Skeleton-Controlled pDNA Delivery of Renewable Steroid-Based Cationic Lipids, the Endocytosis Pathway Analysis and Intracellular Localization

Using renewable and biocompatible natural-based resources to construct functional biomaterials has attracted great attention in recent years. In this work, we successfully prepared a series of steroid-based cationic lipids by integrating various steroid skeletons/hydrophobes with (l-)-arginine headgroups via facile and efficient synthetic approach. The plasmid DNA (pDNA) binding affinity of the steroid-based cationic lipids, average particle sizes, surface potentials, morphologies and stability of the steroid-based cationic lipids/pDNA lipoplexes were disclosed to depend largely on the steroid skeletons. Cellular evaluation results revealed that cytotoxicity and gene transfection efficiency of the steroid-based cationic lipids in H1299 and HeLa cells strongly relied on the steroid hydrophobes. Interestingly, the steroid lipids/pDNA lipoplexes inclined to enter H1299 cells mainly through caveolae and lipid-raft mediated endocytosis pathways, and an intracellular trafficking route of “lipid-raft-mediated endocytosis→lysosome→cell nucleic localization” was accordingly proposed. The study provided possible approach for developing high-performance steroid-based lipid gene carriers, in which the cytotoxicity, gene transfection capability, endocytosis pathways, and intracellular trafficking/localization manners could be tuned/controlled by introducing proper steroid skeletons/hydrophobes. Noteworthy, among the lipids, Cho-Arg showed remarkably high gene transfection efficacy, even under high serum concentration (50% fetal bovine serum), making it an efficient gene transfection agent for practical application.


Introduction
In recent decades, transformation of renewable and biocompatible natural resources [1] into various functional products including chemical reagents, pharmaceutics, oils and fuels, and new functional materials has been highly focused for nurturing the sustainable development. Steroids, a large natural lipids family known as "keys of life", play vital roles including membrane formation, hormone metabolism and cell signal transduction in organelles. Some steroidal compounds possess steroid hydrophobes/skeletons. The pDNA binding affinities of the newly synthesized steroid-based lipids were examined by agarose gel retardation assays. The average particle sizes, surface potentials and morphologies of the steroid-based lipids/pDNA lipoplexes were investigated by dynamic laser light scattering (DLS) and transmission electron microscopy (TEM), respectively. Cytotoxicities of the lipids were evaluated by a thiazoyl blue tetrazolium bromide (MTT) method, and their gene transfection efficiency were investigated by Luciferase and EGFP expression assays in the presence/absence of serum with multiple cell lines. The endocytosis pathway was studied by luciferase expression assays with specific endocytosis inhibitors. The cellular uptake efficiency was evaluated by flow cytometry. Finally, the intracellular localization of the steroid-based lipids/Cy3-pDNA lipoplexes was observed under fluorescence microscopy.

Synthesis Procedures of the Steroids-Based Cationic Lipids
Synthesis procedures of the steroid-based cationic lipids and their important intermediates are described in detail in Supplementary Materials S1.

Nuclear Magnetic Resonance (NMR) and Mass Spectral Measurements
and tetramethylsilane (TMS) was utilized as the internal chemical shift reference. Mass spectra (ESI-MS) were routinely measured on a Varian SATURN 2000 instrument (Palo Alto, CA, USA).

Agarose Gel Retardation Assay
pDNA binding affinity of the new steroid-based cationic lipids was determined by agarose gel retardation assay [31]. The solution of steroid-based cationic lipids/pDNA lipoplexes were firstly prepared under a predetermined +/− charge ratio with 0.01 M PBS and 1.0 µg of pDNA in 20 µL of pure water and kept incubation for 30 min. Then, the lipoplex solution was loaded onto 1% agarose gel (w/v) with EB (0.5 µg/mL in gel) and TAE buffer (pH = 7.4), and the naked pDNA was used as the control. After the gel electrophoresis under +100 V for 1 h, the retardant of pDNA was recorded on a UV transilluminator (UVP, Upland, CA, USA) benchtop 2UV transilluminator system.
Stability of the steroid-based cationic lipids/pDNA lipoplexes were analyzed by agarose gel retardation assay. The lipoplex solution of each steroid lipid was prepared according to the aforementioned method under the N/P charge ratios of 15 and incubated for 40 min. After that, various amount of heparin solution (1 mg/mL) was added into each lipoplex solution and further incubated for 40 min. Then, each solution was loaded onto 1% agarose gel (w/v) containing EB dye (0.5 µg/mL) in TAE buffer solution (pH = 7.4) and run for 40 min at 100 V. The plasmid DNA retardation was illuminated and photographed on a UVP.

