*3.3. Identification of DEGs*

To discover the effects of iron nanoparticles (FeNPs and PBNPs) on the gene expression profiles of HL60 and KG1a cells, a large number of DEGs (*p* < 0.05) were identified using the FPKM (Fragments Per Kilobase of transcript per Million mapped reads) method [20]. Totally, there were 470 (260 upregulated and 210 downregulated) and 1690 (720 upregulated and 970 downregulated) DEGs in the FeNP-treated HL60 and KG1a cells, respectively, and 2008 (1015 upregulated and 993 downregulated) and 2504 (986 upregulated and 1518 downregulated) DEGs in the PBNP-treated HL60 and KG1a cells, respectively. The detailed information of these DEGs is shown in File S1 (Supplementary Materials). The results showed that KG1a had more DEGs than HL60 after the treatment of iron nanoparticles. Especially, after the treatment of FeNPs, KG1a had 3.59 times more DEGs than HL60. These data indicated that KG1a that had higher stemness than HL60 could resist iron nanoparticle-induced ferroptosis by regulating much more genes than HL60, especially encountering FeNPs, a ROS inducer. This is in agreement with the observation that FeNPs induced the significant decrease of cell viability of HL60 but had no significant effect on cell viability of KG1a (Figure 1C).

A four-way Venn analysis displayed the numbers of unique and common DEGs in two cells treated by two kinds of nanoparticles (Figure 2A). It was found that each cell had many unique DEGs under the treatment of two kinds of nanoparticles. Besides those unique DEGs, two kinds of nanoparticles induced some common DEGs in two cells. In comparison, two different nanoparticles induced more common DEGs in a same cell line whereas a same nanoparticle induced less common DEGs in two different cells. There were limited numbers of common DEGs in two cells treated by two kinds of nanoparticles. To provide more detailed information on common DEGs, the common genes (fold change >1.5) regulated by two kinds of nanoparticles in a same cell (Figure 2B) and a nanoparticle in two cells (Figure 2C) were identified. Totally, 52 and 42 genes were commonly regulated by two kinds of nanoparticles in HL60 and KG1a cells, respectively (Figure 2B). In contrast, only 19 genes (fold change > 1.5) were commonly regulated by PBNPs in two cells (Figure 2C) and only 11 genes (fold change > 1.5) were commonly regulated by FeNPs in two cells (Figure 2C). These genes demonstrated the obvious cell- and nanoparticle-specific features in gene expression regulation. Finally, the most important genes that were commonly regulated in two cells treated by two kinds of nanoparticles were identified (Figure 2D). It was found that 14 genes were commonly upregulated in two cells by two kinds of nanoparticles (Figure 2D) and that only 4 genes were commonly downregulated in two cells by two kinds of nanoparticles (Figure 2D). These genes should have a close relationship with the common chemical essence of two kinds of nanoparticles: iron. These genes were closely related with five biological processes, including iron metabolism, antioxidation, lipid metabolism (lysosome dysfunction), vesicle traffic (exocytosis, endocytosis, and phagocytosis), innate immune system, and cytoskeleton. These processes typically reflect what commonly happens in cells when cells are treated by iron nanoparticles no matter their modification and structure.

**Figure 2.** Comparisons of differentially expressed genes (DEGs) in two Leukemia cells treated with two kinds of iron nanoparticles: (**A**) number of upregulated and downregulated DEGs (fold change > 1.0) in the PBNP- and FeNP-treated HL60 and KG1a cells and their relationship, (**B**) common DEGs (fold change > 1.5) in a cell treated by two kinds of nanoparticles, (**C**) common DEGs (fold change > 1.5) in two cells treated by a nanoparticle, (**D**) common DEGs (fold change > 1.0) in two cells treated by two kinds of nanoparticles, and (**E**) schematic of cellular functions of common DEGs (fold change > 1.0) in two cells treated by two kinds of nanoparticles. The detailed information of all DEGs is shown in File S1 (Supplementary Materials). HF, FeNP-treated HL60 cells; HP, PBNP-treated HL60 cells; KF, FeNP-treated KG1a cells; and KP, PBNP-treated KG1a cells.

It was interesting that GBA was significantly downregulated in two cells by two kinds of nanoparticles (Figure 2D). GBA/GBA1 codes for glucerebrosidase (GCase), which plays a central role in the degradation of complex lipids and the turnover of cellular membranes [21]. Deficiency in GCase activity leads to accumulation of glucosylceramide/glucocerebroside in lysosome and compromised lysosomal activity, which would eventually affect lipid metabolism and trafficking [22]. GBA drives autophagy-dependent cell death [23], and the GBA mutation has a close relationship with Gaucher and Parkinson's diseases [24]. Besides GBA1, the expression of other important lysosomal function-related genes was also changed by iron nanoparticles, such as LYST, CLN3, LAMP1, LAMP5, LAPTM5, LAPTM4A, LAPTM4B, and HPS6 in HF, HP, and KP. The compromised lysosomal activity also affects the intracellular iron metabolism because the stored iron in ferritin has to be released as Fe2<sup>+</sup> in lysosome. This is coincident with the iron overload (hyperferritinemia) that occurs in Gaucher cells [25]. The ceramide produced by GCase decomposing of glucocerebroside in lysosome plays a critical role in forming membrane phospholipids and the intracellular matrix. The significant iron nanoparticle-induced downregulation of GBA1 expression thus affects membrane maintenance and repair. This effect may contribute to iron nanoparticles-induced ferroptosis. However, the molecular mechanism of how iron nanoparticles inhibit GBA1 expression still remains unclear.

