Real-time studies of plasma membrane damage trigger cell apoptosis from sulfhydryl nanoparticles to support safety assessment of nanoscale materials

Background: Sulfhydryl groups are present on the surface of nanoparticles in unburned vehicle exhaust and most air pollutants produced by combustion, which raises the risk for exposure of human. Sulfhydride nanoparticles not only penetrate the skin range from the stratum corneum to pass below the dermis, they also entering the systemis circulation from cell endocytosis pass way. The potential risk of skin and body healthy associated from sulfhydride nanoparticles were attach much attentions. It is important to illuminate the underlying toxicity of sulfhydride nanoparticles to humanbody, but the mechanisms underlying the toxicity of nanoparticles on cells remain unclear, especially the relationship from the damage of cells plasma membrane and the cell cycle. Methods: We performed time-response studies and cells-membrane interaction studies in C6 cells to observe the effects of 50nm and 200nm sulfhydryl nanoparticles on the activities, cell metabolism and cell cycle. The cells were exposed to 0, 10 or 20 Nano particles for 12, 36, 24, 48 or 72h to nish the particle- response studies. On the time of treatment, cells were collected to assess the expression of tight junction-associated proteins, P21, FBW7 and cyclin E. To further investigate the mechanisms underlying nanoparticle-induced dysregulation of tight junction-associated protein, we studied the change of lipid bilayers. Sum frequency generation optic spectrum was carried out to study the membrane change. Results: The results show that the smaller particles penetrate the plasma membrane and without bilayer disruption, whereas the larger one will pilled off one leaet of the membrane, they are mostly trapped in endosomes. The larger ones result in slow but unrepairable cell necrosis and caused cell cycle regulation disorders via disturbing the expression of p21, cyclin E, and FBW-7. Conclusion: The results suggest that the destruction of membrane structure by the particles will cause irreversible biological damage, and particles entering cells through protein assisted process will increase the expression of cell cycle related proteins and cells self-repair can be observed from the in vitro experiments. From the interactions between mitochondria lipid model and nanoparticles, we deduced that, the eciencies of nano-scaled drugs could be enhanced by altering the interaction models of nano systems and mitochondria. In the future, mitochondria membrane proteins would also be carefully explored to conrm their roles in the active mitochondrial uptake of nanoparticles and provide new channels for safe and effective mitochondria targeting drug delivery. Real-time studies of plasma membrane damage from sulfhydryl nanoparticles, and analysis the triggering of cell apoptosis, will support safety assessment of nanoscale materials.


Introduction
Nano-objective have much opportunities to contact or be used with organisms. Many drug delivery systems were designed to target to the cytosol or nucleus of cells during the process passage through the cells membranes to reach the target compartment. In this process, e cient delivery and avoiding cytotoxicity must be achieved. In this manner, understanding the mechanisms of nanoparticle-membrane interaction would be the key to designing potent cell-membrane-penetrating drugs.
Most groups studied modi cations on nanoparticle surfaces for cell organ targeting, such as mitochondria targeting. The organelle dynamics in uenced by intrinsic properties of nanoparticles, such as their size, shape, and surface functional groups. But the cell and organelle membrane respond of drug delivery, especially the membrane permeability of nanoparticles, played essential roles when nanoparticles interacted with cells. Drug delivery such as nano particles could enter cell organ through different channels: protein import channels, permeability transition pores, mitochondrial apoptosis related channels (MACs), apoptosis-related ceramide pores, the processes will possible preparative damage to the cell organ membrane. For example, mitochondria-targeting drug delivery was considered related with depolarization of mitochondrial and this process induced therapeutic agents and antioxidants effected with cells and animals [1,2]. However, cell organ depolarization closely related with membrane structure change. Growing numbers of groups studied dynamics process between cells and nanoparticles such as change of lipid bilayer or mitochondria membranes damage. But the mechanism of lipid membrane change from nanoparticles are still hard to be described.
In our former works, the reversible and irreversible mitochondria damages caused by cationic and neutral silica nanoparticles were well discussed respectively. And those mitochondrial injuries nally resulted in the unbalance of the cell cycle regulation network and severe cell cycle arrests [3][4][5]. Considering negatively charged nanoparticles were favorably used as drug deliverer based on mitochondria targeting, further steps for negatively charged nanoparticles were then focused on the presented research [4,5].