Particle Sizes and ζ Potentials of the Steroid-Based Cationic Lipids/pDNA Lipoplexes
Average particle sizes and ζ potentials of the Steroid-based cationic lipids/pDNA lipoplexes were analyzed on a Malvern Zetasizer Nano ZS90 (Worchestershire, UK) at room temperature. The lipoplex solutions were first prepared by mixing steroid-based cationic lipids and pDNA (2.0 µg/mL) under diverse +/− charge ratios in 1 mL pure water, and then laser light at λ = 633 nm was employed at a fixed scattering angle of 90 • for the nanoparticle size analyses.
To investigate the effect of serum, the as-prepared steroid-based cationic lipids/pDNA lipoplexes were incubated with BSA (1 mg/mL) solution for 10 min, and then the average particle sizes and ζ potentials were analyzed on a Malvern Zetasizer Nano ZS90 at ambient temperature.

Morphologies of the Steroid-Based Cationic Lipids/pDNA Lipoplexes by TEM
Morphology of the steroid-based cationic lipids/pDNA lipoplexes were characterized at room temperature on a transmittance electronic microscopy (TEM, JEOL-1230, JEOL Co., Ltd., Tokyo, Japan) with an acceleration voltage of 80 KV. The lipoplexes solution were firstly prepared by mixing steroid-based cationic lipids and pDNA (5.0 µg/mL) under +/− charge ratios of 15 in 1 mL of pure water, and then the lipoplex samples for TEM analysis were dropped onto a 300-mesh carbon-coated copper grid and air-dried at room temperature.

Cytotoxicity of the Steroid-Based Cationic Lipids by MTT and CCK-8 Assays
Cytotoxicity of the steroid-based cationic lipids were evaluated in H1299 (by MTT assay) [31] and HeLa cell lines (by CCK-8 assay) [36]. Cells were firstly seeded into 96-well microplates at 4 × 10 3 cells per well, and were cultivated under 37 • C and 5% CO 2 for 24 h in 100 µL DMEM medium (with 10% FBS). Subsequently, the medium was replaced with fresh DMEM medium (with 10% FBS), and then each steroid-based cationic lipid under varies concentrations was individually added into the wells and further incubated for another 24 h. After that, 20 µL of 5.0 mg/mL MTT (or 10 µL of CCK-8) was added into each well. For MTT assay, after 4 h incubation, the medium was removed, and DMSO (100 µL/well) was added to dissolve the formed MTT formazan with gently shaking for 10 min, then each sample with six replicates (n = 6) was analyzed on a microplate reader (BioTek, ELX800, Winooski, VT, USA) at λ = 490 (λ ref = 630 nm). For CCK-8 assay, after 3 h incubation, each sample with six replicates (n = 6) was directly analyzed on a microplate reader (BioTek, ELX800) at λ = 490 (λ ref = 630 nm). Commercially available branched-polyethylenimine (bPEI-25k) was applied as a reference in both MTT and CCK-8 cytotoxicity assays.

Luciferase Gene Transfection Assay in the Presence and Absence of Serum
Gene transfection efficiency of the steroid-based cationic lipids were evaluated by luciferase gene transfection assay [37]. H1299 and HeLa cells were firstly seeded into 24-well microplates (5 × 10 5 cells/well) and incubated under 37 • C and 5% CO 2 with DMEM (10% FBS) for 24 h. Then, lipoplex solutions were prepared by mixing the steroid-based cationic lipids and luciferase pDNA (1.0 µg/well) in 200 µL FBS-free DMEM medium or DMEM with 10%, 20% and 50% FBS at a predetermined +/− charge ratio and incubated for 30 min. The commercially available bPEI-25k (± = 10) and Lipofectamine2000 (Lipo2000, w/w = 3) were used as references. Then, the prepared lipoplex/complex solutions were added into the 24-well plates and incubated for 4 h (0% FBS) or 24 h (10%, 20%, and 50% FBS). For the transfection with 0% FBS, the medium was replaced with 200 µL fresh DMEM medium (with 10% FBS) and further incubated for 20 h. Consequently, the Luciferase transfection assays were conducted in accordance to the protocol of Promega Luciferase assay system. Finally, total Luciferase protein was measured with BCA assay kit (Applygen Technologies Inc., Beijing, China), and relative light unit per milligram of Luciferase protein (RLU/mg) was calculated to evaluate the cell transfection efficacy of the steroid-based cationic lipids (n = 3).
HeLa cells were seeded into 6-well microplates at a density of 3 × 10 5 cells/well and cultivated in DMEM medium (with 10% FBS) for 24 h. Subsequently, the cells were cultured with fresh serumfree medium, treated with the as-prepared cationic lipids/Cy3-pDNA lipoplexes, and incubated for 4 h. Thereafter, the cells were washed with 1 × PBS for three times and digested by trypsinization. After centrifugation, the harvested cells were re-suspended in 1 mL 1 × PBS solution and were transferred to a flow cytometry (BD FACS Calibur, Mountain View, CA, USA, set as 10,000 cells for each sample). During the FACS analysis, the HeLa cells were gated by sideward scatter versus forward scatter (SSC/FSC) plots, and the Cy3 fluorescence intensities were recorded in the FL2-H channel.

