Intracellular tracing of amyloid vaccines through direct fluorescent labelling

Alzheimer’s disease is a debilitating neurodegenerative condition that progressively causes synaptic loss and major neuronal damage. Immunotherapy utilising Aβ as an active immunogen or via passive treatment utilising antibodies raised to amyloid have shown therapeutic promise. The migratory properties of peripheral blood-borne monocytes and their ability to enter the central nervous system, suggests a beneficial role in mediating tissue damage and neuroinflammation. However, the intrinsic phagocytic properties of such cells have pre-disposed them to internalise misfolded amyloidogenic peptides that could act as seeds capable of nucleating amyloid formation in the brain. Mechanisms governing the cellular fate of amyloid therefore, may prove to be key in the development of future vaccination regimes. Herein, we have developed unequivocal and direct conformation-sensitive fluorescent molecular probes that reveal the intracytoplasmic and intranuclear persistence of amyloid in a monocytic T helper 1 (THP-1) cell line. Use of the pathogenic Aβ42 species as a model antigen in simulated vaccine formulations suggested differing mechanisms of cellular internalisation, in which fibrillar amyloid evaded lysosomal capture, even when co-deposited on particulate adjuvant materials. Taken collectively, direct fluorescent labelling of antigen-adjuvant complexes may serve as critical tools in understanding subsequent immunopotentiation in vaccines directed against amyloidosis and wider dementia.


Results
Assessment of the cellular uptake of a model amyloid antigen via direct fluorescent labelling. Thioflavin T (ThT) fluorescence of the amyloidogenic Aβ 42 peptide was first assessed in R10 cell culture medium, to ensure the formation of mature amyloid fibrils in treatment conditions compatible with subsequent T helper 1 (THP-1) cell culture (see Supplementary methods). Aβ 42 prepared at 4 μM and incubated over 24 h at 37 °C produced a 64.8% increase in fluorescence intensity measured at 482 nm (13.07 ± 0.34, mean ± SD, n = 3), above background of R10 cell culture medium only (4.60 ± 0.85, mean ± SD, n = 3). Amyloid fibril formation propagating as β-sheets at the concentrations stated herein was therefore confirmed in R10 media.
The assessment of the cellular uptake of Aβ 42 was investigated initially in the absence of added aluminium based adjuvant (ABA) nanomaterials. In order to assess cellular uptake of β-pleated sheets of amyloid formed under these conditions, the use of ThT as a pre-labelling reagent (20 μM) was first assessed in treatments containing 8 μM of the peptide only. Aβ 42 pre-incubated for 24 h in R10 medium containing ThT and co-cultured (1:1) with THP-1 cells for 1 h (37 °C, 5% CO 2 ), revealed the cellular uptake of amyloid in a β-sheet conformation as identified through a green fluorescence emission (482 nm) contained in cell cytosol (Fig. 1). ThT-reactive amyloid was identified following 1 h incubation, as evidenced via punctate green fluorescence in focal deposits at the periphery of THP-1 cells. Cellular uptake of amyloid fibrils was also observed in identical experiments over 1, 3, 6 and 24 h (see Supplementary Fig. 1). THP-1 cells co-cultured under identical conditions in the absence of added amyloid produced a uniform weak green fluorescence emission that was found to increase in intensity upon prolonged incubation over 24 h in the presence of ThT (see Supplementary Fig. 2).

The complementary use of transmission electron microscopy confirms the cellular internalisation of fibrillar amyloid.
In order to confirm the cellular internalisation of amyloid, THP-1 cells were co-cultured under identical conditions over 1-24 h, of which cell sections (90 nm) were negatively stained with uranyl acetate (2% w/v). Pre-osmium tetroxide and lead citrate staining were omitted to allow for greater clarity in the detection of potential intracellular amyloid. Aβ 42 aged for 24 h in R10 media and co-cultured with THP-1 cells (at 4 μM) for 1 h, revealed negatively stained deposits of intracellular amyloid fibrils as plaque-like aggregates in cell cytosol (Fig. 2a,e and i). Intracellular clustering of negatively stained fibrils was also identified within the cytoplasm of THP-1 cells following 3 (Fig. 2b,f and j) and 6 h (Fig. 2c,g and k) incubation at 37 °C. Of those fibrillar deposits identified, no aggregates of amyloid were found enclosed in vesicular-like structures in the cytoplasm of THP-1 cells.
