**6. PUFA-Derived Lipid Mediators in Malignant Hematopoiesis**

While omega-3 PUFA have been widely implicated in anti-tumor effects in a variety of solid tumors, albeit with mixed results [81], data in hematological malignancies are sparse. In a review published by Betiati et al., a Scopus and PubMed database search between 1998 and 2012 returned 6 studies published on the subject of the effect of omega-3-PUFAs on hematological malignancies [82]. Since then, only few publications regarding effects of omega-3 PUFAs in hematological malignancies have been published. There is some evidence showing a lower incidence of non-Hodgkin lymphomas (NHLs) in patients on a diet high in omega-3-PUFAs [83]. Another study established higher omega-3-PUFAs in NHL patients in remission as compared to those with active disease [84] and recent data implicate low plasma omega-3-PUFAs as marker of inferior prognosis in diffuse large B-cell lymphoma [85]. However, no reduced risk with higher omega-3 PUFAs for lymphoid and myeloid leukemia could be detected so far [86]. In the context of ALL omega-3-PUFAs were shown to be able to lower treatment-related hypertriglyceridemia [87]. Noteworthy is a small recent study with a total of 22 leukemia or lymphoma patients in which a prolonged overall survival time of patients receiving fish oil (2g/d) was shown. These observations might indicate that EPA and DHA improve the response to treatment with conventional chemotherapy in hematological malignancies [88]. In vitro data show an inhibition of cell growth in AML cell lines by EPA and DHA [89,90]. In the erythrocytes of multiple myeloma (MM) patients, Jurczyszyn et al. demonstrated a decreased n-3/n-6 ratio and lower levels of EPA, despite higher levels of its precursor, α-linolenic acid, were measured. This might suggest an impaired functionality of desaturase and elongase enzymes in these patients [91]. A recent systematic review has assessed the current knowledge regarding the omega-3 PUFA EPA and DHA in the context of cells and models of malignant hematopoiesis [92]. There is a wealth of data gained in different cell models, but generally accepted mechanisms, and applicability in vivo and in humans, are still uncertain.

In contrast to this rather limited experimental evidence regarding omega-3 PUFA in the context of malignant hematopoiesis, the main focus of research in this field has been the omega-6 PUFA derived PGE2. Transcript levels of soluble phospholipase A2 (PLA2) subtypes IB and X have been shown to be upregulated in AML blasts compared to control blood mononuclear cells [93]. This finding might be significant since the enzymatic activity of PLA2 releases the eicosanoid precursor, arachidonic acid (AA) from membrane phospholipids for the generation of COX- and lipoxygenase-derived lipid mediators and is in keeping with the higher levels of free AA observed recently in plasma from AML patients [94].

PGE2 might normally act as a feedback regulator of myelopoiesis as described above [20]. However leukemic cells were shown to be resistant to this feedback mechanism seen in benign hematopoiesis [77,95,96]. Furthermore, insensitivity to PGE2-mediated growth inhibition in marrow cultures from patients with myelodysplastic syndrome preceded the patient's progression to acute leukemia [97]. In keeping with these results, the overall incidence of hematologic malignancies seems not to be reduced by long-term intake of the COX-inhibitor acetylsalicylic acid [98].

In contrast, treatment of mice with indomethacin (a COX inhibitor and thus inhibiting PGE2 synthesis) prior to injection of erythroleukemia cells significantly reduced the number of leukemic cells in both spleen as well as bone marrow [99]. Additionally, Wang et al. could demonstrate a marked reduction of leukemia-initiating cells in a murine limiting dilution transplant assay after treatment with indomethacin [100]. These data indicate a role for COX-metabolites in the proliferation of leukemic cells.

While COX-1 and COX-2 transcripts can be detected by polymerase chain reaction, the COX-2 protein is not present in primary AML and ALL blasts which is concordance with in vitro data from the human promyelocytic leukemia cell line HL-60 [70,101]. However, AML blasts were shown to express COX-1 [101]. Interestingly, constitutive expression of COX-1 can be upregulated by tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) in HL-60 cells. This is accompanied by an increase of PGE2 synthesis and shows a protective effect towards TRAIL-induced apoptosis [70]. Similarly, AML cells treated with doxorubicin showed overexpression of multidrug transporter MDR1 triggered by increased PGE2-formation, and thereby decreased cytostatic efficacy of doxorubicin [102]. These studies suggest that protective mechanisms of the leukemic blasts to avoid eradication are—at least in part—mediated by PGE2.