Intracellular Localization of the Steroid-Based Cationic Lipids/Cy3-pDNA Lipoplexes Observed by Fluorescence Microscopy
H1299 cells were seeded into 6-well microplates (4 × 10 5 cells/well in 1 mL DMEM medium with 10% FBS), and incubated at 37 • C under 5% CO 2 for 24 h. Then, the as-prepared steroid-based cationic lipids/Cy3-pDNA (± = 10) and Lipo2000/Cy3-pDNA (w/w = 3) were added into each well and incubated for 4 h. Before fluorescence imaging, H1299 cells were washed with 1 × PBS three times to diminish the fluorescence background. Then, the cells were fixed with 4% paraformaldehyde in 0.12 M phosphate buffer (pH 7.2) for 10 min, and washed three times with 1 × PBS. Thereafter, DAPI (for cell nuclei staining, 0.5 µg/per well) and Lysotracker (for lysosome staining, 2.5 µg/per well) were added and incubated for 15 min to stain the cell nuclei and lysosome. Finally, the H1299 cells were washed with 1 × PBS three times and the fluorescent images were recorded on a Nikon Ti-S invert fluorescence microscopy.

Synthesis and Characterization of the Steroid-Based Cationic Lipids
Generally, in cationic lipid-based gene carriers, cationic headgroups are connected to hydrophobic lipid tails via certain linkers [39]. As a modular synthetic approach, herein, we prepared four steroid-based cationic lipid analogs (Scheme 1) by facile coupling of the steroid hydrophobes/skeletons (cholesterol, 2H-cholesterol, diosgenin and tigogenin) with Boc-protected (l-)-arginine headgroup through 6-aminocaproic acid, a natural metabolite of (l-)-Lysine. The synthetic procedures and routes are depicted in S1 and Scheme S1 in the Supplementary Materials. Firstly, steroid hydrophobes were reacted with 6-(Boc)-aminocaproic acid with the catalysis of DCC/DMAP [25] to prepare the steroid-caproic lipid precursors ( 1 H NMR spectra shown in Figure S1a). Then, the as-prepared precursors were further coupled with Boc-protected (l-)-arginine (Boc-Arg-OH) in the presence of EDC/DMAP to synthesis the Boc-protected steroid-arginine cationic lipid precursors. Finally, the BOC protection group was removed in excessive trifluoroacetic acid [40] to afford the steroid-based cationic lipids (Cho-Arg, 2H-Cho-Arg, Dios-Arg and Tigo-Arg) as products with the isolation yield of 52-63%. Their structural characterization by NMR and ESI-MS are described in detail in the Experimental Section (Supplementary Materials S1), and the 1 H NMR spectrum of the lipids are shown in Figure S1b. Proton signals at around 7.7, 8.2 and 8.5 ppm were identified as the protonated guanidino groups on the (l-)-arginine. The broad signal around 6.9-7.6 ppm belongs to the protonated α-primary amino group. Cho-Arg and Dios-Arg both showed obvious double bond proton signals at around 5.2-5.3 ppm, which could not be observed on 2H-Cho-Arg and Tigo-Arg lipids due to their hydrogen-saturated steroid skeletons. The results indicated the new steroid-based cationic lipids were successfully prepared with facile and efficient synthetic approaches.

pDNA Binding of the Steroid-Based Cationic Lipids and the Stability of the Lipoplexes
Highly efficient lipid gene carriers should bind/load plasmid DNA at low +/− charge ratios [38]. Herein, pDNA binding affinities of the steroid-based cationic lipids were measured by agarose gel DNA-retardation assay [41]. As shown in Figure 1a, in the presence of 10 mM PBS in TAE buffer solution, the intensity of the migrating free pDNA bands decreases gradually with the increasing amount of steroid-based cationic lipids. Cho-Arg and 2H-Cho-Arg showed higher pDNA binding affinities that could completely retard pDNA at +/− ratio of 4, while Dios-Arg and Tigo-Arg have lower pDNA binding affinity (± > 6). The pDNA binding affinity of the steroid-based cationic lipids was in a sequence of: Cho-Arg (± = 4) ≈ 2H-Cho-Arg (± = 4) > Dios-Arg (± = 6) > Tigo-Arg (± >10), eliciting the pDNA binding affinity was steroid-skeleton dependent. We had revealed that pDNA binding affinity of some cholesterol-based cationic lipids relied largely on the pKa values of headgroups [25]; the above-mentioned results indicated that the pDNA binding affinity could be tuned by choosing certain types of steroid hydrophobes/skeletons for the construction of cationic lipids.