At every time point investigated thereafter over 24 h, electron-dense deposits of negatively stained amyloid fibrils were found internalised in THP-1 cells that varied in size dramatically. Mature amyloid fibrils exhibiting classical negative staining were readily identified following 24 h co-culture with cells (Fig. 2d,h and l). Interestingly, negatively stained amyloid fibrils of morphology not dissimilar to intracytoplasmic Aβ 42 were also found deposited within nuclei of THP-1 cells (see Supplementary Fig. 3). Therefore, TEM of THP-1 cells cultured in the presence of a model Aβ 42 peptide antigen confirmed the presence and intracellular persistence of amyloid β-sheets.
Direct fluorescent labelling reveals the cellular internalisation of simulated amyloid adjuvant vaccines. The direct fluorescent labelling of a model Aβ 42 amyloid antigen (4 μM) and its co-adsorbed ABA (25.0 μg/mL) utilised the fluorophores ThT and lumogallion, respectively. Experiments monitoring cellular uptake of simulated vaccine treatments utilised 20 and 100 μM of the respective fluorophores, throughout. Initially, both fluorescent dyes were added to simulated vaccine treatments at T = 0. All pre-incubated vaccine treatments (24 h) co-cultured 1:1 with THP-1 cells, emitted positive intracellular lumogallion (orange) and ThT fluorescence (green). The broad emission of lumogallion complexed to particulate ABA was, however, found to pass through the longpass fluorescence filter used to collect fluorescence for ThT-reactive amyloid (482 nm), rendering adjuvant and antigen difficult to distinguish from one another (see Supplementary Fig. 4).
In order to isolate the fluorescence of ThT from lumogallion, a single bandpass filter was used (see Supplementary Fig. 5). As such, only emitted light of wavelength 470-500 nm was allowed to pass through to the detector, encompassing the ThT emission maxima of 482 nm when complexed to amyloid. Fluorescence microscopy of aged Aβ 42 (29 μM), prepared as sections (5 μm) and stained with 1 mM ThT for 24 h, revealed a blue versus a green fluorescence emission under single and long bandpass filters, respectively (see Supplementary Fig. 6). While specificity of ThT to mature amyloid fibrils was observed under both filter types, the fluorescence intensity was found to be weaker under single bandpass emission. However, it was found that fluorescence emission for lumogallion and ThT bound to adjuvant and antigen respectively, could be successfully isolated thereby allowing for their intracellular identification within THP-1 cells (see Supplementary Fig. 5). Therefore, single bandpass emission filters were used throughout for the detection of intracellular fluorescence of antigen and adjuvant formulations in simulated vaccine treatments. Optimisation of pre-labelling simulated vaccine formulations focused upon obtaining distinguishable fluorescence of amyloid co-adsorbed to its ABA, over the shortest possible time period. Simulated vaccine treatments containing 8 μM Aβ 42 in the presence or absence of 25.0 μg/mL of an ABA (Alhydrogel ® , Adju-Phos ® or Imject ™ Alum) were first incubated for 24 h, prior to their addition to THP-1 cells. Pre-incubated treatments were subsequently plated 1:1 with cells for 24 h, of which the fluorophores ThT and lumogallion were simultaneously added only 3 h prior to fixation, in order to minimise their influence on the cellular uptake of Aβ 42 and co-adsorbed ABA.
THP-1 cells co-cultured in the presence of pre-labelled Aβ 42 only produced a blue fluorescence emission within the cytoplasm of THP-1 cells. In the absence of added ThT, THP-1 cells co-cultured under identical conditions produced a uniform blue fluorescence emission of lower relative intensity (see Supplementary Fig. 7).
Simultaneous pre-labelling of Aβ 42 in the presence of Alhydrogel ® revealed an intracellular orange fluorescence emission under the lumogallion fluorescence channel, indicative of particulate loading in cell cytosol (Fig. 3a). Upon illumination of the same population of whole cells under the ThT fluorescence channel, an intense intracellular blue fluorescence was observed (Fig. 3b). Merging of both lumogallion and ThT fluorescence filter channels for THP-1 cells co-cultured with Aβ 42 adjuvanted with Alhydrogel ® demonstrated the co-localisation of amyloid and adjuvant material, of which merging of both signals resulted in a pink-purple fluorescence emission (Fig. 3c). Furthermore, overlaying of the bright field image confirmed that ThT and lumogallion reactive material was confined to THP-1 cells (Fig. 3d).
