in vitro Chemical and physical toxicity of polystyrene microplastics in 1 human-derived cells 2

With the increase of plastics production, a variety of toxicological studies regarding the 16 microplastics have been reported since the microplastics could be ingested by the human body and 17 cause serious diseases. However, the previous studies have been mainly focused on the toxicity of 18 sphere type microbeads, which may be different from that of the randomly-shaped microplastics 19 in real environment. Here, we have conducted the in vitro toxicology for randomly-shaped 20 microplastics following the hypothesis that (1) physical cytotoxicity is affected from nano-/micro- 21 size roughness in polystyrene (PS) microplastics and (2) chemical toxicity is caused by chemical 22 reagents from microplastics. We found that the PS microfragments have chemical toxicity. Furthermore, the physical toxicity 38 by PS resulted in cellular membrane damage and correlated with statistically quantified-shape 39 roughness. Therefore, we newly suggested the additional physical toxicity of random shape of 40 microplastics. This provides the evidence of environmental and biological risks on random shape 41 of microplastics. 42

1 vial of PBMCs (approximately 1 × 10 7 cells) were thawed and culture in 5% CO2 incubator at 135 37°C for 1 day using 25 T-flask. Red blood cells (RBCs) were eliminated by RBC lysis buffer for 136 5 min prior to proceed in vitro assay. 137

2.4.Live-dead staining and viability analysis for PBMC 138
Cultured PBMCs were collected and centrifuged at 330 × g for 7 min. After RBC elimination, the 139 cells were seeded onto 96 well-plate at a density of 5 × 10 4 cells per 100 mL. After 1 day, 2 times 140 concentrated PS microfragments were additionally added to each well-plate to get desired 141 concentration (1 mg/mL, 100 µg/mL and 10 µg/mL), respectively. For negative control, 1% of 142 Triton X-100 was treated. PS-treated cells were cultured for 1 day and 4 days. For live-dead 143 staining, LIVE/DEAD TM viability/cytotoxicity kit for mammalian cells (Thermo Fisher Scientific, measured using a SpectraMax 340 PC plate reader (Molecular Devices, San Jose, CA, USA). The 167 cells were treated with 20% DMSO as a positive control. and measured visible wavelength absorbance at 540 nm. For positive control, 1% triton X-100 in 178 PBS was used. 179 For LDH assay, HDFs were seeded on 96 well-plate at a cell density of 1 × 10 4 cells/well. Cells 180 were incubated at 5% CO2 incubator for 1 day. After 1 day, PS microfragments solutions were 181 treated to cells and incubated for 1 day. To get LDH released solution from HDF, cell-PS mixture 182 in well-plate was centrifuged at 1500 rpm for 3 min, then we collected 100 mL of supernatant. The 183 supernatants were mixed with LDH assay reagent which is purchased Takara Bio Inc. (Shiga,184 Japan) and wait for 30 min in dark. The mixture changed to red color by a reaction of LDH was 185

