Aluminum hydroxide adjuvant diverts the uptake and trafficking of genetically detoxified pertussis toxin to lysosomes in macrophages

Abstract Aluminum salts have been successfully utilized as adjuvants to enhance the immunogenicity of vaccine antigens since the 1930s. However, the cellular mechanisms behind the immune adjuvanticity effect of these materials in antigen‐presenting cells are poorly understood. In this study, we investigated the uptake and trafficking of aluminum oxy‐hydroxide (AlOOH), in RAW 264.7 murine and U‐937 human macrophages‐like cells. Furthermore, we determined the impact that the adsorption to AlOOH particulates has on the trafficking of a Bordetella pertussis vaccine candidate, the genetically detoxified pertussis toxin (gdPT). Our results indicate that macrophages internalize AlOOH by constitutive macropinocytosis assisted by the filopodial protrusions that capture the adjuvant particles. Moreover, we show that AlOOH has the capacity to nonspecifically adsorb IgG, engaging opsonic phagocytosis, which is a feature that may allow for more effective capture and uptake of adjuvant particles by antigen‐presenting cells (APCs) at the site of vaccine administration. We found that AlOOH traffics to endolysosomal compartments that hold degradative properties. Importantly, while we show that gdPT escapes degradative endolysosomes and traffics toward the retrograde pathway, as reported for the wild‐type pertussis toxin, the adsorption to AlOOH diverts gdPT to traffic to the adjuvant’s lysosome‐type compartments, which may be key for MHC‐II‐driven antigen presentation and activation of CD4+ T cell. Thus, our findings establish a direct link between antigen adsorption to AlOOH and the intracellular trafficking of antigens within antigen‐presenting cells and bring to light a new potential mechanism for aluminum adjuvancy. Moreover, the in‐vitro single‐cell approach described herein provides a general framework and tools for understanding critical attributes of other vaccine formulations.


| INTRODUC TI ON
Aluminum salts form microparticles of crystalline or amorphous nature that have been used as adjuvants to enhance the immunogenicity of diverse vaccine formulations with outstanding results for several decades now. Potassium aluminum sulfate was the first compound to be shown to boost immunogenicity, utilizing diphtheria toxoid in guinea pigs (Glenny et al., 1926). Currently, aluminum phosphate and aluminum hydroxide are the most used adjuvants in licensed human vaccines. These preformed aluminum salts present several practical advantages over outdated in situ antigen precipitation with potassium aluminum sulphate. This includes the possibility of establishing more standardized preparation protocols, the ability to capture antigens by direct adsorption instead of precipitation, and the improvement of the adsorption/elution performance of vaccines in vivo (Hem & Hogenesch, 2007b). Moreover, current aluminumbased adjuvants are used in combination with microbial-recognition receptors agonists to enhance immunogenicity (Guy, 2007).
Although the antigen adsorption onto aluminum-based adjuvants is mainly mediated by electrostatic and ligand exchange interactions at the surface and crevices of the adjuvant microparticles, other attractive forces may contribute to this process, such as hydrogen bonding, hydrophobic interactions, and van der Waals forces (Hem & Hogenesch, 2007a;Peek et al., 2007). In addition, antigens can be simply entrapped in void spaces within particle aggregates during vaccine preparation, favoring their uniform distribution (Hem & Hogenesch, 2007b;Romero Mendez et al., 2007).
The antigen entrapment and adsorption capacities of aluminum preparations are crucial for their immunopotentiation effect (Hem et al., 2010;Hogenesch, 2012), as changing antigens from soluble to a particulate state may slow down the diffusion of antigens in the site of vaccine administration, a mechanism known as depot effect (Awate et al., 2013), and hence, facilitate their capture and internalization by antigen-presenting cells (APCs) (Hogenesch, 2012;Morefield et al., 2005). Within APCs, the internalized antigens traffic to hydrolytic phagosomes and endolysosomal compartments where peptides are produced and loaded onto major histocompatibility complex type II (MHC-II) receptors for antigen presentation and activation of CD4+ T cells (Ghimire et al., 2012). However, the immunopotentiation effect of aluminum adjuvants has also been associated with the recruitment and activation of APCs at the site of vaccine administration following strong inflammatory stimuli (Awate et al., 2013). This phenomenon leans on aluminum adjuvants' ability to cause oxidative damage of endosomal membranes and the activation of the NLRP3 inflammasome, and consequently, proinflammatory downstream signaling cascades, caused by hydroxyl radicals (Eisenbarth et al., 2008;Hogenesch, 2012;Mold et al., 2016;Reinke et al., 2020). Although broadly accepted, this mechanism for aluminum adjuvancy remains controversial since NLRP3 independent mechanisms have also been reported (Franchi & Nunez, 2008;Marrack et al., 2009;McKee et al., 2009). Moreover, how these two opposing mechanisms, one requiring functional endosomes for antigen presentation, and the other leaning on damaged compartments, concur in an immunopotentiation effect is currently unknown. In fact, the uptake and intracellular trafficking of aluminum adjuvant particles are poorly understood phenomena, despite being central to the adjuvating function of these compounds. Furthermore, it is unknown if aluminum adjuvants can influence the trafficking of antigens such as toxins and viral proteins. These are co-administered in vaccine formulations and antigens are often treated to neutralize their toxicity by either chemical or genetic means, yet it remains possible that some of these proteins may still preserve the ability to escape the endocytic pathway in a mechanism encoded in their molecular structure (do Vale et al., 2016;Johannes & Decaudin, 2005). This is the case of the genetically detoxified mutant of pertussis toxin (gdPT). Pertussis toxin (PTx) is an AB 5 exotoxin produced by Bordetella pertussis, the etiological agent of the acute respiratory infection known as whoopping cough. Upon entering the cell via clathrin-mediated endocytosis, PTx escapes the endocytic pathway to traffic retrogradely to the Golgi apparatus and the endoplasmic reticulum (ER) (el Baya et al., 1997;Plaut & Carbonetti, 2008). It has been proposed that from the ER, the enzymatic A-domain of PTx translocates into the cytoplasm to exert its toxic effect (Banerjee et al., 2016;Locht et al., 2011;Roy et al., 2006). The S1 subunit that forms the A-domain catalyzes the ribosylation of heterotrimeric inhibitory Gi proteins, leading to the increase in cAMP cellular levels and the disruption of signaling by different G protein-coupled receptors (Hsia et al., 1985;Kugler et al., 2007;Locht et al., 2011;Tamura et al., 1983). The genetically detoxified variant of PTx (gdPT) was developed by introducing a double point mutation (R9K/E129G) in the S1 subunit of PTx A domain (Burnette et al., 1988;Dewan et al., 2020;Pizza et al., 1989). These mutations substantially reduced the enzymatic activity of gdTP and completely abolished its toxicity, albeit holding a near-identical structure to that of the wild-type toxin (Ausar et al., 2020;Gregg & Merkel, 2019;Loosmore et al., 1990;Seubert et al., 2014).
Herein, we investigated the cellular uptake and intracellular trafficking of aluminum hydroxide (AlOOH) particles in the RAW 264.7 and bring to light a new potential mechanism for aluminum adjuvancy. Moreover, the in-vitro single-cell approach described herein provides a general framework and tools for understanding critical attributes of other vaccine formulations.

