Titanium Nanoparticle Regulates Innate Immunity

10 Titanium dioxide (TiO₂) nanoparticles, widely incorporated in everyday products such as food additives 11 (E171), cosmetics, confectionary and surgical implants, represent one of the most prevalent forms of 12 engineered metal-based nanoparticles (MNPs) in modern environments. Despite their ubiquity, safety 13 data on their immunological impact remains limited. Of particular concern are sub -5 nm TiO ₂ 14 nanoparticles, which possess the capacity to cross cellular barriers and disrupt protein function. This 15 study investigates the interaction between 3.5 nm triethanolamine-terminated TiO₂ nanoparticles and 16 complement C3, a central protein in the human innate immune system. Using cryo -electron 17 microscopy, we reveal direct binding of TiO₂ nanoparticles to the surface of C3, leading to structural 18 restrictions that hinder its activation and subsequent complement cascade progression. Biochemical 19 and haemolytic assays confirm that these nanoparticles act as alternative pathway-specific inhibitors, 20 impairing normal immune function. These findings underscore the potential immunomodulatory effects 21 of TiO ₂ nanoparticles and highlight the need for thorough evaluation of their biological impact, 22 particularly given their widespread human exposure. 23 24

25 Throughout evolution, humans have been exposed to natural metal and metalloid nanoparticles formed 26 during extreme environmental events such as volcanic eruptions.1 However, engineered nanoparticles, 27 such as silver, gold, zinc and titanium oxides, are products of modern science and have only been 28 introduced into the environment within the last few decades. In particular, titanium dioxide (TiO2) 29 nanoparticles have usage in an immense range of everyday products such as a whitening food 30 additive (E171) , confectionary, paint, hygiene and personal care products. 2-5 Despite the recent 31 increase in anthropogenic metal -based nanoparticles (MNPs) and increased daily exposure , very 32 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint limited safety information is available concerning their biological effect. 6 Synthesised TiO 2 33 nanoparticles can range from a couple of nanometres in diameter through to several hundred 34 nanometres, with each size class capable of eliciting distinct biological responses. 7 Toxicity 35 considerations are particularly relevant for TiO2 nanoparticles <5 nm in diameter due to their ability to 36 cross the blood-brain barrier and penetrate cellular walls.8, 9 A non-exhaustive list of MNP effects on 37 proteins ranges from misfolding, enzymatic inhibition, dysregulation, aggregation and structural 38 alteration.10-13 Regardless of their biological influence, the ultimate fate of MNPs is bio-clearance and 39 destruction by the host immune system. Yet, some metal nanoparticles have shown to compromise 40 important proteins in the immune system, specifically, within the complement system.14-16 41 42 Over the course of 500 million years, the complement system in vertebrates has evolved into the highly 43 advanced and intricate defence mechanism found in humans today. This system involves the 44 coordination of ~50 different immune regulation factors and proteins. 17-19 The complement system 45 functions in tissue, plasma, intracellularly and demonstrates exceptional sensitivity to antibodies, 46 foreign microbial sugars and pathogens. 20 The C3 zymogen, an inactive enzyme precursor that 47 requires activation to function, is perhaps the most significant protein in the entire complement 48 cascade. It exists as the most abundant complement protein found in the blood of healthy humans at 49 concentrations above 1.0 mg/mL.21 For activation of C3 to occur, a small 9 kDa domain (C3a) must be 50 cleaved from the main 176 kDa chain (C3b) by a C3 convertase or serine proteases.22-24 This promotes 51 a significant conformational shift of C3b and the exposure of a highly reactive thioester which is 52 responsible for the opsonisation (tagging) of foreign particles for destruction by downstream immune 53 system machinery.25 54 55 Our groups have previously shown that triethanolamine terminated titania (TATT), consisting of size-56 uniform 3.5 nm TiO2 nanoparticles, upregulate the coagulant system, a system intricately linked with 57 the complement pathways. 26-28 Consequently, these nanoparticles demonstrated excellent wound 58 healing properties resulting from their impact on blood platelet aggregation.28 Other research indicates 59 that TATT can either trigger or inhibit blood platelet aggregation in vitro and in vivo.29, 30 Notably, macro-60 scale titania from implanted surfaces or environmental exposure can be gradually etched away by 61 bodily fluids into smaller nanoparticle oxides, potentially contributing to systemic nanoparticle 62 exposure.31-33 Given the increased human exposure to TiO 2, and the close crosstalk between the 63 complement and coagulant systems, 27, 34 we investigated whether the titania nanoparticles imparted 64 any effect on the structure and function of C3. 65 66 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint Here we present the seminal cryo -EM structure of a TiO 2 nanoparticle bound to wild -type human 67 complement C3. Biochemical and haemolytic assays show that these nanoparticles are responsible 68 for C3 immune activity in blood serum and function as an alternative pathway-specific inhibitor. These 69 findings are further supported by structural investigations which reveal an outer -shell interaction 70 between the nanoparticles and C3, restricting protein motion and impeding the formation of 71 complement complexes necessary for continued immune function. 72 73

74 Biophysical analyses of complement C3 75 Prior to structural analyses, baseline characterisations of native C3 were undertaken, including melting 76 temperature (T m) and hydrodynamic radius (R h) at physiological and extreme conditions. DLS 77 experiments showed that within the range of 25 – 55 °C, the protein maintained a constant diameter of 78 ~12 nm (120 Å) which, surprisingly, is 40 Å smaller than reported inactive C3 structures (Figure 1a).35, 79 36 Polydispersity values were consistently low and the hydrodynamic size remained monodispersed 80 and unchanged after 90 minutes. Thermal stability tests demonstrated that heating the sample 81 incrementally produced notable and rapid C3 aggregation at around 58 °C, with 6 nm cumulative radii 82 protein reaching 100 nm at 62 °C. This result was supported by nanoDSF measurements which 83 showed fast protein denaturation and a resulting Tm of 60 °C (Figure 1b). 84 85 SDS-PAGE and size -exclusion chromatography were used to confirm whether the smaller-than-86 expected protein diameter was due to missing or cleaved domains besides the 9 kDa anaphylatoxin 87 domain of C3a. Despite possessing the same number of domains, the inactive and hydrolysed states 88 of complement C3 have been shown to elute at significantly different times simply owing to 89 considerable inter-domain reorientation.37-39 Following protein digestion under reducing conditions, the 90 gel showed all bands and their corresponding domains were accounted for. The size -exclusion 91 chromatogram further confirmed that all domains were present with the 2 separate elution peaks being 92 attributed to both inactive and hydrolysed C3. Importantly, no smaller peaks or contaminants were 93 detectable after the final C3-associated elution (ESI Figure S1). 