GCLM and NQO1 are representative Nrf2-regulated antioxidant genes [26]. The expression of these two genes was significantly upregulated by two kinds of iron nanoparticles in two cells, suggesting that cell internalization of iron nanoparticles resulted in oxidative stress by the increase of ROS via a Fenton reaction. The cells had to upregulate these antioxidation genes and SLC7A11 to neutralize the increased ROS for maintaining redox balance. In response to the cell internalization of iron nanoparticles, cells also changed the expressions of several key iron metabolism-related genes including FTH, PIR, DNM1, and TRFC. Because iron nanoparticles were internalized into cells by phagocytosis and finally trafficked to lysosome, together with glucocerebroside accumulation in cells resulting from the iron nanoparticle-induced downregulation of the GBA1 gene, cells upregulated several important genes involved in phagocytosis, exocytosis, endocytosis, and vesicular trafficking, such as ABCA1, MCTP2, DNM1, STX3, and BIN2. The interaction of iron nanoparticles with cells also induced upregulation of several genes related to the innate immune system, such as TLR6, BIN2, ADGRG3, and DDX24. This is in agreement with a previous report that iron nanoparticles could induce virus-like immune responses [27].

To further establish a high-confidence gene signature of the two Leukemia cells treated by two kinds of iron nanoparticles, the DEGs that were most significantly regulated by the nanoparticle treatments were screened (Figure 3A). These DEGs had fold changes over 2.0. The top ones of these genes were schematically shown in cells with different treatments (Figure 3B) to indicate their distributions and relationship. It was found that GBA was the only common signature gene highly downregulated in two cells under the treatment of two kinds of iron nanoparticles. CD38 was commonly downregulated and GGNBP2 was commonly upregulated in HF, HP, and KP, respectively. SMACB1 was commonly downregulated in HF, KF, and KP. SMACB1 is a core subunit of the SWI/SNF (BAF) chromatin-remodeling complex and is well-recognized as a tumor suppressor gene, which is inactivated in aggressive cancers such as nearly all pediatric rhabdoid tumors [28,29]. These highly regulated common genes represent the typical gene signatures of various cells and nanoparticles. Besides these common genes, each cell showed some unique highly regulated genes when treated by different iron nanoparticles.

**Figure 3.** Gene signatures of two Leukemia cells treated by two kinds of iron nanoparticles: (**A**) most significantly regulated DEGs (fold change > 2.0) in two cell lines treated by two kinds of iron nanoparticles and (**B**) schematic of signature genes (fold change > 2.0) in two cell lines treated by two kinds of iron nanoparticles. The common genes were placed at the same positions in schematic cells for comparing.

In gene signatures, it is very interesting that CD38 was significantly downregulated in HF, HP, and KP and that LY6E was highly upregulated in HF and KF (Figure 3B). CD38 is an LSC marker gene, and LSC is marked by CD34+CD38 low/−. The above qPCR detection indicated that KG1a expressed high-level CD34 and low-level CD38 whereas HL60 expressed no CD34 and relatively high-level CD38 (Figure 1C), showing that KG1a has much higher stemness than HL60. The treatment of two kinds of nanoparticles all significantly downregulated CD38 in two leukemia cells, especially, PBNPs most significantly downregulated CD38 in HL60 (Figure 3B). These data suggested that both iron nanoparticles preferentially killed non-stemness leukemia cells, which thus increased the relative proportion of stemness leukemia cells in live cells after iron nanoparticle treatment. On the contrary, Ly6E was highly upregulated by FeNPs in two leukemia cells (Figure 3B). The human LY6 genes highly expressed in various cancers represent novel biomarkers for poor cancer prognosis, are required for cancer progression, and play an important role in immune escape [30–35]. More importantly, overexpression of certain Ly6 genes (Ly6D, Ly6E, Ly6K, and Ly6H) turned cancer cells into aggressive stem-like cells or allowed cancer cells to act like cancer stem cells [30]. In agreement with the downregulation of CD38, the significant upregulation of Ly6E in the two FeNP-treated leukemia cells also suggested that FeNPs preferentially killed non-stemness leukemia cells and thus allowed the proportion of stemness leukemia cells to increase in live cells. These data suggested the resistance of LSC to iron nanoparticle-induced cell death such as ferroptosis and nanoptosis [36].
