings underline the reported benefit of GC in acute and severe inflammatory reactions and support their use in early rather than in the later resolving stages of inflammation progression, given the temporal impact of GC on macrophage phenotypes and their capacity to biosynthesize LM such as SPM.

Methods

Monocyte Isolation, Differentiation, and Polarization of Macrophages.

For in vitro studies with monocytes and MDM, the monocytes were isolated from leukocyte concentrates obtained from freshly withdrawn peripheral blood of human volunteers (Institute of Transfusion Medicine, University Hospital Jena, Germany) (53). The experimental protocol was approved by the ethical committee of the University Hospital Jena, and all methods were performed in accordance with the relevant guidelines and regulations. Venous blood was collected in heparinized tubes (16 I.E. heparin/mL blood) from fasted adult (18 to 65 y) male and female registered, healthy volunteers, with written consent. These subjects had no apparent inflammatory conditions, infections, or current allergic reactions, and had not taken antibiotics or anti-inflammatory drugs for at least 10 d prior to blood collection. PBMC were separated using dextran sedimentation of erythrocytes, followed by centrifugation on lymphocyte separation medium (Histopaque®-1077, Sigma-Aldrich). The PBMC were seeded in RPMI 1640 (Sigma-Aldrich) containing 10% (v/v) heat-inactivated FCS, 100 U/mL penicillin, and 100 µg/mL streptomycin (Biochrom/Merck, Berlin, Germany) in cell culture flasks (Greiner Bio-one, Frickenhausen, Germany) for 1.5 h at 37 °C and 5% CO₂ for adherence of monocytes. The monocytes were then differentiated to macrophages and polarized toward M1- and M2-MDM as reported (15). M1-MDM were generated by incubating monocytes with 20 ng/mL GM-CSF (Peprotech, Hamburg, Germany) for 6 d in RPMI 1640 supplemented with 10% FCS, 2 mmol/L glutamine (Biochrom/Merck), and 100 U/mL penicillin, and 100 µg/mL streptomycin to obtain M0-MDM_GM-CSF, followed by 100 ng/mL LPS and 20 ng/mL IFN-γ (Peprotech) treatment for another 48 h. M2-MDM were obtained by incubation of monocytes in the same medium with 20 ng/mL M-CSF (Peprotech) for 6 d to obtain M0-MDM_M-CSF and subsequent treatment with 20 ng/mL IL-4 (Peprotech) for additional 48 h. For studying how Dex or prednisolone affect MDM during polarization, the M0-MDM were pretreated with GC for 15 min, followed by polarization for 48 h with 100 ng/mL LPS and 20 ng/mL IFN-γ into M1_Dex or M1_Pred or with 20 ng/mL IL-4 into M2_Dex or M2_Pred subsets.

SDS-PAGE and Western Blot.

Lysates of monocytes or MDM corresponding to 2 × 10⁶ cells were separated on 10% gels (5-LOX, 15-LOX-1, 15-LOX-2, COX-1, COX-2, and cPLA₂-α) or 16% (FLAP) polyacrylamide gels, and blotted onto nitrocellulose membranes (Amersham™ Protran Supported 0.45 µm nitrocellulose, GE Healthcare, Freiburg, Germany) as reported before (43). Membranes were stained with primary and secondary antibodies; for details on antibodies see SI Appendix. Immunoreactive bands visualized by Odyssey infrared imager (LI-COR Biosciences, Lincoln, NE). Data from densitometric analysis were background corrected.

Cytokine Release.

M0-MDM_GM-CSF or M0-MDM_M-CSF were pretreated with Dex or vehicle for 15 min and then polarized with LPS/IFNγ or IL-4 into different MDM subsets for 48 h as described above. Then, supernatants were collected and centrifuged (2,000 × g, 4 °C, 10 min). The protein levels of TNF-α and IL-6 were analyzed by in-house–made ELISA kits (R&D Systems, Bio-Techne).

Flow Cytometry.

Fluorescent staining of polarized MDM was performed in PBA-E buffer (PBS with 0.5% BSA, 2 mM EDTA and 0.1% sodium azide). To determine cell viability, cells were stained using the Zombie AquaTM Fixable Viability Kit (Biolegend, San Diego, CA) for 5 min at RT. Nonspecific antibody binding was blocked using mouse serum (10 min, 4 °C). Cells were stained by fluorochrome-labeled antibodies mixtures (20 min, 4 °C); for details, see SI Appendix. After staining, MDMs were measured using BD LSR FortessaTM cell analyzer (BD Biosciences), and data were analyzed using FlowJo X Software (BD Biosciences).

Incubations of Monocytes and MDM and LM Metabololipidomics by UPLC-MS/MS.