2.9.TEM analysis 203
To confirm cellular behavior under PS treatment, transmittance electron microscope. For 204 preparation of cells for TEM, we have cultured HDFs in a 100 mm culture dish. At 95% of cell 205 confluency, 100 µg/mL of 5-25 µm PS microfragments were treated to HDFs for 1 day. Then, 206 cells were detached by 0.05% trypsin-EDTA and collected by centrifugation. 4% 207 paraformaldehyde was used for cell fixation.
Numerical data represented in the graphs represents mean value with error bar. Differences 210 between control (Ctrl) and the test groups were compared by unpaired t-test. A p-value > 0.05 was 211 considered non-significant (ns). The *, **, and *** symbols indicate p < 0.05, p < 0.01, and p < 212 0.001, respectively. All the experiments were conducted at least three times. 213 2.11. Statistical analysis of shape in PS microfragment 214 2.11.1. Local curvature 215 In this section, we briefly introduced our methods to define the roughness, along with the statistical 216 methods to compare the roughness measured from the microscopic images. First, let us introduce 217 the definition of the local curvature. In two-dimensional space, local curvature c at a pixel is 218 defined by 219 where * and ** are the first and second derivatives, respectively. In the analysis, the coordinates 221 are rotated so that the downward direction in Eucledean space is toward the center of the debris 222 when calculating the local curvature at every pixel. Local curvature values are obtained at each 223 pixel around the boundary. Negative curvature value at a boundary pixel indicates that the 224 boundary is bent outward (e.g., circle has constant negative curvature value along its boundary) at 225 the pixel, whereas positive one means that it is bent inward (i.e., dent). Hence frequent changes in 226 curvature over the boundary can be interpreted to be more "wiggly." We calculated the values in 227 (2) along the edge of the samples, and analyzed the values obtained the samples using a statistical 228 hypothesis test.  When concluding to reject null hypothesis in the preceeding process, it becomes an interest to find 256 the exact pair having different mean curvature-changes. The procedure is summarized below. Apparently, these tests are not readily available from the usual statistical software. We made an R 273 package named 'microplastics' for analysis. The package can be installed and is publicly available 274 at a webpage 1 along with a detailed description. 275

Characterization of PS microfragments 278
To demonstrate chemical and physical effects of PS microfragments to cells, we have prepared 279 random shape of PS classified according to size range; 5-25 µm, 25-75 µm and 75-200 µm, 280 respectively ( fig. 1). We already have demonstrated the cytotoxicity of spherical PS microparticles, 281 revealed that relatively small size of nanoparticles were shown cytotoxicity, hemolysis and 282 immune response in vitro. For 40 mm and 100 mm of PS microparticles could not happen cellular 283 uptake due to extremely their large size, resulting that it did not induce toxicity and immunity to 284 cells (data not published). However, the random shape of microplastics (fragments) accounts for 285 more than 50% of the total discarded microplastics in the real marine environments [22]; we 286 believed that there has a possibility about additional effects of microfragments to cellular behavior 287 which is different from that of sphere type of microplastics. Therefore, we have prepared random 288 shape microfragments to confirm chemical and physical effects to cells (Fig. 1). 289 to -80°C was demonstrated [24]. Brittle and stiff plastic materials are apt to take place plastic 295 fracture rather than deformation under the abrasive stress; therefore, it helps to make microplastic 296 effectively. The brittle PS pellets were treated to ball mill machine and undergone physically 297 breakage by the abrasive force generated from attrition and impact of ball. The ground PS microplastics by ball mill grinding have random shape of fragments, which are intended to make 299 secondary microplastics that are randomly decomposed by various mechanical stress such as wind, 300 wave action and animal bite [7]. Then, the ground PS microfragments are sorted by 3 different 301 sizes ranges. 302 In this study, we have demonstrated the toxicity of PS microfragments in two perspectives; (1) 303 chemical and (2) physical aspects ( fig. 1B). PS is normally synthesized by free radical 304 polymerization of styrene (ethylbenzene) in a bulk, using benzoyl peroxide (BPO) as initiator [25, 305 26]. According to the previous reports about toxicity of styrene and BPO, these are regarded as 306 toxic, mutagenic, and possibly carcinogenic potential [27, 28, 29, 30]. Therefore, even though 307 purification process undergoes, unpurified reagents in PS plastics could be released. We speculated 308 that these chemicals may induce cytotoxicity (Chemical effects). Additionally, we added one more 309 toxicity variable which could arouse from physical shape of microplastics such as sharpness and We have demonstrated that PS microfragments did not affect the cell death for 1-day culture ( fig.  352 3). At initial stage of cell damage by extrinsic materials, the immune cells release various kinds of 353 cytokines that can start immune response and spread to other cells. Interleukin-6 (IL-6) and tumor 354 necrosis factor-alpha (TNF-a) are commonly well-known pro-inflammatory cytokines that is 355 related to initial step of inflammation [37]. Therefore, we have measured the cellular behavior, 356 especially immune response, for 1-day PS treatment before cell death. In fig. 4, normalized IL-6 357 and TNF-a released from PBMCs were displayed. In overall, the amount of cytokines release from 358 the PBMC is strongly dependent on the concentration of PS microfragements we treated. For 359 positive control, we have treated 5 ng/mL of LPS which is constituent of the outer membrane of 360