pDNA Binding of the Steroid-Based Cationic Lipids and the Stability of the Lipoplexes
Highly efficient lipid gene carriers should bind/load plasmid DNA at low +/− charge ratios [38]. Herein, pDNA binding affinities of the steroid-based cationic lipids were measured by agarose gel DNA-retardation assay [41]. As shown in Figure 1a, in the presence of 10 mM PBS in TAE buffer solution, the intensity of the migrating free pDNA bands decreases gradually with the increasing amount of steroid-based cationic lipids. Cho-Arg and 2H-Cho-Arg showed higher pDNA binding affinities that could completely retard pDNA at +/− ratio of 4, while Dios-Arg and Tigo-Arg have lower pDNA binding affinity (± > 6). The pDNA binding affinity of the steroid-based cationic lipids was in a sequence of: Cho-Arg (± = 4) ≈ 2H-Cho-Arg (± = 4) > Dios-Arg (± = 6) > Tigo-Arg (± >10), eliciting the pDNA binding affinity was steroid-skeleton dependent. We had revealed that pDNA binding affinity of some cholesterol-based cationic lipids relied largely on the pKa values of headgroups [25]; the above-mentioned results indicated that the pDNA binding affinity could be tuned by choosing certain types of steroid hydrophobes/skeletons for the construction of cationic lipids.
For practical applications, the pDNA-loaded lipoplexes should be stable enough to avoid the possible dissociation in physiological conditions before they reached targeting cells/tissues. The stability of steroid-based cationic lipids/pDNA lipoplexes were evaluated by heparin competition assay according to the literature [42]. As shown in Figure 1b, the pDNA migration band appeared from the lipoplexes after addition of different amount of heparin (1.5 mg/mL), due to the heparin induced dissociation of lipoplexes. The lipoplexes stability was in a sequence of: Cho-Arg/pDNA ≈ 2H-Cho-Arg/pDNA > Dios-Arg/pDNA > Tigo-Arg/pDNA. Notably, the trends of lipoplexes stability by heparin competition assays were in good agreement with the results of pDNA binding affinity, indicating that the introduction of flexible hydrophobic tails [43] could enhance the lipoplex stabilities. Moreover, the stability differences among the lipoplexes may bring them different gene transfection efficiencies.

