Conjugation of squalene to gemcitabine as unique approach exploiting endogenous lipoproteins for drug delivery

Once introduced in the organism, the interaction of nanoparticles with various biomolecules strongly impacts their fate. Here we show that nanoparticles made of the squalene derivative of gemcitabine (SQGem) interact with lipoproteins (LPs), indirectly enabling the targeting of cancer cells with high LP receptors expression. In vitro and in vivo experiments reveal preeminent affinity of the squalene-gemcitabine bioconjugates towards LP particles with the highest cholesterol content and in silico simulations further display their incorporation into the hydrophobic core of LPs. To the best of our knowledge, the use of squalene to induce drug insertion into LPs for indirect cancer cell targeting is a novel concept in drug delivery. Interestingly, not only SQGem but also other squalene derivatives interact similarly with lipoproteins while such interaction is not observed with liposomes. The conjugation to squalene represents a versatile platform that would enable efficient drug delivery by simply exploiting endogenous lipoproteins.

1. Lack of originality innovation. The SQ-Gem conjugate and SQ-Gem NPs has been previously reported (Journal of controlled release, 2010, 147(2): 163-170; Nanomedicine: Nanotechnology, Biology and Medicine, 2011, 7(6): 841-849;Chemical Communications, 2014, 50(40): 5336-5338.). The concept of lipopoproteins involved nanoparticles for potentially targeting tumor cells cannot be considered as a novel drug delivery strategy, which is the inherent advantage of SQ-Gem conjugate and SQ-Gem NPs. This paper is a follow-up study of SQ-Gem NPs, and is more related to the pharmacokinetics of SQ-Gem NPs after intravenous administration, e.g. the distribution of SQGem and Gem in human plasma fractions.
2. The SQ-Gem NPs and 3H-SQ-Gem NPs were not well characterized, for example, the data related to the particle size, size distribution and surface charge status are missing. How will lipopoproteins affect the nanostructure of NPs? e.g. the TEM images of SQ-Gem NPs and lipopoproteins-SQ-Gem NPs. There is only speculation, as the results of in silico simulations. Some aspects of this seem nonintuitive to me.
3. The authors concluded that squalene was chosen because of the lipid nature and its structural similarity with cholesterol, which have good affinity with lipoproteins. Whether cholesterol is a better choice to conjugate with gemcitabine? 4. It is not clear to me what is meant by " In this study, a protein-driven dissociation of SQ-Gem nanoparticles into SQ-Gem monomers was demonstrated to occur before cell capture" (line 267 and 268, p 13). Please provide convincing interpretations, maybe with more evidence. If the SQ-Gem NPs were dissociated before cell uptake, how can lipopoproteins facilitate tumor cell targeting? In addition, if the SQ-Gem NPs will be dissociated in blood, what's the advantage of SQ-Gem NPs over SQ-Gem conjugates? Will the PEGylated SQ-Gem NPs also be dissociated in blood? If not, which one is a better formulation for drug delivery, PEGylated SQ-Gem NPs or non-PEGylated SQ-Gem NPs? 5. The organ distribution and the systematic pharmacokinetics profile of SQ-Gem NPs should be examined. How will lipopoproteins in vivo affect the biodistribution in organs and the pharmacokinetics of SQ-Gem NPs?
6. The authors conclude that "It was discovered that endogenous LDL particles may function as carriers for SQ-Gem, thus allowing the indirect targeting of cancer cells displaying high expression and activity of LDL receptors, without the need to functionalize NPs surface with hydrophilic PEG (polyethylene glycol) chains and/or with specific ligands." Actually, PEGylation cannot facilitate the cellular uptake of NPs, even hinder the cellular uptake, but benefit the long circulation in blood. Moreover, the authors didn't set the PEGylated SQ-Gem NPs as a control. How can they draw this conclusion? 7. The manuscript should be carefully checked for grammar mistakes. 92296 Châtenay-Malabry cedex -France 2 The authors thank the reviewer for the positive comments.