K E Y W O R D S
aluminum hydroxide adjuvant, genetically detoxified pertussis toxin, macrophages, macropinocytosis, phagocytosis murine and in the U-937 human macrophage-like cells, and how the adsorption to adjuvant particles affects the intracellular fate of gdPT. Our results show that AlOOH particles are readily internalized by constitutive macropinocytosis, which is characteristic of APCs (Doodnauth et al., 2019;Norbury, 2006). Moreover, the ability of AlOOH to nonspecifically adsorb IgG enhanced the cellular uptake of the adjuvant particles by engaging phagocytosis, which may more closely reflect the fate of AlOOH particles at the site of vaccine administration. Importantly, we found that most of the internalized AlOOH particles traffic to intracellular compartments endowed with endolysosomal degradative properties. Furthermore, we show that while AlOOH-absorbed gdPT traffics with the adjuvant particles to these degradative endosomal compartments, most of the internalized unadjuvanted gdPT localizes to abnormal endosomes that lack hydrolytic activity. Altogether, our findings indicate that the adsorption to aluminum adjuvants may divert antigens from their typical trafficking route toward endolysosomal compartments, therefore bringing to light a simple mechanism for aluminum adjuvanticity.

| Capture and uptake of AlOOH particles by macrophages
To investigate the interaction of AlOOH with macrophages, we first characterized the capture and uptake of antigen-free adjuvant particles by RAW 264.7 murine and U-937 human macrophage-like cells (hereinafter referred to as RAW and U-937 macrophages, respectively) by live cell microscopy. As shown in Videos S1 and S2, RAW and U-937 macrophages undergo continuous membrane ruffling activity and cast filopodial protrusions that scout the cellular periphery. These protrusions drag the AlOOH particles toward the ruffling membranes where the cells swallow them. Figure 1a shows still-images from Videos S1 and S2 that exemplify this process.
The SEM micrographs from Figure 1b show macrophage filopodial protrusions contacting AlOOH particles at a higher resolution. To distinguish AlOOH particles from cell-borne confounders during internalization assays, we resorted to labeling AlOOH for fluorescence microscopy. To this end, we tested labeling AlOOH with the fluorescent compounds lumogallion, morin hydrate, and dextran conjugated with Alexa Fluor Ⓡ dyes. Lumogallion is an aluminum chelator that forms a fluorescent complex with AlOOH (AlOOH-lumo) that can be excited at 490 nm and displays a wide emission spectrum (520-650 nm) that peaks at 580 nm (Mile et al., 2015). Morin forms a highly fluorescent complex with AlOOH (AlOOH-morin) that has its maximum excitation and emission peaks at 418 and 490 nm, respectively (Mile et al., 2015). As an alternative to these methods, we labeled AlOOH particles with Alexa Fluor Ⓡ fluorescent dyes conjugated to the anionic polysaccharide dextran, which have narrower excitation-emission peaks than lumogallion and morin and are available for different fluorescent emissions. Thus, Alexa Fluor® dyes present the technical advantage of being highly combinable with other fluorophores used in fluorescence microscopy. Overnight incubation at room temperature of AlOOH with either dextran-Alexa Fluor 488, lumogallion, or morin resulted in a highly efficient staining of the adjuvant particles through the formation of stable fluorescent complexes, detectable by either microscopy or flow cytometry analysis (Figure 2a-c). Next, utilizing fluorescently labeled AlOOH particles, we tracked the internalization of adjuvant particles by RAW and U-937 macrophages by confocal live-cell microscopy. As shown in Figure 3a and Videos S3 and S4, AlOOH particles were engulfed via two processes. First, uptake of AlOOH particles occurred via filopodial-like structures that seized and dragged the particles toward the cell. Second, AlOOH particles were engulfed at areas of the cell surface that underwent continuous ruffling, a process that requires approximately 6-7 min for completion (Videos S3 and S4 and Figure 3a). Whereas the former process resembles macrophages filopodia capturing bacteria and other particles via phagocytosis (Flannagan et al., 2010;Horsthemke et al., 2017;Moller et al., 2013); the latter concurs better with the ruffling activity associated with constitutive macropinocytosis, characteristic of APCs (Doodnauth et al., 2019). To distinguish whether macropinocytosis and/or phagocytosis were responsible for the uptake of AlOOH particles by macrophages, we treated cells with pharmacological inhibitory compounds with different specificities toward these pathways. To this end, we first implemented a fluorescent labeling method to distinguish between intracellular and extracellular AlOOH particles.
Briefly, macrophages were allowed to internalize AlOOH-lumo particles and at the time points required for the experimental design, the membrane-bound extracellular AlOOH-lumo particles were labeled with Alexa Fluor-conjugated dextran, which is membrane impermeant (refer to experimental procedures for more details). As a result of this procedure, the extracellular AlOOH particles became easily distinguishable from those already internalized by the cell (Figure 3b). This procedure allowed the identification and quantification of intracellular particles by computer analysis obtained from synchronized internalization assays (see experimental procedures), as depicted in Figure 3b,c. For this analysis, the total volume of internalized AlOOH was plotted, instead of particle number, to account for the heterogenicity in the size of AlOOH particles ( Figure S1). The results from Figure 3c shows that the amount of AlOOH internalized by the cells increased over time, which agrees with a process where macrophages capture and ingest particles laying at the cell's surroundings, as could be observed in Videos S1-S4.
We then determined the effect of different inhibitory treatments for macropinocytosis or phagocytosis on the internalization of AlOOH particles. First, we investigated the involvement of integrin and scavenger receptors in the internalization of AlOOH by macrophages, which have broad ligand specificity (Uribe-Querol & Rosales, 2020). As shown in Figure 4a, neither inhibiting integrins, with RGD peptides, nor scavenger receptors, with fucoidan, blocked the uptake of AlOOH by macrophages (Hamasaki et al., 2018;O'Brien & Melville, 2003;Thelen et al., 2010). Furthermore, the internalization of AlOOH was not affected by blocking complement receptor 3 (CD11b/ CD18), involved in opsonin and non-opsonic phagocytosis (Gordon & Rice, 1988;Patel & Harrison, 2008). Although the uptake of AlOOH was not inhibited by blocking phagocytic receptors, the process was strongly inhibited by treating macrophages with the actin depolymerizing compound Latrunculin A, which was shown to inhibit both phagocytosis and macropinocytosis (Canton et al., 2016;Fujiwara et al., 2018) (Figure 4b). Similarly, AlOOH internalization was also inhibited by the phosphoinositide 3-kinase inhibitor, LY294002, which blocks the production of phosphatidylinositol 3,4,5-trisphosphate, a membrane signaling F I G U R E 1 Uptake of aluminum adjuvant particles by macrophages. (a) RAW or U-937 cells were incubated with AlOOH for 5 min at 37°C prior to live cell bright field microscopy. Representative stills showing the capture and internalization of adjuvant particles over time are depicted and individual internalization events highlighted with yellow arrowheads. (b) RAW or U-937 cells were incubated with AlOOH as described above and at 30 min of internalization, cells were processed for SEM. Representative SEM micrographs displaying macrophage filopodial protrusions interacting with aluminum adjuvant (pseudocolor cyan) are shown in middle and right panels (yellow arrowheads). Images are representative of three independent trials. Fifty cells per trial per condition were analyzed. Scale bars, 5 μm
Moreover, the uptake of AlOOH was strongly inhibited by EIPA, a macropinocytosis-specific inhibitor that prevents the activity of the membrane Na + /H + exchanger 1, which is required for the activation of the small Rho GTPases Rac1 and Cdc42 during macropinocytosis (Koivusalo et al., 2010;Lin et al., 2018) (Figure 4b).
Thus, collectively our results indicate that the uptake of AlOOH particles depend on macropinocytosis. However, unlike macropinocytosis of fluid-phase solutes, our observations indicate that macropinocytosis of AlOOH requires filopodial protrusions that bind and drag the particles to the cell surface where macropinocytic events take place. Since none of the treatments applied to block receptors influenced AlOOH uptake efficiency, the binding of AlOOH to macrophage's protrusions may be mediated by the capacity of the adjuvant particles to establish nonspecific binding interactions with multiple different molecular moieties expressed on macrophage surfaces (Flach et al., 2011).
Considering this, we reasoned that opsonic ligand may bind AlOOH in a nonselective fashion, thereby triggering their phagocytosis at the site of vaccine administration. To investigate this hypothesis, and because IgG is the principal immunoglobulin isotype found in interstitial fluids (Janeway et al., 2001), we assessed the effect of adsorbing nonspecific IgG to AlOOH on the uptake of AlOOH particles. As shown in Video S5, when RAW macrophages were confronted simultaneously with IgG-adsorbed and naïve AlOOH particles, the former were captured at higher rates than the latter and via prominent phagocytic cups. Accordingly, the quantitative data from Figure 4c show that treating AlOOH with IgG significantly increased the uptake of adjuvant in both macrophage cell types. Altogether our observations indicate that although AlOOH particles are internalized by macropinocytosis, at the site of vaccine administration the nonspecific adsorption of IgG to AlOOH may favor a more efficient uptake of the adjuvant particles by APCs via phagocytosis, a phenomenon that could also apply to C3b and other opsonins present in interstitial fluids.

