**4. Alterations in macrophage polarization in diabetic wounds**

Innate immune cells including macrophages have been shown to exhibit a proinflammatory phenotype with production and secretion of inflammatory cytokines, factors and chemokines. In diabetic wounds these processes are pathologically exaggerated as a possible contributor to the poorly healing DFU [39]. High susceptibility of diabetic patients to bacterial infections and impaired wound repair is well known [11]. The molecular mechanisms underlying this weakness are not fully understood but have been extensively studied in the past. Now it is believed that disorders in glucose metabolism and related alteration in metabolic pathways may be the reason for this susceptibility [40–42]. Interestingly, for unclear reasons the number of Langerhans and dermal macrophages significantly increases in uninjured diabetic skin [43]. Moreover, the effects of diabetes on macrophages and the related wound healing process are profound, and likely stem from changes in many aspects of the diabetic environment [13, 15, 44–48]. Macrophages that are generated in a high glucose culture system exhibit a reduction in their phagocytic potential and are less capable of clearing an infection [49, 50]. Furthermore, macrophages derived from diabetic mice and human patients appear to have increased responsiveness to inflammatory stimulants and secrete more proinflammatory cytokines than normal, which seems to prevent their later transition into the more reparative M2-like phenotype [51–55]. Indeed, at the initiation of wound healing, the phagocytotic capacity of diabetic M1 macrophages is reduced due to suboptimal differentiation, which happens before the impaired transition of M1 macrophages to M2-like macrophages later on [49, 50]. The failure of their transition or repolarization in the later stages of wound healing prevents regeneration and the repair process, resulting in a delay or even failure to heal [12].

The mechanisms underlying these alterations in the diabetic macrophage phenotype have been extensively studied. Hyperglycemia affects macrophage polarization in vitro and in vivo [46, 56–62]. For example, it has been shown that diabetic mice or human patients have an increased ratio of chemokine (C-C motif) receptor (CCR7) to CD48 3 days after wound formation [46]. CCR7 is an M1 macrophage marker whereas CD48 is an M2 macrophage marker [24]. Moreover, M1 macrophages in diabetes express less matrix metalloproteinases 1 (MMP1) and more pro-inflammatory cytokines like TNFα, resulting in an impairment in keratinocyte migration and subsequent delay of wound repair [46]. Furthermore, a hyperglycemic environment has been shown to lead to an increase in many pro-inflammatory cytokines, including TNFα, IL-1β, IL-12 and IL-6 [63], rendering these M1 macrophages more metabolically active and pro-inflammatory, but less phagocytic [63, 64]. This specific alteration in the phenotype of M1 macrophages under hyperglycemic conditions further increases the sensitivity of macrophages to cytokine stimulation and starts a vicious cycle that maintains M1 macrophage polarization and leads to a prolonged inflammation during the wound healing process [64, 65].

The role of interleukins in macrophage differentiation and polarization has been recently studied [66–73]. Some interleukins have been targeted in a therapeutic modality, exhibiting a significant impact on treatment outcomes. For example, depletion of a pro-inflammatory cytokine, IL-23, causes a significant increase in M2 macrophage polarization through loss of IL-17, which leads to improvements in diabetic wound healing [74]. Similar results have been obtained using IL-17-knockout mice or using antisera against IL-17 [74]. IL-1β is highly expressed in activated M1 macrophages in a hyperglycemic environment [63, 64]. Interestingly, experiments have shown that IL-1β expression is regulated by a protein complex called NOD-, LRR- and pyrin domain-containing protein 3 (NLRP3), which is an inflammasome [75] that controls the dimerization and activation of caspase-1, leading to the subsequent transformation of the IL-1β precursor (pro-IL-1β) into its activated form IL-1β to be secreted [76]. Knockdown of NLRP3 with siRNA-mediated gene silencing reduced the production and secretion of IL-1β, which is beneficial to diabetic wound healing [77].

Recent research has also shed light on epigenetic alterations in macrophages in a hyperglycemic environment. These epigenetic changes can induce enhanced expression of proinflammatory cytokines to promote and sustain M1 macrophage polarization [78]. Now it is believed that epigenetic modifications are the main cause of the alterations in macrophage phenotype in diabetes [79]. The epigenetic modifications include histone modifications, DNA modifications and other post-transcriptional controls like microRNAs [10]. Moreover, regulation of macrophage polarization requires interactions with other cell types such as adipocytes, keratinocytes, fibroblasts and other immune cells (Neutrophils, T-cells, dendritic cells, etc.) that secrete factors to modulate macrophage polarization [14]. In the setting of diabetes, these cell-cell interactions are altered, leading to a suboptimal polarization of macrophages [46, 80–82].