Reviewer #2:
When nanoparticles (NPs) systemically administrated into an organism they will immediately interact with variety of biological molecules (proteins, lipids and lipoproteins), which might influence their biological fate and clearance mechanisms. In this manuscript, the authors show that, after i.v. administration, NPs made of the squalene derivative of the gemcitabine (SQGem) strongly interacted with lipopoproteins (LPs), which indirectly enabled targeting of cancer cells with high LP receptors expression. The manuscript is well written and interesting however, the claims should be supported by the data. In this version of the manuscript they are still claims without strong, comprehensive evidence for the suggested mechanism. The concept of designing SQ NPs is not novel and this strong group has published elegant papers in this field including a high profile Nature Nanotech paper a few years ago. Here, they claim that they have insights into the mechanism by which these NPs are taken up and home to cancer cells highly expressing LP receptors. The work is fairly solid but there is lack of robust, killer experiments to show directly the authors claims. If the authors can address the concerns below the paper should be published in Nature Comm.
Major concerns that needs to be address prior to publication: 1. The particles bind to HDL in the serum according to the radioactive data. This could be misleading, since the NPs has lipids properties can can be easily found in this fraction anyhow. The authors need to demonstrate selectivity by other means and specificity. Answer: As suggested by the reviewer additional experiments have been performed to further demonstrate the specific interaction between SQGem and the lipoproteins. First, to check if NPs can be (or not) easily found in the same fraction anyhow, we have carried out a control ultracentrifugation in NaBr density gradient. Briefly, 3 H-SQGem NPs have been added to a solution of 1.25 g mL -1 NaBr and incubated at 37 °C for 5 minutes. Then, 1 mL of this mixture has been placed at the bottom of a centrifuge tube (exactly in the same conditions than with plasma collected from rats 5 minutes post iv administration of 3 H-SQGem NPs). It was clearly observed that NPs were not stable in the 1.25 g mL -1 NaBr solution since they rapidly formed a white precipitate. Following the ultracentrifugation, this precipitate localized at the level of the fraction 12. On the contrary, when the experiment was performed with plasma collected after intravenous administration of the 3 H-SQGem NPs to rats, the peak of radioactivity corresponded to the fractions 8-11, containing also the highest amount of cholesterol and precipitates were never observed. (Revised manuscript: Figure 4, page 11). These results highlight that the localization of the radioactivity in the HDL fractions is the result of a specific interaction and did not result from an artifact due to NPs density.
The specificity of the interaction between SQGem NPs and LDL was then confirmed by isothermal titration calorimetry experiments performed with human lipoproteins separated from the blood of healthy volunteers. The heat flows were determined when SQGem NPs were added to LDL dispersion, HDL dispersion or albumin solution placed in the titration cell. ITC thermograms revealed the existence of a strong interaction between the SQGem and the LDL, while it was not observed either with HDL or with albumin. The specificity of the recorded signal was confirmed after dilution of SQGem NPs in PBS ( Supplementary Fig. S8). These results are in agreement with those obtained after incubation of the SQGem NPs with human blood (Revised Manuscript: Figure 1, page 8) and provide a further proof of the existence of a specific interaction with the human LDL fraction.

In the in vitro experiment -it is better to selectively completely knockout the HDL receptor (or comprehensively silence it) and show it is really the mechanism behind this observation.
