Poly(lactic-co-glycolic) acid (PLGA)–polyvinyl alcohol (PVA) nanoparticles (NPs) were synthesized by the double-emulsion solvent evaporation method described by Vasir and Labhasetwar (15). Briefly, PVA (1 mg/mL in 2-Morpholinoethanesulfonic acid monohydrate (MES) buffer) (Sigma-Aldrich Co., St. Louis, MO, USA; m.w. 30,000–70,000, 87%–90% hydrolyzed) was covalently linked to MBB2ΔCH2 (300 μg) through the EDC/sulfo-NHS Crosslinking protocol (Thermo-Fisher Scientific, Waltham, MA, USA) and then added to 10 mL of 2.0% (w/v) PVA solution in cold Tris-EDTA pH 8 (TE buffer). Afterward, 30 mg of PLGA (Sigma-Aldrich Co.; m.w. 30,000–60,000, lactide:glycolide 50:50) were dissolved in 1 mL of chloroform. To start the synthesis, 300 μL of bovine serum albumin (BSA) or Fluorescein isothiocyanate (Fitc)-BSA (Sigma-Aldrich Co.; 20 mg/mL) were added to the PLGA solution and vortexed and sonicated for 30 s at 15 W of the maximum power using Bandelin Sonopuls HD 2070 (Bandelin electronic GmbH & Co. KG, Berlin, Germany) to form the primary emulsion. The emulsion was added to the PVA solution and sonicated for 1 minute as described above to form a secondary emulsion. The resulting emulsion was stirred (750 rpm) overnight to allow for chloroform evaporation. NPs were collected by centrifugation at 12,000g for 30 minutes at 4°C, and the pellet was washed in TE buffer. Finally, NPs were resuspended in 2 mL of phosphate-buffered saline (PBS) filtered 0.2 μm and stored at 4°C. Untargeted NPs were produced following the same protocol, skipping the antibody coupling step.

Nanoparticle tracking analysis (NTA) was performed using NanoSight LM10 (Malvern Panalytical, Malvern, UK) diluting NP samples 1:4,000 (v/v) in Milli-Q water immediately before the measurements.

Hydrodynamic diameter (Z-Ave), polydispersity index (PDI), and surface charge (zeta potential) analyses were performed with the dynamic light scattering technique using Zetasizer Nano ZSP (Malvern Panalytical, Malvern, UK). NPs were diluted 1:200 (v/v) in H₂O Milli-Q filtered 0.2 μm, and measurements were performed at 37°C and a scattering angle of 173°.

The morphology of NPs was analyzed by scanning electron microscopy (SEM). NPs were diluted 1:50 v/v in H₂O Milli-Q filtered 0.2 μm. Then, a drop of the sample was deposited on a glass coverslip and left to dry at room temperature. Glass coverslips were mounted on aluminum stubs coated with double-sided carbon tape. Samples were carbon coated using the Q150T ES plus sputter coater (Quorum, Sacramento, CA, USA). Samples were analyzed by scanning electron microscopy using a Gemini 300 SEM (Zeiss, Oberkochen, Germany), working in secondary electron mode, at an acceleration voltage of 2 kV and a working distance of 3 mm.

Functional activity of complement was determined by measuring the activity of the classical pathway (CH50) on sensitized mutton erythrocytes (EA, Microbiol s.r.l., Cagliari, CA, Italy) as previously described (16, 17). Briefly, NHS (100 μL) and NPs (8 μL) were incubated for 2 hours at 37°C under shaking (750 rpm). The samples were centrifuged for 5 minutes at 8,000g, and the supernatants were diluted in complement fixation diluent (CFD; NaCl 142 mM, Na-5-5-diethylbarbiturate 5 mM, MgCl₂ 0.5 mM, agar 0.05%, NaN₃ 0.01%, and Ethyleneglycol- bis(β-aminoethyl)-N,N,Nʹ,Nʹ-tetraacetic Acid (EGTA) 10 mM) at 1:50, 1:100, 1:200, and 1:400 (v/v) in a final volume of 200 μL and incubated with 50 μL of antibody-sensitized erythrocytes (EA) 1% for 30 minutes to activate the classical complement pathway. The total lysis is represented by EA 1% diluted in H₂O, while the blank is represented by EA 1% diluted in CFD. The reaction was then blocked by the addition of EDTA 20 mM. The % lysis was measured at 415 nm with ELISA Reader TECAN Infinite M200 (Tecan Italia, Milan, Italy) after centrifugation at 12,000g for 1 minute. Complement activation was checked by preparing a standard curve using NHS as a positive control. The % lysis was calculated using the following formula:

The % lysis versus the serum dilution was plotted, and the dilution required for 50% hemolysis was calculated. The CH50 (U/mL) (hemolytic complement 50) provides the number of hemolytic units present in mL of NHS.