For those cells co-cultured with vaccine treatments containing Aβ 42 in the presence of Imject ™ Alum, lumogallion-reactive particulates were identified in cell cytosol via an intensive orange fluorescence emission (Fig. 3e). ThT-reactive material was additionally identified intracellularly via the observation of a blue fluorescence emission confined to THP-1 cells (Fig. 3f). As with cells treated with amyloid and adjuvanted with Alhydrogel ® , both lumogallion and ThT fluorescence was found co-localised in the cytoplasmic compartment of cells (Fig. 3g). Merging of bright field and fluorescence channels indicated that ThT and lumogallion positive fluorescence was confined to THP-1 cells (Fig. 3h).
Simulated Aβ 42 vaccine treatments adjuvanted with 25.0 μg/mL Adju-Phos ® and co-cultured with THP-1 cells, produced an intensive and uniform intracellular orange fluorescence emission, under the lumogallion filter channel. Due to the inability to distinguish lumogallion reactive material in whole cells, the concentration of Adju-Phos ® was halved in cell culture conditions. Subsequent analyses of co-cultured THP-1 cells highlighted the presence of a punctate intracellular orange lumogallion-reactive fluorescence emission (Fig. 3i). A blue fluorescence emission was also observed in identical cells under the ThT fluorescence channel and in close proximity to those structures producing lumogallion fluorescence (Fig. 3j). Interestingly ThT fluorescence was identified infrequently and of a lower intensity in comparison to those cells co-cultured in the presence of Alhydrogel ® and Imject ™ Alum, respectively. Merging of fluorescence channels revealed the close proximity of ThT and lumogallion fluorescence within the cytoplasm of THP-1 cells (Fig. 3k), of which overlaying of the bright field image confirmed their co-deposition in cells (Fig. 3l).

Confirmation of the cellular internalisation of amyloid and aluminium based adjuvant materials via transmission electron microscopy. The complementary method of TEM was performed in
addition to fluorescence microscopy, in order to support the observations of intracellular Aβ 42 co-adsorbed to its respective ABA. THP-1 cells were co-cultured with simulated amyloid vaccines under near-identical conditions and pre-labelling regimes, of which the ABA concentration was maintained at 50.0 μg/mL across treatment conditions. This concentration of ABA has proven optimal in our previously published studies of adjuvant uptake in a THP-1 cell line 27,28 and hence, was selected herein.
THP-1 cells co-cultured with simulated amyloid vaccines adjuvanted with Alhydrogel ® revealed semi-crystalline and electron dense needles found solely within vesicular-like compartments within cell cytoplasm (Fig. 4a). Higher magnifications revealed negatively stained amyloid fibrils found deposited within cell cytosol. Interestingly, fibrillar deposits of apparent amyloid were identified both in the absence and presence of co-deposited needles of the adjuvant (Fig. 4d). Although rarely observed, negatively stained amyloid fibrils were found localised in cell nuclei, as with those experiments performed in the presence of the peptide only (see Supplementary Fig. 3). THP-1 cells co-cultured in simulated vaccine treatments containing Imject ™ Alum, demonstrated the intracellular presence of large amorphous plate-like particulates of the adjuvant, spanning several microns in width. Those electron-dense aggregates identified were contained in vesicular-like compartments and found solely within the cytoplasm of THP-1 cells (Fig. 4b). Negatively stained amyloid fibrils were observed only in the cell cytoplasm of THP-1 cells loaded with particulate adjuvant. The direct co-localisation of fibrillar amyloid and amorphous plates of particulate adjuvant material were clearly identified via negative staining with uranyl acetate (Fig. 4e). Finally, analyses of THP-1 cells co-cultured with Adju-Phos ® in the presence of Aβ 42 revealed the clear presence of positively stained amorphous aggregates in endosomal-like compartments (Fig. 4c). While particulate ABA was clearly contained intracellularly, classical negative staining of fibrillar amyloid was elusive. However, fibril-like projections formed between adjacent particulates were identified contained within the cytoplasm of THP-1 cells.

