Direct comparison of two kinds of linoleic acid-docetaxel derivatives: in vitro cytotoxicity and in vivo antitumor activity

Rational designed lipid-drug derivatives provide a favorable approach to improve the druggability of highly hydrophobic prototypes. It has been regarded as common sense that good cytotoxicity is the guarantee of superior anticancer efficacy for candidate derivatives screening. However, does it apply to lipid-drug conjugate-based self-assembled nanoparticles? Here, we established the above two derivatives and a non-correlation between the cytotoxic activity in vitro and drug efficacy in vivo was found. The IC50 of DSL NPs (DTX-S-LA nanoparticles) and DL NPs (DTX-LA nanoparticles) were 4.02 and 209.6 ng/mL (DTX equivalent concentration), respectively. However, DL NPs unexpectedly showed stronger tumor inhibition abilities than DSL NPs. To explain the non-positive correlation between cytotoxicity and anticancer efficacy, more experiments were carried out in depth. Remarkably, the drug release studies in blood and PK study both suggested that the DL NPs were more stable to remain the structural integrity in circulation, which resulted in more accumulation in tumor sites. As verified by the bio-distribution study, DL NPs performed a superior target effect than DSL NPs in tumors. Our data indicated that the biological fates of so-called smart bond inserted derivatives in vivo are complicated; thus, simple cytotoxicity is not enough for derivatives screening, and the comprehensive understanding of both in vitro and in vivo behaviors is essential.


Introduction
Malignant tumor is a deadly threat to human health and life. At present, chemotherapy, supplemented by surgical and radiation treatment, is still the main treatment regimen for cancers [1][2][3][4]. Sadly, the clinical translation and usage of the vast majority of chemical substances have been largely limited due to its' fast elimination and hydrophobicity. Lipid-drug derivatives synthesized by conjugating hydrophobic chemotherapeutic drugs with fatty acid via various linkers have been widely studied to improve the druggability [5][6][7][8][9]. Fatty acid as adjuvant matrix not only produces cytotoxic agents though lipid peroxidation effect [10][11][12][13] but also renders double bonds and structural flexibility to facilitate the transformation of hydrophobic moieties into nanoassemblies. The conjugate nanoassemblies, with high drug loading, are potential treatment paradigms to break the bottleneck of prototype as reported [14][15][16][17][18][19].
For the massive drug candidates, tireless researches had been paid to find a simple favorable method for drug screening. 2D or 3D cultures of immortalized cancer cells have been widely used as primary in vitro tumor models in the high throughput screening of anticancer parent drugs owing to cytotoxicity is generally considered to have a positive relationship with the anticancer efficacy [20][21][22][23]. Nowadays, lipid-drug conjugate-based self-assembled nanoparticles have received increasingly attention, and DHA-paclitaxel have entered clinical research. A series of conjugates have been synthesized by changing lipid species and then screened for efficacy study in vivo through IC50 from cytotoxicity in vitro. Wang et al. [24] established six lipid-drug derivatives and conducted subsequent cytotoxicity assay on several cell lines. After screening by IC 50 , the strongest cytotoxic lipid-drug derivative was chosen to perform the anticancer efficacy.
However, is this cell-based drug screen also predictive for lipid-drug derivatives? Many critical issues which have been neglected lies in that the cytotoxic activity is structure-related and inhibition effect is affected by manifold causes [25,26]. Some previous literatures may support this view [22,27,28]. Ansell et al. [29] synthesized nine paclitaxel derivatives with succinate or diglycolate cross-linkers. He found that the efficacy correlated well with the PK profiles rather than cytotoxicity. The greatest antitumor response in vivo was not seen in the derivative with the lowest IC 50 . Apart from this, our group previously loaded disulfide bond and mono thioether bond insertions (DTX-ss-VE and DTX-s-VE) into liposomes. Though similar IC 50 , these two liposomes possessed completely different inhibition effects in vivo. The anticancer efficacy of DTX-ss-VE loaded liposomes was comparable with DTX solutions, while there is no inhibition effect was observed in DTX-s-VE loaded liposome group [30].
Although the relationship between cytotoxicity and anticancer efficacy especially for derivatives is of great importance, misuse of this relationship would bring about inappropriate drug identification and failure clinical trials. Thus, solving this doubt is conducive to more appropriate experimental design and data interpretation, and thereby perfecting the drug selection during preclinical phase. The increased investment in preclinical analysis could remarkably promote the effective bench-to-bed translation and largely reduce the rate of attrition of drugs in clinical settings. This paper makes a systematic comparative study of cytotoxicity in vitro and efficacy in vivo of lipid-drug conjugates compounds for the first time, so as to provide a reference for the development of lipid-drug conjugate-based nanoassemblies. In this work, the underlying relationship between cytotoxicity and anticancer efficacy of two kinds of lipid-drug derivatives, which utilized ester bond and mono thioether bond to conjugate DTX and linoleic acid (termed DTX-LA and DTX-S-LA), was thoroughly investigated. Conjugate-induced self-assemble process was used to fabricate nanoassemblies. Inhibition abilities were evaluated in both cellular and animal to investigate the correlation in vitro-in vivo behaviors. The release behaviors, PK, and bio-distribution study were further investigated to provide interpretation in depth. (RPMI-1640), trypsin, 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT), and fetal bovine serum (FBS) were purchased from Dalian Meilun Biotechnology Co., Ltd, China. DTX-LA and DTX-S-LA were synthesized and characterized by former work in our lab [14,31].