Answer: All the in vitro cell culture experiments have been performed on human cancer cells and not in rodent cancer cells. Since LDLR is involved in cholesterol uptake in humans and to meet the reviewer recommendation, we have looked for LDLR knockout human cells. Unfortunately, we failed to obtain cells displaying complete inhibition of LDL receptor expression. In addition, it was mandatory to have a cell line allowing relevant comparison with MDA-MB-231 used in this article. Nevertheless, with the aim to further support the hypothesis, we have performed a supplementary experiment and have investigated the contribution of the LDLR in the uptake of 3 H-SQGem NPs on MDA-MB-231 and MCF-7 breast cancer cells, since those cells share similar features (they are both human breast adenocarcinoma cells) but display high and low level of LDLR expression, respectively ( Supplementary Fig. S16). After 6 h of incubation with 3 H-SQGem NPs at 37 °C, a higher radioactivity signal was detected in MDA-MB-231 cells (high expression of LDLR) compared to the MCF-7 ones (low expression of LDLR) ( Supplementary Fig. S21). These results showed that a higher LDLR expression induced greater uptake which confirmed the key role of the LDLR in the cell internalization of the SQGem bioconjugates. 3. Many of the experiments are not "strong enough to state the current statement. It is wised to perform a comprehensive proteomic analysis to comprehensively demonstrate that HDL are directly interacting with the SQ NPs. Answer: As suggested by the reviewer, we have performed proteomic analysis to identify and quantify the proteins adsorbed at the nanoparticles surface. The commonly used experimental setting requires the incubation of NPs with plasma, several washing steps and finally, the desorption of the proteins by treatment with dithiothreitol and SDS solution (2 h, 60 °C). However, when this protocol has been applied to the SQGem NPs (n=2) no trace of protein corona was detected (see lines 2 and 5 in the figure below, which clearly showed the absence of protein corona after incubation of SQGem NPs with rat plasma. Lines 3 and 6 correspond to the total plasma proteins, while lines 1 and 4 correspond to the HDL fraction) 5, rue Jean-Baptiste Clément 92296 Châtenay-Malabry cedex -France 5 These results are in agreement with our previous conclusion that the squalene moiety drives the insertion of SQGem molecules into LDL, which supposes that SQGem NPs undergo first a quick disassembly in plasma. As a result of the NPs loss of integrity, no stable protein adsorption at the surface of the SQGem NPs with subsequent formation of a protein "corona" could occur. Thus, we have further investigated the disassembly of the SQGem NPs in blood by performing new experiments with FRET SQGem NPs. These NPs have been prepared by labelling SQGem NPs with the squalene derivatives of the cyanine 5.5 (SQCy5.5, 0.6% w/w) and the cyanine 7.5 (SQCy7.5, 0.6% w/w), which behave as FRET donor and acceptor, respectively. For details concerning the preparation of these FRET nanoparticles, see Supplementary Information pages 13-14. Fluorescence emission spectrum of these FRET NPs is reported in Supplementary Fig. S12. Taking advantage of the dependence of FRET signal on the distance between the FRET pair (SQCy5.5/SQCy7.5), this approach enabled to monitor the integrity of SQGem NPs over time. Stability has been evaluated at 37 °C after opportune dilution of NPs in (i) water, (ii) rat blood and (iii) ethanol where SQGem is soluble. Thus, the dilution in ethanol has been used as a control of the complete disassembly of FRET SQGem NPs and consequently absence of energy transfer between the donor and acceptor dyes. Samples were imaged at regular times post incubation using the IVIS Lumina imaging system. After excitation at 640 nm (SQCy5.5 excitation wavelength), the emission was collected at 695-770 nm (donor signal, D) and 810-875 nm (FRET signal, A). Following signal quantification, the FRET proximity ratio, providing a semiquantitative measurement of the FRET efficiency has been calculated according to the equation A/(A+D).(Preus, S. & Wilhelmsson, L. M.. Chembiochem, 2012Chembiochem, , 13, 1990Chembiochem, -2001 While NPs were stable in water up to 24 h, the rapid drop of the FRET signal in blood clearly indicated a fast disassembly of the NPs in this medium. (Supplementary Fig. S13). These results explain the absence of the protein corona after proteomic analysis and support the hypothesis that the interaction between SQGem NPs and lipoproteins involves SQGem molecules and not intact nanoparticles.
Finally, we have investigated the interaction between SQGem NPs and lipoproteins by transmission electron microscopy. Briefly, SQGem NPs have been incubated with human LDL for 5 minutes at 37 °C and then images of this physical mixture have been acquired. SQGem NPs only and LDL particles only have been used as controls. In the physical mixture, shape modified SQGem NPs, possibly in the course of a disassembled process (yellow arrows), and LDL particles (pink arrows) are visible but also SQGem NPs displaying lipoproteins on their surface (blue arrows).  Fig. S15) This microscopy image reveals how the first encounter between SQGem NPs and LDL occurs. Of note, it provides only a static image, while the interaction between the SQGem and lipoproteins is rather a dynamic phenomenon. Indeed, as shown in the FRET experiments detailed above, the establishment of this preliminary interaction is followed by the disassembly of the SQGem NPs and by the release of single SQGem bioconjugates that will further insert into the lipoproteins (which is obviously not observable by TEM).