| Impact of AlOOH particle size on its uptake
AlOOH-containing vaccine formulations typically consist of antigens adsorbed to a population of particles that is highly heterogeneous in size, typically ranging from 1 to 10 μm in diameter (Hem & Hogenesch, 2007a). To investigate the effect of particle size on the uptake of AlOOH by macrophages, we tested two different preparations of AlOOH with distinct particle size distributions: a preparation displaying a median diameter d(0.5) of 6.6 μm (control), and a preparation of AlOOH treated by repeated freeze-and-thaw cycles (freeze/thaw) to favor the formation of large particles, displaying a median diameter d(0.5) of42.2 μm ( Figure S1), which has been reported to lead to the loss of vaccine potency (Clapp et al., 2014).
Next, AlOOH from control and freeze/thaw preparations were labeled with lumogallion and utilized for internalization assays with U-937 macrophages. After 4 h of incubation with either preparation of AlOOH-lumo, macrophages were detached from their wells and labeled with DAPI and anti-CD11c antibodies for flow cytometry analysis (Figures 5a and S1). As shown in Figure 5b,c, 87 ± 7.5% of the cells incubated with the control AlOOH preparation internalized AlOOH-lumo particles. However, only 34 ± 15% of macrophages incubated with the freeze /thaw AlOOH preparation contained intracellular AlOOH-lumo. Most of the internalized AlOOH were smaller particles present in the freeze/thaw preparation, while the large particles remained extracellular, as revealed by the microscopy imaging data from Figure 5b. Altogether, these results confirm that particle size is a critical attribute for adjuvant internalization by APCs.

| Internalized AlOOH traffic to functional endolysosomal compartments
We next sought to characterize the intracellular compartment occupied by AlOOH particles in macrophages, which will hereafter be referred to as AlOOH-containing compartments (ACCs). AlOOH immune adjuvanticity has been associated with the rupture of endosomes, caused by AlOOH-induced oxidative damage of endosomal membranes, and leading to NLRP3-inflammasome activation (Eisenbarth et al., 2008;Hogenesch, 2012;Mold et al., 2016;Reinke et al., 2020). Although this may trigger inflammatory responses and the recruitment and activation of APCs at the site of vaccine administration, it may also render endosomal compartments non-degradative, affecting antigen processing and the presentation capacity of APCs (Trombetta et al., 2003). To shed light on this conundrum, we investigated if ACCs could complete their endosomal maturation and develop endolysosomal degradative properties. As shown in Figure 6a,b, the vast majority of AlOOHlumo in U-973-cells localized to ACCs positive for endolysosomal markers, as also seen for RAW cells ( Figure S2). ACCs were positive for Lysosomal-associated membrane protein 1 (Lamp-1) and fused with late endosomes and lysosomes labeled with pre-loaded Alexa Fluor®-conjugated dextran, indicating they successfully reach the late endo-lysosomal maturation stage. This was further confirmed by assessing their capacity to acidify. The acidification of endolysosomes relies on the vacuolar ATPase (v-ATPase) H + pump, and the low permeability to H + of endolysosomal membranes to sustain a H + gradient. Figure 6a,b show that most of the ACCs analyzed were acidic, since they accumulated the acidotropic fluorescent compound LysoTracker, and hence indicate being delimited by intact membranes able to sustain H + gradients. Moreover, as expected for functional degradative endolysosomal compartments, ACCs tested positive for lysosomal protease activity, as per the chromogenic protease substrates Magic Red and DQ-red BSA (Figure 6a,b).
Next, we assessed whether ACCs could remain degradative due to the action of autophagy-mediated repair mechanisms counterbalancing AlOOH's putative capacity to induce membrane damage (Chauhan et al., 2016;Maejima et al., 2013). To investigate this possibility, we assessed the recruitment of the autophagy protein Microtubule-associated protein 1A/1B-light chain 3 (LC3) to ACCs since LC3 binds damaged endocytic compartments, as they are targeted for autophagy-mediated repair. Figure 6c and S3) (Maejima et al., 2013;Thiele & Lipsky, 1990;Uchimoto et al., 1999). To further confirm these results, we resorted to assessing the recruitment of mCherry-galectin 8 and GFP-lysenin to ACCs, which also detect damaged compartments ( Figure S3a), albeit being sensitive to different levels of membrane damage in endocytic compartments. Galectin-8 binds to sugar moieties exposed in damaged compartments to target them for selective autophagy. The lysenin probe binds sphingomyelin, a lipid in the luminal leaflet of endolysosomal membranes that readily translocate to the cytosolic face upon small membrane disruptions that are undetectable by the other probes (Thurston et al., 2012). Figure S3b,c shows that neither mCherry-galectin-8 nor GFPlysenin, associated with ACCs, unless LLOMe was applied to the cells. Altogether our results demonstrate that ACCs are functional degradative endolysosomal compartments.