Preparation of DL NPS and DSL NPS
One-step precipitation method was used to prepare the nanoassemblies. In short, 4 mg of DTX-LA or DTX-S-LA and 20% (w/w) DSPE-PEG 2K were accurately weighed and dissolved in 100 μL ethanol. Then this miscible solvent containing formulation components was cautiously added dropwise into 2 mL deionized water and continuously stirred for 2 min (800 rpm, K-MSH-Pro-6A, JKI, Shanghai, China). Apply vacuum-rotary evaporation procedure for almost 10 min to remove ethanol, and finally, volume with deionized water to 2 mL. Particle size and polydispersity index (PDI) of conjugate NPs were measured by a Zetasizer (Nano ZS, Malvern, UK) in triplicate.

Cell lines and cell culture
The murine breast cancer cell line (4T1) was bought from the cell bank of Chinese Academy of Medical Sciences (Beijing, China). 4T1 cells were cultured in RPMI-1640 medium supplemented with 10% FBS, penicillin (30 mg/L), and streptomycin (100 mg/L) in a humid atmosphere containing 5% CO 2 at 37 °C.

In vitro cell viability assay
To explore whether conjugate NPs could effectively inhibit tumor cells' growth in vitro, and more importantly, to provide guidance for in vivo experiments, MTT assay was performed in 4T1 cells. A certain density of 4T1 cells (1000 cells/100 μL/well) were incubated in 96-well plates for 12 h to allow cell attachment. Then fresh medium containing a series of concentrations of DTX solution, DL NPs, and DSL NPs was added to each well to replace old RPMI-1640 medium. Cultivated for another 48 h or 72 h, the drug-contained medium was replaced by 100-μL fresh medium and 20-μL MTT solutions (5 mg/mL) which would be discarded after incubating for 4 h. A total of 200 μL DMSO was added to each well to dissolve the formazan. The absorbance value of each hole in 96-well plates at 570 nm was selected for measurement on a microplate reader (Model500, USA). Equation 1 was utilized to calculate the inhibition rate. And the half maximal inhibitory concentrations (IC 50 ) were evaluated by nonlinear regression analysis.

In vivo antitumor efficacy study
To further explore whether these conjugate nanoassemblies could improve the therapeutic index of DTX in vivo, anticancer efficacy study should be performed. This study was performed in the female BALB/c mice weighed 18-22 g. In short, the 4T1 cells, suspended in PBS (5 × 10 6 cells per mouse), were subcutaneously injected into the right auxiliary flank of mice to build tumor-bearing mice models. The mice were divided evenly into 4 groups (n = 5) when the tumor volumes reached almost 120-150 mm 3 . Then each group of mice were subjected to treat every 2 days with saline, Home-made Taxotere (10 mg/kg), DL NPs (10 mg/ kg, DTX equivalence), and DSL NPs (10 mg/kg, DTX equivalence) via tail vein injection, respectively. The tumor volumes and body weights were monitored and recorded every 2 days. Tumor volumes were calculated by Eq. 2. After 4 times administration followed by 2-day observation, the mice were sacrificed to collect its' tumors and calculate the tumor burden after the last treatment by Eq. 3.

Physical stability and drug release behavior studies
The physical stability of nanoassemblies was carried out at 4 °C for 3 months. The particle size and PDI were determined on a specific time point to monitor the variations. The drug release behavior of DL NPs and DSL NPs was performed in plasma. Briefly, the certain concentration of NPs solution was incubated with mice plasma in an air (a represents the shortest width and b represents the longest length) bath (CHA-S, Guohua Electric Applance Co., Ltd. Jiangsu, China) with the shaking of 100 rpm at 37 °C. Plasma samples, at pre-determined time interval, were withdrawn and extracted by acetonitrile for the sedimentation of protein.