94 95 Although all biophysical and purity analyses showed the C3 sample as homogenous, conflicting 96 evidence suggested the protein was significantly smaller than the crystallographic and cryo -EM 97 structures already reported. To explain this discrepancy, we turned to single -particle cryo-EM and 98 determined the structure of native complement C3 under the same physiological conditions as all 99 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint previous analyses. To our surprise, we observed several states of the complement protein achieving 100 an extreme range of inter-domain motion. 101 102 Full-range of C3 flexibility revealed 103 The most compact state , shown in Figure 2a , appears as the dominant species in solution and 104 represents the inactive state conformation. This arrangement shows the thioester domain (TED) 105 compressed and adjacent to the macroglobulin ring (MG) domains of MG2 and MG6. The interactions 106 between the TED and MG domains are facilitated mainly through dipole contacts rather than strong 107 ionic interactions. This feature allows for the movement experienced by the MG domains to proceed 108 freely. In fact, of the three interacting domains in this region, there are almost no electrostatic 109 interactions or hydrogen bonds formed. Instead, a select few complementary hydrophobic and 110 hydrophilic pockets of amino acids are enough to stabilise this conformation (Figure 3). 111 112 3D Flexibility analysis shows two distinct states attributed to MG ring domains 1 - 6 which possess the 113 ability to locally swivel and pivot in -place (ESI Figure S2). This was an unexpected finding, as the 114 structurally well-characterised macroglobulin ring domain is reported as rigid.40, 41 Combining both 3D 115 variability and 3D flexibility analyses, the global hinge -like motion between the contracted and 116 expanded states could be resolved in addition to the flexible local motions experienced by each 117 domain throughout the full movement. The MG domain movement between each conformational 118 endpoint was found to reach up to 28 Å, with the MG2 and MG6 connecting arm region shifting by 12 119 Å. Minor classes found in 2D classification suggest that the extension may actually be greater than 120 this, however these views were too rare to be reliably incorporated into the 3D variability constructions. 121 In biological systems, the thioester domain acts as a “warhead” to connect to the detected foreign 122 particle. The current structural series shows the MG2 and MG6 domains assist with this tagging 123 process by uncovering the thioester region and exposing it for binding. These domains act as an 124 extending arm which moves the bulky MG regions away from the TED by roughly 30°. Furthermore, 125 the MG7 domain exhibits a slight bending motion which affords more solvent-accessible space for the 126 linker domain to extend into. 127 128 Each “compact” and “stretched” conformation could be reconstructed to a FSC0.143 of 3.0 Å and 4.0 Å, 129 respectively (ESI Supplementary Table 1). The local resolution range varied more depending on the 130 degree of flexibility demonstrated in each conformational state. The alpha chain of C3 remained rigid 131 with local resolution reconstructions of 2.5 Å possible. The flexible MG ring domains could be 132 reconstructed to approximately 5 Å in the extended pose while a higher resolution of 3.2 Å was 133 achievable in the contracted, more rigid conformation. 134 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint Biophysical assays suggest TiO2 nanoparticles interact with human Complement C3 135 The alternative pathway of complement consists of continuously activated C3b and inactive C3 at low 136 concentrations. Therefore, it presented as an attractive testing ground for assessing the effect of TATT 137 nanoparticles on the immune system. Repeating the DLS and thermal unfolding experiments on TATT 138 in the same buffered system as C3 proved unsuccessful as the presence of salt promoted the near -139 immediate aggregation and gelation of the nanoparticle colloid throughout a range of concentrations 140 (ESI Figure S3a). Instead using only filtered water, the nanoparticles showed no signs of aggregation 141 which was confirmed by DLS as the TiO 2 particles were found to be of a homogenous diameter and 142 stable at room temperature for 2 hours (ESI Figure S3b). 143 144 After confirming that C3 was also amenable in only filtered water, several concentrations of TiO 2 145 nanoparticles were added to differing aliquots of 0.4 mg/mL C3 for DLS and nanoDSF analysis (ESI 146 Figure S3c). DLS results showed no appreciable indications of protein or nanoparticle aggregation. 147 NanoDSF showed near identical Tm inflection points for C3 in both water and buffered solution over a 148 range of temperatures. Importantly, the addition of TiO 2 resulted in an increase of Tm by ~7 °C for 149 several separate concentrations of nanoparticle, demonstrating a meaningful interaction and 150 significant thermostabilising effect of the TiO2 species on C3. 151 152 Biochemical assays show TiO2 nanoparticles downregulate C3 activity 153 To test the effect TATT may have on C3 function, the formation of the alternative pathway (AP) 154 convertase, C3bBb was considered. In the presence of Mg2+ ions, factor B (FB) binds to C3b forming 155 the foundation of an AP convertase, allowing for the action of factor D (FD) to cleave FB into fragments 156 Ba and Bb. In turn, C3bBb is formed with a substrate specificity for native C3, the convertase cleaves 157 off C3a generating C3b. Using C3b as a positive control, native C3 (1 µM) was incubated with and 158 without TATT (1 µM) prior to the addition of FB and FD, relying on C3(H 2O) contamination or 159 spontaneous hydrolysis occurrence to enable convertase formation and Bb generation. To analyse 160 the effect of TATT, the ability for FB binding to native C3 was assessed by comparing peak area values 161 of FB and fragment Bb generation detected by a monoclonal antibody against Factor Bb neoepitopes. 162 The capillary electrophoresis data generated combined area for Factor B and Factor Bb that were 163 converted to percentage area in the Compass for Simple Western (v 6.1.0) programme after peak 164 naming. Figure 4A conveys the percent of Factor B remaining, indicating a reduction in the AP 165 convertase formation and activity, reducing Factor Bb generation. In the presence of TiO2, after 5 166 minutes, 95.1% FB remains (Native C3; 71.8%, C3b; 64.3%), 30 minutes 69.6% remains (Native C3; 167 13.9%, C3b; 15.6%) and 60 minutes 18.5% remains (Native C3; 11.9%, C3b; 8.9%). 168 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint Haemolysis of red blood cells by C3 is inhibited with TiO2 169 Building on the AP inhibition, the capacity for native C3 to maintain its haemolytic properties was 170 tested. Native C3 was incubated with incrementally increasing concentrations of TATT (0.5, 1, 2, 4, 8 171 and 16 µM). Rabbit red blood cells (RBCs) were used in this experiment as they fail to inhibit human 172 complement and therefore become rapidly opsonised with C3b, enabling AP convertase and 173 subsequent membrane attack complex formation (MAC) assembly, leading to haemolysis. In this 174 serum dependant assay, an absorbance at 540 nm was recorded from native serum (total lysis), C3 175 depleted serum (blank) and the test samples of C3 with or without nanoparticles. The test samples are 176 reported as a percentage of total haemolysis and show a significant reduction in haemolysis as 177 nanoparticle concentration increases compared to neat native C3 (4 µM TATT p= 0.0187; 8 µM TATT 178 p= 0.0268; 16µM TATT p= 0.003) as shown in Figure 4B. 179 180 Direct observation of TiO2 nanoparticles interacting with human C3 181 Using these optimal conditions, C3 and TiO2 at differing ratios were applied to Quantifoil EM grids and 182 plunge frozen in liquid ethane in preparation for cryo -EM. Despite their modest size of 3.5 nm, the 183 nanoparticles were readily visible as intense white spots on the micrograph with the ~185 kDa protein 184 in close proximity , barely visible as a blurred grey signal (ESI Fig ure S4a). Additional sample 185 processing in RELION 5.0 provided more discernible particles of both C3 protein, TiO2 nanoparticles 186 as well as classes which showed both in close proximity (ESI Figure S4b, S4d). Several ATLAS and 187 eucentric height scans on different grids suggested that both particle distribution and contrast were 188 favourable in thicker ice. It was immediately obvious from the micrographs that the nanoparticles 189 further contributed to sample heterogeneity as they did not appear to interact with the protein in a 190 uniform manner. This became evident from particle picking and 2D classification as several potential 191 classes showed up to 3 nanoparticles may be interacting with a single protein, although at low 192 alignment resolutions, this may be attributed to increased particle densities in the thick ice regions. 