It is also important to understand that other types of NPs made from other materials may not interact the same. Therefore and in order to generalize the concept -it is important to use as control at least two other types of NPs made from different materials and show that they also interact in the same manner.
Answer: As suggested by the reviewer additional experiments have been performed in order to demonstrate that (i) the interaction with the lipoproteins is a feature common to different squalene-based nanoparticles and not restricted to SQGem NPs and that (ii) the lipidic nature of a drug nanocarrier does not imply the capacity to establish an interaction with circulating lipoproteins.
Thus, with the aim of generalizing the proposed approach, we have investigated the behavior of nanoparticles made by the self-assembly of two other squalene derivatives: 3 H-Squalene-adenosine ( 3 H-SQAd) NPs (chosen because of the already demonstrated pharmacological activity of SQAd in stroke, see Gaudin et al., Nature Nanotechnology, 2014) and Squalene-Cy5.5 (SQCy5.5) NPs (chosen because this squalene conjugate allowed another analytical detection, ie. fluorescence). The interaction with lipoproteins of these two structurally different squalene derivatives has been determined by using the same experimental setting already established for SQGem NPs. Thus, 3 H-SQAd NPs and SQCy5.5 NPs were intravenously injected to rats and their distribution among plasma fractions was assessed. First, it was observed that the distribution profile of 3 H-SQAd NPs perfectly overlapped that already observed after administration of SQGem NPs, and the highest percentage of 3 H-SQAd has been found in the HDL fraction (63%), followed by the LDL (15%) and the VLDL (7%) fractions. Only 15% of the 3 H-SQAd distributed in the LPDF fraction. Similarly to the free gemcitabine, also the free adenosine did not interact significantly with the lipoproteins and was mainly recovered at the bottom of the tube, in the LPDF. As already observed with free gemcitabine ( Supplementary Fig. S7), this localization could be attributed to the physico-chemical properties of the molecule, rather than to interactions with soluble plasma proteins. Secondly, the administration of SQCy5.5 NPs resulted also in comparable distribution among the plasma fractions, thus confirming that the conjugation to squalene led to a specific interaction with lipoproteins. In this case, due to detection limits, quantification has been carried out in the pool of fractions corresponding to LDL/VLDL (fractions 1-6), HDL (fractions 7-13) and LPDF (fractions 14-20). The highest % of SQCy5.5 (expressed as relative to the concentration in the whole plasma) has been found again in the HDL fractions (50%). (Supplementary Fig. S10 On the whole, these results clearly demonstrated that the interaction with the HDL fraction, the main cholesterol transporting particles in rats, was not exclusive of SQGem but represented a more general concept common to different squalene derivatives, independently of the molecule conjugated to the squalene. Nevertheless, as suggested by the reviewer, we additionally investigated whether such a specific interaction with the major cholesterol transporting particles in rats could also occur (or not) with other types of nanocarriers. Thus, we have prepared 3 H-Gemcitabine loaded liposomes ( 3 H-LipoGem) and we have administered them intravenously to rats. Following the same experimental protocol as already described for 3 H-SQGem NPs, 3 H-SQAd NPs and SQCy5.5 NPs, blood has been collected, five minutes after injection, by cardiac puncture and the different plasma fractions separated by an ultracentrifugation in NaBr density gradients. The majority of the radioactivity, expressed as relative to the total amount in plasma, was found in the fractions 5-8, which could a priori correspond to the HDL1 subpopulation, also suggesting an interaction of liposomes with this subpopulation. However, very importantly, a control experiment carried out by replacing the plasma with a 1.25 g mL -1 NaBr solution incubated for 5 minutes with 3 H-GemLipo, revealed that 3 H-GemLipo accumulated in the same 5-8 fractions even in the absence of any lipoprotein. In addition, it was observed that, contrarily to the different squalene derivatives, the LipoGem only partly overlapped with the cholesterol content in the different fractions. Accordingly, it was concluded that the distribution of 3 H-GemLipo in fractions 5-8 simply resulted from their accumulation in the region of the gradient which displayed the same density than the liposomal formulation and was not a consequence of the interaction with the lipoproteins. (Supplementary Fig. S11). In a nutshell, all these additional experiments confirmed that the interaction of the squalene bioconjugates with main cholesterol-transporting lipoproteins in rats is a generic concept driven by the squalene moiety by virtue of its biorelation to cholesterol and is not simply dependent on the lipidic nature of the nanocarrier.