Direct erythrocyte cell lysis was performed by incubating EA 1% with NPs (8 μL) diluted in CFD for 2 hours at 37°C under shaking (750 rpm). The % lysis was calculated as described in the section “CH50 screening assay”.

NPs (4 μL/well) and normal human plasma (NHP; 80 μL/well) were seeded in a 96-well plate. To assess the clotting capacity of the nano-devices, CaCl₂ 20 mM was added to initiate the clotting reaction. The eventual coagulation causes an increase in the turbidity of the well, which has been read at 405 nm every 2 minutes for 70 minutes with ELISA Reader TECAN Infinite M200 (Tecan Italia S.r.l.).

Serum samples were obtained from five APS patients ( Table 1 ) with medium–high titer antibodies to domain I of β2GPI after obtaining informed consent and were previously shown to induce clot formation in rats (18). Normal human serum was the pool of at least 30 different healthy donors. NHP was the pool of at least 30 different healthy donors using EDTA as an anti-coagulant. The ethical committee of Istituto Auxologico Italiano approved the study; the patients/participants provided written informed consent to participate in this study. Total IgG were purified from human sera using protein A affinity chromatography and then labeled with Fitc, as already described by our group (19).

MBB2ΔCH2 was produced as previously described (14). Briefly, stable transfected CHO cells were cultured in a serum-free medium, and the recombinant antibody was purified from cell supernatant by protein A affinity chromatography.

In vivo experimental models were established in groups of three male Wistar rats (290–320 g) kept under standard conditions in the Animal House of the University of Trieste, Italy. The in vivo procedures were performed in compliance with the guidelines of European (86/609/EEC) and Italian (Legislative Decree 116/92) laws and were approved by the Italian Ministry of University and Research (Prot. No. 910/2018PR). The study was conducted in accordance with the Declaration of Helsinki.

Rats were intraperitoneally injected with lipopolysaccharide (LPS) (2.5 mg/kg) and then anesthetized with Zoletil (Virbac, Carros, France; 25 mg/kg) and xylazine (Rompun; 7.5 mg/kg). Afterward, they received a slow infusion of Rhodamine 6G (Sigma) into the jugular vein to stain leukocytes and platelets. Targeted or untargeted BSA-Fitc (Sigma)-filled NPs (400 μL of a suspension of approximately 1.9 × 10¹² NPs/mL) were then infused into the carotid artery. The deposition of NPs on mesenteric endothelium was monitored by intravital microscopy for 90 minutes. The amount of NPs bound to the endothelial surface was evaluated by measuring the fluorescence of BSA-Fitc encapsulated in NPs and tNPs. First, four different regions of interest (ROIs) were drawn in the background of the picture, and the mean fluorescence intensity of the background was calculated. This value was subtracted from the starting picture, and then this was converted into a binary picture in order to create a different ROI set on the fluorescent signal. Finally, the Fitc and the background signals were calculated and employed to measure the corrected total cell fluorescence (CTCF):

Rats were intraperitoneally injected with LPS (2.5 mg/kg) and then anesthetized with Zoletil (Virbac, 25 mg/kg) and xylazine (Rompun, 7.5 mg/kg). Afterward, they received a slow infusion of Rhodamine 6G (Sigma) into the jugular vein to stain leukocytes and platelets. Targeted or untargeted NPs (400 μL of a suspension of 1.9 × 10¹² NPs/mL) or saline (400 μL) were then infused into the carotid artery. Ten minutes later, they were then infused into the carotid artery with 6.5 mg of Fitc (Sigma)-labeled IgG purified from five APS patients with medium–high titer antibodies to the Domain I (DI) domain of β2GPI. The deposition of IgG on mesenteric endothelium was monitored by intravital microscopy for 90 minutes. The rate of IgG bound to endothelium was evaluated by comparing the fluorescence intensity on the endothelial surface to the fluorescence intensity of the vascular lumen. First, four ROIs were drawn in the background of the picture, and the mean fluorescence intensity of the background was calculated. This value was subtracted from the starting picture, and then this was converted into a binary picture in order to create a different ROI set on the fluorescent signal. After that, four ROIs were drawn on the blood vessel lumen, and as performed before, this signal was subtracted from the original picture in order to isolate only the most fluorescent spots representing the deposited IgG. Finally, the Fitc and the background signals were calculated and employed to measure the spots-CTCF.