193 Using a non-standard processing strategy in RELION 5.0 (Materials and Methods), we obtained a low-194 resolution electron density map suitable for protein and nanoparticle fitting. 195 196 Despite a loss of resolution, the C3 core particle density, comprising of the thioester and surrounding 197 domains, matched very well with the previously-collected apo-C3 (amino acids ~550-1641). Similarly 198 to apo-C3, the macroglobulin ring density remained challenging to capture, however the location of 199 the missing domains could be approximated as the MG2 and MG6 domain densities connecting the 200 TED and MG ring, could be accounted for. The predominant pose of the C3 protein appears to be in 201 the most compressed state, where the MG2 and MG6 domains rest directly parallel to the thioester 202 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint domain. Several charged residues along the linker region form complementary interactions between 203 oppositely charged residues on the external face of the thioester domain, further stabilising this 204 conformation (Figure 5A, B). The vast majority of 2D classes also support this as we were unable to 205 locate any rare views of extended protein in close proximity to nanoparticle density. In fact, it seems 206 as though these extended forms of C3 are only observable in the absence of TiO2. 207 208 Despite the lack of a defined binding site, protein flexibility and alignment challenges from spurious 209 classes, a clear density corresponding to the diameter of TiO2 was located adjacent to the C3 electron 210 density. Owing to the signal thresholding during data processing, the most intense signal from the 211 interior of the nanoparticle has been eliminated, leaving just the periphery which resulted in a density 212 with a hollow interior. The pseudo-spherical density can be confidently attributed to the nanoparticle 213 rather than the missing macroglobulin ring domains as the TiO2 density not only appears as the most 214 intense after map contouring compared to protein signal, but is approximately 40 Å from the expected 215 location of the macroglobulin ring. Furthermore, this region is situated well outside the latent axis of 216 motion displayed by the apo-C3 structure. 217 218 Overlaying the reported inactive structure with the nanoparticle-bound C3 shows that the majority of 219 the structure is unchanged, however producing a calculated RMSD of 4.27 Å relative to the MG7 220 domains of the native and TiO2 structures (Figure 6A, B). This variability is largely due to the apparent 221 outward shift of the MG7 domain (residues 807 -911), as the RMSD of the remaining protein is 0.912 222 Å. This macroglobulin domain is connected by two linker regions and is reported to move during 223 activation of C3, demonstrating a rotation of approximately 35°. This movement is facilitated by the 224 excision of the polypeptide chain between MG7 and MG8 by factor I, thus allowing the two MG 225 domains to nearly exchange positions. 36 Within an inactive C3, this domain is situated between the 226 linker and thioester domains, and has already been shown to move within the inactive state in the apo-227 C3 structure. The location of MG7 in the current structure shows an outward shift of approximately 9 228 Å in the opposing direction to MG8. As the linkers are still intact, a slight rotation of this domain is 229 afforded to form an interaction with the nanoparticle which is positioned 6 Å adjacent to this domain. 230 Given the significant electronegative nanoparticle, it was anticipated that any interaction formed 231 between the species would occur at an acidic or phosphorylated region of the protein. A close 232 inspection of the solvent-accessible residues on MG7 indeed confirm that the nanoparticle is forming 233 an outer-sphere complexation with several electropositive residues (His846, Lys857, Arg858, Arg859, 234 His860) in addition to hydrogen bonding interactions with several more residues (Cys851, Thr855, 235 Thr856, Gln862). 236 237 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint Discussion/Conclusion 238 At present, this work represents the first example of a protein-nanoparticle interaction using a protein 239 which lacks an engineered binding site specificity for the nanoparticle substrate. As such, this had 240 evident repercussions on particle alignment and 2D classification challenges for non -specific 241 interactions between the protein and nanoparticles.42-44 Following a non-standard processing strategy 242 for the C3 -TiO2 particles, a low -resolution reconstruction shows the interaction between C3 and a 243 single TiO2 nanoparticle. Despite the lack of a dedicated binding site for the nanoparticle, the location 244 of TiO2 does not appear to be random as it is situated adjacent to an electropositive, solvent-accessible 245 region of the MG7 domain. This location demonstrates a charge complementarity between the highly 246 electronegative, oxygen-coated, surface of the titania nanoparticle and the solvent-accessible arginine 247 and lysine residues on MG7. Although the resolution is not sufficient to unambiguously determine the 248 conformation of the amino acids participating in the interaction, the location of the nanoparticle 249 throughout all classification steps remained consistent, indicating this interaction as non-transient and 250 stable. Notably, the C3 has remained in an inactive conformation as the thioester domain, during 251 activation, would produce a C3 diameter of ~160 Å. In the current structure, the compact density 252 produced a diameter of 100 Å, much more similar to the size of reported inactive C3.36 These structural 253 findings were supported through the c ombination of Wes and haemolysis results : a significant 254 functional deactivation of the C3 is imparted by the nanoparticle. In the context of the alternative 255 pathway, this can be explained by the binding site obstruction for the C3 convertase, C3bBb. In the 256 reported C3 convertase structure (2WIN),45 the MG7 domain and C-terminal domain of C345C form a 257 junction where Factor B is able to bind. Analysis of these inter-domain interfaces shows that they are 258 stabilised mainly through hydrophilic interactions, particularly hydrogen bonds and polar side chain 259 contacts, rather than by hydrophobic or charged interactions. Following this, Factor D cleaves bound 260 into Bb and Ba, with the immediate expulsion of Ba. This leaves Bb attached to C3 to form the full C3 261 convertase of C3bBb, prepared to activate neighbouring C3 zymogens. In the current structure, the 262 titania nanoparticle is situated almost exactly where