The authors would like to thank the reviewer for his/her interesting comments which have allowed to improve the manuscript.

The authors have carried out and report here a study of the use of squalene to induce drug insertion into LPs for cancer cell targeting. After intravenous administration, SQ-Gem nanoparticles strongly interacted with lipopoproteins, which may facilitate active targeting due to high LP receptors expression in tumor cells. The topic is
interesting. However, the whole study is not well designed and several major points were missed. The paper is premature for publication in Nature Communications. I recommend this paper should be submitted to a more specific journal. And several major concerns need to be addressed before resubmission. 1. Lack of originality innovation. The SQ-Gem conjugate and SQ-Gem NPs has been previously reported (Journal of controlled release, 2010, 147(2): 163-170;Nanomedicine: Nanotechnology, Biology and Medicine, 2011, 7(6): 841-849;Chemical Communications, 2014, 50(40): 5336-5338.)

5, rue Jean-Baptiste Clément 92296 Châtenay-Malabry cedex -France 11
Answer: It is the opinion of the authors that the submitted manuscript does not represent a simple additional study on SQGem NPs. We believe that this article will have a higher impact, because it demonstrates, for the first time, the use of a squalene moiety for driving drug insertion into lipoproteins, which are thus exploited as indirect natural drug carriers. Importantly, this interaction occurs spontaneously, without the need of complex LDL isolation or recombinant LDL synthesis and/or drug loading. This approach represents therefore a novel concept in drug delivery. The additional experiments performed in this revised version of the manuscript further increase the strength of the proposed approach, as the covalent coupling to squalene of other molecules, with therapeutic or imaging activities (i.e., adenosine, NIR dye), endows the resulting bioconjugates with the capacity to interact with the lipoproteins too. The ability of squalene to mediate the insertion into lipoproteins (likely by virtue of its bio-relation to cholesterol) is unique since in additional experiments, we also demonstrated that gemcitabine-loaded liposomes (chosen as an example of another lipidic nanocarrier) did not interact with lipoproteins. In the light of these results, we believe that we have demonstrated that the conceptual approach has a generic character, represents a real novelty in the drug delivery field and opens perspectives for a large application of the squalene-based NPs.

The SQ-Gem NPs and 3H-SQ-Gem
NPs were not well characterized, for example, the data related to the particle size, size distribution and surface charge status are missing. How will lipopoproteins affect the nanostructure of NPs? e.g. the TEM images of SQ-Gem NPs and lipopoproteins-SQ-Gem NPs. There is only speculation, as the results of in silico simulations. Some aspects of this seem non-intuitive to me. Answer: As suggested by the reviewer, data on the characterization of the SQGem NPs (and on the other squalenebased nanoparticles and liposomes used in this study) have been added to the revised manuscript. (See Supporting  Table S1, page 43).
In addition, as suggested, we have investigated the interaction between SQGem NPs and lipoproteins by transmission electron microscopy. Briefly, SQGem NPs have been incubated with human LDL for 5 minutes at 37 °C and then images of this physical mixture have been acquired. SQGem NPs only and LDL particles only have been used as controls. In the physical mixture, shape modified SQGem NPs, possibly in the course of a disassembled process (yellow arrows), and LDL particles (pink arrows) are visible but also SQGem NPs displaying lipoproteins on their surface (blue arrows). (Supplementary Fig. S15) This microscopy image reveals how the first encounter between SQGem NPs and LDL occurs. Of note, it provides only a static image, while the interaction between the SQGem and lipoproteins is rather a dynamic phenomenon. Indeed, the establishment of this preliminary interaction is followed by the disassembly of the SQGem NPs (demonstrated by additional experiments using FRET nanoparticles, see Supplementary Fig. S12 and S13) and the release of single SQGem bioconjugate molecules that will further insert into the lipoproteins (which is obviously not observable by TEM). To still convince the reviewer about the reality of the interaction between SQGem NPs and LDL, additional experiments have been carried out by isothermal titration calorimetry (ITC), using human lipoproteins separated from the blood of healthy volunteers. The heat flows were determined when SQGem NPs were added to LDL dispersion, HDL dispersion or albumin solution placed in the titration cell. ITC thermograms revealed the existence of a strong interaction between the SQGem and the LDL, while it was not observed either with HDL or with albumin. The specificity of the recorded signal was confirmed after dilution of SQGem NPs in PBS (Supplementary Fig. S8). These results are in agreement with those obtained after incubation of the SQGem NPs with human blood (Revised Manuscript: Figure 1, page 8) and provide a further proof of the existence of a specific interaction with the human LDL fraction, not only based on in silico simulations.

