Rapid transport of deformation-tuned nanoparticles across biological hydrogels and cellular barriers

To optimally penetrate biological hydrogels such as mucus and the tumor interstitial matrix, nanoparticles (NPs) require physicochemical properties that would typically preclude cellular uptake, resulting in inefficient drug delivery. Here, we demonstrate that (poly(lactic-co-glycolic acid) (PLGA) core)-(lipid shell) NPs with moderate rigidity display enhanced diffusivity through mucus compared with some synthetic mucus penetration particles (MPPs), achieving a mucosal and tumor penetrating capability superior to that of both their soft and hard counterparts. Orally administered semi-elastic NPs efficiently overcome multiple intestinal barriers, and result in increased bioavailability of doxorubicin (Dox) (up to 8 fold) compared to Dox solution. Molecular dynamics simulations and super-resolution microscopy reveal that the semi-elastic NPs deform into ellipsoids, which enables rotation-facilitated penetration. In contrast, rigid NPs cannot deform, and overly soft NPs are impeded by interactions with the hydrogel network. Modifying particle rigidity may improve the efficacy of NP-based drugs, and can be applicable to other barriers.

g (line 323) does not really belong in the same plot as the simulation results. This confused me when I read the manuscript. I think it would be better to place it in a separate figure.
In addition, I think that the paragraph on experimental imaging of the nanoparticles (lines 351 to 361) should be placed before the paragraph on MD simulations and appropriate modifications to those paragraphs made. The MD simulations should probably be presented as supporting the experimental results, not the other way.
Given that the effectiveness of the semi-elastic particles seems to be related to their ability to deform into ellipsoidal shapes, there is a question of whether shape or elasticity is more important. Are rigid ellipsoidal particles as effective as spherical semi-elastic particles? Although it may not be feasible to construct real ellipsoidal particles of the same type found in this paper, rigid ellipsoidal particles could be simulated. If the authors do not choose to simulate rigid ellipsoidal particles, they should at least mention the possibility that they might also be effective or why that is unlikely and provide supporting references.
The equations for MSD at time t and diffusivity (lines 469 and 553) should indicate somehow that an average is taken over all of the particles.
For fitting the simulation MSDs, it is stated that the MSDs were fit starting from time 0 (line 563). Generally, there is a short time over which the MSD is not expected to be linear. Therefore, fitting should usually not start from time 0. This nonlinearity at short times can be seen in Figure 6 d. Given the length of the trajectories, throwing away this short time will not have a large effect on the results. However, one way to determine this time is to plot the derivative of the log of MSD with respect to the log of time. Once that begins fluctuating around a value of 1, then the MSD has become linear. See the attached plot, where the nonlinear part lasts between 10 and 15 ps and the fitting should not be started until after that.
Reviewer #2: Remarks to the Author: In this paper, the authors tested their hypothesis that particle diffusion in biological hydrogels would be dependent upon particle rigidity. In particular, they formulated core-shell PLGA-lipid NPs possessing a range of rigidity, as determined by Young's moduli, and demonstrated that semielastic NPs were most efficient at mucus and tumor penetration. This paper is relevant to the readership of the journal and easy to read, and experiments are generally well designed and executed. However, the connection between their experimental observation and datat interpretation is relatively loose and there are a few critical issues to be addressed, as listed below, prior to its consideration for publication. blood pharmacokinetics rather than their abilities to penetrate tumor tissue. C omparing the NP distribution following local administration would be a more relevant experiment to demonstrate superior ability of semi-elastic NPs to diffuse in tumor tissues. 9. Why soft NPs retained much shorter in tumor in vivo compared to other NPs if they readily adhere to the tumor tissue as suggested by the authors (Figure 6b)? 10. Rhodamine B does not "label" the water, but visualize the volume occupied by water.