gram-negative bacteria and promotes inflammatory responses effectively [38], indicating that 361
23.58 times (for IL-6) and 87.61 times (for TNF-a) higher responses were occurred. When we 362 added PS microfragments, the inflammation got induced at higher concentration; specially almost 363 similar IL-6 release amount was presented for 1 mg/mL of 75-200 µm PS as that of positive 364 control. Consequently, we have demonstrated that PS microfragments induce release of found that larger size of PS microfragments promotes more cytokine release, which is also 368 confirmed the PBMC viability at day 4 ( fig. 3B). It is implied that roughness or sharpness edge of 369 PS which is main factor for physical toxicity is not directly related to PBMC toxicity; chemical 370 effect is considered as main contributor. 371

Cellular survival rate in vitro analysis under PS microfragments treatment 372
The cellular survival rates for human dermal fibroblast (HDF) and cervical cancer cells (HeLa) are 373 displayed in fig. 5. Similar as viability of PBMCs, both cells were exhibited non-toxic behavior at 374 day 1, getting worse their viability at day 4. 55-60% and 64-67% of cellular viability were shown 375 in case of HeLa cells and HDFs, respectively. Surprisingly, some PS microfragments did not have 376 cytotoxicity at higher concentration, which is due to the fact that small hydrophobic 377 microfragments tend to form cluster in culture media condition at high concentration. Therefore, 378 the PS clusters at high concentration might not effectively impact on cellular viability which means 379 that physical toxicity of PS microfragments were reduced. However, the PBMC, which is 380 immature immune cells, is extremely sensitive against extrinsic materials and pathogens than HeLa 381 and HDF, resulting in toxicity at higher concentration of PS microfragments ( fig. 3)

. Consequently, 382
it is clearly confirmed that the PS microfragments have in vitro toxicity at long term culture 383

regardless of their concentration and size. 384
We also demonstrated the in vitro toxicity of PS to KATO III cells, the gastric cancer stem cells. When the PS is synthesized, styrene and BPO are commonly used. According to the previous 406 studies about relationship between oxidative stress and styrene or BPO [42,43]. Therefore, we 407 have considered that the impurities in PS microfragments may occur mitochondria damages at 408 short-term cultivation (1 day), resulting in ROS generation. It is noted that the viability of HDF at 409 day 1 is not concentration dependent and still remain high level ( fig. 5B), which is due to the fact 410 that oxidative stress by PS is not main factor to induce cell-death. Also, a moderate oxidative stress could induce mitochondrial metabolism is reported [44]. Therefore, it is not surprising that 412 moderate oxidative stress to cells get lead to mitochondrial metabolism, resulting that inverse 413 relationship between ROS generation and viability (measured by CCK-8 which basically confirms 414 level of mitochondria metabolism) was revealed. 415 To directly demonstrate the cellular response by PS microfragments, we have treated 5-25 mm 416 size PS to HDFs at a concentration of 100 µg/mL for 1 day and observed by TEM ( fig. 7). In case 417 of control (non-treated cells), though vacuoles and apoptotic structure of cells were detected, most 418 of cells are seemed to be healthy. In the contrary, cells occurred apoptosis and there are have 419 abnormal vacancies (blue arrow) or autophagosome (red arrow) in cytosol for PS treated cells. We 420 have considered that autophage is happened to overcome starvation and oxidative stress conditions 421 [45]. We did not find the microfragments in cells due to the size is over 5 µm. and it was confirmed that all three sizes of microfragements have different roughness and 500 sharpness, and accordingly it showed different physical cell toxicity. Finally, to demonstrate 501 toxicity of microplastics, we have shown that chemical and physical properties of microplastics 502 should be considered. 503

Availability of data and materials 512
All data generated or analyzed during this study are included in this published article and its 513 supplementary information files. 514 515