Discussion
We have investigated the fate of amyloid as a model antigen in a cellular model of vaccination. Through the use of direct fluorescent labelling, the intracytoplasmic accumulation of Aβ 42 propagating as β-sheets was found in a monocytic THP-1 cell line. The use of ThT as a conformation-specific fluorophore for β-sheets of Aβ 42 revealed that cytosolic loading of the amyloidogenic peptide occurred within 1 h and persisted intracellularly over 24 h, as revealed by a green intracellular fluorescence emission 24 .
A recent study monitoring the cellular uptake of Aβ 42 identified the clathrin-independent endocytic route of macropinocytosis that allowed for the entry of the amyloidogenic peptide into a neuroblastoma SH-SY5Y cell line 21 . As a known pathway for the cellular uptake of protein antigens, solute molecules and nutrients, internalisation via this route is rapid as with phagocytosis occurring within minutes upon exposure to cultured cells 29,30 . The subsequent fusing of macropinocytotic vesicles with late lysosomes results in an acidic protease-rich environment, capable of degrading their vesicular cargo 30 .
As a high resolution and complementary technique to fluorescence microscopy, TEM of sectioned THP-1 cells successfully identified intracytoplasmic negatively stained amyloid fibrils. Interestingly, while the latter has been shown intraneuronally in vivo, their enclosure in endosomal-like vesicles was not observed under the dosing regimen utilised herein 20,31 . Therefore, our results indicate the absence of mature autophagolysosomes, typically demonstrated through the presence of membrane-enclosed vesicles via TEM 27,31 .
In support of our new findings, recent studies have shown that while the cellular uptake of amyloid fibrils mechanistically initiates autophagy, the resultant degradation of amyloid is hindered through rupturing of lysosomal compartments 32  More recently, fibrillar forms of the Parkinson's disease (PD) related amyloidogenic protein, α-synuclein, were found to induce rupture of intracellular vesicles, following endocytosis 33 . Interestingly, amyloid fibrils produced from both wild-type and mutant familial related variants of the protein were observed to rupture lysosomal vesicles whether co-cultured with neuroblastoma SH-SY5Y or human dopaminergic neuronal cell lines. Vesicle rupture of the former was also noted when co-cultured with the human AD-related microtubule-associated protein, tau 33 . Therefore the propensity of all amyloidogenic proteins to undergo structural conversion from their soluble native α-helical form to a conformation rich in β-sheet structure, pre-disposes their fibrillar structure to rupture intracellular lysosomal vesicles upon uptake 32,33 .
Aβ 42 of increasing size was additionally identified in a β-pleated sheet conformation within cell nuclei over prolonged incubation periods. The presence of nuclear inclusions containing amyloid in neurodegenerative disorders has revealed co-localisation of the eukaryotic protein, ubiquitin 17,34 . As an important initiator of the proteasomal machinery in intracellular environments, covalent conjugation of the protein to misfolded proteins would normally lead to their degradation, in vivo 17 . In AD, ubiquitin is known to persist in such nuclear protein inclusion bodies of which the accumulation of long-chain ubiquitin molecules only has been shown to form amyloid-like fibrils, potentially exacerbating the condition 17 .
The persistence of intracytoplasmic or intranuclear amyloid may also be explained through disruptions in protein homoeostasis 31 . The mammalian target of rapamycin complex 1 (mTORC1) is a key regulator of autophagy of which its inactivation allows for nutrient metabolism and the removal of non-functional and misfolded proteins 35 . In AD however, downstream mTORC1 signalling through phosphoinositide-3-kinase (PI3K), leads to the accumulation of autophagic vacuoles 36 . Therefore, our observation of cellular accumulation of the peptide antigen only depositing as mature amyloid fibrils, possibly owed to the inability of THP-1 cells to degrade these aberrant deposits.
The direct fluorescent labelling of a model Aβ 42 peptide antigen formulated in the presence of clinically relevant and experimental ABA preparations revealed their co-localisation in the cytosol of THP-1 cells. Lumogallion is an established and sensitive fluorescent molecular probe for the identification of aluminium in both cells and tissues and was utilised herein to monitor the fate of ABA, co-adsorbed to its target peptide antigen 28,37,38 . The complementary technique of TEM supported the observations from fluorescence microscopy, revealing the co-uptake of ABA and aggregates of β-pleated sheet amyloid antigen materials, in sectioned THP-1 cells.
