IgA and IgM protein primarily drive plasma corona‐induced adhesion reduction of PLGA nanoparticles in human blood flow

Abstract The high abundance of immunoglobulins (Igs) in the plasma protein corona on poly(lactic‐co‐glycolic) acid (PLGA)‐based vascular‐targeted carriers (VTCs) has previously been shown to reduce their adhesion to activated endothelial cells (aECs) in human blood flow. However, the relative role of individual Ig classes (e.g., IgG, IgA, and IgM) in causing adhesion reduction remains largely unknown. Here, we characterized the influence of specific Ig classes in prescribing the binding efficiency of PLGA nano‐sized VTCs in blood flow. Specifically, we evaluated the flow adhesion to aECs of PLGA VTCs with systematic depletion of various Igs in their corona. Adhesion reduction was largely eliminated for PLGA VTCs when all Igs were removed from the corona. Furthermore, re‐addition of IgA or IgM to the Igs‐depleted corona reinstated the low adhesion of PLGA VTCs, as evidenced by ∼40–70% reduction relative to particles with an Igs‐deficient corona. However, re‐addition of a high concentration of IgG to the Igs‐depleted corona did not cause significant adhesion reduction. Overall, the presented results reveal that PLGA VTC adhesion reduction in blood flows is primarily driven by high adsorption of IgA and IgM in the particle corona. Pre‐coating of albumin on PLGA VTCs mitigated the extent of adhesion reduction in plasma for some donors but was largely ineffective in general. Overall, this work may shed light into effective control of protein corona composition, thereby enhancing VTC functionality in vivo for eventual clinical use.

glycol coating (PEG) coating on the particle surface. 11,13,14 These observations were attributed to the significant presence of immunoglobulins (Igs) in the corona formed on the NP surface, which prevents adhesive interaction of the targeting ligands with their corresponding cell or surface receptors. 8,[11][12][13][14] However, the precise role of individual Ig protein classes in this process has yet to be studied. Since protein properties vary widely, it is plausible that corona-derived effects are linked to a single or combination of a few proteins with certain features (e.g., large molecular weight, low off-rate, and conformational stability) rather than an aggregate effect of the complex corona. For instance, IgG is known to be highly opsonic in nature, and this feature has been linked to its significant presence on NP surface, leading to rapid hepatic uptake and clearance in vivo. 15,16 This work seeks to evaluate (a) which Ig class(es) in a vascular- here as our prior work demonstrated the high sensitivity of this material to human plasma and due to its ubiquitous use for construction of many drug delivery systems. 17 We investigate the individual impact of IgG, IgA, or IgM since these Ig classes collectively constitute the majority (98%(m/m)) of plasma Ig content and thus are expected to play significant roles in driving PLGA adhesion reduction.
We explored the covalent attachment of albumin to PLGA VTCs given the availability and common use of this dysopsonin protein in increasing circulation time of NPs in the bloodstream by counteracting opsonin (e.g., IgG, complement) adsorption and silencing recognition by the mononuclear phagocytic system (MPS). 7,[18][19][20][21][22] Finally, we focused on non-PEGylated VTCs to emphasize the effect of the base material (PLGA) in any corona-derived observations, which is expected to offer critical insight into the amplitude of surface modifications required to abate or alter plasma protein adsorption. Despite important strides made with their use to achieve non-fouling NPs, 23 PEG coatings on NPs still cannot completely prevent protein adsorption. 18 Furthermore, the formation of anti-PEG antibodies that is expected to accelerate PEG-NP clearance rate upon re-administration in vivo remains a major limitation. 24 Thus, there is a need for more research to elucidate further the mechanism of protein corona-induced altering of VTCs targeting, which would provide valuable information for the development of novel, NP-based therapeutics.

| Particle size and concentration characterization
Carboxylated, 500 nm PLGA particles were obtained from Phosphorex, Inc. (Hopkinton, MA). Carboxylated PLGA particles were dispersed in PBS11 containing 1%(wt/vol) bovine serum albumin (BSA) and then washed with PBS. Particles were incubated in 50 mM MES at pH 7 for-20 hr prior to DLS measurement of size distribution (Table 1), corresponding to the time required for the NeutrAvidin conjugation.