The APS model (7) was established in rats that were intraperitoneally injected with LPS (2.5 mg/kg) and then anesthetized with Zoletil (Virbac, 25 mg/kg) and xylazine (Rompun, 7.5 mg/kg). Afterward, they received a slow infusion of Rhodamine 6G (Sigma) into the jugular vein to stain leukocytes and platelets; tNPs, NPs (400 μL of a solution of 1.9 × 10¹² NPs/mL), and saline were then infused into the carotid artery. Ten minutes later, 1 mL of pooled sera from five APS patients containing antibodies to β2GPI (18) was infused into the carotid artery to cause thrombus formation. The evolution of hemodynamics in the mesenteric vessels was monitored by intravital microscopy for 90 minutes. The number of thrombi observed in the vascular tree and the percentage of occluded vessels were evaluated by measuring the blood flow. A fourth group of rats received 400 μL of a solution of 1.9 × 10¹² tNPs/mL intravenously 1 day before the above-described procedure, and the same parameters were evaluated.

Statistical significance was determined using the Prism (GraphPad) software. Statistical analysis of means was performed using the two-way ANOVA for the in vivo experiments regarding the ability of tNPs to bind activated endothelyum and to block thrombus formation or one-way ANOVA for the in vitro tests of NPs and TNPs safety and for the in vivo experiment about the ability of tNPs to block APS' IgG endothelial deposition. Variance was similar between groups, and p-values <0.05 were considered statistically significant.

PLGA–PVA uncoated and targeted nanoparticles (indicated as NPs and tNPs, respectively) were produced using a standardized approach that guarantees a final concentration of approximately 1.9 × 10¹² NPs/mL, as measured by NTA. Both preparations consisted of an aqueous inner core and a polymeric outer shell, covalently decorated with MBB2ΔCH2, targeting β2GPI, in tNP (20). The core was loaded with Fitc-BSA, to allow visualization of NPs bound to blood vessels by fluorescence microscopy, or BSA, to have comparable nanostructures in experiments aimed at evaluating their therapeutic effect. The physicochemical characterization of NPs is reported in Figure 1A . All preparations exhibited a similar size with an average diameter lower than 300 nm (279 ± 19.6 nm for NP and 283 ± 6.63 nm for tNP), a negative surface charge (−13 mV for NP and −10 mV for tNP), and a good index of heterogeneity as confirmed by the low PDI (0.15 for NP and 0.17 for tNP). The regular round shape of the NPs was verified by SEM ( Figure 1B ), showing that MBB2ΔCH2 coupling does not significantly alter NP diameter and morphology. All formulations exhibited very good stability when stored at 4°C for 6 months; no variation in size and zeta potential and no NP aggregation were detected when analyzing preparations by dynamic light scattering (DLS) (data not shown).

PLGA–PVA NPs were tested in vitro to evaluate their potential safety in circulating blood ( Figure 2 ). The possibility that NPs may cause mechanical disruption of red blood cells was investigated by incubating an erythrocyte suspension with NP or tNP for 2 hours at 37°C. No lysis was observed in the supernatant collected at the end of the experiment, thus excluding a direct damaging effect of NPs on red cells. The possible effect of NPs and tNPs on complement activation was then evaluated. Briefly, pooled normal human serum was incubated with each NP at a concentration similar to that used in in vivo experiments, and the residual complement activity was analyzed by a hemolytic assay (CH50). The results showed a modest, non-significant reduction of the complement activity, and no difference was found in the effect of the two NPs. A turbidity assay was finally performed to ascertain whether NPs interfere with blood clot formation. To this end, NPs were incubated with pooled NHP, and the sample turbidity was analyzed over time following the addition of a solution containing Ca²⁺ to trigger the coagulation process. The data revealed a similar trend either in the presence or in the absence of NPs, and no significant difference was found between NPs and control groups in the half coagulation time extrapolated from the curves.

Preliminary experiments were conducted to verify the in vivo ability of anti-β2GPI antibodies (recombinant or present in APS patients) to bind to activated endothelium. Serum samples were obtained from five APS patients selected for the presence of medium–high titer IgG to cardiolipin, β2GPI, and, in particular, domain I of β2GPI ( Table 1 ); the sera were pooled before use.