Furthermore, TEM confirmed the presence of needle-like crystals of Alhydrogel ® , plate-like aggregates of Adju-Phos ® and amorphous deposits of Imject Alum ™ , all found enclosed within endosomal intracellular compartments. Therefore, the cellular uptake of both the clinically approved and experimental adjuvant formulations was in agreement with our previous work, even when coadministered with a model Aβ 42 peptide antigen 27 .
Negatively stained amyloid fibrils of the model antigen were clearly distinguishable in cell cytosol when co-administered with Alhydrogel ® and Imject Alum ™ . Interestingly, mature fibrillar aggregates co-deposited with Alhydrogel ® or directly attached to Imject Alum ® were identified on the periphery of endocytosed particulate adjuvant materials. However, intracellular aggregates of mature Aβ 42 were elusive in vaccine formulations adjuvanted with Adju-Phos ® in which the potential co-localisation of amyloid fibrils and ABA could only be detected at higher magnifications. Intracytoplasmic vesicles noted for Aβ 42 formulated in the presence of ABA, indicated that endocytosis was the most likely route of entry through autophagic processing to lysosomes 27,37,39 . As with THP-1 cells co-cultured with Aβ 42 only however, mature amyloid fibrils whether co-localised or co-adsorbed to their target ABA, were not observed to enter endosomal-like vesicles. Therefore, cellular uptake of Aβ 42 as a model peptide antigen and ABA are suggested to be internalised via the differing endocytic pathways of macropinocytosis and autophagy, respectively 21,29,32,33 . To our knowledge, this is the first report of the apparent lack of lysosomal vesicular enclosure of the pathogenic Aβ 42 species in a simulated cell-based vaccination model of AD.
The thermodynamic saturation constant of Aβ 42 predicts that spontaneous amyloid fibril formation occurs at concentrations at or exceeding 2 μM 40 . As such, catalysis of Aβ through metal interactions, glycoprotein binding and seeding have been implicated in the induction of auto polymerisation of nanomolar concentrations of the peptide, typically found in vivo [40][41][42] .
Furthermore, templating of soluble Aβ 40 in vitro has been found to undergo polymerisation to a mature fibrillar form, upon the addition of brain extracts containing amyloid fibrils from transgenic AD mice 42 .
More recently, corrupting of host Aβ in vivo in the brains of transgenic murine models of AD has been demonstrated through intracerebral inoculation with brain extracts containing fibrillar amyloid 8,43 . Interestingly, Aβ injected intraperitoneally was found to elicit a Trojan horse like mechanism, delivering Aβ as seeds to the brain of transgenic murine models of AD 8 . Regardless of the mechanism by which self-aggregation occurs, repeated nucleation of the soluble alpha-helical form of the peptide through nucleation-driven seeding results in the self-propagation of amyloid to a conformation rich in β-sheet structure 1,3,18 .
Therefore, our results support the uptake and potential transport of amyloid in a β-pleated sheet conformation that persists intracellularly, speculatively through lysosomal degradation 32,33 accumulating and propagating as plaque-like fibrillar deposits. Circulating blood-borne monocytes are known to enter the central nervous system (CNS) in response to neuroinflammation and are thought to promote tissue repair and support the production of neurotrophic growth factors 44 . The intracellular persistence of mature amyloid fibrils through evasion of autophagy or proteasomal-degradation mechanisms may therefore also allow for the silent entry of fibrillar amyloid across the blood-brain barrier (BBB).
Monocytes are one of the first phagocytic antigen presenting cell (APC) types to be recruited to the site of injection, following vaccination 45 . As such, mechanisms governing the immunoreactivity and cellular fate of amyloid may prove key in the development of vaccination regimes targeting these aberrant deposits. In summary, we herein demonstrate the ability for β-pleated sheet-rich amyloid fibrils to persist intracellularly, of which their translocation away from the injection site could be enhanced in the presence of clinically relevant ABA. Alhydrogel ® has previously been demonstrated as a more effective adjuvant for translocation away from the injection site, through its heightened cellular loading in THP-1 cells versus Adju-Phos ® . The apparent lack of mature amyloid fibrils found for cells co-cultured with the latter would further implicate such in vivo 27 .