Factor B would normally interact with C3 to 263 activate the zymogen and form the convertase. Furthermore, the metal oxide nanoparticle, albeit not 264 much larger than a Bb domain, is an extremely electronegative species as the titanium core is coated 265 by oxygen. This results in a steric and electrostatic repulsive effect on the binding of Factor B to C3, 266 producing no interaction between the two species. This was supported with a consistent melting 267 temperature from nanoDSF of Factor B in water, buffer and in the presence and absence of TATT 268 nanoparticles (ESI Figure S5). Additionally, the close proximity of the titania nanoparticles to MG7, a 269 central domain involved in protein flexibility, has implications for the native motion of C3. The presence 270 of the nanoparticles may sterically hinder conformational shifts in MG7 or disrupt intramolecular 271 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint interactions necessary for the propagation of structural changes, potentially altering the activation or 272 regulation of the complement cascade. Relatedly, a recent report managed to deny the translocation 273 of C3a and rearrangement of C3 by binding a specific nanobody to native C3.46 To achieve this, the 274 nanobody was found to bind at the MG7 domain in the same position as the TiO2 nanoparticles. These 275 interferences with the dynamic regions upon the protein could impair the ability of C3 to undergo the 276 structural rearrangements required for its conversion to C3b or subsequent binding events. Although 277 the C3-TiO2 data are low resolution, it is clear by the predominant protein conformation that C3 indeed 278 remains in a native state. Recent evidence from our lab suggests that C3 can undergo conformational 279 shifts and adopt a C3b-like state even in the absence of Factors B and D, which promote cleavage. 280 Evidence of this can be extrapolated from both DLS and cryo-EM results. Particle radii of C3 from DLS 281 measurements remained reliably compact in the presence of nanoparticles when compared to apo 282 C3. This was further supported by cryo-EM electron densities, as all obtained classifications, both 2D 283 and 3D, showed the inactive (compact) form of C3. No evidence from the microscopy or biophysical 284 data suggests titania nanoparticles are associated with C3 in the extended conformation or that they 285 promote a major conformational shift to an active C3b -like state. Accordingly, given the inhibitory 286 biochemical and structural results, it is likely that these small, TATT nanoparticles are recognised as a 287 ligand inhibitor by C3, rather than as a foreign surface. 47, 48 Further experiments are necessary to 288 confirm whether this perception is due to the size, composition or charge of the MNPs. Interestingly, 289 the Compstatin family of cyclic peptide inhibitors, specific for C3, are shown to bind in a completely 290 separate area of the protein. These organic inhibitors are all documented to bind between the MG4 291 and MG5 domains, approximately 60 Å away from the observed TiO2 binding site at the MG7 domain. 292 Yet, Compstatin binding does not arrest the conformation of C3, but instead prevents cleavage by 293 convertases ( C4b2a in the classical/lectin pathway and C3bBb in the alternative pathway). 294 Accordingly, the titania nanoparticles may act to stabilise C3 in a compact, inactive form by 295 preferentially binding to flexible domains like MG7, thereby biasing the conformational landscape 296 toward a closed state. Such stabilisation could have functional implications, potentially dampening C3 297 activation by limiting the structural transitions required for convertase formation or cleavage. 298 299 The complement and coagulation systems are deeply intertwined, functioning not as isolated 300 cascades but as a coordinated network that responds dynamically to tissue injury. Crosstalk between 301 these systems ensures rapid haemostasis, immune cell recruitment, and tissue repair - functions that 302 are tightly regulated to prevent chronic inflammation or thrombosis. Our findings of TATT nanoparticles 303 inhibiting C3 activity introduce a novel mechanism by which these particles may influence the balance 304 between inflammation, coagulation, and regeneration, for example at wound sites. Our data suggest 305 that TiO₂ exerts active immunomodulatory effects by suppressing complement activation at the level 306 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint of C3. Furthermore, these results raise pertinent questions about the immunological effects these 307 nanoparticles may have on human immune systems. The inhibition of C3 by TATT nanoparticles 308 implies a downstream dampening of both the classical and alternative complement pathways, which 309 could translate into decreased anaphylatoxin generation (C3a, C5a), reduced opsoni sation, and 310 eventual diminished membrane attack complex (MAC) formation. This modulation could directly alter 311 immune cell recruitment and activation at injury sites, potentially shifting the local environment from a 312 pro-inflammatory to a more regenerative state. 313 314 These findings reframe TiO ₂ not just as a scaffold or surface -modifying agent, but as a bioactive 315 nanomaterial capable of modulating complex immune -coagulation networks. Our findings align with 316 emerging evidence that other metal or metal oxide-based nanoparticles, such as silver, zinc oxide, 317 cerium oxide, and gold, also exhibit immunomodulatory properties through diverse mechanisms.49-51 318 These insights support a paradigm shift in the use of nanomaterials in regenerative medicine, 319 positioning them not merely as passive scaffolds or antimicrobial agents, but as tuneable modulators 320 of host immune and haemostatic responses. 321 322

323 C3 isolation, purification and storage 324 Native C3 was isolated and purified in-house from human plasma according to previous reports.52 The 325 final step of purification involved cation exchange chromatography which was performed using a Mono 326 S 5/50 GL column (GE Healthcare, Bio-Sciences AB, Uppsala, Sweden). The column was equilibrated 327 with 5 column volumes (CV) buffer A (20 mM phosphate buffer pH 6.8), followed by sample application 328 (0.25 CV). The column was washed with 3 CV of buffer A and then the elution was started using a 329 linear gradient of buffer B (20 mM phosphate buffer pH 6.8, ranging from 0 to 0.85M NaCl). The protein 330 elution was monitored with an UV detector at 280 nm and fractions of 0.5 mL were collected during 331 the elution procedure. C3 was precipitated in a low salt and pH buffer for storage by dialysis against 332 a 40 mM phosphate buffer with ionic strength 0.05 (mS) and pH 6.0 at +4°C with a buffer volume at 333 least 10-fold higher than the volume of C3. After 24 h the buffer was replaced with fresh buffer and 334 dialysis continued for another 24 h until a visible precipitate was formed. The stock solution of 335 precipitated native C3 was then suspended (without centrifugation) and aliquoted before storage at -336 80°C. In order to dissolve the precipitated C3, 1-part VB++ (5x stock solution) was added to 4-parts of 337 precipitated C3 immediately before use.37 338 339 340 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint Preparation of sol-gel titania 341 The synthesis of the uniform titania colloids used in this work was performed by following a previously 342 reported preparation.53 The precursor of titanium ethoxide (5 mL) was dissolved in anhydrous ethanol 343 (5 mL) followed by the addition of 1.5 mL of triethanolamine during continuous stirring. A hydrolysing 344 solution (1 mL) was produced by mixing nitric acid (0.5 M) with ethanol (2 mL). The resulting lightly -345 yellow, transparent solution contained 120  mg/mL TiO 2 according to TGA measurements. 