Reviewer #3: Remarks to the Author: In their paper "Rapid Transport of Deformation-Tuned Nanoparticles across Biological Hydrogels and C ellular Barriers", the authors show that particles of different rigidities but identical surface chemistry and size can have very different diffusion properties within biological hydrogels. The experiments are straightforward and convincing, notwithstanding some quibbles with Fig. S4 described below. Overall, it is my recommendation that the manuscript be accepted for publication with minor revisions. More specifically, to me it seems like the results of this paper suggest that a nanoparticle with a 70nm solid core and 200nm shell diffuses essentially as though it were 70nm solid nanoparticle, because it can deform to slip through a 70nm pore by presenting a 70nm cross section. Or more generally, a PLGAx-Lip-F1275% particle could diffuse essentially like a solid nanoparticle of diameter x, except that for small PLGA cores this trend stops holding because the liposome can deform around gel polymers. If this is true, then the natural comparison looking at nanoparticle size changes in Fig. S4 would be to compare PLGA70-Lip-F1275% to hard, rather than soft, smaller nanoparticles. As it is, I am not entirely sure what the motivation for the Fig. S4 experiment is. At any rate, the discussion may benefit from clarifying the advantages of semi-elastic nanoparticles in potential applications. C ompared to hard nanoparticles, this paper shows that semi-elastic particles have a lower effective diameter for diffusion (which is good!), but also have a lower uptake (which is bad!) possibly also essentially due to a smaller effective diameter, so semielastic particles are basically equivalent to smaller hard particles. Is the idea just that you can fit more stuff in the semi-elastic particles, which can be bigger than the equivalent smaller, solid particle? That's interesting! And could perhaps be stated explicitly. But overall, the benefits of semi-elastic larger nanoparticles over smaller, hard nanoparticles seem somewhat marginal. However, it appears that adding a solid core to (soft) liposomes would be helpful on many levels, from diffusion to uptake. That's very interesting. So it seems as though the utility of this paper will come largely from applications where a liposome, rather than a solid nanoparticle, is desired for the biological effect. If this is true it is potentially worth stating directly. Some minor comments: -I think the capitalization on the title is off, the copy editor can presumably confirm this -Line 672: it's the Hanes group, not the Justin group -Line 175: I think the word "frequency" should be deleted here: "Distributions of the logarithms of individual…" seems fine. -Line 366: to stay consistent it should be " Fig. S5" (the period was left out) -Figures S4 and S5 are mentioned after Fig. S6-S8, so they should be reordered. - Figure S9 is never referenced in the main text

Response to Review
We are grateful to the Editorial Office and reviewers for their constructive comments and suggestions. We have carefully revised the paper according to the reviews' comments. The changes are highlighted in the revised manuscript. The following are our detailed responses: Reviewer #1 (Remarks to the Author): This paper certainly presents important results and provides an explanation for those results by exploring a simple simulation model. Since I am not familiar with most of the literature related to nanoparticle transport through mucus, I cannot say if it is original or not. Assuming that it is sufficiently original, I recommend it for publication with a few changes.
Response: Thank you for your positive assessment of our work. Response: Thank the reviewer for making these insightful suggestions, which we have adopted. Figure 6g has been placed in a separate Figure 6 in the revised manuscript.
The MD simulations are conducted to reveal the mechanisms that underlie the experimental findings, so we have adopted the reviewer's suggestion, and the paragraph on experimental imaging of the nanoparticles has been placed before the paragraph on MD simulations in the revised manuscript.
Given that the effectiveness of the semi-elastic particles seems to be related to their ability to deform into ellipsoidal shapes, there is a question of whether shape or elasticity is more important. Are rigid ellipsoidal particles as effective as spherical semi-elastic particles? Although it may not be feasible to construct real ellipsoidal particles of the same type found in this paper, rigid ellipsoidal particles could be simulated. If the authors do not choose to simulate rigid ellipsoidal particles, they should at least mention the possibility that they might also be effective or why that is unlikely and provide supporting references.
Response: Thank the reviewer for this constructive suggestion. We have adopted reviewer's suggestion and constructed new simulation model to simulate the diffusion of rigid ellipsoidal NPs in the polymer network as comparison. In the simulations, two rigid ellipsoidal NPs with different aspect ratios (2:1 and 3:1) were used, which had the same volume as the spherical NPs. The results indicated that the diffusivities of the rigid NPs (one spherical NP and two ellipsoidal NPs) were influenced by the aspect ratio of the NPs. The ellipsoidal NPs diffused faster than the rigid spherical NPs. We further found that the diffusivity of spherical semi-elastic NPs lied between those of the rigid spherical NPs and rigid ellipsoidal NPs with aspect ratio 2:1 (see Fig.   R1). These results further provided evidence that the semi-elastic NPs would change its shape from a sphere to an ellipsoid during diffusion. We have added these results in our revised manuscript and Supplementary Information. Response: In response to this comment, we have corrected the sentence in the revised manuscript as: "Particle-averaged mean squared displacement (MSD) and effective diffusivities (D eff ) were calculated using the following equations: We thank the reviewer for her/his positive evaluation of our manuscript and the constructive comments/suggestions which have helped us improve the manuscript.
Reviewer #2 (Remarks to the Author): In this paper, the authors tested their hypothesis that particle diffusion in biological hydrogels would be dependent upon particle rigidity. In particular, they formulated It has been reported that the mesh size of mucus and ECM varies in a wide range of 10-1000 nm 1, 2, 3, 4, 5, 6 . In our experiment we found that the average pore size of the mucus was about 100-240 nm 7 , and the hydrodynamic diameter of the NPs was about 200 nm, which is close to the size of the mucus pore. To mimic experiment, we modeled the dimensions of particles and mesh pore sizes in the simulation to be 10:16.