In conclusion, we have demonstrated the use of direct fluorescent labelling in the cellular monitoring of amyloid when administered in simulated vaccine formulations. The ability to detect both an intracellular amyloid antigen and its co-adsorbed ABA following their internalisation, provide proof of concept for the transport of both fluorophores across the cell membrane of THP-1 cells 37 . Furthermore, the conformation-specific fluorophore, ThT, clearly demonstrated the cellular persistence of misfolded amyloid deposits in both cytosolic and nuclear inclusions. Overall, direct fluorophore-labelling and high-resolution TEM approaches omitting usual lead citrate staining may provide invaluable tools in the assessment of existing and future vaccinations directed against amyloid and other relevant neuropathological hallmarks of wider dementia.

Methods
Preparation of Aβ 42 stock solutions. All chemicals were purchased from Sigma Aldrich, UK unless otherwise stated. The amyloidogenic peptide, Aβ 42 was purchased from Bachem as the lyophilised salt and was reconstituted in 0.01 M NaOH to prepare 100 μM stocks. Under these highly alkaline conditions (ca pH 12), the fully dissolved peptide only exists in a monomeric form 25,46 . Subsequent stock solutions were aliquoted and stored frozen (−20 °C) and the peptide was thawed fully, immediately prior to use. Peptide concentrations in the final stock solutions were determined prior to their dilution in complete R10 medium. Briefly, thawed stocks were centrifuged at high speed (5 min, 15000 g) and their protein content determined using absorbance at 280 nm using a NanoDrop 1000 spectrophotometer (Thermo Scientific, UK). The final measured peptide concentration was adjusted based on percentage purity stated by the manufacturer and was used to prepare subsequent dilutions of Aβ 42 . Initial experiments, monitoring the cellular uptake of Aβ 42 only (longpass emission), were formulated in R10 medium containing 20 μM ThT at T = 0. Subsequent treatments prepared in this manner were added to THP-1 cells following 24 h incubation. Experiments utilising treatments containing both Aβ 42 and an additional ABA were plated with THP-1 cells and incubated for 21 h at 37 °C (5% CO 2 ). Dyes for pre-labelling utilised thioflavin T (ThT) at 100 μM and lumogallion at 1 mM, both prepared in ultrapure water and filtered through 0.22 μm syringe filters. ThT and lumogallion were added to cells co-cultured with simulated vaccine treatments at a final concentration of 10 and 50 μM respectively and treated cells were incubated for a further 3 h, prior to fixation.

Preparation
Fluorescence microscopy. Whole non-sectioned THP-1 cells were prepared on poly-lysine coated slides, as described in the Supplemental methods. Fluorescence micrographs were obtained by use of an Olympus BX50 microscope equipped with a BX-FLA reflected light fluorescence attachment (mercury source) and a vertical illuminator. Micrographs obtained at X 1000 magnification utilised an X 100 Plan-Fluorite oil immersion objective (Olympus, UK) using low auto-fluorescence immersion oil (Olympus immersion oil type-F).
Transmission electron microscopy. Spurr-resin embedded agar-cell blocks were sectioned using cut glass knives at 90 nm, by use of an automated REICHERT-JUNG Ultracut E ultramicrotome. Any sections containing ABA materials were sectioned at 100 nm by use of a Leica ultracut UCT ultramicrotome, equipped with a 45° Diatome diamond knife (30-US, Electron Microscopy Sciences). Sections were mounted on G2002 200 mesh thin bar copper grids (Athene, Agar Scientific, UK), pre-treated with a COAT-QUICK "G" grid coating pen (Daido Sangyo Co. Ltd. Japan) for adherence. Grids were allowed 24 h drying time prior to analysis. Cell sections for TEM were viewed on a JEOL 1230 transmission electron microscope operated at 100 kV (spot size 1). A 10 μA activated field emission was used to increase the standing current to 67-68 μA during image acquisition. Electron micrographs were captured using a Megaview III digital camera from Soft Imaging Systems (SiS), using the iTEM universal TEM imaging platform software. Data availability. All data generated or analysed during this study are included in this published article and its Supplementary Information files.