In this work, SiO 2 nanoparticles with different size (50nm and 200nm) were functionalized with sulfhydride groups for negative surface charges. The two sizes of nanoparticles presented different cellnanoparticle interactions models, especially different processes when interacting with the mitochondrial membrane. Furthermore, cell fate changes, including cell cycle changes and cell cycle regulation network uctuations were also detected. P21, cyclin E, and FBW-7 abundances were specially measured as identi ers of G1 and S phase progresses. This work will address the mechanisms of nanoparticle toxicity, based on NPs and mitochondrial membrane interactions, to support safety assessment of nanoscale materials.

Nanoparticles properties
Cell internalization process of nanoparticle In this part we analysis cellular internalization of nanoparticles and the relationship with their properties.
The shape, size, and surface charges of 50 nm and 200 nm sulfhydride nanoparticles were detected to determine the cell internalization process of nanoparticles. IR spectrum shown the sulfhydride groups on nanoparticles (Figure 1a), dynamic laser light scattering shown the size distribution of the nanoparticles were 50 ± 5nm and 200 ± 20 nm ( Figure S1). ζ-potentials of the nanoparticles were -38.033 ± 5 mV and -24.900 ± 5 mV, the surface charge densities were -4.845 10 -3 mV/nm 2 and -1.982 10 -4 mV/nm 2 .
With the scattering light from nanoparticles, we quanti ed the intercellular nanoparticles (Figure 1b), the maximum number of nanoparticles can be found at 8 h, but the amount of nano particles decreased when the time increased from 8 h to 32 h, mappings of scattering light results can be found in gure S1.
As it was shown in the literature [3][4][5], with the cell's differentiation, the number of internalized nanoparticles gradually decreased. It was based on the cells proliferation and the relative quantity decrease of the particles. And in this experiment, the internalized particles with the size of 50 nm was found have larger quantities than the 200 nm sized nanoparticles at 8 h but they metabolized faster than 200 nm sized nanoparticles in 32 h.
Then cell internalization processes of nanoparticles were assessment, from gure 1c, we found that both of the nanoparticles with different size will enter cells membrane and they can be found in the cells cytoplastic. From the confocal images as show in gure 1c, nanoparticles with the size of 50 nm will accumulation around the cells nucleus, far from the cells membrane. In the confocal image, particles with diameter of 50 nm can not be found from the top scan but they can be observed from the side scan. The nanoparticles with the size of 200 nm can be found close to the cells membrane, and some of the particles are on the top of cells membrane, that can be clear observed from top scan of confocal. We suggest that the different distribution and metabolism of the nanoparticles in cells was from the different endocytosis and exocytosis processes of nanoparticles. The interactions between the particles and cell membranes will be important to understanding the process. So, sum frequency generation spectrum (SFG) was used as a monolayer lipid analysis method for the interactions of nanoparticles/cells membrane studies.

SFG assay of the interactions between nanoparticles and model lipid membrane
Nanoparticle entering organelle such as mitochondria against bio-membrane potential were considered dependent on the membrane ip-op process (protein-free) or transport proteins (protein-assisted) on mitochondria. To detect the transmembrane movements of nanoparticles between the two leaflets of a cell membrane from the protein-free or protein-assisted manners, sum-frequency generation vibrational spectroscopy was used to observe the direct interactions between nanoparticles and lipid membrane ( Figure S2).
Previous studies had shown that sum-frequency generation (SFG) vibrational spectroscopy provided a direct method to examine phospholipid ip-op without labeling [11][12][13][14]. The wavelength ranges from 2000~2300 ( Figure 2a) was from d-DSPC ( Figure S2), it was obtained to avoid the noisy from the inner lea et of DSPC ( Figure S2) and to avoid the noisy from the particles. From the results it can be observed that when the particles interacted with the lipid bilayer, the signals of d-DSPC dropped off with the quantities of particles increasing from 20µg to 60µg.
To know whether 200 nm nanoparticles had the similar in uences on the phospholipid membrane of the liquid phase, the in uences of 200 nm nanoparticles on POPC/POPC phospholipid membrane were studied (Figure 2b). The symmetrical two lea ets of POPC/POPC phospholipid could not show SFG intensity with their symmetry structures (Figure 2b blue line). However, when the lipid membrane interacted with the nanoparticle, one lea et of the bilayer were damaged by the nanoparticles the other lea et of lipid bilayer keep in order, and asymmetry structures of phospholipid were formed, SFG signal enhancement could be found ( (Figure 2b red line). The results suggested that tilts, turn or other actions happened on one of the lea ets when the lipid bilayer interacted with the nanoparticles.
To further understanding the interaction process between nanoparticles and the lipid bilayers, we quanti ed the speed of the lipid bilayer ip-op under the in uence from nanoparticles.