Particle Size, Surface Charge and Morphology of the Lipoplexes
Prior works revealed that gene transfection and cellular trafficking manners of lipoplex nanoparticles relied greatly on their size, shape, and surface charge [44], which depended on structural factors such as cationic headgroups, lipid hydrophobes, etc. [45]. Herein, the hydrodynamic particle size and surface charge of the steroid-based cationic lipids/pDNA lipoplexes were measured by DLS. As shown in Figure S2a, the lipoplexes showed different average particle sizes. Cho-Arg and 2H-Cho-Arg formed smaller lipoplexes (92-110 nm), whereas much bigger particle sizes were observed on Dios-Arg (190-205 nm) and Tigo-Arg (318-390 nm) lipoplexes, indicating the steroid hydrophobes/skeletons played important roles in lipoplexes formation. Meanwhile, different lipoplex surface charges were observed ( Figure S2b). Cho-Arg/pDNA and 2H-Cho-Arg/pDNA lipoplexes demonstrated a relatively rapid surface charge increasing from −18-−15 mV to +28-+36 mV when the +/− ratio increasing from 2 to 30, whilst Dios-Arg/pDNA and Tigo-Arg/pDNA showed comparatively slower charge increasing (from −18-−15 mV to +23-+28 mV) within the same +/− range, elucidating that the surface charge could be tuned by selection of particular steroid skeletons. Moreover, the steroid-lipids/pDNA lipoplex nanoparticles For practical applications, the pDNA-loaded lipoplexes should be stable enough to avoid the possible dissociation in physiological conditions before they reached targeting cells/tissues. The stability of steroid-based cationic lipids/pDNA lipoplexes were evaluated by heparin competition assay according to the literature [42]. As shown in Figure 1b, the pDNA migration band appeared from the lipoplexes after addition of different amount of heparin (1.5 mg/mL), due to the heparin induced dissociation of lipoplexes. The lipoplexes stability was in a sequence of: Cho-Arg/pDNA ≈ 2H-Cho-Arg/pDNA > Dios-Arg/pDNA > Tigo-Arg/pDNA. Notably, the trends of lipoplexes stability by heparin competition assays were in good agreement with the results of pDNA binding affinity, indicating that the introduction of flexible hydrophobic tails [43] could enhance the lipoplex stabilities. Moreover, the stability differences among the lipoplexes may bring them different gene transfection efficiencies.
Morphologies of the steroid lipids/pDNA lipoplexes were further investigated by TEM. In Figure S2c, the steroid lipids/pDNA lipoplexes (± = 15) were observed as near-spherical nanoparticles [24]. Cho-Arg and 2H-Cho-Arg/pDNA lipoplexes were seen as small nanoparticles (40-110 nm) and Dios-Arg/pDNA were bigger nanoparticles (120-210 nm), whereas Tigo-Arg/pDNA showed the biggest particle sizes of 180-340 nm. Likewise, Yu et al. [46] recently reported that the lipoplexes particle size of a biotin-conjugated diosgenin cationic lipid were larger than their cholesterol-containing analogs. The trend of steroid-based lipids/pDNA lipoplexes morphology measured by TEM was in accordance with the DLS data. Compared to DLS results, smaller lipoplex particle size was observed by TEM, which could be interpreted as a result of nanoparticle shrinking in the drying process during TEM sample preparation [47].
It had been disclosed that the interaction of cationic gene carriers with serum proteins could affect their gene transfection properties [18]. To further examine the interaction between the lipoplexes and serum proteins, the average particle size of the steroid-based lipids/pDNA lipoplexes (± = 15) were measured in the presence of bovine serum albumin [48] (BSA, 1 mg/mL) after incubation for 20 min ( Figure S2d). The average particle sizes of Dios-Arg/pDNA and Tigo-Arg/pDNA found largely significantly increased (49-57 nm), whereas Cho-Arg/pDNA and 2H-Cho-Arg/pDNA only showed slight particle size increases (8-15 nm). The results elicited the steroid-based cationic lipids may have different gene transfection properties under serum environments.

Cytotoxicity of the Steroid-Based Cationic Lipids by MTT/CCK-8 Assay
Developing highly biocompatible cationic gene carriers was essential for safe gene delivery/transfection. We evaluated the in vitro cytotoxicity of the steroid-based cationic lipids in H1299 and HeLa cells by MTT and CCK-8 assays with bPEI-25k as the control. As shown in Figure 2a,b, very high cytotoxicity of the bPEI-25k control was observed in both H1299 and HeLa cell lines (IC 50 = 28.8 µg/mL in H1299; and IC 50 = 23.7 µg/mL in HeLa), due to its strong cell membrane disruption effect [49].
Notably, the steroid-based lipids have different cytotoxicities within the concentration range from 0 to 150 µg/mL. Cho-Arg (IC 50 = 88.5 µg/mL in H1299; and IC 50 > 150.0 µg/mL in HeLa) and 2H-Cho-Arg (IC 50 = 92.7 µg/mL in H1299; and IC 50 > 150.0 µg/mL in HeLa) showed moderate cell viability. Dios-Arg showed much higher cytotoxicity (IC 50 = 83.5 µg/mL in H1299; and IC 50 = 74.1 µg/mL in HeLa), which may be due to the inhibition effect of diosgenin on 3-hydroxy-3-methylglutaryl CoenzymeA (HMG-CoA) reductase, a rate-limiting enzyme involved in cholesterol biosynthesis [50]. Tigo-Arg showed much lower cytotoxicity than the other three lipids in H1299 and HeLa cells, but the mechanism is not clear yet. Similar trends of the cytotoxicity were also observed in COS-7 cell line ( Figure S3). Interestingly, we noticed that there is only a double bond difference between the structures of Dios-Arg and Tigo-Arg, indicating the molecular structure (even a double bond) of the steroid hydrophobes could greatly affect the apparent cytotoxicity. Likewise, Yu et al. revealed that diosgenin-cyclen cationic lipid has a higher cytotoxicity than its cholesterol-cyclen counterpart [51]. In previous work, we have revealed that the apparent cytotoxicity of some cholesterol-based cationic lipids greatly relied on their cationic headgroups [25]. These results suggested that the apparent cytotoxicity could be controlled by selecting certain steroid hydrophobes/skeletons for the construction of cationic lipid gene carriers.
2H-Cho-Arg/pDNA only showed slight particle size increases (8-15 nm). The results elicited the steroid-based cationic lipids may have different gene transfection properties under serum environments.