| Biomolecule surface conjugation
Particles were first conjugated with NeutrAvidin via covalent carbodiimide chemistry followed by linkage to biotinylated sialyl-Lewis a (sLe a ) (Glycotech Corporation; Gaithersburg, MD) as described elsewhere. 25 Briefly, 9.1 3 10 9 mm 2 /ml particles were incubated with 5 mg/ml Neu-trAvidin in MES buffer for 15 min, followed by addition of an equal volume of 75 mg/ml EDAC. For particles with both albumin and NeutrAvidin attached, human serum albumin (Sigma-Aldrich) was added in a 4:1 ratio with NeutrAvidin during the incubation step. The solution pH was then adjusted to 7.4 and placed on an end-to-end rotator for 20 hr. The conjugation reaction was halted with 7.5 mg glycine/ml, and the protein-coated particles were washed with and re-suspended in 1 ml of PBS. Attachment of the targeting ligand, sLe a , on PLGA particles was performed by suspending NeutrAvidin or NeutrAvidin 1 Albumin particles in 100 mL of biotinylated sLe a (diluted in PBS11 1%(wt/vol) BSA) at a concentration of 0.5-3 mg/ml at a surface area to volume ratio of 1.5 3 10 9 mm 2 /ml for 45 min.

| Quantification of biomolecule surface density
NeutrAvidin density was quantified using an Attune flow cytometer (Applied Biosystems) via staining conjugated particles with biotin-PE for 20 min, followed by washing with PBS11 1%(wt/vol) BSA.
Quantification of the number of NeutrAvidin and sLe a sites on the particle surface was achieved via use of Quantum R-PE MESF or FITC calibration beads (Bangs Laboratories) as previously described. 25 Particle sLe a density ( Table 2) was determined via staining with anti-CLA-PE (Miltenyi Biotec; San Diego, CA). Rat-IgM-PE (Fisher Scientific) was used as the isotype control. Albumin surface density was tested in a similar manner via goat-anti-albumin-FITC (human). Particles stained with goat-IgG-FITC at the same concentration served as the isotype control. Table 3 lists the albumin and NeutrAvidin surface densities obtained for PLGA particles in this work.

| Preparation of blood and buffer mediums
Human whole blood (WB) was obtained from healthy human donors according to a protocol approved by the University of Michigan Internal Review Board. A written informed consent was obtained from all subjects before blood collection. Acid-citrate dextrose (ACD) anticoagulant was added at a ratio of 0.14 ml/ml of WB. To obtain plasma, ACD WB was centrifuged at 2,250 g for 20 min at 48C and again at 6,797 g for 5 min to ensure removal of red and white blood cells and platelets. Plasma was then filtered (0.45 mm pore size) before use.
Depletion of all Igs from plasma was performed using the Pure-  to 100% (i.e., undiluted) plasma, the extent of the adhesion reduction was saturated with no additional reduction observed up to 60 min incubation time. 11 Since the depletion kits limited the amount of depleted plasma available, incubation of diluted plasma was employed, and particles were incubated for 1 hr (Supporting Information Figure   S1). The 1 hr incubation time with 25% plasma was chosen here since similar levels of adhesion reduction were achieved with this condition, as compared to 5 min 100% plasma incubation, 11 BSA and 1.4%(wt/vol) dextran 13,29 ) medium as previously described. 12 The hematocrit was fixed to 38%(vol/vol) for all experiments. Particles in RBCs-in-VB were introduced to the flow chamber for 5 min at a shear rate of 200/s. After the flow experiment, the particle adhesion density (#bound/mm 2 ) to HUVEC was obtained by fluorescent imaging along the width of the monolayer at a fixed position from the channel entrance.