The authors concluded that squalene was chosen because of the lipid nature and its structural similarity with cholesterol, which have good affinity with lipoproteins. Whether cholesterol is a better choice to conjugate with gemcitabine?
Answer: Being the squalene a precursor in cholesterol's biosynthesis, this bio-relation drives the insertion of the SQGem bioconjugates into lipoprotein particles. We agree with the logic hypothesis of the reviewer that the replacement of the squalene by a cholesterol moiety would further increase the affinity toward lipoproteins and the transport to cells with high lipoprotein receptor activity. We had also considered this possibility but we did not follow this approach, because cholesterol-based nanomedicines may represent a cardiovascular risk factor that would compromise further possible clinical application. In spite of this limitation, the comment of the reviewer has motivated us to verify whether a cholesterol-gemcitabine bioconjugate could self-assemble in water as nanoparticles and exert a cytotoxic activity.
Thus, we have synthesized the CholGem bioconjugate according to the following reaction scheme.
CholGem NPs were then prepared by nanoprecipitation. Briefly, CholGem was weighted and dissolved in ethanol (2 mg mL -1 ). The resulting solution was then added dropwise under magnetic stirring into 1 mL of MilliQ ® water (ethanol/water 0.5/1 v/v), followed by ethanol evaporation under vacuum. The obtained suspension contained CholGem NPs (final concentration 1 mg mL -1 ) with a mean diameter of 134 ± 12 nm and a narrow size distribution (PdI 0.10 ± 0.02). NPs were negatively charged with a mean zeta potential value of -48 mV. Once demonstrated that nanoparticles could be prepared, their cytotoxicity was evaluated in vitro using the 3-(4,5dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) test on MDA-MB-231 breast cancer cells and compared to that of SQGem NPs. Results are shown in the figure below. It was observed that CholGem NPs did not display any significant cytotoxicity and 70% cell viability was observed even after 72 h incubation with NPs at 200 µM concentration. On the other hand, when cells were exposed to SQGem NPs complete cell death was observed already at the concentration of 25µM. In conclusion, although it was demonstrated that the CholGem bioconjugate was also able to self-assemble as NPs, it did not display any advantage compared to the SQGem NPs due to absence of pharmacological activity.
The design and in vitro evaluation of the CholGem NPs being out of the scope of the submitted article, these data have not been added to the revised manuscript.

It is not clear to me what is meant by " In this study, a protein-driven dissociation of SQ-Gem nanoparticles into SQ-Gem monomers was demonstrated to occur before cell capture" (line 267 and 268, p 13). Please provide convincing interpretations, maybe with more evidence. If the SQ-Gem NPs were dissociated before cell uptake, how can lipopoproteins facilitate tumor cell targeting? In addition, if the SQ-Gem NPs will be dissociated in blood, what's the advantage of SQ-Gem NPs over SQ-Gem conjugates? Will the PEGylated SQ-Gem NPs also be dissociated in blood? If not, which one is a better formulation for drug delivery, PEGylated SQ-Gem NPs or non-PEGylated SQ-Gem NPs?
Answer: To meet these reviewer comments, additional experiments have been carried out in order to provide a stronger demonstration that the interaction with the lipoproteins involves single SQGem bioconjugate molecules as a consequence of the disassembly of the SQGem NPs in biological medium.