In response to the reviewer's concern, we have considered decreasing the pore size to 14σ , in which case all the NPs were trapped in the network oscillating within one grid cell (Fig. R2). We also increased the pore size to 18σ , and observed an increase in the diffusivity due to loosened restriction of the polymer network. Still, the semi-elastic NPs diffused faster than the soft and hard NPs (Fig. R2). To properly reflect the polymer affinity, we conducted additional atomistic simulations in which the interactions between the lipid-coated NPs / F127-coated NPs and polymer chains were extracted as follows.
In our experiments, the liposomes were coated with F127 molecules. To obtain the interaction energy between NPs and mucin fibers, we constructed two model systems: (1) one bilayer interacting with one mucin glycoprotein chain; and (2) R3(a-b)). The upper bound of interaction represents another extreme case in which the network has strong interaction with the NPs. It was seen that all the NPs were trapped within the grid cell and cannot freely move ( Fig. R3(g-h)). This case is at odds with our experimental observation that the NPs could move from one grid cell to another. When the parameter was tuned to an intermediate one,  NPs (although statistical analysis is not provided) ( Figure 6). This discrepancy again raise a concern for using the simulation to interpreter the authors' experimental data.
Response: Thank the reviewer for pointing out this issue. The discrepancy between experiment and simulation is due to the selection of the interaction parameter between the NPs and the polymer, which could influence the diffusivity of NPs greatly. As seen in Fig. R3, the diffusivity of soft NPs is even larger than those of hard and semi-elastic NPs when the interaction is small. Only when the interaction is in an appropriate range, could the semi-elastic NPs attain the highest diffusivity. In response to the reviewer's concern, in our revised manuscript, we have determined the interaction parameters as 0.  3. The authors claimed that soft NPs exhibited lowest diffusivity as they "deformed excessively and attached themselves to the polymer", which indicates that particle diffusion is largely hindered by adhesive interactions rather than flexibility itself. In addition, we agree with the reviewer that exhibiting similar z-potentials among different NP population does not ensure similar/identical NP surface properties. So we have elected to use liposomes as the shells of the three types of NPs (soft, semi-elastic and hard NPs) in our study, as shown in Fig. 1b. The compositions of the liposomal shells were the same, and all the shells represented phospholipid bilayer with a width of 8 nm (Fig. 1b) (Table R1). MPT were then used to observe the transport dynamics of NPs in freshly obtained rat intestinal mucus. As shown in Fig. R5, there is only a minor additional increase in MSD of NPs as F127 content increased from 5% to 10%, indicating that 5% F127 density was sufficient for uniform penetration in biological hydrogel. The surfaces of NPs we synthesized were well shielded, and NPs are not elongated to intertwine with gel fibers.  The reviewer also raised a question about the viscous drag. Traditionally, diffusion of particle in a continuous medium is described by the Stoke-Einstein relation, which shows that the diffusivity is inversely proportional to the solvent viscosity and the size of the particle. We have carefully fabricated the control particles according to previous work 7,8 in the revised manuscript, and now the z-potential of PLGA-PEG NPs is approximate -5 mV. We have conducted the "Mucus penetrating particle (MPP) transport in rat intestinal mucus ex vivo" again, and the results are comparable to previously reported data. On a time scale of 1 s, the <MSD> value of PLGA-PEG NPs was approximately 1.2671, which is consistent with the reported result 8 .
In addition, we have added a statistical analysis for the different formulations.
6. Several necessary details are missing throughout the method section, which makes it hard to precisely gauge the validity of this study and to reproduce the observations here by an independent group. For example, why two different pluronics (i.e. F68 and F127) were used for the formulation of core-shell NPs? how were the Young's moduli of NPs calculated? What was the properties of intestinal mucus collected from rodents (e.g. mucin concentration, etc.) and how many animals were sacrificed to collect mucus? What was the concentration of NPs added to mucus for particle tracking studies? How many movies were taken and analyzed? How the MPP-control particles were made; did the authors strictly follow published methods of formulation and characterization and/or collaborate with relevant groups? How the particle tracking data was processed/averaged? Was the presence of mucus confirmed in the cell culture studies? Were qualitative image-based studies repeated to corroborate the authors' findings or could the authors conduct image-based quantification?
Response: We thank the reviewer for these constructive criticisms. We have added detailed methods in our revised manuscript.
It is reported that pluronic F127 is a good candidate for decorating NPs to enhance mucus diffusivity 13,14,15,16,17,18 . We adopted pluronic F127 to modify the shell of the NPs (liposome) to facilitate their mucus penetration capacity. For the fabrication of PLGA cores, we found that pluronics F68 was more suitable to make the particle size controllable. So we applied these two different pluronics for the formulation of core-shell NPs.