| Unadjuvanted gdPT disrupts and escapes the endolysosomal pathway
Since ACCs are endowed with endolysosomal degradative properties, it is conceivable that the trafficking of AlOOH-adsorbed antigenic proteins to ACCs may favor their degradation and processing for antigen presentation, hence contributing to AlOOH adjuvant function. This could be critical for vaccine antigenic components that have the capacity to escape the endo-phagocytic route, as could be the case of viral particles and bacterial exotoxins (Uribe-Querol & Rosales, 2017). To investigate this hypothesis, we compared the intracellular fate of free and AlOOH-adsorbed gdPT in U-937 and RAW macrophages. This firstly required the characterization of gdPT trafficking, which has not been determined thus far. To follow gdPT in macrophages, we tagged gdPT with Alexa Fluor® dyes (gdPT-AF; see experimental procedures and Figure   S4a,b for details). Briefly, U-937 macrophages were incubated with gdPT-AF at 4°C for 30 min to allow for the binding of gdPT-AF to the cells. The cells were then chased at 37°C, to trigger gdPT-AF endocytosis, synchronously. As shown in Video S6 and Figure   S4b, gdPT-AF binds neatly to the surface of U-937 cells (0 h) and is rapidly internalized when conditions are permissive to endocytosis. Next, we ran a similar pulse and chase experiment, but in this case, the cells were fixed at the indicated times and processed for immunofluorescent labeling to assess the trafficking of gdPT-AF to different organellar compartments. As shown in Figure Figure S6b). Thus, the association of gdPT-AF with endosomal, Golgi, and ER compartments strongly suggest that gdPT follows the retrograde pathway toward the ER, as has previously reported for the wild-type toxin (Carbonetti, 2010;Kugler et al., 2007;Plaut & Carbonetti, 2008). Nevertheless, most of the gdPT-AF localized within late endosomal compartments by 6 h after the onset of the internalization assay (Figure 7b), which prompted us to investigate the state of maturation of this gdPT-AF containing endosomal compartments. As shown in Figures 8 and   9, these Lamp-1 positive gdPT-AF compartments were depleted from endolysosomal degradative properties. Indeed, they chiefly exclude the lysosomal protease cathepsin D (Figures 8a and 9a) and did not acidify, as per the lack of accumulation of the acidotropic dye LysoTracker (Figure 9b, right graph). Thus, our results indicate that gdPT hinders the ability of endolysosomal compartments to mature into degradative organelles, a phenomenon that may diminish macrophages' capacity for processing and presenting gdPT-derived antigens.

| AlOOH divert gdPT to the endo-lysosomal pathway
We next investigated the trafficking of gdPT adsorbed to AlOOH (AlOOH-gdPT-AF) in U-937 macrophages. The AlOOH-gdPT-AF were produced by incubating gdPT and AlOOH particles for 30 min at room temperature in tris-buffered saline. The AlOOH-gdPT-AF particles recovered were the product of electrostatic interactions, as their formation was reduced in the presence of increasing concentrations of potassium phosphate ( Figure S4c,d). Nevertheless, the AlOOH-gdPT-AF complex was resistant to multiple washes with PBS F I G U R E 2 Fluorescent labeling of aluminum adjuvant particles. (a) AlOOH preparations were conjugated with lumogallion, morin or dextran conjugated with Alexa Fluor® dye overnight at 4°C prior to visualization by spinning disk confocal microscopy. Representative micrographs showing each fluorophore-adjuvant conjugate by differential interference contrast (DIC) and fluorescent microscopy are depicted. (b-c) Fluorophore-AlOOH conjugates were processed by flow cytometry. Representative scatterplots showing the labeling efficiency for AlOOH-lumo, AlOOH-morin, and AlOOH-Dex-AF are shown in (b). The mean fluorescence intensity profiles for AlOOHlumo (em: 561 nm), AlOOH-morin (em: 490 nm), and AlOOH-Dex-AF (em: 491 nm) from the selected regions in (b) are shown in histograms depicted in (c). Spinning disk confocal images are a merge of z-stacks. Images are representative of three independent trials. Scale bars, 5 μm buffer and immunostaining conditions and thus, appropriate for our cellular studies. When the AlOOH-gdPT-AF particles were internalized by either U-937 or RAW macrophages, they chiefly localize in