In the absence of external interference, the lipid bilayer ipped spontaneously at a stable rate, which could be determined directly from the SFG intensity of CH 3 asymmetric stretching vibration on lipid bilayers. To understanding the spontaneously ipping of the membrane will be very important for us to understanding the process of membrane ip-op under affection from nanoparticles. The time-dependent CH 3 asymmetric stretching vibration intensity could be simpli ed to the following function [15]: Where "IR, max" and "IR, min" were the maximum and minimum resonant CH 3 asymmetric stretching intensity, and k was the intrinsic rate of lipid ip-op, which was determined by tting function (1) to the CH 3 asymmetric stretching vibration intensity. Conversely, if the decay curve of CH 3 asymmetric stretching vibration intensity could be t by an exponential function, the lipid must be ipped at a stable rate.
The kinetics of DSPC ip-op were determined by continuously monitoring the CH 3 asymmetric stretching vibration intensity from the lipid DSPC as a function of time in the presence of nanoparticles. Figure 2c showed a representative decay curve for 60 µg of 200 nm nanoparticles added to the d-DSPC/DSPC bilayer at 0 s to 2000s. To obtain the kinetic rate of DSPC ip-op, the decay curve was t by a single exponential function: y = 9.74708 e -0.00314x + 0.82359 As it was reported, in room temperature the d-DSPC / DSPC phospholipid would be very stable and intrinsic rate k would be hard to be observed. When the temperature increased to 41 ○ C, d-DSPC/DSPC phospholipid will turn ip-op and the intrinsic rate k was 1. Comparing with that, the spontaneous ip-op rate of d-DSPC/DSPC phospholipid in 41 ○ C was an orderof-magnitude slower. So, it could be considered that in room temperature, the phospholipid membrane ip-op was only in uenced by the particles, while the spontaneous ip-op rate was negligible. From the result, we found that 200 nm nanoparticles tended to ip the d-DSPC/DSPC phospholipid membrane at a stable rate and signi cantly alter the rate of trans bilayer movement. Details of d-DSPC/DSPC phospholipid signals were described by the SFG curve ( Figure 2 c and d).
The time-dependent CH 3 intensity of POPC-POPC phospholipid membrane was shown in Figure 2 d. It was clear that 200 nm nanoparticles had great in uences on the rate of POPC-POPC phospholipid membrane ip-op. To verify the intrinsic rate of membrane ip-op, we made a simple coordinate transformation, as shown in Figure 2 d*. The curve was t by exponential function: The result showed that the POPC / POPC membrane in uenced by 200 nm nanoparticles was ipped at a stable rate, and the rate k was 5.68 x 10 -4 s -1 , which was also much larger than that of phospholipid membrane spontaneous ip-op.
Deduced from SFG result, 200 nm nanoparticles changed the monolayer of the phospholipid by ipping or peeling one of the lea ets and led to intensity signals increasing or dropping. In summary, 200 nm nanoparticles could signi cantly alter the rate of phospholipid membranes movement. And for phospholipid membrane of solid or liquid phases, turn, tilt, and other disturbs both can be found on one of the lea ets.
50 nm nanoparticles were also studied, the results showed that 50 nm nanoparticles did not ip the DSPC-dDSPC phospholipid membrane at a stable rate and the uctuation of CH 3  In conclusion, SFG results proved that 50 nm nanoparticles can not in uence the lipid membrane, while, when the phospholipid membrane interacted with 200 nm nanoparticles, disturbance on one lea et of the phospholipid membrane could be found. But the 200 nm nanoparticles could not disturb both lea ets of the lipid membrane from ip-op processes. 200 nm nanoparticles would scrape the mitochondria membrane violently. It can be deduced from the results that the transmembrane delivery process will be different for different size nanoparticles. It will led to different cell dynamics and stress responses. It would be very interesting to further study the cellular network and cellular metabolisms when they underwent the treatments of different types of nanoparticles. In next step we tried to determine the cell toxicology mechanism of different nanoparticles by combining SFG optical analysis and biochemistry studies.
Subcellular localization of 50 nm and 200 nm nanoparticles 50 nm and 200 nm sulfhydride silica nanoparticles were co-cultured with C6 cells for 8 h, 16 h, 24 h, and 32 h. Side scattered light (SSC) from ow cytometry assay revealed that the cellular uptake of both kinds of nanoparticles increased with time extending within 24 h. It can be deduced from the results ( Figure S1, S3, S4) that more 50 nm nanoparticles were internalized than 200 nm ones within the same period. However, SSC results also show that small quantities of 200 nm particles lead to much serous organelle damages and organelle fragmentation rarely can be found with large quantities of 50nm nanoparticles.