28 The 346 obtained product was of high-enough purity to not require additional purification. 347 348 Biophysical measurements 349 DLS, thermal unfolding and nanoDSF measurements were performed on Prometheus Panta 350 (NanoTemper, Munich, Germany) in technical triplicate. For apo -protein experiments, 0.5 mg/mL of 351 C3 in filtered MQ water or buffered solution was added to separate capillaries at room temperature. 352 For TiO2 experiments, nanoparticles in a stock solution of 1 mM were suspended in filtered MQ water, 353 briefly incubated with C3 and added to capillaries. All raw data were exported for analysis and graph 354 generation in GraphPad Prism version 10.0. 355 356 Biochemical assays 357 For capillary electrophoresis analysis by Wes immunoassay and alternative pathway haemolysis, 358 native C3 (1 µM) was preincubated with varying micromolar concentrations of TiO2 nanoparticles in 359 VBS++ for 30 minutes at 37°C. 360 361 For capillary electrophoresis, 70 µg/mL of C3, C3b and C3 incubated with nanoparticles were added 362 to a preparation of Factor B (100 µg/mL) (Complement Technology Inc) and Factor D (0.5 µg/mL) 363 (Complement Technology Inc) in VBS ++ and incubated over a time course of 5, 30 and 60 mins. 364 Individual time points were stopped with the addition of EDTA to a final concentration of 10 mM. The 365 Wes immunoassay was performed under reducing conditions with a primary mouse monoclonal anti-366 human complement Factor Bb antibody (2 µg/mL) (Bio-Rad Laboratories). 367 368 The alternative pathway haemolysis assay was set up and adapted according to previous methods,54, 369 55 where 25 µL of C3 (200 µg/mL) preincubated with 0.5, 1, 2, 4, 8 and 16 µM of TiO 2 nanoparticles 370 were added to 50 µL C3 depleted serum (Complement Technology Inc) followed by the addition of 371 100 µL 50% rabbit erythrocytes (v/v) in GVB Mg EGTA. These samples were then incubated in a 372 shaking water bath for 20 mins at 37°C at 200 rpm before being stopped with the addition of cold VB-373 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint EDTA. The samples were then centrifuged at 860 g for 10 mins at 4°C and absorbance measured at 374 540 nm on a CLARIOstar Plus (BMG Labtech). 375 376 Cryo-EM structure determination of native C3 377 C3 protein buffer was exchanged from VB++ to filtered MQ water and concentrated to 0.4 mg/mL before 378 grid freezing. Cryo -EM samples were prepared on UltrAuFoil R1.2/1.3 300 mesh holey gold grids 379 (Quantifoil Microtools). Once the grids were glow discharged with a PELCO easiGlow instrument (20 380 mA, 0.4 mBar, 1.5 min), 3 µL of C3 was applied to the glow -discharged grid s. Vitrification was 381 performed with a Vitrobot Mark IV (Thermo Fisher Scientific) at 4 °C and 95 % humidity. Grids were 382 blotted in duplicate for 3 seconds and plunged into liquid ethane. Following vitrification, the grids were 383 clipped into autogrid cartridges (Thermo Fisher Scientific) for use with autoloader systems. 384 385 All grids were ATLAS screened on a Glacios 200 kV cryo-electron microscope mounted with a Falcon 386 III direct electron detector. Data were acquired in EER format at a pixel size of 0.7463 Å/pixel with a 387 dose rate of 0.91 electrons pixel-1 sec-1. Approximately 1200 micrographs had 2x binning applied and 388 were processed using exposures fractionated into 50 frames for motion correction. This enabled for a 389 rapid assessment of the sample quality, particle density and distribution before high-resolution data 390 were collected. For full data collections, a 300 kV Titan Krios G2 (Thermo Scientific) transmission 391 electron microscope equipped with a K3 detector (Gatan) and a 20 eV BioQuantum energy filter 392 (Gatan), was used. Data were collected with the following settings: 0.648 Å/pixel (190 kx 393 magnification), 51 e−/Å2 total electron dose, −0.5 to −2.5 μm defocus range. Automated collection was 394 performed using the EPU software (Thermo Scientific) with 30° stage tilting. A total of 17,420 movies 395 were collected. 396 397 During data collection, cryoSPARC live (cryoSPARC v4.1.1) was employed for on -the-fly data 398 processing for the first few hundred micrographs using the same settings as for the initial screening 399 dataset. Processing was performed in cryoSPARC v4.6.056 and included initial Patch Motion Correction 400 and Patch CTF being performed before micrograph curation. Micrographs were manually curated 401 based on CFT estimations, total frame motion, defocus, ice thickness and contamination. Using the 402 obtained template from the screening dataset, all particle picking was performed with template picker 403 and the Topaz 57 wrapper template picker in cryoSPARC . Once data collection was completed, the 404 template-based picks totalled ~5 million and were 2x downsampled and extracted with a 400 x 400 405 box size before being Fourier-cropped to 250 x 250 pixels . Particles were classified into 150 2D 406 classes (10 Å minimum separation distance, 170 - 200 Å circular mask inner -diameter, 2 final full 407 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint iterations, 40 online -EM iterations, 400 batch size per class). Following 2D class selection, this 408 classification/selection procedure was repeated twice more which resulted in ~ 1.9 million particles 409 suitable for Ab-initio reconstruction. Ab-initio reconstruction was followed by heterogenous refinement 410 using a few poor density “dummy” classes as seeds for junk particles. Two further rounds of 411 heterogenous refinement seeding resulted in 818,343 particles which displayed discrete 412 heterogeneity, which was elucidated using 3D variability analysis. 58 Masked refinement and 413 reconstruction jobs reliably produced densities at resolutions of ~2.5 Å. 3D variability analysis 414 demonstrated the inherent continuous disorder of the particles and was used to separate individual 415 latent dimensions of motion attributed to approximately 3 states experienced by the particles. Particles 416 responsible for each state were isolated and reconstructed independently to produce each individual 417 C3 pose. 418 419 A model was constructed by using the reported inactive C3 structure (PDB 2A73)36 as a starting model. 420 Owing to the large conformational differences between the current particles and the reported structure, 421 the domains were individually docked and fitted into the electron density with UCSF ChimeraX version 422 1.9.59 These hand-fitted domains were then quantitatively fitted into the density and refined using the 423 DockEM60 and Refmac Servalcat61 packages within CCP-EM 1.6.0.62 The Coot63 program was used for 424 manual rebuilding and inspection of the model in the electron density before a final real -space 425 refinement run in PHENIX. 64 The relevant structural validation metrics are shown in supplementary 426 Table 1. 427 428 Cryo-EM structure determination of C3-TiO2 429 As previous, C3 protein buffer was exchanged from VB++ to filtered MQ water and concentrated to 0.4 430 mg/mL before grid freezing. Cryo-EM samples were prepared on UltrAuFoil R1.2/1.3 300 mesh holey 431 gold grids (Quantifoil Microtools). While the grids were glow discharged with a PELCO easiGlow 432 instrument (20 mA, 0.4 mBar, 1.5 min), C3 was incubated with nanoparticles at an equimolar 433 concentration for 5 minutes on ice before 3 µL of the mixture was applied to the glow-discharged grids. 434 Vitrification was performed with a Vitrobot Mark IV (Thermo Fisher Scientific) at 4 °C and 95 % humidity. 