| DISCUSS ION
Despite the critical role that aluminum-based adjuvants play in immune response augmentation, the cellular mechanisms underlying their interactions with APCs, i.e., particle internalization and intracellular trafficking, are poorly understood (Marrack et al., 2009).
To investigate the uptake and trafficking of aluminum-based adjuvants in macrophages, and to determine the impact of adjuvantadsorption on the intracellular fate of the pertussis vaccine antigen in vitro, we utilized AlOOH and gdPT as an antigen-adjuvant model.
We show that macrophages can internalize a wide range of AlOOH particle sizes, instead of undergoing abortive phagocytosis as previously reported for antigen-loaded aluminum adjuvants (Flach et al., 2011), and that most of these internalized particles follow the endo-lysosomal trafficking, endowing their compartments with a degradative lumen. Importantly, the internalization efficiency of AlOOH in macrophages was drastically reduced when the average size of the particles was increased by freeze and thaw cycles, which induces AlOOH aggregation. This size-dependent effect may explain the reduction in vaccine immunogenicity observed after subjecting vaccines to freeze-thaw cycles (Clapp et al., 2014), thereby demonstrating that particle size is a critical determinant for AlOOH uptake by macrophages, and confirming previous studies showing the importance of adjuvant particle size in vaccine immunogenicity (Clausi et al., 2008;Shardlow et al., 2017).
We show that the fate of gdPT within macrophages is strictly determined by whether it is adsorbed to AlOOH particles. Hence, several attributes critical to antigen-adjuvant formulation were identified and a simpler explanation for AlOOH-based adjuvantation was proposed.
By following the uptake of AlOOH by live-cell microscopy and utilizing pharmacological inhibitors, we demonstrated that macrophages internalize non-opsonized AlOOH particles via macropinocytosis. Furthermore, video microscopy revealed that AlOOH particles are internalized at areas of the macrophages surface undergoing copious and continuous membrane ruffling and where the formation of macropinosomes is evident, which is a behavior consistent with the constitutive form of macropinocytosis reported for macrophages and dendritic cells (Doodnauth et al., 2019). Through constitutive macropinocytosis, APCs ceaselessly sample extracellular fluids surveying for antigens (Canton et al., 2016;Sallusto et al., 1995). However, unlike for the macropinocytosis of extracellular fluid, we herein show that AlOOH particles are dragged by filopodia or filopodial-like extensions toward the ruffling membrane.
This particle capturing mechanism has been previously described assisting the opsonic and non-opsonic phagocytosis of microbes and disparate particles (Flannagan et al., 2010;Jain et al., 2019;Sallusto et al., 1995), and has been shown to be dependent on filopodial pulling and contraction mechanisms controlled by various myosin actin motors and actin tread-milling (Alieva et al., 2019;Horsthemke et al., 2017;Kress et al., 2007). Multiple sensing and binding receptors mediate filopodia environmental scouting and particle-capturing functions, including integrins, cadherins, and phagocytic receptors (Alieva et al., 2019;Chang et al., 2016;Horsthemke et al., 2017). In this regard, it has been reported that AlOOH-polymer nanoparticles can bind the scavenger receptor A (Jiang et al., 2018). Nevertheless, in our hands, the blocking of integrins, scavenger receptors, or complement receptor 3 (CD11b/CD18) impacted neither the binding (data not shown) nor the internalization of AlOOH particles by RAW and U-937 macrophages. Consequently, we hypothesize that F I G U R E 3 Uptake of fluorescent aluminum adjuvant and quantitative analysis. (a) RAW cells stably transfected with life-act-RFP or U-937 cells pretreated with SiR-Actin probe were incubated with dextran-AF or lumo-conjugated AlOOH, respectively, for 5 min at 37°C prior to live cell imaging. Representative stills from Video S3 showing internalization of AlOOH-Dex-AF488 or AlOOH-lumo are depicted. White arrowheads point to aluminum adjuvant particles undergoing internalization. "Fi" labels macrophage filipodial protrusions. (b) RAW cells were incubated with AlOOH-lumo for 0 or 2 h at 37°C and subsequently processed for differential labeling of internal/external aluminum adjuvant particles, as described in experimental procedures. Representative confocal micrographs show internal versus external adjuvants associated with RAW cells. To allow for cell delineation, F-Actin was stained with blue phalloidin (pseudocolored gray). Cells associated with adjuvant particles either attached (green and red) or internalized (green) are outlined with white dotted lines. (c) Volocity® quantifications of the total volume (μm 3 ) of AlOOH-lumo particles internalized per cell over time, in both RAW and U-937 cells, are represented in the scatter plots, where each red dot corresponds to the mean of an independent experiment and each dashed line represents the average ± SEM. *p ≤ .05, ****p ≤ .0001. Spinning disk confocal images represent a merge of z-stacks. Images are representative