Further studies from confocal microscope were carried out to explore the subcellular distributions of different nanoparticles we tried to determine the cell toxicology mechanism of different nanoparticles from biochemistry studies.
As is shown in gure 3, 50 nm nanoparticles were mainly observed in the cytoplasm within 32 h. Though lysosome was not found co-localize with 50 nm nanoparticles, they were injured or swollen (white dash circles in Figure 3) after 24 h and 32 h co-incubation with these nanoparticles. Moreover, 50 nm nanoparticles were found to keep adhering to mitochondria, and a few of them also entered mitochondria after 24 h and 32 h co-incubation (yellow arrows in Figure 3) and this phenomenon was considered caused mitochondria deformation as well (yellow dash circles in Figure 3).
From gure 3, 200 nm nanoparticles were also mainly found in the cytoplasm within 32 h. However, different from 50 nm ones, 200 nm nanoparticles entered lysosomes after 16 h co-incubation (white arrows in Figure 3) and induced increasingly serious lysosome injuries during 16 h -32 h (white dash circles in Figure 3). Interestingly, 200 nm nanoparticles was found kept close attachments to mitochondria but can not be found internalization by mitochondria. From the results of SFG optical analysis, it was proved that, large sized nanoparticles (200nm) can damage one lea et of lipid bilayer in a speed way. We suggested that will be the main reason the large particles can entry the lysosomes composed with monolayers at a short time. At the same time, SFG results proved that they can not damage both lea et of the lipid bilayers, so they hardly can be internalization by mitochondria, since the barrier of monolayer on mitochondria membrane blocked the particles entering. However, the damage of monolayer at mitochondria resulted in their deformation that can be found at 24h and 32h.
Comprehensively, 50 nm and 200 nm nanoparticles presented different subcellular localizations, since they underwent different subcellular transport pathways. It was well known that, lysosome is a low-pH organelle, which was always used as a target organelle for pH-triggered nanoparticles. In our experiments, it was found that 50 nm nanoparticles tended to escape from the lysosome uptake process due to their smaller sizes. The particles with 200nm would reach lysosomes at the end of the endocytosis process, however, this process can not result from the pH-triggered property, but the kinetic process that was proved damaged monolayer of phospholipid membrane.
The interactions between nanoparticles and mitochondria were more complex. 50 nm and 200nm nanoparticles can be found aggregated around mitochondria, it was considered that the protons pump process, H+ transport and proton gradient of mitochondria appear critical to the particle aggregation. After 24h, nanoparticles of 50nm were found entered mitochondria against membrane potential. The SFG results show that small particles of 50nm can not ip the DSPC-dDSPC phospholipid membrane. Without ipping or peeling any lea et of the phospholipid bilayer, the only way that the 50nm particles could be internalized into mitochondria was considered to be via proteins assistant process. As for 200 nm nanoparticles, from light scattering and co-location images it can be found that 200 nm nanoparticles can hardly penetrated the mitochondria. From SFG results, 200 nm nanoparticles just can peel off one lea et of the membrane that is why we can not nd 200 nm nanoparticles penetrated the mitochondria. With the in uences from 200 nm nanoparticles, the mitochondria fusion could be observed in Figure 3, which must result from one lea et of the phospholipid membrane peeled or ipped by 200 nm nanoparticles.
In uence of different sized sulfhydride nanoparticles and their toxicity for cells should be deeply illuminated. We analysis the ROS, cell apoptosis and cell cycle in further studies.

ROS production
Reactive oxide species (ROS) production in cells was detected to estimate the oxidative stress caused by 50 nm and 200 nm sulfhydride nanoparticles. Flow cytometry was carried out to illustrate ROS productions under the in uence of two kinds of nanoparticle treatment. And uorescence intensities within 102-103 were counted as strong ROS release. To eliminate false-positive caused by washing and centrifuging, the average uorescence intensity of within 102-103 of the control group was deducted in experiment groups.
According to Figure 4a, both 50 nm and 200 nm nanoparticles induced instant ROS production within 1 h, and 200 nm nanoparticles induced more ROS release than the 50 nm ones. After 24 h, ROS productions caused by 50 nm nanoparticles did not change signi cantly, while ROS production caused by 200 nm decreased to similar to 50 nm nanoparticle group. According to the results, cells reacted rapidly to both 50 nm and 200 nm nanoparticles, and 200 nm nanoparticles could induce more violent ROS release. Moreover, the ROS response induced by the two kinds of nanoparticles would converge and maintain for at least 24 h.