The rats received an intraperitoneal injection of LPS in order to induce binding of β2GPI to endothelial cells, as previously described (21). Fitc-labeled MBB2ΔCH2 or Fitc-labeled APS IgG were administered into the carotid artery, and their localization on mesenteric vessels was monitored over time. Intravital fluorescence microscopy analysis showed a patchy distribution of the fluorescent signal on the mesenteric endothelium of rats receiving either fluorescent MBB2ΔCH2 or aPL IgG ( Figure 3 ). As expected, aPL IgG caused vascular occlusions ( Figure 3 ), while MBB2ΔCH2, which does not activate the complement system, was able to target β2GPI on endothelial cells but failed to trigger the coagulation process and the vessel occlusion.

A similar distribution pattern of BSA-Fitc loaded tNPs on the mesenteric endothelial cell was observed in rats treated with LPS, which stimulated cell deposition of β2GPI, recognized by the recombinant anti-β2GPI antibody on NP surface. On the contrary, no relevant fluorescent signal was seen on the mesenteric vessels of rats receiving fluorescent untargeted NPs ( Figure 4A ). The deposition of both types of nanoparticles on vascular endothelium was analyzed by measuring the fluorescence signal of BSA-Fitc encapsulated in the NPs. The results presented in Figure 4B show a progressive increase in the deposition of tNPs on the endothelium of blood vessels that started within a few minutes from NP injection, reaching the highest level in 30 minutes and slightly decreased up to 90 minutes. A marginal localization of untargeted NPs was observed on vascular endothelium, and the extent of their deposition was significantly lower than that of tNPs.

It is important to underline that neither NPs nor tNPs were able to activate the coagulation cascade in blood, and intravascular thrombi were not documented during the analysis.

To examine the effect of tNPs in preventing the binding of aPL to blood vessels, LPS-treated rats received targeted or untargeted NPs, or saline as a control, followed by a pool of Fitc-labeled IgG purified from five APS patients, containing antibodies to β2GPI. The deposition of IgG on mesenteric blood vessels was monitored over time by intravital microscopy ( Figure 5A ). The analysis showed again an irregular distribution of Fitc-labeled IgG on endothelial cells, and the sites along the vessels, where the labeled IgG were deposited, had decreased in number in rats treated with tNPs compared to those seen in rats receiving saline or untargeted NPs ( Figure 5A ). Aware of the fact that the number of deposits alone was not exhaustive in describing the phenomenon observed, a quantitative analysis of fluorescence intensity was performed as indicated above. As shown in Figure 5B , the deposits of aPL IgG on endothelial cells were markedly reduced and nearly undetectable 90 minutes after the administration of tNP compared to those seen in animals receiving saline. Interestingly, some inhibition of IgG binding to endothelial cells was observed in rats treated with untargeted NPs, most likely due to non-specific absorption of circulating IgG by the NPs rather than the result of competitive inhibition of IgG binding to endothelial cells.

The ability of NPs to prevent the thrombotic activity of deposited IgG anti-β2GPI was then examined. To this end, rats primed with LPS, to promote deposition of β2GPI on endothelial cells, received either targeted or untargeted NPs, or saline as a control prior to administration of pooled sera from five APS patients containing antibodies to β2GPI to trigger coagulation. Analysis of the mesenteric vasculature, monitored over time by intravital microscopy, revealed a marked decrease in the number of thrombi in tNP-treated rats compared to animals receiving saline ( Figure 6A ). A reduced number of blood clots were also seen following administration of untargeted NPs, consistent with the results of passive absorption of circulating aPL IgG shown in Figure 5B , although the decrease was less marked than that observed in tNP-treated rats ( Figure 6A ). Interestingly, the analysis of the blood clot dimension, which may be responsible for vascular occlusion, revealed that only tNPs have a significant effect in reducing the number of occluded vessels while untargeted NPs were almost ineffective ( Figure 6B ).

To investigate whether the tNPs, administered several hours prior to patients’ aPL IgG, circulate in sufficient numbers to prevent coagulation, the rats were challenged with tNPs or saline 24 hours before LPS priming followed by administration of aPL IgG. The results presented in Figures 6C, D show that tNPs were still able to significantly reduce both the number of blood clots and the percentage of occluded vessels.