of three independent trials. Fifty cells per trial per condition were analyzed. Scale bars, 5 μm F I G U R E 4 Aluminum adjuvant particles are internalized via macropinocytosis. (a and b) RAW or U-937 cells were incubated with AlOOHlumo for 2 h at 37°C in the presence of the following inhibitory molecules or controls, respectively: CD11b blocking antibody (0.01 mg/ ml), fucoidan (100 μg/ml), RGD peptides (0.1 mg/ml), EIPA (100 μM), LY294002 (50 μM), Latrunculin A (0.25 μM), IgG isotype control, DMEM media, methanol or DMSO. Subsequently, cells were processed for differential labeling of internal/external aluminum adjuvant particles. Scatter plots show the particle uptake index (refer to experimental procedures for details), for each treatment. **p ≤ .01, ****p ≤ .0001. (c) RAW or U-937 cells were incubated with AlOOH-lumo (control) or previously opsonized with human IgG for 2 h at RT, and subsequently processed for differential labeling of internal/external aluminum adjuvant particles. The particle uptake index for each condition was calculated as described above and represented in the scatter plots. One hundred cells for each condition in each of the three independent experiments were analyzed. *p ≤ .05, **p ≤ .01 We show that AlOOH particles traffic to intracellular compartments that we termed ACCs to distinguish them from macropinosomes, which designates compartments carrying bulk liquid, and Assessing the impact of AlOOH-adsorption on gdPT intracellular trafficking required first to characterize the trafficking of the unadjuvanted gdPT in macrophages. We found that unadjuvanted gdPT, despite having its ribosylation-based toxicity disarmed, is still capable of disrupting endosomal maturation and subverting the trafficking retrogradely to the Golgi cisternae and then to the endoplasmic reticulum, likely following the pathway reported for the wild-type PTx (el Baya et al., 1997;Plaut & Carbonetti, 2008). To the best of our knowledge, our study is the first one describing the trafficking of gdPT and reporting its capacity of disrupting endosomal maturation. On the other hand, the adsorption of gdPT to AlOOH hampers its ability to disrupt and escape the endocytic pathway, and this could be the consequence of AlOOH keeping gdPT from interacting with the ACCs membrane. Consequently, different from unadjuvanted gdPT containing compartments, AlOOH-gdPT containing compartments are degradative organelles, and hence, possibly capable of antigen processing and MHC-II presentation (Ghimire et al., 2012). This indicates that by simply adsorbing antigens that bear the capacity to escape and/or disrupt the endocytic pathway to AlOOH, these antigens can be re-routed to degradative compartments and eventually presented to T cells as MHC-II-peptide complexes. Importantly, these putative mechanisms could also apply to other vaccine antigens with the potential of disrupting intracellular trafficking. Nevertheless, more studies are required to understand the mechanism behind this phenomenon and how it contributes to aluminum adjuvancy.
In our study, gdPT was readily adsorbed to AlOOH particles.
However, it has been reported that immunopotentiation does Our results show that there are several attributes likely to be critical for an antigen-AlOOH formulation that may need to be monitored and controlled throughout vaccine development. These include: (i) the size of the AlOOH particle which may affect whether a particle can be taken up at all; (ii) the capacity to adsorb IgG or other opsonins nonspecifically; and (iii) the stability of adjuvant-antigen absorption during formulation and after vaccine administration.
Adsorption of antigen can be specifically controlled during the formulation process and could have a profound impact on the type of immune response elicited, especially for antigens like gdPT. Control of such attributes will provide better predictability of the vaccine response in terms of safety and immunogenicity. While the mechanistic role of aluminum-based adjuvants has been the subject of much discussion (Oleszycka & Lavelle, 2014), we speculate that the simple process of particle size range optimization and the ability to direct immunity by uptake and trafficking toward degradative endocytic compartments might be a central mechanism underpinning the AlOOH immunopotentiation effect at the cellular level. Indeed, others have demonstrated the capacity of AlOOH to induce immunity independently of NLRP3/caspase1 (Franchi & Nunez, 2008;Marrack et al., 2009;McKee et al., 2009;Oleszycka & Lavelle, 2014).
Nevertheless, our findings do not rule out a contribution of NLRP3 inflammasome to AlOOH adjuvancy, since we detected AlOOH lysosomalytic activity in unprimed macrophages affecting a reduced F I G U R E 6 Adjuvant containing compartments acquire endolysosomal properties. (a) U-937 cells were incubated with AlOOH-morin or AlOOH-lumo for 2 h at 37°C to study the association of fluorescent AlOOH particles with the following endolysosomal markers: Lamp-1 (immunofluorescence), LysoTracker deep red, pre-loaded 10 kDa Dex-AF647, magic red cathepsin L (MRCL) or DQ-red BSA, as described in experimental procedures. Scale bars, 5 μm. (i) and (