As is widely accepted, mitochondria damages are highly correlated with oxidative stress. 200 nm nanoparticles were observed to peel one of the lea ets in the mitochondrial lipid bilayer, as well as to induce acute and violent ROS production. So, it was suggested that ip-op on one lea et lipid of the mitochondrial membrane caused mitochondria damages and mitochondria-derived ROS aggravated. For the 50 nm nanoparticles, injuries of mitochondria from particle-invasions could be found without lipid membrane disturbance. But the ROS productions from 50 nm nanoparticles were signi cantly slighter than those caused by 200 nm nanoparticles (Figure 4), even though varying degrees of mitochondria damages can be found in both of the groups (yellow dashed circles in Figure 3). These results deduced that protein-dependent mitochondrial penetration reduced acute ROS release and mitochondria injuries at the same time the damage of mitochondria membrane will increase ROS release. In conclusion, 200 nm nanoparticles induced more mitochondria-derived ROS, due to peeling one lea et of the mitochondria membrane and causing serious mitochondrial damage. Mitochondria injuries and mitochondria-derived ROS would attack biomacromolecules such as DNA and protein as well as organelle dysfunction, which manifested cell metabolism changes and even necrosis and apoptosis [17].

Cell apoptosis assay
Cell apoptosis assay after 24 h and 48 h caused by 50 nm and 200 nm sulfhydride silica nanoparticles was conducted via ow cytometry (Figure 4b). According to the results, 50 nm caused late apoptosis and severe cell necrosis. From 24 h to 48 h, both late apoptosis and necrosis decreased obviously. For 200 nm nanoparticles, during 24 h, both of the late apoptosis and cell necrosis were not severe as those caused by 50 nm nanoparticles. After 48 h, the late apoptosis caused by 200 nm nanoparticles almost did not change, while necrosis increased rapidly.
The execution and propagation of cell necrosis were well described to be correlated with ROS, which also provoked damages to proteins, lipids, and DNA [18]. Interestingly, 50 nm nanoparticles caused lower ROS production from 1 h to 24 h, however, they induced higher cell necrosis in 24 h. The results indicated that, 50 nm nanoparticles actively internalized by mitochondria and lead to mitochondria dysfunction following mitochondria deformation (yellow dashed circles in Figure 3) triggering apoptosis. From 24 h to 48 h, decreasing cell death was observed caused by 50 nm nanoparticles, suggested that necrosis induced by 50 nm nanoparticles could be gradually alleviated. We deduced from the results that the biomolecule damages along with cell recovering were also partially repaired. Differently, 200 nm nanoparticles induced more severe ROS production but less severe necrosis than 50 nm nanoparticles did. We suggest that one lea et of the lipid ip-op on the mitochondrial membrane caused mitochondria damages and mitochondria-derived ROS aggravated. At the same time, lysosome deformation (white dashed circles in Figure 3) also contributes to the quick ROS release. At 48 h, a large number of necrosis cells can be found deduced that mitochondria-derived ROS aggravated caused by 200 nm nanoparticles could result in slower but unrepairable mitochondria damages, so that the affected cells underwent slow but deteriorating necrotic pathways.

Cell cycle assay
Most cells adjusted themselves to survive under the pressure of 50 nm and 200 nm sulfhydride nanoparticles. Cell cycle changes after 24 h and 48 h caused by 50 nm and 200 nm sulfhydride silica nanoparticles were assessed by owcytometry, and results were shown in Figure 5. Treated with 50 nm nanoparticles for 24 h, cells presented obvious G1 phase prolongation, but S phase did not change sharply. After 48 h treatment, G1 phase restored, while S phase extended signi cantly. Treated with 200 nm nanoparticles for 24 h, cells presented obvious G1 phase prolongation and slight S phase shorten. After 48 h treatment, G1 phase arrest alleviated slightly but S phase prolonged remarkably. Comprehensively, 50 nm nanoparticles caused signi cant G1 phase arrest within 24 h. In this stage, limited cells were able to pass the G1/S checkpoint, because large quantities of cells were arrested in the G1 phase due to undesired intracellular environment, including organelle injuries, ROS release and DNA damages. From 24 h to 48 h, the G1 phase recovered obviously while S phase prolonged, which means in this stage, with some self-recovery mechanism, cells arrested in the G1 phase overpassed the G1/S checkpoint and entered the S phase. Nevertheless, some severe genome and protein damages caused by ROS release in mitochondria and lysosome injuries were not fully healed, so these cells still could not nish the DNA replication and were arrested in S phase. From the results, we suggested that the 50 nm nanoparticles, which were actively internalized by mitochondria, caused partially repairable metabolism disorders, and cells would initiate an obvious self-repair process.
