Introduction

Anti-MAG neuropathy is a slowly progressing polyneuropathy associated with IgM monoclonal gammopathy of neurological significance1. The peripheral neuropathy is caused by pathogenic IgM autoantibodies that bind the human natural killer-1 (HNK-1) epitope, a carbohydrate present on Schwann cells of the peripheral nerve fibers2. The binding of anti-MAG autoantibodies induces demyelination of peripheral nerves and causes slowly progressing and predominately distal sensory ataxia, neuropathic pain, weakness, tremor, and disability3. There is a high unmet medical need for more specific options for the therapy of this chronic disease, since the currently used off-label treatments often lack efficacy and non-specifically modulate the immune system. Since the progression of this chronic neuropathy is closely related to anti-MAG IgM autoantibody levels and their relative reduction correlates well with a clinical improvement, a specific and efficient removal of the autoantibodies from circulation could be highly beneficial2,4,5,6,7,8,9,10. In previous studies, we demonstrated the efficient removal of anti-MAG IgM antibodies in an immunological mouse model with PPSGG11, a glycopoylmer consisting of a biodegradable poly-l-lysine backbone of approximately 400 lysine units, of which 35% are linked to a mimetic of the HNK-1 glycoepitope. The remaining 65% of the lysine side chains are coupled with thioglycerol substituents to increase solubility of PPSGG. This glycopolymer of 194 kDa binds to the anti-MAG IgM in a highly selective and multivalent manner1.

The monovalent mimetic of the HNK-1 glycoepitope shows only millimolar affinity to anti-MAG IgM autoantibodies of patients. However, when the glycoepitope is presented polyvalently the affinity increases up to 230,000-fold11, firstly because of the high local density of the antigen leading to fast rebinding after dissociation and secondly because of simultaneous binding to multiple binding sites of the pentameric IgM12,13,14. Targeting lectins with multivalent glycopolymers has gained increasing therapeutic interest, such as in the field of biological toxins, e.g. cholera toxin, shiga toxin, and ricin15,16,17,18,19,20. Furthermore, multivalent glycopolymers were applied to remove host antibodies against the α(1-3)-galactose epitope in xenotransplantation21.

Comparable polyvalent epitope presentations by nanoparticles are also applied therapeutically. They are used for tissue-specific targeting, gene delivery, immunomodulatory purposes, as tracers for medical imaging, and carriers of drug loads14,22,23,24,25,26,27,28. However, common challenges of nanoparticles are toxicological implications and quick metabolic degradation29,30,31,32,33. Another problem related to nanoparticles is their short plasma half-life caused by phagocytosis, an obstacle that can be addressed by chemical modifications, most often by pegylation33. A comparable short plasma half-life was also observed for PPSGG.

In this communication, we investigated the mode of action of PPSGG by characterizing its shape and surface properties, its binding stoichiometry with the anti-HNK-1 IgM antibodies, as well as the size of the PPSGG/IgM complex. Furthermore, we studied tissue distribution of PPSGG and the PPSGG/IgM complexes, the specific cellular uptake mechanism and the elimination pathway.

Results and discussion

The linear glycopolymer PPSGG preferably binds one or two anti-HNK-1 IgM antibodies in vitro

For a better understanding of the novel mode of action as well as related potential risks, we studied the morphological characteristics, tissues distribution, and route of elimination of PPSGG alone and the PPSGG/anti-HNK-1 IgM complex.