Thus, we have further investigated the disassembly of the SQGem NPs by performing new experiments with FRET SQGem NPs. These NPs have been prepared by labelling SQGem NPs with the squalene derivatives of the cyanine 5.5 (SQCy5.5, 0.6% w/w) and the cyanine 7.5 (SQCy7.5, 0.6% w/w) which behave as FRET donor and acceptor, respectively. For details concerning the preparation of these FRET nanoparticles, see Supplementary Information on pages 13-14. Fluorescence emission spectrum of these FRET NPs is reported in Supplementary Fig. S12. Taking advantage of the dependence of FRET signal from the distance between the FRET pair (SQCy5.5/SQCy7.5), this approach enabled to monitor the integrity of SQGem NPs over time. Stability has been evaluated at 37 °C after opportune dilution of NPs in (i) water, (ii) rat blood and (iii) ethanol where SQGem is soluble. Thus, the dilution in ethanol has been used as a control of the complete disassembly of FRET SQGem NPs and consequently absence of energy transfer between the donor and acceptor dyes. Samples were imaged at regular times post incubation using the IVIS Lumina imaging system. After excitation at 640 nm (SQCy5.5 excitation wavelength), the emission was collected at 695-770 nm (donor signal, D) and 810-875 nm (FRET signal, A). Following signal quantification, the FRET proximity ratio, providing a semiquantitative measurement of the FRET efficiency, has been calculated according to the equation A/(A+D). (Preus, S. & Wilhelmsson, L. M.. Chembiochem, 2012, 13, 1990-2001

Now, what is the advantage of SQ-Gem NPs over SQ-Gem conjugates? Although the interaction involves single
SQGem bioconjugate molecules, their formulation in form of nanoparticles is required to allow their intravenous administration in vivo. Indeed, the SQGem bioconjugates are soluble in organic solvents only, while they are not soluble in water, forming nanoparticles in aqueous medium. These NPs can be easily administered intravenously after adjustment of the tonicity (addition of 5% dextrose).

Which one is a better formulation for drug delivery, PEGylated SQ-Gem NPs or non-PEGylated SQ-Gem NPs?
The SQGem NPs of this study were not PEGylated. However, as pointed by the reviewer, whether the PEGylation could be responsible of a different behavior of the NPs, was worth to be investigated. Accordingly, we have prepared PEGylated SQGem NPs, and with the aim to assess their stability over time in different media, we have labelled these nanoparticles with SQCy5.5 and SQCy7.5 for allowing FRET experiments. Briefly, 2 mg of SQGem mixed with SQPEG (10% w/w), SQCy5.5 (0.6% w/w) and SQCy7.5 (0.6% w/w) were dissolved in ethanol. The resulting solution was added dropwise under magnetical stirring into MilliQ water (ethanol/water 0.5/1). Ethanol was then evaporated under reduced pressure and a suspension of PEG_FRET NPs (final concentration of SQGem 2 mg mL -1 ) was obtained. These nanoparticles displayed a mean diameter of 158 ± 22 nm, a narrow size distribution (PdI 0.06) and a negative surface charge (zeta potential -31 ± 5 mV). Stability of PEG_FRET NPs was evaluated as described above for the FRET NPs. As displayed in Supplementary Fig.  S14, the FRET efficiency profiles of PEGylated and non PEGylated SQGem nanoparticles perfectly overlapped. These results clearly indicated that once administered in vivo, the SQGem bioconjugates were released similarly from both kind of NPs (PEGylated or not) and could be further available as molecules for the interaction with circulating LPs.
On the basis of these results, it was clear that the PEGylation of the SQGem NPs would just increase the complexity of the nanoformulation, without providing any advantage in terms of drug delivery. Moreover, since several concerns about the safety of PEG have been recently reported, the possibility to achieve a long circulation time in blood without surface modification with PEG chains, but as a simple result of the spontaneous interaction with the circulating lipoproteins, represents once again an original approach. Supplementary Fig. S14. In vitro stability of FRET NPs vs PEG_FRET NPs. In vitro stability study of FRET NPs and PEG_FRET NPs after dilution with (i) MilliQ ® water, (ii) rat blood or (iii) ethanol. Results are expressed as mean ± standard error of the mean (s.e.m.) (n=2).
These already published data provide the reviewer with the information that the conjugation to squalene induces, indeed, a modification of the drug behavior and a different distribution pattern comparatively to the free Gem.