| Plasmids and transfection
The construct GFP-LyseninW20A was kindly provided by Dr.

| Fluorescent labeling of AlOOH particles and gdPT
AlOOH-lumo was prepared as described elsewhere (Mile et al., 2015). Fluor 647 10,000 MW, anionic, fixable, cat# D22914, Thermo Fisher Scientific), both at a stock concentration of 0.5 mg ml −1 in 1 ml trisbuffered saline (TBS) 1X for 24 h in a rotating platform at RT. After labeling, each preparation was washed three times with phosphatebuffered saline (PBS) 1X at 12,000g for 3 min each and stored for several months at 4°C without losing fluorescence.
Purified genetically detoxified pertussis toxin (gdPT) was produced by Sanofi Pasteur Ltd. Canada. gdPT was labeled with Alexa Fluor dyes (gdPT-AF). Briefly, for Alexa Fluor 488-labeled gdPT, 470 μg of gdPT was incubated with 3 μl of Alexa Fluor reagent (Alexa F I G U R E 7 Intracellular trafficking of gdPT. (a) U-937 cells were allowed to internalize gdPT-AF488 as described in experimental procedures and fixed at the indicated time points. Cells were immunostained against Lamp-1, TGN46, and calnexin. Spinning disk confocal images correspond to a single z-plane. Framed areas are enlarged in (i-iii). Specifically, (i) represents the merge while (ii-iii) represent single fluorescent channels. Images are representative of three independent trials. 50 cells per trial per condition were analyzed. Scale bars, 5 μm. (b) U-937 cells were incubated with gdPT-AF488 for 6 h and then processed for immunofluorescence against calnexin, GM130, TGN46, or Lamp-1. For each free toxin containing compartment, the Manders' coefficient (M 2 ) was determined and if M 2 was greater than 0.7, and the particle was considered positive for that marker. Data are presented as mean ± SEM of a representative experiment, where at least 10 cells were quantified for each condition F I G U R E 8 Intracellular trafficking of free and adjuvanted gdPT. U-937 cells were incubated with either gdPT-AF488 (a) or AlOOH-gdPT-AF488 (b) as described in experimental procedures and fixed at different time points. Cells were subsequently processed for immunostaining against Lamp-1 and cathepsin D. spinning disk confocal images correspond to a single z-plane. Framed areas are enlarged in (i-iv). Specifically, (i) represents the merge while (ii-iv) represent single fluorescent channels. Images are representative of three independent trials. Over 100 cells per trial per condition were analyzed. Scale bars, 5 μm

| Effect of phosphate buffer on the adsorption of gdPT to AlOOH particles
To investigate the effect of electrostatic interactions, the adsorption of gdPT to AlOOH was measured in the presence of increasing concentration of sodium phosphate buffer pH 7.4. Samples of gdPT and AlOOH were mixed on an orbital mixer for 30 min at RT. The samples were then centrifuged for 5 min at 4000g. The supernatants containing the non-adsorbed protein were collected and the gdPT F I G U R E 9 Quantification of the association of free toxin, adjuvanted toxin or adjuvant alone with various intracellular markers. (a) U-937 cells were incubated with gdPT-AF488 or AlOOH-gdPT-AF488 for 6 h and then immunostained against Lamp-1 and cathepsin D. For either AlOOH-gdPT or free gdPT compartment, the Manders' coefficient (M 2 ) was determined and if M 2 was greater than 0.7, the particle was considered positive for that marker. Data are presented as mean ± SEM of three independent experiments, where 10 cells were quantified for each condition per independent experiment. p-Value ≤0.05. (b) U-937 macrophages were incubated with AlOOH-lumo, gdPT-AF594, or AlOOH-gdPT-AF594 for the indicated time points. Percent association of compartments with AlOOH-lumo, gdPT-AF594, or AlOOH-gdPT-AF594 with dextran (left) or LysoTracker (right) is depicted in the bar graphs. Data are presented as mean ± SEM of three independent experiments. *p ≤ .05, **p ≤ .01, *** p ≤ .001