435 Grids were blotted in duplicate for 3 seconds and plunged into liquid ethane. Following vitrification, 436 the grids were clipped into autogrid cartridges (Thermo Fisher Scientific) for use with autoloader 437 systems. 438 439 A 300 kV Titan Krios G2 (Thermo Scientific) transmission electron microscope equipped with a K3 440 detector (Gatan) and a 20 eV BioQuantum energy filter (Gatan), was used for the full data collections. 441 Data were collected with the following settings: 0.648 Å/pixel (190 kx magnification), 51 e −/Å2 total 442 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint electron dose, −0.5 to −2.5 μm defocus range. Automated collection was performed using the EPU 443 software (Thermo Scientific) and a total of 17,706 movies were collected. Processing was performed 444 in cryoSPARC v4.6.0 and included initial Patch Motion Correction and Patch CTF being performed 445 before micrograph curation. Micrographs were manually curated based on CFT estimations, total 446 frame motion, defocus, ice thickness and contamination. 447 448 A visual inspection of the micrographs and blob-picked classes in cryoSPARC suggested there was 449 a clear picking bias which made further classification and alignment impossible (ESI Figure S4a). As 450 the electron dosage charged the TiO 2, the nanoparticles appeared as bright hotspots on the 451 micrographs which almost completely overpowered signals attributed to protein. Attempts to use the 452 previous C3 reconstruction as a picking template were unsuccessful due to the picking algorithm 453 being unable to locate any signal related to protein. Analysis of the protein and TiO2 relative intensities 454 in EMAN265 showed that there was approximately a 1:10 intensity disparity between the protein and 455 nanoparticles, respectively (ESI Figure S4c). Attempts to enforce non-negativity, solvent-clamping and 456 template picking using native C3 prior to 2D classification proved unsuccessful as the picking was 457 strongly biased toward the most intense signals of the nanoparticles. To the best of our knowledge, 458 cryoSPARC does not have the ability for the user to manually threshold and normalise intensities, and 459 for this reason we attempted to complete the data processing using RELION 5.0. 66 Following a 460 standard pre -processing strategy up to CTF estimation, Topaz was trained on a small subset of 461 particles and once training converged, particles were extracted from all micrographs. The extraction 462 job settings were not default: particle box size 384 pixels, diameter background circle 40 pixels, 463 stddev white dust removal -1.0, stddev black dust removal 3.0, with normalisation on and a minimum 464 FOM of 1.0. The relion image handler tool was used to threshold a series of incremental values from 465 1.0 – 3.0 in order to find which cutoff yielded the best SNR of both protein and nanoparticle. After 2D 466 classification was run on each thresholded particle stack, normalisation was run using the relion stack 467 create and relion preprocess tools. Performing this step was found to improve particle alignment as 468 well as provide an intensity distribution favoured by RELION. A low-resolution map was obtained from 469 the thresholded and normalised particle stack after running through 2D classification , 3D 470 reconstruction and masked refinement. 471 472 The obtained density showed most domains present excluding those in the macroglobulin head region 473 (MG1-4). Excluding this missing area, the domains were individually docked and fitted into the electron 474 density with UCSF ChimeraX version 1.9, which conformed to the inactive state C3 conformation. 475 476 477 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint Structural and statistical image preparation 478 All cryo-EM images were prepared using ChimeraX. All biophysical and biochemical graphs and 479 analyses have been prepared, fitted and displayed using the program GraphPad Prism version 10.0. 480 481

and funding 482 We gratefully acknowledge the Cryo -EM Uppsala facility for the use of the Glacios microscope, 483 computational facilities and the expert technical support of Daniel Larsson, all of which is funded by 484 the Department of Cell and Molecular Biology, the Disciplinary Domains of Science and Technology 485 and of Medicine and Pharmacy at Uppsala University. This work was supported by the SciLifeLab & 486 Wallenberg Data Driven Life Science Program, Knut and Alice Wallenberg Foundation (grants: KAW 487 2020.0239 and KAW 2017.0003), and by the National Bioinformatics Infrastructure Sweden (NBIS) at 488 SciLifeLab. We would like to especially thank Dustin Morado, T im Schulte and Piotr Draczkowski for 489 their expert assistance with data collection and interpretation. The computations and data handling 490 were additionally enabled by resources provided by the National Academic Infrastructure for 491 Supercomputing in Sweden (NAISS), partially funded by the Swedish Research Council through grant 492 agreement no. 2022-06725. This work was supported by the Swedish research council grant number 493 2022-03971_VR. 494 495 Author contributions 496 M.S, K.N.E, B.N and V.G.K conceptualised the research. G.A.S synthesised the pure titania colloidal 497 nanoparticles. D.E and K.F purified the C3 protein and performed all biochemical and haemolysis 498 assays. J.J.W performed all biophysical characterisations with C3 and TiO 2. J.J.W prepared all TEM 499 grids, performed data collection, processing, model refinement and validation of all cryo -EM 500 structures. J.J.W and D.E compiled and analysed the data, wrote the original manuscript and prepared 501 all figures and tables. Editorial assistance was provided by V.G.K, G.A.S, M.S, B.N and K.N.E. All 502 authors have reviewed, edited and approved the final version of the manuscript. 503 504 Competing interests 505 The authors declare no conflicts of interest. 506 507 Data availability 508 Original data points are shown in the manuscript. Additional information can be obtained from the 509 corresponding author upon request. 510 511 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint

512 1. Ermolin, M. S.; Ivaneev, A. I.; Fedyunina, N. N.; Fedotov, P. S. Nanospeciation of metals and 513 metalloids in volcanic ash using single particle inductively coupled plasma mass spectrometry. 514 Chemosphere 2021, 281, 130950. DOI: 10.1016/j.chemosphere.2021.130950. 515 2. Zhang, C.; Derrien, M.; Levenez, F.; Brazeilles, R.; Ballal, S. A.; Kim, J.; Degivry, M. C.; Quere, 516 G.; Garault, P.; van Hylckama Vlieg, J. E.; et al. Ecological robustness of the gut microbiota in 517 response to ingestion of transient food-borne microbes. ISME J 2016, 10 (9), 2235-2245. DOI: 518 10.1038/ismej.2016.13. 519 3. Sadeghi, R.; Rodriguez, R. J.; Yao, Y.; Kokini, J. L. Advances in Nanotechnology as They 520 Pertain to Food and Agriculture: Benefits and Risks. Annu Rev Food Sci Technol 2017, 8, 467-521 492. DOI: 10.1146/annurev-food-041715-033338. 522 4. Baranowska-Wojcik, E. Factors Conditioning the Potential Effects TiO2 NPs Exposure on 523 Human Microbiota: a Mini-Review. Biol Trace Elem Res 2021, 199 (12), 4458-4465. DOI: 524 10.1007/s12011-021-02578-5. 525 5. Baranowska-Wojcik, E.; Szwajgier, D.; Oleszczuk, P.; Winiarska-Mieczan, A. Effects of 526 Titanium Dioxide Nanoparticles Exposure on Human Health-a Review. Biol Trace Elem Res 527 2020, 193 (1), 118-129. DOI: 10.1007/s12011-019-01706-6. 528 6. Shim, K. H.; Hulme, J.; Maeng, E. H.; Kim, M. K.; An, S. S. Analysis of zinc oxide nanoparticles 529 binding proteins in rat blood and brain homogenate. Int J Nanomedicine 2014, 9 Suppl 2 (Suppl 530 2), 217-224. DOI: 10.2147/IJN.S58204. 531 7. Liao, F.; Chen, L.; Liu, Y.; Zhao, D.; Peng, W.; Wang, W.; Feng, S. The size-dependent 532 genotoxic potentials of titanium dioxide nanoparticles to endothelial cells. Environ Toxicol 2019, 533 34 (11), 1199-1207. DOI: 10.1002/tox.22821. 