6. The authors conclude that "It was discovered that endogenous LDL particles may function as carriers for SQ-Gem, thus allowing the indirect targeting of cancer cells displaying high expression and activity of LDL receptors, without the need to functionalize NPs surface with hydrophilic PEG (polyethylene glycol) chains and/or with specific ligands." Actually, PEGylation cannot facilitate the cellular uptake of NPs, even hinder the cellular uptake, but benefit the long circulation in blood. Moreover, the authors didn't set the PEGylated SQ-Gem NPs as a control. How can they draw this conclusion? Answer: As stated by the reviewer, the surface modification of nanoscale drug delivery systems with poly(ethyleneglycol) (PEG) has been successfully used to interfere with NPs opsonization and their rapid uptake. Indeed, the PEGylation endows nanoparticles with long circulating properties (ie., "stealthness") and the capacity to passively accumulate at level of the tumors, taking advantage of the so-called enhanced permeability and retention (EPR) effect. However, and we agree with the reviewer, the presence of the PEG chains could also represent an obstacle to the nanoparticle cell internalization. Thus in the revised version of the manuscript the sentence has been rephrased as follows (see revised manuscript page 6): "It was discovered that endogenous LDL particles may function as carriers for SQ-Gem, thus allowing the indirect targeting of cancer cells displaying high expression and activity of LDL receptors, without the need to functionalize NPs surface with specific ligands." In the additional experiments already mentioned before, PEGylated SQGem NPs (that were missing in the first version of the manuscript) have been prepared and compared with the SQGem NPs. As described above, we have used the FRET principle to monitor the stability of these NPs both in water and in rat blood. Results revealed the same behavior, (Supplementary Fig. S14) thus clearly demonstrating that both non-PEGylated and PEGylated NPs disassembled in blood at the same rate, leading to the release of individual SQGem bioconjugate molecules capable of interaction with lipoproteins. Accordingly, to meet the 3Rs guiding principles (reduction, refinement and replacement) for the use of animals in research, the PEGylated nanoparticles have not been tested in vivo.
compatibility of SQPEG with SQGem which has been acknowledged also by the reviewer. Importantly, the use of SQPEG for the preparation of these PEGylated SQGem NPs has been reported in details previously and this nanoformulation was certifyed in terms of colloidal properties and surface modification (see F. Bekkara-Aounallah et al., Novel PEGylated nanoassemblies made of self-assembled squalenoyl nucleoside analogues, Adv. Funct. Mater. 2008, 18, 3715-3725).
Nevertheless, to meet the reviewer comments, we have carried out additional experiments to prepare PEGylated SQGem NPs using the DSPE-PEG 2000, instead of the SQPEG. SQGem_DSPE-PEG NPs were prepared by nanoprecipitation. Briefly, 2 mg of SQGem mixed with DSPE-PEG (3% mol) were dissolved in ethanol. The resulting solution was added dropwise under magnetical stirring into MilliQ water (ethanol/water 0.5/1 v/v). Ethanol was then evaporated under reduced pressure and a suspension of SQGem_DSPE-PEG NPs (final concentration of SQGem 2 mg mL -1 ) was obtained. These nanoparticles displayed a mean diameter of 114 ± 5 nm and a polydispersity value of 0.16 ± 0.01. The zeta potential value (-41 ± 3 mV) confirmed surface modification of SQGem NPs (-22 mV). Once demonstrated that such nanoparticles could be prepared, their cytotoxicity was evaluated in vitro using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) test on MDA-MB-231, the breast cancer cell line already used in the submitted article. And cytotoxicity of SQGem_DSPE-PEG NPs was compared to that of SQGem NPs. Results are shown in the figure below.
The cytotoxity profile of SQGem NPs and SQGem_DSPE-PEG NPs perfectly overlapped, thus confirming once again that the PEGylation did not modify the behavior of the SQGem NPs. Of note, the cell culture medium contained 10% serum with lipoproteins. As previously observed with SQPEG, the introduction of the DSPE-PEG didn't provide any advantage in terms of drug delivery efficiency.
Since the authors feel that the design and in vitro evaluation of the SQGem_DSPE-PEG NPs remains a little out of the scope of the submitted article, they suggest not to add these new data to the revised manuscript which could confuse the major scientific message of the paper (ie. that le linkage of the squalene moiety to a drug allows incorporation into LDL, after intravenous administration, which indirectly confers targeting capacity towards LDLR expressing cells, incl. cancer cells).