Regarding 200 nm nanoparticles, they also caused G1 phase arrest within 24 h, which indicated that organelle injuries, ROS release and DNA damages were also caused by the 200 nm nanoparticles. From 24 h to 48 h, G1 phase arrest recovered slightly, which suggested that though some of the cells arrested in the G1 phase passed the G1/S checkpoint, most of these cells were still arrested in the G1 phase. Simultaneously, the cells entered the S phase also could not complete DNA replication and be detained in the S phase. These disordered G1/S transition and S phase progress were attributed to the DNA and protein damages, which followed mitochondrial membrane disturbance (Figure 1

Western Blot assay
The changes in the cell cycle regulation system could be the underlying reasons for the cell cycle changes of C6 cells under the effects of 50 nm and 200 nm sulfhydride silica nanoparticles. Therefore, the abundances of p21, cyclin E, and FBW-7, which were key mediators in the G1 phase and S phase, were measured and presented in Figure 6.
Regarding p21, its abundance in normal C6 cells increased from 8 h to 24 h and then dropped steeply from 24 h to 32 h. After co-incubated with 50 nm nanoparticles, p21 abundance was signi cantly lower than the control group in 8 h. And then p21 abundance increased from 8 h to 24 h but decreased slightly from 24 h to 32 h. Differently, after co-incubated with 200 nm nanoparticles for 8 h, p21 abundance was only slightly lower than the control group. And then p21 abundance increased from 8 h to 24 h but decreased signi cantly from 24 h to 32 h.
As for cyclin E, in the control group, cyclin E doubled from 8 h to 24 h and then decreased steeply from 24 h to 32 h. After being co-incubated with 50 nm nanoparticles, cyclin E abundance was incredibly lower than the control group in 8 h. And then cyclin E abundance increased to almost 10-folds from 8 h to 24 h, and then decreased slightly from 24 h to 32 h. Besides, affected by 200 nm nanoparticles, cyclin E abundance was lower than the control group in 8 h. And then cyclin E abundance increased incredibly from 8 h to 24 h, and decreased steeply from 24 h to 32 h.
Regarding FBW-7 in the control group, FBW-7 abundance increased obviously from 8 h to 24 h, and then FBW-7 abundance decreased three quarters from 24 h to 32 h. Under the disturbance of 50 nm nanoparticles, FBW-7 abundance was obviously lower than the control group in 8 h, And then FBW-7 abundance increased 4-folds from 8 h to 24 h and then decreased one half from 24h to 32h. Differently, being affected by 200 nm nanoparticles, FBW-7 abundance was signi cantly lower than the control group in 8 h. And then FBW-7 abundance increased quickly from 8 h to 24 h, and then dropped steeply from 24 h to 32h.
Cell cycle changes and cell cycle regulator expression changes implicated how cells responded to and tried to overcome the effects of 50 nm and 200 nm nanoparticles. Among the three measured proteins, cyclin E was selected as a signi cant label of G1/S transition. 8 h was considered to be a time point of the middle G1 phase when cyclin E began to be synthesized, 24 h could be recognized as a time point of the late G1 phase or early S phase. In 32 h, cyclin E was mostly degraded in the middle-late S phase stage [19].
Thus, after co-incubated with 50 nm nanoparticles, p21 was consumed, and the synthesis and degradation of cyclin E were both retarded for hours. These results explained the G1 phase arrest in 24 h and S phase arrest in 48 h after cells were treated with 50 nm nanoparticles, and further proved that 50 nm nanoparticles could retard cell cycle progress. Therefore, 50 nm nanoparticles did cause mitochondria deformation (yellow dashed line in Figure 3), ROS release (Figure 4a), but they triggered a p21-mediated self-repair process to prolonged G1 and S phase for cells to repair the damaged proteins and genome. So, the cell necrosis and G1 phase arrest relived in 48 h. Differently, after co-incubated with 200 nm nanoparticles, the hardly changed p21 abundance p21 had no contribution in the cellular responses to 200 nm nanoparticles. However, FBW-7 was downregulated to maintain cyclin E abundance and force cells to pass the G1/S transition, so that the un-repaired organelle and biomolecule damages were no longer repairable, and nally resulted in the increasing cell necrosis from 24 h to 48 h (Figure 4b).
Therefore, it could be deduced that when mitochondria internalized nanoparticles actively, cells could initiate an effective self-repair process, and the organelle and biomolecule damages were partially repairable. Whereas, when nanoparticles caused defects on the mitochondria membrane violently, cells could not initiate an effective self-repair process, and the organelle and biomolecule damages were nonrepairable.
In the further step we culture zebra sh with different types of nanoparticles to study the teratogenesis from different cell damage and repair processes. It will be essential for understanding early zebra sh development especially in the process of neuro-development.