In vitro data with AML-mesenchymal stroma cells (AML-MSC) co-cultures show a greatly increased COX-2 expression in MSC and induced PGE2 production in dependence of IL1β and ARC (apoptosis repressor with caspase recruitment domain). ARC is a protein that regulates leukemia microenvironment interactions through NFκB/IL1β and was shown to be an adverse prognostic marker in AML [103]. The COX-2 derived elevation of PGE2 from stromal cells seem to support AML chemoresistance through the expression of β-catenin which regulates ARC [104]. These data indicate

that PGE2 production in the microenvironment takes part in a mechanism of an antiapoptotic action and microenvironment-mediated chemoresistance in certain subgroups of AML.

Downstream of the COX enzymes the last step in the synthesis of PGE2 is performed by the prostaglandin E synthase. This enzyme was shown to be present in normal tissues in minor amounts but is strongly upregulated in neoplastic cells [105,106]. In AML cell lines the prostaglandin E synthase was shown to be upregulated and specific inhibition of the enzyme resulted in an inhibition of proliferation [107] (Figure 2).

As shown for solid tumors, the inhibition of PGE2 receptors EP1, EP2, and EP4 allows for inhibition of cancer-associated inflammation and tumor growth. Mice deficient of these PGE2 receptors display decreased tumorigenesis as demonstrated in various experimental settings mainly for solid tumors [57,108–110]. Concordantly, expression levels of EP1 and EP2 have been demonstrated to be increased in cancerous tissues [111]. In AML, Ross et al. as well as Yagi et al. could demonstrate elevated transcript levels of EP2 in AML blasts in a pediatric cohort [112,113] and Denizot et al. could show that AML blasts express functional EP2 receptors [25,114] (Figure 2).

Interestingly, an omega-3 PUFA-derived lipid mediator has been implicated in anti-leukemia effects: The cyclopentenone prostaglandin Δ12-PGJ3, produced through cyclooxygenase action from the omega-3 PUFA EPA, was able to decrease leukemia burden in two murine models of leukemia [115] by selectively targeting leukemia stem cells (LSCs).

Concerning other eicosanoids and their role in hematopoiesis, only limited data has been published. For instance, lipoxins, which are produced by the lipoxygenases (as well as by the COX enzymes if acetylated by ASA), were shown to suppress tumor angiogenesis [116]. Actually, defective lipoxin synthesis was found in leukemia [117], indicating a stabilizing role for lipoxin in benign hematopoiesis (Figure 2). Additionally, a landmark study by Kode et al. demonstrated that an activating β-catenin mutation in osteoblasts can induce the development of leukemia by activating Notch signaling in hematopoietic precursors [118]. Here, osteoblasts exhibited increased expression of the Notch ligand Jagged 1. Conversely, there is some evidence suggesting that lipoxin A4 might decrease the expression of Jagged 1 [119]. Generally, however, the relevance of the lipoxygenase-pathway in benign as well as malignant hematopoiesis is yet to be explored further.

Concerning 5-lipoxygenase, Gal et al. could show elevated transcript levels of 5-lipoxygenase in the CD34+/CD38+ fraction of AML blasts compared to the less mature CD34+/CD38- fraction of the same patients [2]. In a model utilizing mice deficient in 12/15-lipoxygenase it was found that this enzyme is required for the maintenance of long-term HSC quiescence as well as self-renewal [120].

#### **7. Conclusions**

Recent treatment approaches in AML focus increasingly on immune therapy. One of the challenges in the field is to eliminate or reprogram the immune suppressive microenvironment often created by tumors [56]. Interestingly, the otherwise pro-inflammatory lipid mediator PGE2 seems to play a major role in mediating some of these suppressive effects by either direct inhibition of effector T-cell function or indirectly by increasing the frequency of immunosuppressive cell types. Particularly approaches to lower PGE2 might thus serve to enhance immune therapy approaches Current experimental data confirm the importance of this aspect also in the context of AML.

**Author Contributions:** Conceptualization, T.K. and K.H.W.; writing—original draft preparation, T.K. and A.L.; writing—review and editing, A.P., E.R. and K.H.W.; visualization, T.K.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