Pharmacokinetic properties study
The pharmacokinetic (PK) experiment was carried out on Sprague-Dawley (SD) rats weighed 200-220 g which were divided into 3 groups (n = 5). Administrated with a single intravenous injection of Home-made Taxotere, DL NPs, and DSL NPs to deliver a DTX equivalent dose of 5 mg/kg, blood samples were withdrawn at pre-determined time intervals via orbital venous plexus. The plasma was collected by centrifugation at 3000 rpm for 10 min. Precipitation of protein method was applied to extract the drugs. The quantitative analysis was assessed by HPLC-MS/MS with C18 column (100 mm × 2.1 mm, 5 μm). Acetonitrile/water (95:5, v/v) as mobile phase at 0.2 mL/min was used to analyze DTX-LA and DTX-S-LA. The concentration of DTX was determined by elution: 0-0.5 min, 70% water; 0.51-2.5, 5% water; and 2.6-3.0, 70% water.

Bio-distribution study
BALB/c mice bearing 4T1 malignant tumor were used to fulfill biodistribution study. The mice model was built according to antitumor efficacy study's method. The mice were randomly divided into 3 groups (n = 6) and treated with Home-made Taxotere (10 mg/kg), DL NPs (10 mg/kg, DTX equivalence), and DSL NPs (10 mg/kg, DTX equivalence) via tail vein injection, respectively. After four and twenty-four hours postinjection, three mice of each group were sacrificed to harvest major organs (heart, liver, spleen, lung, and kidney) and tumors. Then the free DTX in tissue homogenate were quantified by HPLC-MS/MS on an ACQUITY UPLC system (Waters Corp). The methods of sample extraction and quantification were in accordance with that of PK study as mentioned above.

Preparation of conjugate NPs
One-step precipitation method was used to prepare the nanoassemblies (Scheme 1). The particle size confirmed the feasibility of self-assembled nanoparticles. As predicated, once these highly hydrophobic derivatives added to water, self-assemble process would occur to yield an ordered nanoprecipitation. As shown in Fig. 1, the average diameter of DL NPs and DSL NPs were both about 100 nm and the PDI was always below 0.2. The DL NPs and DSL NPs had spherical shapes with a uniform size of 100 nm.

In vitro cell viability assay
DTX solution, DL NPs and DSL NPs were examined against 4T1 cell lines to determine the cytotoxicity. As shown in Fig. 2, more cancerous cells were inhibited as the incubation time prolonged from 48 to 72 h suggesting the conjugate NPs had time-dependent inhibition capacity. This could be attributed to that more exposure time was needed to release the active parent drug. Notably, DL NPs displayed the lowest toxicity against the 4T1 cell lines. The IC 50 of DSL NPs and DL NPs at 48 h were 4.02 and 209.6 ng/mL, respectively.

In vivo antitumor efficacy study
Anticancer ability study was conducted in 4T1 tumor bearing mice. The BALB/c mice were treated with Home-made Taxotere, DL NPs, and DSL NPs every 2 days (Fig. 3A). Taxotere is the commercial formulation of DTX which utilized Tween 80 and 13% (w/w, ethanol/ water) solution of ethanol as solvents. Surprisingly, despite of 50-fold difference of cytotoxicity, the tumor volumes of DL NPs group were even smaller than that of DSL NPs at the same dose level (10 mg/kg, DTX equivalence). After the last treatment, the average tumor volumes of Home-made Taxotere, DL NPs, and DSL NPs groups were about 338 ± 65, 245 ± 42, and 329 ± 41 mm 3 , respectively. There was an extremely significant difference between saline group and DL NPs group (Fig. 3B). Besides, as shown in the Fig. 3C, the last 4 days witnessed a notable side effects in the group of Home-made Taxotere with the body weight slumped about 7.8%, whereas that of the conjugate NPs groups maintained the same level off at 20 g. The tumor burden after the last treatment of Home-made Taxotere, DL NPs, and DSL NPs were about 0.98%, 0.44%, and 0.77%, respectively (Fig. 3D). All considered, the outcomes of DL NPs far exceeded any expectations.  D Tumor burden after the last treatment. *P < 0.05, **P < 0.01, and ***P < 0.001 between the groups indicated Physical stability and drug release behavior studies DL NPs and DSL NPs as potential high drug loading nanotherapeutics, maintaining the integrity of preparation and molecular structure, were the prerequisites for them to have a strong anticancer efficacy. Thus, to explore the reasons of non-positive correlation between cytotoxicity and anticancer efficacy, it was a must to study the stability of conjugate nanoparticles. As shown in the physical stability curve (Fig. 4), there were no any significant variations in particle size and PDI implying conjugate NPs could remain considerably stable within 3 months in deionized water at 4 °C. To further explore the release behavior under more physiological conditions, the conjugate nanoassemblies were placed in plasma samples at 37 °C with continuously shaking. Figure 5A illustrates that more than 80% DTX-S-LA conjugates were hydrolyzed within 24 h in plasma, whereas only about 30% DTX-LA degraded in the same medium. Additionally, < 5% DTX was converted from DTX-LA conjugates. Based on the values, it was clear that DL NPs had significantly slower release behavior than DSL NPs when incubated with blood samples (Fig. 5B) . This different release behavior might ascribe to the mono thiother bond inserted conjugates that were more susceptible to hydrolysis leading to less stability than ester linked conjugates [32].