| SDS-PAGE
SDS-PAGE was performed using 4-20% Tris-Glycine precast gels from Bio-Rad Laboratories or Life Technologies as previously described in. 12 For the solutions prepared in the SDS-PAGE experiments, conditions were diluted to 2% plasma solution, and therefore the total protein content during loading is expected to be similar across the different conditions. For the gels characterizing the corona proteins stripped from the NP surface, loading was normalized across the different experimental conditions by fixing the total particle surface area (typically 2.54 3 10 8 mm 2 /ml plasma). Furthermore, the same batch of particles exposed to the same amount of plasma was used on a given day for all conditions to reduce any deviations in expected particle concentration due to counting different particle stocks.

| ELISA
Sandwich ELISA was employed for measurement of human plasma IgG,  3 | R E S U L T S

| Assessment of Ig depletion column specificity
It is critical to assess the specificity of the commercially obtained depletion kits employed in this work for the removal of IgG, IgA, and other Igs from human plasma. Both IgG and IgA in blood plasma are 150 kDa in size, as confirmed via SDS-PAGE of commercially obtained solutions (Supporting Information Figure S2). Table 4 lists the % IgG, IgM, and IgA retained in plasma post exposure to the different columns/ beads relative to plasma as obtained via sandwich ELISA. The SDS-PAGE samples in Figure 1 show visual depletion of the 150 kDa band for IgG and Igs-depleted plasma samples ( Figure 1A), as well as IgA-depleted samples ( Figure 1B) To confirm that a significant amount of Igs were depleted from the corona when exposed to plasma devoid of Igs, an SDS-PAGE was performed on the proteins removed from the corona of PLGA particles incubated in buffer, native plasma, and Igs 1 albumin-depleted plasma ( Figure 1C). An approximate fourfold decrease in the intensity of the 150 kDa band (obtained via ImageJ analysis) was observed in the corona of particles exposed to 25% Igs 1 albumin-depleted plasma relative to 25% native plasma. Furthermore, the intensity value of the   We then probed whether IgG, which is abundant in human plasma (75%(m/m)), drives the observed PLGA NP adhesion reduction upon plasma exposure by exclusive depletion of this protein from the native plasma. On average, PLGA NPs exposed to IgG-depleted plasma retained 64 6 9% of the adhesion observed relative to buffer control; this was significant compared to the relative adhesion observed for particles with a native plasma corona. However, when observing individual donors, a significant recovery in adhesion was only seen for NPs exposed to IgG-depleted plasma from one donor, donor A, relative to the native plasma from the same donor ( Figure 2B). However, it is possible that variation in IgG affinity for PLGA across donors also contributes to the observed deviations in particle adhesion between donors in the absence of IgG. As such, SDS-PAGE was performed for NPs incubated in native plasma from donors A, C, and D.
The PLGA corona formed from donor A showed a noticeably heavier adsorption of the 150 kDa Ig band (12-fold increase in intensity) when compared head-to-head with corona formed from donor C or D plasma (Supporting Information Figure S4). Since donors A and C have a similar plasma level of IgG, the high presence of the 150 kDa band in the corona acquired from "A" relative to C plasma suggests that IgG affinity for PLGA also contributes to the significant adhesion recovery observed upon removal of IgG from donor A's plasma.