Cytotoxicity of the Steroid-Based Cationic Lipids by MTT/CCK-8 Assay
Developing highly biocompatible cationic gene carriers was essential for safe gene delivery/transfection. We evaluated the in vitro cytotoxicity of the steroid-based cationic lipids in H1299 and HeLa cells by MTT and CCK-8 assays with bPEI-25k as the control. As shown in Figure  2a,b, very high cytotoxicity of the bPEI-25k control was observed in both H1299 and HeLa cell lines (IC50 = 28.8 µg/mL in H1299; and IC50 = 23.7 µg/mL in HeLa), due to its strong cell membrane disruption effect [49]. Notably, the steroid-based lipids have different cytotoxicities within the concentration range from 0 to 150 µg/mL. Cho-Arg (IC50 = 88.5 µg/mL in H1299; and IC50 > 150.0 µg/mL in HeLa) and 2H-Cho-Arg (IC50 = 92.7 µg/mL in H1299; and IC50 > 150.0 µg/mL in HeLa) showed moderate cell viability. Dios-Arg showed much higher cytotoxicity (IC50 = 83.5 µg/mL in H1299; and IC50 = 74.1 µg/mL in HeLa), which may be due to the inhibition effect of diosgenin on 3-hydroxy-3-methylglutaryl CoenzymeA (HMG-CoA) reductase, a rate-limiting enzyme involved in cholesterol biosynthesis [50]. Tigo-Arg showed much lower cytotoxicity than the other three lipids in H1299 and HeLa cells, but the mechanism is not clear yet. Similar trends of the cytotoxicity were also observed in COS-7 cell line ( Figure S3). Interestingly, we noticed that there is only a double bond difference between the structures of Dios-Arg and Tigo-Arg, indicating the molecular structure (even a double bond) of the steroid hydrophobes could greatly affect the apparent cytotoxicity. Likewise, Yu et al. revealed that diosgenin-cyclen cationic lipid has a higher cytotoxicity than its cholesterol-cyclen counterpart [51]. In previous work, we have revealed that the apparent cytotoxicity of some cholesterol-based cationic lipids greatly relied on their cationic headgroups [25]. These results suggested that the apparent cytotoxicity could be controlled by selecting certain steroid hydrophobes/skeletons for the construction of cationic lipid gene carriers.

Luciferase Gene Transfection Assay of the Steroid-Based Cationic Lipids in the Absence/Presence of Serum
Luciferase gene transfection efficiency of the steroid-based cationic lipids was evaluated in H1299 and HeLa cells in the presence/absence of serum [18]. The pDNA-loaded lipoplexes were firstly prepared by mixing the steroid-based cationic lipids with pDNA under various +/− charge ratios. Commercially available Lipo2000 and bPEI-25k were used as controls. As shown in Figure  3a,b, Cho-Arg showed the highest luciferase gene transfection capability (maximum: 3.5 × 10 8 RLU in