Pharmacokinetic properties study
Derivatives and preparation strategies would dramatically alter the PK profiles in circulation. PK parameters could be suitable values to explore the internal metabolism and provide useful information and reference for explaining the raised questions. The PK study was carried out on SD rats. As shown in the drug concentration-time curve (Fig. 6) and Table 1, the PK profiles could be largely improved by conjugate NPs. The t 1/2 of Home-made Taxotere, DTX-LA of DL NPs, and DTX-S-LA of DSL NPs were 3.21, 4.03, and 4.01 h, respectively. The t 1/2 had been enlarged 1.3 times by lipid-drug derivative nanoparticles. Remarkably, the AUC of DL NPs (DTX-LA) even achieved 2 times higher than that of DSL NPs (DTX-S-LA). This value was crucial important because the pharmacokinetics had a tremendous impact on drug exposure in tumors. Notably, it was not surprising that the proportion of DTX derived from DL NPs was considerably less than that of DSL NPs, and this phenomenon was consistent with the results of drug release experiment. The more unstable structure of DTX-S-LA might be the reason that caused the premature DTX release from vesicle in circulation.
Apart from that, the AUC of DTX derived from DSL NPs was 3 times higher than that of Home-made Taxotere, but higher AUC did not always mean better anticancer efficacy. Actually, there was no difference in anticancer efficacy between DSL NPs and Home-made Taxotere. This might partly due to the type of surfactant composition. Herein, the concentration of DTX measured contained the unencapsulated drug form fraction and encapsulated form fraction, and the unencapsulated drug form also contained free drug form and plasma protein bonded form. It should be kept in mind that only free form of DTX was bioactive. However, it was very difficult to distinguish from one another because the three forms are in dynamic equilibrium. Besides, some reports clarified that the surfactants were biologically and pharmacologically active and thereby indeed had an influence on the PK profiles or cell uptake [32,33]. In fact, the administration mode of DTX was different, Homemade Taxotere was a bolus injection, while for DSL NPs and DL NPs, the DTX was released from DTX-linoleic Fig. 4 Long-term stability of DL NPs (A) and DSL NPs (B) after store at 4 °C for 90 days acid derivatives at a different rate during the circulation. All above-mentioned reasons complicated the correlation between pharmacokinetic parameters and the efficacy; these need more in-depth investigation in the future.

Bio-distribution study
Considering the biodistribution behavior had an influence on anticancer efficacy and safety profiles, the distribution and bioactivation behavior of conjugate NPs were investigated in BALB/c mice. The highest DTX concentration in tumor tissues belonged to mice of DL NPs group (Fig. 7), suggesting DL NPs could successfully accumulate in tumor sites through passive targeting and get activated by esterases in the tumors. By contrast, the leakage of DSL NPs in blood circulation was adverse to delivery to the tumor tissues, and the fraction of DTX of DSL NPs in tumor was significantly less than that of DL NPs. For Home-made Taxotere, a considerable DTX disposition in spleen and lung was found both at 4 h and 24 h post injection.