| Evaluation of how adsorption of specific Ig classes to plasma corona affects PLGA NP adhesion
The lack of a full recovery in the absence of IgG, unlike observed with the deletion of all Igs from above, suggests that other Ig types also play a role in the plasma corona-induced blocking of PLGA NP adhesion.
Since it was not possible to exclusively remove the other main Ig types as performed with IgG, we examined the inverse experiment-total removal of Ig proteins followed by a systematic re-addition of a particular class (IgG, IgA, or IgM) to the PLGA corona. It is important to note here that solely IgA1 was tested in the ELISA experiments since 80-90%(m/m) of IgA in blood consists of IgA1 (monomeric structure). 31 The composition and concentration of the different Ig solutions used in the re-addition experiments were denoted as follows: "Iso-IgG" (94% (m/m) IgG, 5 mg/ml), "Iso-IgA" (70%(m/m) IgA1, 0.6-0.8 mg/ml) (See Supporting Information Figure S5) and "Purified-IgM". Iso-IgA and Purified-IgM were re-added at physiological levels given their relatively pure composition and to best represent the individual impact of these Ig classes on PLGA adhesion reduction in vivo. Physiological concentrations of IgA1 and IgM were based on the ELISA measurements (Supporting Information Figure S6). Conversely, a high Iso-IgG concentration (5 mg/ml for 25% plasma solution) was employed in the re-addition experiment to mimic donor A's plasma conditions, where exclusive removal of IgG from the corona resulted in a major, significant increase in adhesion efficiency. Figure 3A shows SDS-PAGE of the various protein solutions in which particle coronae were formed. Figure 3B shows the relative adhesion levels obtained for PLGA NPs incubated in the various solutions characterized in Figure 3A.
Overall, the removal of all Igs plus re-addition of the 160 kDa (IgArich) band corresponds to significant adhesion reduction, 35 6 4%, relative to the removal of all Igs without re-addition (Supporting Information Figure S3). Furthermore, this level of particle adhesion obtained with re-addition of IgA was not significant from the level obtained for particles exposed to native plasma (38 6 4% relative to the removal of all Igs). When the Igs-depleted NP corona is enriched with Iso-IgG, adhesion was not significantly different from the adhesion observed for NPs with just an Ig-depleted corona (p 5 .0915), re-affirming the minimal role of IgG in causing adhesion deficiency of PLGA particles.
When Purified-IgM was re-added to the Igs-depleted corona, NPs again exhibited a slight, but significant, reduction in adhesion level, 64 6 5%, relative to NPs with just an Igs-depleted corona. This level of adhesion was, again, not significantly different (p 5 .103) than the adhesion of NPs with a native plasma corona.

| Albumin protein attachment to mitigate Iginduced PLGA adhesion reduction
To determine if the Ig corona-induced reduced adhesion efficiency of PLGA particles could be neutralized, we evaluated the flow adhesion of albumin-conjugated PLGA (PLGA 1 Albumin) particles exposed to native plasma and buffer and compared the results to the adhesion of standard PLGA particles (Figure 4). On average, there was no significant difference in the extent of adhesion (relative to buffer) between standard PLGA-sLe a and PLGA 1 Albumin-sLe a particles (8,000 sites/mm 2 ) with a native plasma corona when averaged across several donors ( Figure   4A). However, analysis of the adhesion of standard and albuminconjugated particles between individual donors revealed that the albumin-conjugated particles performed at 54 6 13% higher adhesion efficiency relative to that of the standard PLGA when the coronae were derived from donor C's plasma ( Figure 4B). Furthermore, an SDS-PAGE gel analysis of the particle corona upon exposure to donor C plasma shows a twofold decrease in intensity of the 150 kDa band (Ig-rich) band for the corona stripped from the albumin-conjugated relative to the standard PLGA particles (intensity normalized to respective particle PBS corona; Supporting Information Figure S7). PLGA 1 Albumin particle adhesion was also slightly increased relative to standard PLGA when exposed to donor E (20 6 8% increase); however, this was not significant when tested via one-way ANOVA (p 5 .4620). In the case of particle exposure to donor A and D plasma, no significant improvement in adhesion efficiency was observed between standard and albuminconjugated PLGA particles (donor A, p 5 .1930; donor D, p > .9999).