Monocytes and MDM subsets (2 × 10⁶/mL) were incubated in 1 mL PBS containing 1 mM CaCl₂ with E. coli (O6:K2:H1; ratio = 1:50) or 0.5% ECM at 37 °C. After the indicated incubation periods, the supernatants (1 mL) were transferred to 2 mL of ice-cold methanol containing 10 µL of deuterium-labeled internal standards (200 nM d₈-5S-HETE, d₄-LTB₄, d₅-LXA₄, d₅-RvD₂, d₄-PGE₂ and 10 µM d₈-AA; Cayman Chemical/Biomol GmbH, Hamburg, Germany) to facilitate LM quantification and sample recovery. Solid phase extraction of LM, sample preparation, and UPLC-MS-MS analysis of LM was conducted exactly as reported in Jordan et al. (43).

To consider reduced numbers of adherent MDM after GC-treatment (17) during stimulation with E. coli for LM analysis, the total protein amount was determined for all samples using a modified Lowry Protein Assay employing DC Protein Assay Reagents Package (#5000116, Bio Rad, Hercules, CA). LM formation, given in pg, was normalized to total protein amount adjusted to 0.15 mg (corresponding to the mean protein amount of 2 × 10⁶ plated MDM 48 h prior determination) or 1 mg, as indicated.

Participants, Study Approval, Study Design, Blood Collection, and Sample Processing of COVID-19 and IBD Patients.

Study design – COVID 19.

Patients hospitalized at the Jena University Hospital due to acute COVID-19 pneumonia were eligible for this study (SI Appendix, Fig. S6A, for details see SI Appendix, Table S7). Patients who needed additional oxygen support were treated with 6 mg Dex, i.v., daily, according to the standard of care (3). The patients were divided into two groups: Dex vs. control group. Peripheral blood was taken at the day of admission and before the initial administration of Dex (sample d0), at day 3 (sample d3) and day 5 (sample d5) after Dex treatment for analysis of LM, mRNA, and proteins. Independently, additional blood was withdrawn for routine clinical diagnostics addressing differential blood cell count, cytokines, and CRP on day 4 (median). To assess the severity of disease, the WHO Ordinal Scale of COVID-19 Disease Severity was performed at time of sample collection by a clinician. Additionally, the patients’ characteristics including age, sex, and laboratory parameters were collected. Anonymized demographics and laboratory parameters are summarized in Fig. 5A and SI Appendix, Table S7.

Study design – IBD.

Patients with acute flares of IBD with at least moderate disease activity [defined by partial Mayo Score (pMS) ≥ 5 in Ulcerative Colitis and Harvey-Bradshaw Index (HBI) ≥ 8 for Crohn’s Disease; for details see SI Appendix, Table S8] were eligible for the study and were treated with prednisolone (1 mg/kg body weight, per os, per day) at Jena University Hospital (SI Appendix, Fig. S6B). Peripheral blood was taken before the first dose of prednisolone and 4 d (median) after. Patients’ data on age, sex, IBD type, manifestation, and current activity level of IBD, the IBD-related medication, and laboratory parameters were gathered. Anonymized demographics and laboratory parameters are summarized in Fig. 5E and in SI Appendix, Table S8.

Study approvals.

The COVID-19 study was approved by the Ethics Committee of the Jena University Hospital (no. 2020-1711-Material). Written informed consent was obtained from patients or their legal surrogates prior to participation, following the ethical guidelines of the 1975 Declaration of Helsinki. The IBD study was approved by the local Ethics Committee (2018-1242_1-BO) and written informed consent was obtained from all patients before inclusion in the study and following the ethical guidelines of the 1975 Declaration of Helsinki.

Blood collection, cell isolation, and processing.

For the COVID-19 study, peripheral venous blood was collected in 9 mL K3-EDTA (ethylenediaminetetraacetic acid), 2.7 mL Serum-Gel and 2.6 mL Citrate 3.2% Monovettes (Sarstedt, Nümbrecht, Germany). The Serum-Gel and citrate-stabilized blood samples were centrifuged (10 min, 800 × g, 20 °C) to separate the serum or plasma fraction. Serum (1 mL) samples were stored at −80 °C for cytokine measurement using ELISA. Citrate-plasma samples (1 mL) were immediately frozen and stored at −80 °C for LM profiling. For experimental details on the isolation of PBMCs, monocytes, and neutrophils, see SI Appendix.

Statistical Analysis.

Results are expressed as mean ± SEM and/or as -fold change to vehicle control of n observations, where n represents the number of experiments with separate donors. Analyses of data were conducted using GraphPad Prism 9 software (San Diego, CA). For in vitro experiments parametric statistical tests were used. Two-tailed ratio paired t test was used for comparison of two groups. For multiple comparison, repeated measures ANOVA with Bonferroni post hoc tests was applied. For in vivo/ex vivo experiments nonparametric statistical tests were used, as tests for normal distribution of datasets (Shapiro–Wilk test, Kolmogorov–Smirnov test) frequently failed. Wilcoxon matched pairs signed rank test was used for comparison of two groups. For multiple comparison, Friedman test with uncorrected Dunn´s test was applied. The criterion for statistical significance is P < 0.05.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.