Animal experiments
The effects of 50 nm and 200 nm nanoparticles on the embryonic development of zebra shes were presented in gure 8 and gure S5. Figure 8a demonstrated the mutation frequencies of zebra sh eggs treated with 10 μg/ mL 50 nm and 200 nm sulfhydride silica nanoparticles. According to the results, both 50 nm and 200 nm nanoparticles caused zebra sh egg mutations. The probable process of hatching enzyme inhibition were related to the nanoparticles internalizing through the chorion pore canals or the blocking of chorion pores due to the nanoparticles accumulation on the chorion surface and leading to oxygen supply shortage. From the SEM of Zebra sh eggs we found that both of the nanoparticles can attached on the surface of the eggs and entered into them result into different damage. While, 200 nm nanoparticles caused much more and severer mutations than 50 nm ones did. The results supported that 50 nm nanoparticles caused repairable cell injuries, especially genome and protein damages, in zebra shes.
However, the cell injuries, as well as genome and protein damages caused by 200 nm nanoparticles, were unrepairable. Figure 8b presented the different innate abnormity of newly hatched zebra shes under the effects of 50 nm and 200 nm nanoparticles. It was presented that after treated with 50 nm nanoparticle, multiple tumidities of abdomens can be found in zebra shes (red rectangular in Figure 8b). However, after 200 nm nanoparticle treatment, zebra shes had both tumidities in their abdomens (blue rectangular in Figure 8b) and nerve malformation at their back heads (dashed blue rectangular in Figure 8b). The results suggested that both 50 nm and 200 nm nanoparticles had toxicity during the embryonic development process, but 200 nm nanoparticles presented their toxicities on both neural system and viscus development, while 50 nm nanoparticles, presented their toxicities on viscus development.

Discussion
In conclusion, different interaction processes between the nanoparticles and lipid membranes will result into different endocytosis kinetics. As show in the cartoon (Fig. 7), we mimic the intercellular pass way for different sized nanoparticles, 200 nm nanoparticles would scrape the mitochondria membrane violently, at the same time the 50nm nanoparticles could not ip or peel the monolayer of phospholipid membrane. The results of subcellular localization of the particles show that even though 200nm nanoparticles scrape the mitochondria membrane, they cannot penetrate into the cells mitochondrion, but for 50nm nanoparticles, they can be found in the cells mitochondrion without damage of the membrane. From the results, we deduced that 50nm nanoparticles entered into the mitochondrion from protein dependent pass way. From the results of cell apoptosis and western blot we found that when the particles scrape the mitochondria membrane (200nm), much serious ROS and large quantities of necrosis cells can be found. The non-repairable organelle and metabolism disorder resulted from decreasing of P21 should related with monolayer ip-op of mitochondria membrane (Figs. 6 and 7, a). The cells interacted with endocytosis process of the 50 nm nanoparticles show better state (Figs. 4 and 5). Self-repair process of cells can be found, it was suggested from abundant accumulation of P21 (Figs. 6 and 7, b). At the same time cell cycle can be found tend to similar with the control group after 24h. Cells metabolism disorder process was reparable when intracellular nanoparticles entered into cells by protein depended pass way.
Experiments of zebra sh helped us further understanding the toxicity effect of different particles, Organelle and biomolecule damages caused by 200 nm nanoparticles were non-repairable, that may relate to the damage of mitochondria membrane disturb. We found that the toxicity of 200 nm nanoparticles could not be eliminated and presented rstly on the neural system, the early developed system, and then on viscus, the later developed systems. We deduce from the results that the toxicities of 200 nm nanoparticles were maintained in the whole embryonic development, since the mitochondria membrane damage. However, the organelle and biomolecule damages caused by 50 nm nanoparticles were partially repairable, the toxicity of 50 nm nanoparticles was weakened in the stage of embryonic development. We suggested from the results that 50 nm nanoparticles enter mitochondria from the protein-assisted passway, that did not cause malignant neurodevelopmental abnormity but caused obvious viscus malformation as show in Fig. 8. Optical strategy in this work helped us deeply understand the toxicity mechanism of different sized nanoparticles and to support safety assessment of nanoscale materials.

Method
Characterization of sulfhydride nanoparticles Size, shape, and surface charges of 50 nm and 200 nm sulfhydride silica nanoparticles were rstly con rmed for further experiments. Size distributions and ζ-potentials in PBS of 50 nm and 200 nm sulfhydride silica nanoparticles were analyzed using Zetasizer Nano ZS (Malvern Panalytical). Infrared Spectroscopy (NICOLET iS50 FT-IR) was used to distinguish the sulfhydride groups on the SiO2 particle samples. For IR examination, aminated and alkylated nanoparticles were dehydrated under an infrared lamp, and then the dried particles were embedded with potassium bromide (KBr) and tableted to ~ 100 nm thick lms for further Infrared Spectroscopy. Infrared Spectra was acquired at 2 cm-1 resolution.