| DISCUSSION
Due to improved insight into vascular disease progression as well as NP interaction with blood flow components, vascular targeting remains (B) Adhesion of sle a -targeted PLGA particles incubated for 1 hr in 25% plasma (native), 25% plasma (Igs-depleted), 25% plasma (Igsdepleted) 1 5 mg/ml Iso-IgG, 0.6-0.8 mg/L Iso-IgA, and 0.2 mg/ml commercial IgM prior to a parallel plate flow chamber assay in RBCsin-VB (38% hematocrit) at 200/s. * 5 p < .01 compared to 25% plasma (native) trial, concentration 1 3 10 6 # particles/ml a viable alternative to invasive surgical procedures and poorly localizing therapies that cause a host of detrimental side effects to patients. 1,3,32 Recently, the drug carrier's plasma protein corona has been identified to exert an adverse effect on particle adhesion efficiency. 8,11,13,14 Thus, it is of great interest to the field to determine whether corona-induced adhesion reduction is driven by specific proteins (e.g., size, surface affinity), as this would offer key insight into the intelligent design of highly efficient VTCs. Moreover, a growing body of literature has focused on "exploiting" the protein corona by attracting a high presence of specific protein types into the corona to improve cell targeting and uptake efficiency. 10,[33][34][35][36] The data presented in this study shows that PLGA adhesion Given the high abundance of IgG in plasma and its well-known opsonic nature, it was surprising that the removal of IgG from the corona did not significantly restore adhesion efficiency except one donor-A. We hypothesize that PLGA adhesion was effectively restored when the corona was depleted of donor A's IgG due to this donor having a relatively high plasma IgG concentration ( Figure 2C) and a seemingly higher affinity for the PLGA corona compared with IgG in other donors (Supporting Information Figure S4). As such, we conclude that donor A may simply be an anomaly, especially as the re-addition of iso- serum, which is plasma depleted of clotting factors along with fibrinogen and to some extent fibronectin, to plasma and showed the serum acquired corona conferred a stronger adverse effect on PLGA particle adhesion. 12 Thus, it is unlikely that these proteins contribute to the depletion effects reported in this work.
Given the broad use of PLGA for drug delivery applications and prior FDA approval status, strategies to mitigate the Ig-induced reduction in targeted adhesion to the vascular endothelium are of interest.
While PEGylation of NP surfaces has been extensively explored, and shown to reduce protein adsorption, increase circulation time in vivo, and improve adhesion kinetics in blood flow, 46,47 it is also widely known that PEGylation does not eliminate protein adsorption. 11,14,18,23 Furthermore, corona-induced adhesion reduction of PLGA and other materials has previously been shown to persist in the presence of high surface PEG densities. Attachment of dysopsonin material coatings (e.g., albumin) may serve as a promising avenue to mitigate the extent of corona-induced adhesion reduction given the utility of this approach in the drug delivery space and recent FDA approval of albumin nanoparticle formulations for cancer treatment (e.g., Abraxane ® ). 15,[48][49][50][51] Specifically, dysopsonin coatings like albumin are known to silence interaction with the MPS system, increase circulation time in vivo, and reduce opsonic protein adsorption-similar to the action of PEG. 7,15,[18][19][20][21][22]48,52 Thus, covalent attachment of albumin to PLGA NPs was explored here as an avenue to abate Ig-induced adhesion reduction. However, pre-conjugating PLGA with albumin largely had a minimal effect in restoring PLGA adhesion efficiency upon plasma exposure, except donor C plasma, suggesting that the positive adhesion effect of albumin may be donor-dependent. Indeed, the corona obtained from donor C native plasma contained a 12-fold reduction in intensity of the 150 kDa band compared to donor A (Supporting Information Figure S4). Thus, it is likely easier for the albuminconjugated NPs to resist IgA adsorption in donor C plasma compared to donor A. Overall, given that albumin was only effective for donor C, other avenues for achieving non-fouling PLGA VTCs still needs to be explored. One possibility may be the use of zwitterionic functionalities on PLGA surfaces, as these materials have shown remarkable promise in eliminating protein adsorption to the particle surface. 53,54 Though, it remains largely unknown whether these biocompatible coatings will eliminate corona-induced reduced particle adhesion efficiency to HUVEC in the context of the complex, dynamic blood flow environment. 53,55-58

| C O NC LU S I O N S
This study revealed that IgA and IgM proteins primarily drive adhesion reduction of plasma-exposed PLGA NPs. These observations will hopefully shed light into the design methodology of high-efficient VTCs capable of evading corona-induced adhesion reduction in blood flow.
Given previous studies, which have demonstrated the inherent limitations of PEG coatings, it will be of great interest to explore employment of zwitterionic or carbohydrate coatings to achieve high-efficient binding PLGA NPs. Finally, this study is inherently limited given its use of in vitro assays, which fail to capture the added complexity of the in vivo corona. Indeed, future studies have will transition toward probing differential effects of protein corona fouling between in vitro and in vivo exposed particles. LITERATURE CITED