Luciferase Gene Transfection Assay of the Steroid-Based Cationic Lipids in the Absence/Presence of Serum
Luciferase gene transfection efficiency of the steroid-based cationic lipids was evaluated in H1299 and HeLa cells in the presence/absence of serum [18]. The pDNA-loaded lipoplexes were firstly prepared by mixing the steroid-based cationic lipids with pDNA under various +/− charge ratios. Commercially available Lipo2000 and bPEI-25k were used as controls. As shown in Figure 3a,b, Cho-Arg showed the highest luciferase gene transfection capability (maximum: 3.5 × 10 8 RLU in H1299; and 2.6 × 10 7 RLU in HeLa, ± = 5-10), 1.5-87-fold greater than that of its non-double bond analog 2H-Cho-Arg (maximum: 2.0 × 10 8 RLU in H1299; and 2.9 × 10 5 RLU in HeLa cells). Meanwhile, Dios-Arg (maximum: 8.5 × 10 7 RLU in H1299; and 9.3 × 10 4 RLU in HeLa cells) and Tigo-Arg (7.1 × 10 7 RLU in H1299; and 4.2 × 10 4 RLU in HeLa cells) showed much lower transfection capability. Similar trends were also observed in pEGFP gene transfection assays ( Figure S4). The results indicated that the steroid hydrophobes/skeletons played key roles in determining the gene transfection efficiency. Moreover, it can be noted that all the steroid lipids showed low cytotoxicity (>91.1% cell viability) within the pDNA transfection dose range (about 1.4-29.2 µg/mL, for lipoplexes ± 1-20) in H1299 and HeLa cell lines, indicating the steroid-based cationic lipids could be employed as comparatively safe carrier biomaterials for gene delivery application. For practical application, we further evaluated the Luciferase gene transfection efficiency in the presence of fetal bovine serum (10% FBS) [52]. As shown in Figure 4a,b, the luciferase gene transfection efficiency of Dios-Arg, Tigo-Arg and bPEI-25K drastically decreased in the presence of 10% FBS, while Cho-Arg, 2H-Cho-Arg and Lipo2000 do not show obvious decreasing of the gene transfection efficiency. Moreover, Cho-Arg and 2H-Cho-Arg could maintain high transfection efficiency even under higher serum concentrations (20-50% FBS, Figure 5). In the aforementioned results, Cho-Arg and 2H-Cho-Arg lipoplexes showed higher stability and smaller nanoparticle sizes For practical application, we further evaluated the Luciferase gene transfection efficiency in the presence of fetal bovine serum (10% FBS) [52]. As shown in Figure 4a,b, the luciferase gene transfection efficiency of Dios-Arg, Tigo-Arg and bPEI-25K drastically decreased in the presence of 10% FBS, while Cho-Arg, 2H-Cho-Arg and Lipo2000 do not show obvious decreasing of the gene transfection efficiency. Moreover, Cho-Arg and 2H-Cho-Arg could maintain high transfection efficiency even under higher serum concentrations (20-50% FBS, Figure 5). In the aforementioned results, Cho-Arg and 2H-Cho-Arg lipoplexes showed higher stability and smaller nanoparticle sizes than their Dios-Arg and Tigo-Arg counterparts, which may lead to their higher transfection efficiency. Moreover, Cho-Arg and 2H-Cho-Arg possess great serum-compatibility, similar to that of the multi-component transfection agent Lipo2000 [53] and surface "shielded" cationic polysulfobetaine PDMAEMA copolymers [54]. The results suggested that highly serum-compatible gene transfection could be achieved on single-component lipid gene carriers through molecular engineering approach. Notably, the optimized transfection efficiency of Cho-Arg was much higher than that of bPEI-25k and reached 1/3 of Lipo2000, which make it a potential gene transfection agent for practical application.

Endocytosis Pathway Analysis of the Steroid-Based Cationic Lipids
It had been revealed that the gene carriers/pDNA complexes entered the cell mainly through several endocytosis pathways, including clathrin-mediated endocytosis (inhibitor: Chlorpromazine, CPZ), caveloae-mediated endocytosis (inhibitor: Genistein, GENI), lipid-raft-mediated endocytosis (inhibitor: methyl-β-cyclodextrin, MBCD), micropinocytosis (inhibitor: Amiloride, AMIL) and microtubule-assisted phagocytosis (inhibitor: Nocodazole, NOCO). These pathways are directly relevant to the gene transfection efficiency [32,56,57]. To understand the endocytosis mechanisms of the steroid-based cationic lipids/pDNA lipoplexes, we analyzed the luciferase transfection in the presence of various endocytosis inhibitors, and the luciferase gene expression without addition of endocytosis inhibitors were set as 100%. As shown in Figure 7, the relative luciferase gene expression drastically decreased to 3.3-9.2% in the presence of lipid-raft mediated endocytosis inhibitor methyl-β-cyclodextrin (MBCD) [58], and also decreased to around 3.1-37.8% with the addition of caveolae-mediated endocytosis inhibitor genistein (GENI) [38], whilst the relative luciferase gene expression did not show dramatic decline in the presence of chlorpromazine (clathrin mediated endocytosis inhibitor) [59], amiloride (macropinocytosis inhibitor) [60] and nocodazole (microtubule inhibitor) [61], indicating the steroid-based cationic lipids/pDNA lipoplexes were mainly taken up by H1299 cells via caveolae and lipid-raft mediated pathways, which are regarded as cholesterol-dependent [10,62]. Notably, all the steroid-based cationic lipids/pDNA lipoplexes entered the H1299 cells via similar cholesterol-dependent endocytosis pathways, which suggests that changing the steroid hydrophobes/skeletons in the cationic lipids would not significantly alter the endocytosis pathway of their lipoplexes.

