**4. Discussion**

Using the methods recommended in the European Community Directives, we found that all the physical and chemical indexes results of Xinjiang sa fflower honey samples were in line with the requirements. Hydroxymethylfurfural is a substance produced by honey during storage or heating. It is an important criterion for evaluating the quality of honey [43,44]. As the honey storage time becomes longer or the heating temperature rises higher, the more its content increase. Amylase is an important indicator of freshness in honey [1]. Honey enzymes will inactivate (or the value will drop) when honey is heated to a temperature higher than 40 ◦C. The amylase value measured in our experiment was 21.1 ± 0.3 mL/g honey, which is much higher than the 8 mL/g honey of EU standard (2001/110/EC). We did not detect hydroxymethylfurfural nor sucrose in the sa fflower honey samples. The moisture complied with EU standard 2001/110/EC to not be higher than 20 g/100 g. The sa fflower honey results for moisture were better than carob honey from Morocco as studied by Redouan et al. [45], The carob honey moisture was 19.5 g/100 g. Therefore, the Xinjiang sa fflower honey samples were fresh and mature honey.

As polyphenols endowed honey with distinct color, taste, flavor and biological activities. Polyphenol analysis is therefore considered as an important tool for determining the quality of honey. A total number of 14 polyphenolic compounds were preliminary characterized using the HPLC-Q-TOF/MS analysis, and the sum of each component content was 14.83 mg/kg honey. Solid phase extraction (SPE) using the Amberlite XAD-2 resin is one of the most widely used extraction methods for isolating individual phenolic and flavonoid compounds in honey [46]. In the present study, we didn't measure the phenolic compounds recovery, which is a major limitation. Nevertheless, the honey was reported with a mean recovery of 43.7% (ranging from 6.7%, gallic acid, to 65.2%, naringin) [47]. In addition, the discrepancies of the extraction recoveries have been found among di fferent types of honey, which might be attributed to sugars as well as some other matrix constitutions. Among them, eight phenolic acid components, and the vanillic acid and p-hydroxybenzoic acid contents were higher, which were 3.196 mg/kg honey and 1.524 mg/kg honey, respectively, accounting for 65.6% of the total phenolic acid content. Protocatechuic acid, gallic acid, ca ffeic acid, ferulic acid, and other active components were also detected. In addition, seven flavonoid components were detected, among which quercetin and myricetin were higher, at 5.342 mg/kg honey and 1.021 mg/kg honey, respectively, accounting for 83.4% of the total flavonoid content. Rutin, quercetin-3-O-glucoside, kaempferol and naringenin were lower. Protocatechualdehyde, kaempferol-3-O-glucoside, morin, luteolin, diosmetin, chrysin, pinocembrin, galanin, and CAPE, were not detected. We propose the first reason is that di fferent honey plants, di fferent regions, and other environments have a grea<sup>t</sup> impact on the polyphenol composition. Secondly, our polyphenol extraction method has a low extraction rate of flavonoids. In a previous study, Bangladesh honey samples (Mustard flower honey, Kalijira honey, Padma flower honey, and Teel/sesame honey) had higher levels of ca ffeic acid and benzoic acid in phenolic acid substances, while gallic acid, kaempferol, myricetin, and naringenin were also detected [18]. Sundarban, Bangladesh, which is a multifloral honey collected from the largest mangrove forest in the world, showed various phenolic compounds, including gallic acid, vanillic acid, pyrogallol, trans-cinnamic acid, and other phenolic substances, as well as naringin, rutin, and quercetin flavonoids. Among all these phenolic compounds, trans-cinnamic acid and naringin had the highest contents [48]. There are more than 30 kinds of polyphenols in honey. Naringenin and hesperetin typically only appear in specific honeys, such as orange honey [17].

The DPPH· and ABTS+ free radical scavenging capacity experiments used the IC50 (semi-inhibitory rate, i.e., the concentration of the analyte required to scavenge 50% of free radicals) as the criterion for evaluating the antioxidant activity. The literature showed that due to the rich polyphenol content, honey has a very significant antioxidant e ffect. Five types of honey in Croatia showed high DPPH· free radical scavenging activity (black locust honey 125.48 mg/mL, lime honey 42.77 mg/mL, sage honey 25.04 mg/mL, chestnut honey 16.02 mg/mL, and honeydew honey 8.69 mg/mL [49]), In eight kinds of carob honey from di fferent areas of Morocco, the DPPH free radical IC50 values ranged from

12.54 mg/mL to 23.52 mg/mL [45] and seven kinds of Turkish single flower honey (Heather, Oak, Chestnut, Pine, Astragalus, Acacia, Lavender) demonstrated, ABTS<sup>+</sup>·free radical IC50 values from 0.06 g/mL–3.68 g/mL [50]. In our experiment, the DPPH and ABTS<sup>+</sup>·free radical IC50 values of ECH were 68.23 ± 0.40 μg/mL and 81.88 ± 0.54 μg/mL, respectively, and they had high antioxidant activity.