Discussion
DTX tethered with unsaturated fatty acids had the ability to self-assemble into nanoprecipitations in water. Self-carrier nanoparticles yielded remarkably improved drug loading (61.9% for DL NPs and 53.4% for DSL NPs). DSPE-PEG 2K as an amphipathic stabilizer could forbid the adsorption of serum proteins enabling NPs remain stable in bloodstream. Nanostructures without Tween 80 significantly alleviated excipient-induced toxicity with reserving therapeutic efficacy.
Cytotoxic model was a common method for screening anti-tumor agents. Some previous studies [20][21][22][23] also indicated that the lower IC 50 , the better therapeutic efficacy. In this work, however, different results were found. In spite of, inferior cytotoxicity, DL NPs exhibited unexpected slightly better anticancer efficacy when compared with Homemade Taxotere and DSL NPs. This non-positive correlation between cytotoxicity in vitro and anticancer efficacy in vivo could be explained by manifold reasons as below.
The release behavior of conjugate nanoassemblies was performed in plasma sample mediums. The release kinetics revealed that DL NPs owned much more delayed release behavior suggesting ester bond linked conjugates were considerably more stable than redox sensitive conjugates. That is probably because the mono thioether bond could be easily oxidized to electron-withdrawing sulfone or sulfoxide bearing ROS or esterases, and triggered release [34]. Being stable in extra-cellular environment was important for maintaining the structural integrity. A rapid drug loss was occurred in the group of DSL NPs when incubated with rat plasma indicating the molecule structure was not stable enough to resist esterases or ROS species in plasma. The relatively unstable structure of DTX-S-LA would reduce the drug exposure in tumor sites. Uncontrolled drug release upon administration would prevent a conjugate candidate from exerting its therapeutic activity. This phenomenon might partly explain the limited antitumor response of DSL NPs.
The PK study showed that Home-made Taxotere had the shortest t 1/2 indicating conjugate nanoassemblies could prolong the PK behavior. DL NPs had the higher AUC (DTX-LA), and less DTX release compared with DSL NPs which evidenced the ester bond inserted conjugates were more stable in circulation. The phenomenon was consistent with the above release studies in vitro. DL NPs served as a reservoir to inhibit the premature release of parent drug during systemic circulation and thereby would definitely promote the tumor accumulation. And the results of tissue distribution had provided valuable support to this inference.
In conclusion, the relationships between the release profiles, in vitro cytotoxicity, and in vivo antitumor efficacy were of great importance in the lipid-drug selection. The intermediate linker played a vital important role in the results of drug conversion rate. The prodrug-like conjugates should remain structure intact in circulation but get rapid released in tumor sites. Immature release happened in the process of in vivo delivery could lead to inadequate drug exposure in tumor sites. In other words, this mono thioether bond linked conjugates were much more vulnerable to hydrolyze than DTX-LA in cytotoxicity assays in vitro. Thus, the cell inhibition capacity of DSL NPs was much stronger in comparison to that of DL NPs in vitro. However, tumor microenvironment was a complex system involving different cellular and non-cellular elements.
The cytotoxicity experiments which used growing cells in coated, two-dimensional Petri dishes and glass slides were limited to reproduce the in vivo conditions of tumor system. Varied enzymes and lower pH in tumor microenvironment would promote the hydrolysis of ester bonds. Accordingly, more active release could happen in tumor site rather than monotonous adherent cells in vitro. Besides, higher AUC and passive target significantly prompted more DL NPs to accumulate in tumor sites which would certainly contribute to the good anticancer efficacy. However, the MTT assay neglected the circulation process thereby further causing this discrepancy between cytotoxicity and anticancer efficacy. The in vivo disposition was closely interrelated with safety profiles. Poor bio-tolerance was found in the group of Home-made Taxotere with steadfast plunge of body weight. The poor potency of Taxotere to accumulate in tumor sites could explain the off-target toxicity. Compared with the group of Taxotere, the safety of DL NPs and DSL NPs had significantly improved. Drastically alleviated toxicity could be explained by the fact that not only excipient-associated side effects had been avoided, but also the tissue distribution behavior had been changed by the NPs. Additionally, this higher safety profiles further promised a dose escalation for better anticancer efficacy. (Supplementary material)

Conclusions
In this study, two kinds of lipid-drug derivative nanoassemblies had been constructed through the expedient fabrication process. The high drug loading had been considered to be the most notable merits of conjugate NPs. The nanoassemblies, with very little or no exogenous excipients, provided a possibility to mediate drug delivery in a tumor tissuespecific manner and enhance anticancer effect of prototype. Reduced from potential carriers-induced toxicity, conjugate NPs had excellent detoxification effect. This nanoplatform for cancer therapy was an invaluable reference for designing more potent drugs for clinical use.
This study would also help researchers increasingly aware that cytotoxic activity was structure-related. For lipid-drug derivatives, no directly positive correlation existed between the cytotoxicity in vitro and anticancer efficacy in vivo. Even small changes in the structure of drugs or formulations might cause huge alternation behavior in vivo. The results of experiments in vitro could not fully represent or replace in vivo experiments. As for the principle of linker design and drug selection, more related studies should be carried out rather than cytotoxicity alone.