Intracellular Localization of the Steroid-Based Cationic Lipids
To get more insight into the cellular fate, the intracellular localization of the Cy3-pDNA-loaded lipoplexes in H1299 cells were observed by fluorescence microscope, with Lipo2000/Cy3-pDNA as the control. The cell nuclei and lysosome were separately labeled with fluorescent dyes DAPI (blue) and Lysotracker (green). As shown in Figure 8, after 4 h transfection with Cho-Arg/Cy3-pDNA and 2H-Cho-Arg/pDNA lipoplexes, strong red fluorescent spots were found within or around the

Intracellular Localization of the Steroid-Based Cationic Lipids
To get more insight into the cellular fate, the intracellular localization of the Cy3-pDNA-loaded lipoplexes in H1299 cells were observed by fluorescence microscope, with Lipo2000/Cy3-pDNA as the control. The cell nuclei and lysosome were separately labeled with fluorescent dyes DAPI (blue) and Lysotracker (green). As shown in Figure 8, after 4 h transfection with Cho-Arg/Cy3-pDNA and 2H-Cho-Arg/pDNA lipoplexes, strong red fluorescent spots were found within or around the DAPI-stained nucleus and part of them colocalized with the lysotracker-stained lysosome; Lipo2000/Cy3-pDNA lipoplexes showed similar localization effects. Meanwhile, with the transfection of Dios-Arg/Cy3-pDNA and Tigo-Arg/Cy3-pDNA lipoplexes, comparatively weaker red fluorescent spots were observed dispersed around or within the cell nuclei and lysosome, indicating their low cellular uptake capability, which may result in their low gene transfection efficiency. In prior works, we disclosed the lipoplexes of some cholesterol-disulfide cationic lipids (CHOSS) possess perinucleic localization effect [24], which may be due to the cholesterol-induced perinucleic localization. Moreover, we revealed that some (l-)-arginine-rich peptide mimics could enable the nuclei-oriented pDNA localization [38]. Likewise, Ma et al. showed that the "proton sponge" feature of (l-)-arginine could lead to lysosome localization effect [63]. Thus, in this case, the perinucleic, nucleic and lysosome localization effects of the steroid-based cationic lipids/Cy3-pDNA lipoplexes may be inferred as the synergistic contributions of the steroid hydrophobes/skeletons and (l-)-arginine headgroups. Accordingly, we proposed that an intracellular trafficking route of "lipid-raft-mediated endocytosis→lysosome→cell nucleic localization" may be involved in the intracellular trafficking of the lipoplexes. The results suggest that molecular structural factors including steroid hydrophobes/skeletons and (l-)-arginine headgroups could remarkably affect the intracellular localization of the pDNA-loaded lipoplexes, which also elicited a possible way to control the intracellular behaviors/mechanisms of the lipid gene delivery systems via molecular engineering approaches. (l-)-arginine headgroups. Accordingly, we proposed that an intracellular trafficking route of "lipid-raft-mediated endocytosis→lysosome→cell nucleic localization" may be involved in the intracellular trafficking of the lipoplexes. The results suggest that molecular structural factors including steroid hydrophobes/skeletons and (l-)-arginine headgroups could remarkably affect the intracellular localization of the pDNA-loaded lipoplexes, which also elicited a possible way to control the intracellular behaviors/mechanisms of the lipid gene delivery systems via molecular engineering approaches.

Conclusions
In summary, we successfully prepared a series of new steroid-based cationic lipids by incorporation of modular steroid hydrophobes and (l-)-arginine headgroup via facile and efficient synthetic approach. The study on the physico-chemical properties showed that the pDNA binding

Conclusions
In summary, we successfully prepared a series of new steroid-based cationic lipids by incorporation of modular steroid hydrophobes and (l-)-arginine headgroup via facile and efficient synthetic approach. The study on the physico-chemical properties showed that the pDNA binding affinity of the steroid-based cationic lipids, average particle size, ζ potential, morphology and serum stability of the steroid-lipids/pDNA lipoplexes greatly depend on the steroid hydrophobes/skeletons. On the other hand, the biological features such as cytotoxicity, gene transfection efficiency, serum-compatibility and cellular uptake capability also strongly relied on the steroid hydrophobes/skeletons. Among the lipids, Cho-Arg showed remarkably high optimized pDNA transfection efficacy, even under the presence of 50% FBS, making it a latent transfection reagent for in vivo gene transfection. Interestingly, the steroid lipids/pDNA lipoplexes inclined to enter H1299 cells mainly through lipid-raft-mediated endocytosis pathway, following proposed intracellular trafficking routes of "lipid-raft-mediated endocytosis→lysosome escape→cell nucleic localization". Compared to our previous works [24,25,31], this study provides deeper understanding of the structure-function relationships for the steroid-based cationic lipid gene carriers, especially the correlation between the lipid skeletons/structures and their endocytosis pathways and intracellular trafficking/localization, which may offer a new route to further design "endocytosis pathway maintainable/adjustable lipid nanomaterials". Moreover, it also suggests that high-performance steroid-based lipid gene carriers could be achieved by optimizing the steroid hydrophobes/skeletons through molecular engineering approaches.