All images of NPs obtained from AFM were processed by the software NanoScope Analysis (Bruker), and the Young's modulus was calculated by processing the images of NPs in PeakForce DMT mode (n=3). The detailed computation principle and method are shown in Fig. R6. Figure R6 | Determination of modulus from force curve -DMT model.
For mucus collection, we have adopted methods reported by the Hanes group 19,20,21,22 . Briefly, the small intestine was excised after euthanizing the rats, and approximately 1.5-2 mL of mucus from each fasted rat was collected. The properties of intestinal mucus collected from rodents have been reported in our previous work 9 , which is in consistent with previous reports 5, 23,24 . The average size of the mucus pore was approximately 200 nm, and the majority of pores were 100-240 nm in diameter.
In this study, ten rats were sacrificed to collect mucus for all particle tracking studies.
The concentration of NPs added to mucus for particle tracking studies was 100 μg/ml in PBS, and fifteen movies were taken and analyzed for each NP. We have added these supplements in the revised manuscript.
We have adopted the reported methods by the Hanes group 7,8 to fabricate the MPP-control particles in the revised manuscript, and we have properly cited the literatures in the manuscript.
The particle tracking movies were processed by the software ImageJ, which could transform the variation of coordinate position into motion trajectories of particles. We then collected the tracking data of 300 particles followed by calculating time-averaged mean square displacement (MSD) and effective diffusivities (D eff ).
Prior to the experiment, E12 cells were incubated for one week to ensure the formation of mucus. We have applied a method reported previously 25   7. The authors used student's t-test for statistical analyses, but multiple comparisons should be made with ANOVA. In addition, statistical analysis is missing throughout the manuscript, particularly for particle tracking studies.
Response: In response to the reviewer's suggestion, we have updated the methods for statistical analyses. We have also added the statistical analysis for particle tracking studies.
8. In Figure 4, it is unclear whether semi-elastic NPs indeed provided improved For the study of NP penetration in tumor tissue, we thank the reviewer for this insightful suggestion. We agree that the rigidity of NPs would affect their blood pharmacokinetics. So we have adopted the suggestion and conducted an experiment in which all the NPs were locally administrated. As shown in Fig. R8a, the soft and hard   Molecular simulations and stimulated emission of depletion (STED) microscopy revealed that the rotational diffusion of ellipsoid NPs within the complex mesh structure plays an active role in such diffusion enhancement. In this study, we presented evidences that semi-elastic NPs could change their shape to ellipsoidal ones, and this thin shape further facilitated their diffusion in biological hydrogels.
At any rate, the discussion may benefit from clarifying the advantages of semi-elastic nanoparticles in potential applications. Compared to hard nanoparticles, this paper shows that semi-elastic particles have a lower effective diameter for diffusion (which is good!), but also have a lower uptake (which is bad!) possibly also essentially due to a smaller effective diameter, so semi-elastic particles are basically equivalent to smaller hard particles. Is the idea just that you can fit more stuff in the semi-elastic particles, which can be bigger than the equivalent smaller, solid particle? That's interesting! And could perhaps be stated explicitly. But overall, the benefits of semi-elastic larger nanoparticles over smaller, hard nanoparticles seem somewhat marginal. However, it appears that adding a solid core to (soft) liposomes would be helpful on many levels, from diffusion to uptake. That's very interesting. So it seems as though the utility of this paper will come largely from applications where a liposome, rather than a solid nanoparticle, is desired for the biological effect. If this is true it is potentially worth stating directly.
Response: We again thank the reviewer for these constructive comments. It is true that the semi-elastic particles could load more stuff, which would benefit drug delivery. We have adopted the reviewer's suggestion and added some discussion in the revised manuscript: "In addition, adding a solid core with suitable size to (soft) liposomes would be helpful on many levels, from diffusion to uptake, and further for drug delivery especially for overcoming the multi-biological barriers." Some minor comments: -I think the capitalization on the title is off, the copy editor can presumably confirm this -Line 672: it's the Hanes group, not the Justin group Response: We thank the reviewer for pointing out this typo, and we have corrected the sentence: "the Hanes Group has fabricated mucus penetrating particles (MPPs)".
-Line 175: I think the word "frequency" should be deleted here: "Distributions of the logarithms of individual…" seems fine.
Response: We thank the reviewer for pointing out this typo, and the word "frequency" has been deleted in the revised manuscript.
-Line 366: to stay consistent it should be " Fig. S5" (the period was left out) Response: We have added the period in the revised manuscript.
- Figures S4 and S5 are mentioned after Fig. S6-S8, so they should be reordered.
Response: Thank the reviewer for bringing up this issue. We have reordered the  Supplementary Information and Fig. S12 and S13."