534 8. Brun, E.; Carriere, M.; Mabondzo, A. In vitro evidence of dysregulation of blood-brain barrier 535 function after acute and repeated/long-term exposure to TiO(2) nanoparticles. Biomaterials 536 2012, 33 (3), 886-896. DOI: 10.1016/j.biomaterials.2011.10.025. 537 9. Liu, X.; Sui, B.; Sun, J. Size- and shape-dependent effects of titanium dioxide nanoparticles on 538 the permeabilization of the blood-brain barrier. J Mater Chem B 2017, 5 (48), 9558-9570. DOI: 539 10.1039/c7tb01314k. 540 10. Shemetov, A. A.; Nabiev, I.; Sukhanova, A. Molecular interaction of proteins and peptides with 541 nanoparticles. ACS Nano 2012, 6 (6), 4585-4602. DOI: 10.1021/nn300415x. 542 11. Saptarshi, S. R.; Duschl, A.; Lopata, A. L. Interaction of nanoparticles with proteins: relation to 543 bio-reactivity of the nanoparticle. J Nanobiotechnology 2013, 11, 26. DOI: 10.1186/1477-3155-544 11-26. 545 12. Li, J. J.; Zou, L.; Hartono, D.; Ong, C. N.; Bay, B. H.; Lanry Yung, L. Y. Gold Nanoparticles 546 Induce Oxidative Damage in Lung Fibroblasts In Vitro. Advanced Materials 2007, 20 (1), 138-547 142. DOI: 10.1002/adma.200701853. 548 13. Monopoli, M. P.; Aberg, C.; Salvati, A.; Dawson, K. A. Biomolecular coronas provide the 549 biological identity of nanosized materials. Nat Nanotechnol 2012, 7 (12), 779-786. DOI: 550 10.1038/nnano.2012.207. 551 14. Chen, F.; Wang, G.; Griffin, J. I.; Brenneman, B.; Banda, N. K.; Holers, V. M.; Backos, D. S.; 552 Wu, L.; Moghimi, S. M.; Simberg, D. Complement proteins bind to nanoparticle protein corona 553 and undergo dynamic exchange in vivo. Nat Nanotechnol 2017, 12 (4), 387-393. DOI: 554 10.1038/nnano.2016.269. 555 15. Simberg, D.; Park, J. H.; Karmali, P. P.; Zhang, W. M.; Merkulov, S.; McCrae, K.; Bhatia, S. N.; 556 Sailor, M.; Ruoslahti, E. Differential proteomics analysis of the surface heterogeneity of dextran 557 iron oxide nanoparticles and the implications for their in vivo clearance. Biomaterials 2009, 30 558 (23-24), 3926-3933. DOI: 10.1016/j.biomaterials.2009.03.056. 559 16. Chowdhury, S. M.; Kanakia, S.; Toussaint, J. D.; Frame, M. D.; Dewar, A. M.; Shroyer, K. R.; 560 Moore, W.; Sitharaman, B. In vitro hematological and in vivo vasoactivity assessment of 561 dextran functionalized graphene. Sci Rep 2013, 3, 2584. DOI: 10.1038/srep02584. 562 17. Nesargikar, P. N.; Spiller, B.; Chavez, R. The complement system: history, pathways, cascade 563 and inhibitors. Eur J Microbiol Immunol (Bp) 2012, 2 (2), 103-111. 564 18. Ricklin, D.; Hajishengallis, G.; Yang, K.; Lambris, J. D. Complement: a key system for immune 565 surveillance and homeostasis. Nat Immunol 2010, 11 (9), 785-797. DOI: 10.1038/ni.1923. 566 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint 19. Wang, Z.; Hood, E. D.; Nong, J.; Ding, J.; Marcos-Contreras, O. A.; Glassman, P. M.; Rubey, K. 567 M.; Zaleski, M.; Espy, C. L.; Gullipali, D.; et al. Combating Complement's Deleterious Effects on 568 Nanomedicine by Conjugating Complement Regulatory Proteins to Nanoparticles. Adv Mater 569 2022, 34 (8), e2107070. DOI: 10.1002/adma.202107070. 570 20. Merle, N. S.; Church, S. E.; Fremeaux-Bacchi, V.; Roumenina, L. T. Complement System Part I 571 - Molecular Mechanisms of Activation and Regulation. Front Immunol 2015, 6, 262. DOI: 572 10.3389/fimmu.2015.00262. 573 21. Zarantonello, A.; Revel, M.; Grunenwald, A.; Roumenina, L. T. C3-dependent effector functions 574 of complement. Immunol Rev 2023, 313 (1), 120-138. DOI: 10.1111/imr.13147. 575 22. Tack, B. F.; Harrison, R. A.; Janatova, J.; Thomas, M. L.; Prahl, J. W. Evidence for presence of 576 an internal thiolester bond in third component of human complement. Proc Natl Acad Sci U S A 577 1980, 77 (10), 5764-5768. DOI: 10.1073/pnas.77.10.5764. 578 23. Gros, P.; Milder, F. J.; Janssen, B. J. Complement driven by conformational changes. Nat Rev 579 Immunol 2008, 8 (1), 48-58. DOI: 10.1038/nri2231. 580 24. Markiewski, M. M.; Nilsson, B.; Ekdahl, K. N.; Mollnes, T. E.; Lambris, J. D. Complement and 581 coagulation: strangers or partners in crime? Trends Immunol 2007, 28 (4), 184-192. DOI: 582 10.1016/j.it.2007.02.006. 583 25. Ricklin, D.; Reis, E. S.; Mastellos, D. C.; Gros, P.; Lambris, J. D. Complement component C3 - 584 The "Swiss Army Knife" of innate immunity and host defense. Immunol Rev 2016, 274 (1), 33-585 58. DOI: 10.1111/imr.12500. 586 26. Dzik, S. Complement and Coagulation: Cross Talk Through Time. Transfus Med Rev 2019, 33 587 (4), 199-206. DOI: 10.1016/j.tmrv.2019.08.004. 588 27. Ekdahl, K. N.; Teramura, Y.; Hamad, O. A.; Asif, S.; Duehrkop, C.; Fromell, K.; Gustafson, E.; 589 Hong, J.; Kozarcanin, H.; Magnusson, P. U.; et al. Dangerous liaisons: complement, 590 coagulation, and kallikrein/kinin cross-talk act as a linchpin in the events leading to 591 thromboinflammation. Immunol Rev 2016, 274 (1), 245-269. DOI: 10.1111/imr.12471. 592 28. Seisenbaeva, G. A.; Fromell, K.; Vinogradov, V. V.; Terekhov, A. N.; Pakhomov, A. V.; Nilsson, 593 B.; Ekdahl, K. N.; Vinogradov, V. V.; Kessler, V. G. Dispersion of TiO(2) nanoparticles improves 594 burn wound healing and tissue regeneration through specific interaction with blood serum 595 proteins. Sci Rep 2017, 7 (1), 15448. DOI: 10.1038/s41598-017-15792-w. 596 29. Akinola, P. O.; Lateef, A.; Asafa, T. B.; Beukes, L. S.; Hakeem, A. S.; Irshad, H. M. 597 Multifunctional titanium dioxide nanoparticles biofabricated via phytosynthetic route using 598 extracts of Cola nitida: antimicrobial, dye degradation, antioxidant and anticoagulant activities. 599 Heliyon 2020, 6 (8), e04610. DOI: 10.1016/j.heliyon.2020.e04610. 600 30. Nemmar, A.; Melghit, K.; Ali, B. H. The acute proinflammatory and prothrombotic effects of 601 pulmonary exposure to rutile TiO2 nanorods in rats. Exp Biol Med (Maywood) 2008, 233 (5), 602 610-619. DOI: 10.3181/0706-RM-165. 603 31. Chouirfa, H.; Bouloussa, H.; Migonney, V.; Falentin-Daudre, C. Review of titanium surface 604 modification techniques and coatings for antibacterial applications. Acta Biomater 2019, 83, 37-605 54. DOI: 10.1016/j.actbio.2018.10.036. 606 32. Seisenbaeva, G. A.; Daniel, G.; Nedelec, J. M.; Kessler, V. G. Solution equilibrium behind the 607 room-temperature synthesis of nanocrystalline titanium dioxide. Nanoscale 2013, 5 (8), 3330-608 3336. DOI: 10.1039/c3nr34068f. 609 33. Mudunkotuwa, I. A.; Grassian, V. H. Citric acid adsorption on TiO2 nanoparticles in aqueous 610 suspensions at acidic and circumneutral pH: surface coverage, surface speciation, and its 611 impact on nanoparticle-nanoparticle interactions. J Am Chem Soc 2010, 132 (42), 14986-612 14994. DOI: 10.1021/ja106091q. 613 34. Eriksson, O.; Mohlin, C.; Nilsson, B.; Ekdahl, K. N. The Human Platelet as an Innate Immune 614 Cell: Interactions Between Activated Platelets and the Complement System. Front Immunol 615 2019, 10, 1590. DOI: 10.3389/fimmu.2019.01590. 616 35. Fredslund, F.; Jenner, L.; Husted, L. B.; Nyborg, J.; Andersen, G. R.; Sottrup-Jensen, L. The 617 structure of bovine complement component 3 reveals the basis for thioester function. J Mol Biol 618 2006, 361 (1), 115-127. DOI: 10.1016/j.jmb.2006.06.009. 619 36. Janssen, B. J.; Huizinga, E. G.; Raaijmakers, H. C.; Roos, A.; Daha, M. R.; Nilsson-Ekdahl, K.; 620 Nilsson, B.; Gros, P. Structures of complement component C3 provide insights into the function 621 and evolution of immunity. Nature 2005, 437 (7058), 505-511. DOI: 10.1038/nature04005. 622 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint 37. Adler, A.; Manivel, V. A.; Fromell, K.; Teramura, Y.; Ekdahl, K. N.; Nilsson, B. A Robust Method 623 to Store Complement C3 With Superior Ability to Maintain the Native Structure and Function of 624 the Protein. Front Immunol 2022, 13, 891994. DOI: 10.3389/fimmu.2022.891994. 