| Adsorption of fluorescent gdPT to AlOOH particles
gdPT-AF was adsorbed to AlOOH particles (AlOOH-gdPT-AF) by coincubating 0.6 mg ml −1 AlOOH with 40 μg ml −1 gdPT-AF in 1 ml TBS (1X) at RT for 30 min on a rotating platform. AlOOH-gdPT-AF were washed with 1 ml TBS (1X) three times at 12,000g for 3 min each, to remove the excess of unbound gdPT, and subsequently stored at 4°C for no more than 1 week. Following internalization assays, immunostaining, and mounting of coverslips, slides were visualized within 1 week otherwise fluorescent toxin leach out from their original compartment.

| Freeze and thaw treatment of adjuvant particles
To analyze the impact of particle size on AlOOH internalization by macrophages, AlOOH particles were subjected to five consecutive cycles of freezing/thawing (−80°C for 15 min/37°C for 5 min) and when required, this was followed by lumogallion labeling as described above. F I G U R E 1 0 Proteolytic activity in free or adjuvanted gdPT-containing compartments. U-937 cells were allowed to internalize either gdPT-AF488 or AlOOH-gdPT-AF488 for the indicated time points and processed for the visualization of cathepsin activity by live cell confocal microscopy using the fluorogenic substrate, magic red, as described in experimental procedures. Representative micrographs of 2 and 6 h of incubation with free or adjuvanted gdPT are shown and are representative of three independent trials. Spinning disk confocal images correspond to a single z-plane. Framed areas are enlarged in (i-iii). Specifically, (i) represents the merge while (ii-iii) represent single fluorescent channels. Fifty cells per trial per condition were analyzed. White arrowheads point to adjuvanted gdPT positive for the magic red probe. Scale bars, 5 μm

| Inhibitor treatments
To study the internalization mechanism of AlOOH by mac-

| Quantifications of microscopy images with Volocity®
To To determine the percentage of adjuvant-gdPT-AF488 and gdPT-AF488 positive for Lamp-1, cathepsin, calnexin, GM130, or TGN46, Manders' co-occurrence analysis was performed as indicated above. Images were deconvolved (90% confidence limit) prior to analysis. Touching AlOOH-gdPT or free gdPT were separated using an object size guide of 0.29 μm 3 . Particles were considered to co-occur with the markers when the M 2 (channel 2 corresponding to gdPT fluorescence) was determined to be greater than 0.7. The average percentage of AlOOH-gdPT and gdPT containing compartments positive for Lamp-1, cathepsin, calnexin, GM130, or TGN46 was reported and based on 10 cells for each marker per independent experiment.
To determine the effect of the different inhibitor treatments on the internalization of AlOOH-lumo particles by macrophages, the adjuvant particle uptake index was calculated by normalizing the average volume of adjuvant internalized in treated cells by the corresponding value from untreated cells. Briefly, Volocity Ⓡ was set to identify AlOOH-lumo but negative for Dex-AF647 fluorescence (i.e., internalized particles), utilizing an object size guide of 0.20 μm 3 , estimated by determining the average volume of adjuvant particles for at least 50 images. A Volocity Ⓡ algorithm calculated the total volume of intracellular particles in each cell by determining the number of voxels within each particle and the voxel's volume of the microscope system.

| SDS-PAGE and in-gel fluorescence
3-5 μg of gdPT-AF488 were mixed with Laemmli's sample buffer (cat# 1610747, Biorad, Canada) and 2 mM dithiothreitol (DTT, cat# DTT-RO, Millipore Sigma), and heated at 100°C for 5 min prior to electrophoresis in 16% tris-glycine polyacrylamide gels for 1.5 h at 120 V. Gels were washed once with milliQ water for 10 min prior to in-gel fluorescence using the PharosFX imaging system (Biorad).
Data was analyzed using the Quantity One software. Afterwards, protein bands were stained in the same gels with InstantBlue™ (cat# ab119211, Abcam) overnight at RT. After 5 washes with milliQ water for 10 min each, protein bands were visualized using the GelDoc imaging system (Biorad). Data was analyzed using the Image Lab software.

| Statistical analysis
Unless otherwise stated, data are presented as mean ± SEM of three independent experiments. Statistical analysis was carried out using the Prism 9.0.2 software (GraphPad, La Jolla, CA). Data were assumed to be normally distributed. Two conditions were statistically compared using an unpaired, two-tailed Student's t test or a nested t test, while multiple conditions were compared utilizing a one-way ANOVA with Tukey's post hoc test, a one-way ANOVA with Dunnett's post hoc test or a nested one-way ANOVA with Dunnett's post hoc test. For all the statistical tests performed, p-values ≤ .05 were considered statistically significant.

ACK N OWLED G M ENTS
We would like to acknowledge Durga Acharya and/or Bruno  Terebiznik wrote the paper. All authors gave approval to the final version of the paper.

E TH I C S S TATEM ENT
The studies described obtained are exempted from ethics approval by Institutional Review Board as no animals or human samples were used, except cell lines.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request from the authors.