In this experiment, ECH can significantly reduce the mRNA levels of the related inflammatory factors iNOS, IL-1β, TNFα and MCP-1, while up-regulating the expression of the antioxidant genes HO-1, TXNRD and NQO. Due to the expression of iNOS when suppressed, the amount of NO released is reduced accordingly. Through the detection of IκB α phosphorylation and the NF-κB signaling pathway, we found that ECH was likely to inhibit the phosphorylation of IκBα to produce P-IκBα, thereby inhibiting the activation of the NF-κB signaling pathway and reducing inflammation. The Nrf-2/HO-1 signaling pathway demonstrated a protective e ffect against oxidative stress. Nrf-2 can regulate the gene expression of HO-1 at the transcription level, while HO-1 can inhibit the production of inflammatory factors in LPS-stimulated Raw 264.7 cells and protect the cells under oxidative stress [27]. We measured the expression of Nrf-2 and HO-1, which increaseds in line with the improvement of the ECH concentration. The 2.5 μg/mL and 5 μg/mL ECH anti-inflammatory and anti-oxidant e ffects were superior to the positive control group Dexamethasone, indicating that ECH has good anti-inflammatory and anti-oxidant e ffects.

ECH is rich in phenolic acids and flavonoids and also has good anti-oxidation and anti-inflammatory effects. Thus, we speculated that the existence of these compounds provides a material basis for the antioxidant and anti-inflammatory activities of sa fflower honey extract. Vanillic acid can play an anti-inflammatory and analgesic role through anti-oxidation and inhibiting the production of pro-inflammatory cytokines associated with NF-κB [51]. Protocatechuic acid and p-hydroxybenzoic acid showed significant antioxidant activity in thyme honey [52]. The O3-H15 bond of gallic acid is easy to break to scavenge free radicals, providing antioxidant e ffects [53]. Gallic acid can also block LPS-induced activation of TLR4/NF-κB (Toll-like receptor 4/nuclear factor-κB) to inhibit the inflammation of Raw 264.7 cells [54]. Ca ffeic acid reduces LPS-induced neuroinflammation through the regulation of cytokine networks, down-regulates NF-κB-dependent pro-inflammatory genes, and reduces oxidative stress [55]. Ferulic acid can exert anti-inflammatory activity by inhibiting nitroso-oxidative stress and pro-inflammatory cytokine production [56]. We found that the greatest common denominator of these substances is that they all contain di fferent amounts of hydroxyl groups; thus, we infer that the vast majority of the substances involved in antioxidant activities are all related to these hydroxyl groups. It was reported that quercetin can scavenge active oxygen, inhibit damage caused by oxidative stress, and prevent the development of TNFα and the secretion of iNOS and IL-1β in Raw 264.7 cells induced by lipopolysaccharide (LPS) [57]. Myricetin exerted anti-inflammatory effects by inhibiting NF-κB-p65 in the NF-κB pathway [58]. Rutin [59] and naringenin [60] alleviate oxidative stress and exert anti-inflammatory e ffects by regulating mechanisms related to the Nrf-2/HO-1 signaling pathway. Quercetin-3-O-glucoside can inhibit the expression of TNFα, IL-1β and IL-6 and the activation of NF-κB, and simultaneously up-regulate the expression of Nrf-2 and HO-1, and inhibit the inflammation and oxidation to resist the induction of cisplatin of acute kidney injury in mice [61]. Kaempferol can up-regulate the expression of iNOS in rat articular chondrocytes stimulated by IL-1β, and can inhibit the degradation of IκB α and the activation of NF-κB in rat articular chondrocytes stimulated by IL-1β [62]. Based on this, we propose that these polyphenolic compounds in ECH are the main contributors to the anti-oxidation and anti-inflammatory e ffects of sa fflower honey.

In the present study, we must explain the limitations of our experiment. First, as the identified polyphenols might had multiple isoforms and exist in the honey in a very complex form, their quantifications in our honey samples might have some bias. Second, phenolic compounds recovery from XAD-2 resin could to be estimated which can partly counteract the strong matrix e ffects in the honey. Third, in this study we only preliminary detected several polyphenolic compounds in our sa fflower honey samples, we still did not propose a "marker" compound in our honey samples, which is with grea<sup>t</sup> importance for the standardization and quality control on this special honey type. Nevertheless, our research group are keeping working on the isolation and identification on the marker compound of the safflower honey, which is helpful to discriminate from other commercial honey types. In such work accurate recovery rate will be examined on the single maker compound.