625 38. Hack, C. E.; Paardekooper, J.; Van Milligen, F. Demonstration in human plasma of a form of C3 626 that has the conformation of "C3b-like C3". The Journal of Immunology 1990, 144 (11), 4249-627 4255. DOI: 10.4049/jimmunol.144.11.4249. 628 39. Pangburn, M. K.; Schreiber, R. D.; Muller-Eberhard, H. J. Formation of the initial C3 convertase 629 of the alternative complement pathway. Acquisition of C3b-like activities by spontaneous 630 hydrolysis of the putative thioester in native C3. J Exp Med 1981, 154 (3), 856-867. DOI: 631 10.1084/jem.154.3.856. 632 40. Janssen, B. J.; Gros, P. Structural insights into the central complement component C3. Mol 633 Immunol 2007, 44 (1-3), 3-10. DOI: 10.1016/j.molimm.2006.06.017. 634 41. Nishida, N.; Walz, T.; Springer, T. A. Structural transitions of complement component C3 and 635 its activation products. Proc Natl Acad Sci U S A 2006, 103 (52), 19737-19742. DOI: 636 10.1073/pnas.0609791104. 637 42. Yu, G.; Yan, R.; Zhang, C.; Mao, C.; Jiang, W. Single-Particle Cryo-EM and 3D Reconstruction 638 of Hybrid Nanoparticles with Electron-Dense Components. Small 2015, 11 (38), 5157-5163. 639 DOI: 10.1002/smll.201500531. 640 43. Montesano-Roditis, L.; Glitz, D. G.; Traut, R. R.; Stewart, P. L. Cryo-electron microscopic 641 localization of protein L7/L12 within the Escherichia coli 70 S ribosome by difference mapping 642 and Nanogold labeling. J Biol Chem 2001, 276 (17), 14117-14123. DOI: 643 10.1074/jbc.M008430200. 644 44. Chen, C.; Daniel, M. C.; Quinkert, Z. T.; De, M.; Stein, B.; Bowman, V. D.; Chipman, P. R.; 645 Rotello, V. M.; Kao, C. C.; Dragnea, B. Nanoparticle-templated assembly of viral protein cages. 646 Nano Lett 2006, 6 (4), 611-615. DOI: 10.1021/nl0600878. 647 45. Rooijakkers, S. H.; Wu, J.; Ruyken, M.; van Domselaar, R.; Planken, K. L.; Tzekou, A.; Ricklin, 648 D.; Lambris, J. D.; Janssen, B. J.; van Strijp, J. A.; et al. Structural and functional implications of 649 the alternative complement pathway C3 convertase stabilized by a staphylococcal inhibitor. Nat 650 Immunol 2009, 10 (7), 721-727. DOI: 10.1038/ni.1756. 651 46. Gadeberg, T. A. F.; Jorgensen, M. H.; Olesen, H. G.; Lorentzen, J.; Harwood, S. L.; Almeida, A. 652 V.; Fruergaard, M. U.; Jensen, R. K.; Kanis, P.; Pedersen, H.; et al. Cryo-EM analysis of 653 complement C3 reveals a reversible major opening of the macroglobulin ring. Nat Struct Mol 654 Biol 2025, 32 (5), 884-895. DOI: 10.1038/s41594-024-01467-4. 655 47. Pangburn, M. K.; Morrison, D. C.; Schreiber, R. D.; Müller-Eberhard, H. J. Activation of the 656 alternative complement pathway: recognition of surface structures on activators by bound C3b. 657 The Journal of Immunology 1980, 124 (2), 977-982. DOI: 10.4049/jimmunol.124.2.977. 658 48. Bajic, G.; Yatime, L.; Sim, R. B.; Vorup-Jensen, T.; Andersen, G. R. Structural insight on the 659 recognition of surface-bound opsonins by the integrin I domain of complement receptor 3. Proc 660 Natl Acad Sci U S A 2013, 110 (41), 16426-16431. DOI: 10.1073/pnas.1311261110. 661 49. Moyano, D. F.; Liu, Y.; Ayaz, F.; Hou, S.; Puangploy, P.; Duncan, B.; Osborne, B. A.; Rotello, 662 V. M. Immunomodulatory effects of coated gold nanoparticles in LPS-stimulated in vitro and in 663 vivo murine model systems. Chem 2016, 1 (2), 320-327. DOI: 10.1016/j.chempr.2016.07.007. 664 50. Hirst, S. M.; Karakoti, A. S.; Tyler, R. D.; Sriranganathan, N.; Seal, S.; Reilly, C. M. Anti-665 inflammatory properties of cerium oxide nanoparticles. Small 2009, 5 (24), 2848-2856. DOI: 666 10.1002/smll.200901048. 667 51. Poon, W. L.; Alenius, H.; Ndika, J.; Fortino, V.; Kolhinen, V.; Mesceriakovas, A.; Wang, M.; 668 Greco, D.; Lahde, A.; Jokiniemi, J.; et al. Nano-sized zinc oxide and silver, but not titanium 669 dioxide, induce innate and adaptive immunity and antiviral response in differentiated THP-1 670 cells. Nanotoxicology 2017, 11 (7), 936-951. DOI: 10.1080/17435390.2017.1382600. 671 52. Hammer, C. H.; Wirtz, G. H.; Renfer, L.; Gresham, H. D.; Tack, B. F. Large scale isolation of 672 functionally active components of the human complement system. Journal of Biological 673 Chemistry 1981, 256 (8), 3995-4006. DOI: 10.1016/s0021-9258(19)69557-8. 674 53. Kessler, V. G.; Seisenbaeva, G. A.; Unell, M.; Hakansson, S. Chemically triggered biodelivery 675 using metal-organic sol-gel synthesis. Angew Chem Int Ed Engl 2008, 47 (44), 8506-8509. DOI: 676 10.1002/anie.200803307. 677 54. Fromell, K.; Adler, A.; Aman, A.; Manivel, V. A.; Huang, S.; Duhrkop, C.; Sandholm, K.; Ekdahl, 678 K. N.; Nilsson, B. Assessment of the Role of C3(H(2)O) in the Alternative Pathway. Front 679 Immunol 2020, 11, 530. DOI: 10.3389/fimmu.2020.00530. 680 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint 55. Nilsson, U. R.; Nilsson, B. Simplified assays of hemolytic activity of the classical and alternative 681 complement pathways. J Immunol Methods 1984, 72 (1), 49-59. DOI: 10.1016/0022-682 1759(84)90432-0. 683 56. Punjani, A.; Rubinstein, J. L.; Fleet, D. J.; Brubaker, M. A. cryoSPARC: algorithms for rapid 684 unsupervised cryo-EM structure determination. Nat Methods 2017, 14 (3), 290-296. DOI: 685 10.1038/nmeth.4169. 686 57. Bepler, T.; Morin, A.; Rapp, M.; Brasch, J.; Shapiro, L.; Noble, A. J.; Berger, B. Positive-687 unlabeled convolutional neural networks for particle picking in cryo-electron micrographs. Nat 688

2019, 16 (11), 1153-1160. DOI: 10.1038/s41592-019-0575-8. 689 58. Punjani, A.; Fleet, D. J. 3D variability analysis: Resolving continuous flexibility and discrete 690 heterogeneity from single particle cryo-EM. J Struct Biol 2021, 213 (2), 107702. DOI: 691 10.1016/j.jsb.2021.107702. 692 59. Meng, E. C.; Goddard, T. D.; Pettersen, E. F.; Couch, G. S.; Pearson, Z. J.; Morris, J. H.; 693 Ferrin, T. E. UCSF ChimeraX: Tools for structure building and analysis. Protein Sci 2023, 32 694 (11), e4792. DOI: 10.1002/pro.4792. 695 60. Zou, J.; Zhang, W.; Hu, J.; Zhou, X.; Zhang, B. DockEM: an enhanced method for atomic-scale 696 protein-ligand docking refinement leveraging low-to-medium resolution cryo-EM density maps. 697 Brief Bioinform 2025, 26 (2). DOI: 10.1093/bib/bbaf091. 698 61. Yamashita, K.; Palmer, C. M.; Burnley, T.; Murshudov, G. N. Cryo-EM single-particle structure 699 refinement and map calculation using Servalcat. Acta Crystallogr D Struct Biol 2021, 77 (Pt 10), 700 1282-1291. DOI: 10.1107/S2059798321009475. 701 62. Wood, C.; Burnley, T.; Patwardhan, A.; Scheres, S.; Topf, M.; Roseman, A.; Winn, M. 702 Collaborative computational project for electron cryo-microscopy. Acta Crystallogr D Biol 703 Crystallogr 2015, 71 (Pt 1), 123-126. DOI: 10.1107/S1399004714018070. 704 63. Emsley, P.; Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr D 705 Biol Crystallogr 2004, 60 (Pt 12 Pt 1), 2126-2132. DOI: 10.1107/S0907444904019158. 706 64. Afonine, P. V.; Poon, B. K.; Read, R. J.; Sobolev, O. V.; Terwilliger, T. C.; Urzhumtsev, A.; 707 Adams, P. D. Real-space refinement in PHENIX for cryo-EM and crystallography. Acta 708 Crystallogr D Struct Biol 2018, 74 (Pt 6), 531-544. DOI: 10.1107/S2059798318006551. 709 65. Tang, G.; Peng, L.; Baldwin, P. R.; Mann, D. S.; Jiang, W.; Rees, I.; Ludtke, S. J. EMAN2: an 710 extensible image processing suite for electron microscopy. J Struct Biol 2007, 157 (1), 38-46. 711 DOI: 10.1016/j.jsb.2006.05.009. 712 66. Kimanius, D.; Dong, L.; Sharov, G.; Nakane, T.; Scheres, S. H. W. New tools for automated 713 cryo-EM single-particle analysis in RELION-4.0. Biochem J 2021, 478 (24), 4169-4185. DOI: 714 10.1042/BCJ20210708. 715 716 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted June 13, 2025. ; https://doi.org/10.1101/2025.06.13.658543doi: bioRxiv preprint