Sum frequency generation (SFG) assay
We then studied model cell membrane and SiO2/NPs interactions by SFG and supported lipid bilayers. Details regarding SFG experiment set-up and SFG theory are published and will not be repeated here To avoid the overlap of signals in the C-H stretching frequency region we developed symmetric double layer of 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) lipid bilayer, chemical structures were show in Figure S2. SFG only detects signals from media without inversion symmetry. SFG intensity can hardly be observed from POPC-POPC due to their approximate inversion symmetry. After SiO 2 particles were added into the sub-phase in contact with the lipid bilayers, the lipid bilayers will be interacted with the particles. Therefore strong SFG signals were detected in the C-H stretching frequency range (between 2800 and 3000 cm − 1 ).  ([ab109199], Abcam, USA)), respectively as the primary antibodies overnight, and then incubated with the HRP* Goat anti-Rabbit IgG secondary antibody (ImmunoWay, USA) for 3h. After incubation, the bands were then detected using the ECL Prime Western Blotting Detection System (BioRaD, USA), and the grey-scale was analyzed by Gel-pro analyzer 4.0.

In vitro analysis
In vivo experiments AB wild-type strain was used in this study. All lines were purchased from China Zebra sh Resource Center (CZRC, China). All studies involving animal manipulations were approved by the Institutional Animal Use and Care Committee of Nanjing Ruiying Runze Biopharmaceutical Technology Co., Inc. and followed the National Institutes of Health guidelines for the care and use of animals. 50 nm and 200 nm sulfhydride silica nanoparticles were co-cultured with zebra sh eggs at a concentration of 10 µg/ mL respectively. The shes were hatched out after 48 h co-incubation with the two kinds of nanoparticles, and continually these newly hatched zebra shes were co-incubated with 10 µg/ mL 50 nm and 200 nm nanoparticles respectively for seven days. The survival and mutation of eggs and shed were observed under the microscope.

Conclusion
Mitochondria was highly related to the origin and development of various metabolic disorders, and mitochondria targeting nano-system designing as drug and drug deliverer, and it is an emerging strategy for various complex diseases. Therefore, deep digging of the interactions between nanoparticles and mitochondria would be illuminating for nano-scale drug design targeting to mitochondria. When the nanoparticles interact with cells, cell metabolic arrest and repair process were related with the particles size, surface charge but the possible mechanisms have not been widely investigated. This work we studied the interactions between mitochondria membrane and different sized nanoparticles. We found that when interacted with 200nm particles, one lea et of the cells membrane were teared off and this process led to cell metabolic block. Defects on the mitochondria membrane could not initiate a p21mediated self-repair process, nor be repaired by another mechanism, Irreversible cell damage and animal injury can be observed. Differently, 50 nm nanoparticles did not disturb the mitochondrial membrane structure but entered mitochondria in a mitochondrial-protein-dependent manner. Slight organelle and biomolecule damages can be found from in vivo experiments, but from the in vitro experiments, it can be observed that the damages could be partially repaired by p21 mediated self-repair process. The results suggest that the destruction of membrane structure by the particles will cause irreversible biological damage, and particles entering cells through protein assisted process will increase the expression of cell cycle related proteins and cells self-repair can be observed from the in vitro experiments.
This work demonstrated that the interaction models of mitochondria and nanoparticles could determine further cellular responses to nanoparticles, so the e ciencies of nano-scaled drugs could be enhanced by altering the interaction models of nano-systems and mitochondria. In the future, mitochondria membrane proteins would also be carefully explored to con rm their roles in the active mitochondrial uptake of nanoparticles and provide new channels for safe and effective mitochondria targeting drug delivery.   The confocal images of C6 cells after being co-incubated with 50 nm ( rst and second row) and 200 nm (third and fourth row) sulfhydride nanoparticles for 8 h, 16 h, 24 h, and 32 h. Nanoparticles were labeled with red, and lysosomes were labeled with green ( rst and third row), and mitochondria were labeled with green (second and fourth row). White dashed circles presented the lysosome deformations, and white arrows pointed to the co-localizations of nanoparticles and lysosomes. Yellow dashed circles presented the mitochondria deformations, and yellow arrows pointed to the co-localizations of nanoparticles and mitochondria. Scale bar = 10 μm.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.