A specific role for histone modification in the control of macrophage polarization has been recently highlighted. In eukaryotes, DNA and histones gather together to generate units called nucleosomes [83]. When histones are tightly wrapped with DNA, the access to transcriptional machinery is blocked to prevent transcription [84]. However, when DNA-histone machinery is disassembled, transcriptional binding is allowed. An N-terminal "tail" with lysine (K) residues on histones can be modified by some enzymes through catalyzing methylation and acetylation [84]. Histone methylation and demethylation are controlled by histone methyltransferases (HMTs) and histone demethylases (HDMs), respectively, which regulate macrophage differentiation and polarization [84] and are responsible for the M1 to M2 repolarization during wound healing [85]. For example, mixed-lineage leukemia 1 (MLL1) is a methyltransferase that catalyzes H3K4me3 deposition to affect macrophage polarization and the induction of expression of proinflammatory genes in macrophages [86]. Mechanistically, MLL1 is found to regulate changes in macrophages partially via Toll-like receptor 4 (TLR4) in both diabetic humans and diabetic mice [87, 88]. On the other hand, Suppressor of variegation, Enhancer of Zeste, Trithorax and myeloid-Nervy-DEAF-1 domain-containing protein 3 (SMYD3), which is another H3K4me3 methyltransferase, has been shown to regulate M2-like polarization of macrophages [89]. Besides HMTs, HDMs also play a critical role in macrophage polarization during wound healing. For example, Jumonji domain-containing protein 3 (JMJD3) is a H3K27 demethylase that regulates a context-dependent polarization of macrophages towards either a proinflammatory M1-like or an anti-inflammatory M2-like macrophage phenotype [90–95]. JMJD3-mediated release of H3K27me3 is compromised in diabetic wound macrophages, resulting in enhanced and sustained expression of genes associated with inflammation [96, 97]. However, lipopolysaccharides (LPS) and IL-4 have been shown to induce JMJD3 for directing M2-like macrophage polarization [96]. Together, a lot of data have demonstrated the importance of histone methylation and demethylation by HMTs and HDMs in controlling macrophage polarization during wound repair [98, 99]. Transcriptional repression is often regulated by DNA methylation, which is catalyzed by DNA methyltransferases (DNMTs) to transfer a methyl group to the cytosine ring of DNA at clusters of CpG islands [100]. The potential binding of transcription factors to a promoter region is significantly altered via methylation of CpG islands on the promoter [101]. For example, DNMT1 has been shown to induce M1-like polarization of macrophages [102]. Moreover, genetic

depletion of DNMT1 or chemical suppression of DNMT1 by 5-aza-2′-deoxycytidine promotes M2-like macrophage polarization [103] and improves wound healing in diabetic mice [104]. Macrophage polarization during wound healing has also been shown to be affected by histone acetylation and deacetylation. Transcriptional activation is enhanced by acetylation of the lysine residue on the histone tail, for which an acetyl group is transferred from acetyl CoA to the lysine residue catalyzed by histone acetyltransferases (HATs) [105]. In diabetes, it has been shown that histone deacetylase 6 (HDAC6) alters the phenotype of macrophages through IL-1β but not IL-10 [106].

Post-transcriptional control is also an important regulator of macrophage polarization in diabetes. For example, microRNAs have been shown to be important regulators of gene expression during macrophage polarization [9]. MicroRNAs (miRNAs) are non-coding small RNAs about 20 base pairs in length. miRNAs control protein levels of an expressed gene through Watson-Crick pairing to the 3′-untranslated region (3′- UTR) of the mRNA of a specific gene, resulting in altered protein translation [107]. It has been reported that M2 macrophages express high levels of miRNA-146, while M1 macrophages express low levels of miRNA-146 [108, 109], and the levels of miR-146 appear to alter macrophage polarization and their production and secretion of proinflammatory and anti-inflammatory cytokines [108, 109]. It has also been shown that miR-155 can induce an M1-like macrophage polarization through suppressing antagonists of proinflammatory cytokines [110]. Moreover, miR-33 was shown to favor M2 macrophage polarization through suppressing NLRP3 [111], a key inducer of IL-1β and inflammation [76, 77]. Similarly, long noncoding RNAs (lncRNAs) have an emerging role in regulating macrophage polarization [112–116]. Besides miRNAs, long non-coding RNAs also play essential roles in macrophage phenotypic determination and control of inflammation [117]. The role of lncRNA GASS in macrophage polarization and associated wound healing has been reported recently [113].