For the assessment of the colloidal stability and aggregate formation of PPSGG (Fig. 1), anti-HNK-1 IgM and the PPSGG/IgM complex, dynamic light scattering (DLS) measurements were performed. For PPSGG a single peak with a hydrodynamic diameter of 64.02 nm (± 3.34 nm SD) and a polydispersity index (PdI) of 0.357 (± 0.007 SD) (Fig. 2A-1) were obtained. The hydrodynamic radius of anti-HNK-1 IgM amounts to 46.18 nm (± 11.12 nm SD) with a PdI of 0.279 (±0.032 nm SD) (Fig. 2A-2) and is consistent with the elution behavior observed on size exclusion chromatography (SEC) during the Multi Angle Light Scattering (MALS) experiments (Supporting Information Fig. S1). The peak broadness, however, results in a high PdI and is based on the polydispersity of diffusion coefficients in the sample, requiring contributions from a broader range of size classes, making exact mass estimates inaccurate. When PPSGG was co-incubated with anti-HNK-1 IgM to form the PPSGG/anti-HNK-1 IgM complex, a peak size of an average of 52.33 nm (± 11.01 nm SD) with a PdI of 0.271 (± 0.038 SD) were measured (Fig. 2A-3 to 6). Increasing concentration of IgM, did not further affect the peak size. Thus, although the multivalency of both glyopolymer and pentameric anti-HNK-1 IgM could potentially lead to crosslinked complexes and the formation of large aggregates, none of them were observed according to DLS data. For more accurate size and shape determination, transmission electron microscopy (TEM) was applied. PPSGG, anti-HNK-1 IgM, and mixtures thereof were incubated at room temperature for 30 min and stained with 2% uranyl acetate for analysis. PPSGG showed a rod-like linear polymer structure with an approximate length of 100 nm (Fig. 3A), whereas the pentameric anti-HNK-1 IgM exhibited a diameter of approximately 50 nm (Fig. 3B). When PPSGG was co-incubated with anti-HNK-1 IgM, spherical structures were formed with a diameter similar to the IgM alone (Fig. 3C), confirming the results of the DLS and SEC-MALS experiments.

Figure 1
figure 1

PPSGG: PLL bearing the HNK-1 mimic (PLL400, loaded with 35% HNK-1 mimic, 194 kDa); PPSGG-pHAb: PPSGG labeled with the pH-sensitive dye pHAb; PPSGG-sulforhodamine: PPSGG labeled with the fluorescent dye sulforhodamine 101. For protocols of the syntheses see Supporting Information.

Full size image
Figure 2
figure 2

Hydrodynamic diameter, surface charge of PPSGG and stoichiometry of PPSGG/anti-HNK-1 IgM complex. (A) DLS plots of PPSGG (A1), anti-HNK-1 IgM (A2), and a titration of increasing concentrations of anti-HNK-1 IgM [0.13 mg/ml (A3), 0.2 mg/ml (A4), 0.3 mg/ml (A5), and 0.33 mg/ml (A6) with PPSGG [0.267 mg/ml (A3), 0.2 mg/ml (A4), 0.1 mg/ml (A5), and 0.067 mg/ml (A6)] showing broad but monodisperse peaks remaining stable at around 50 nm. PPSGG and anti-HNK-1 IgM share similar hydrodynamic diameters and remain stable with increasing concentrations of anti-HNK-1 IgM. No formation of larger aggregates was observed. (B) PPSGG was incubated with anti-HNK-1 IgM at RT for 30 min before measuring the fluorescence at a rotor speed of 30,000 rpm on a Beckman Coulter XL-l AUC. The mass was calculated based on the frictional ratio of human IgM (f/f0, 1.82). Results show two peaks at 22 S and 33 S, which correspond to a molecular weight of 1.3 MDa and 2.3 MDa. Given the 950 kDa mass of IgM and approx. 200 kDa mass of PPSGG, the data indicates a 1:1 and 2:1 binding stoichiometry that is shifted towards 2:1 complexes in presence of larger amounts of IgM. (C) PPSGG (1 mg/mL) was diluted in 10 mM NaCl and the zeta potential measured at 25 °C using an applied voltage of 150 V. The average (five measurements) zeta potential values for PPSGG measured at different pH values (pH 4–10). The zeta potential averaged − 46 mV (± 3.7 mV SD) over the whole pH range. A comparison with a reference polymer (undisclosed) with only half the negative charges per epitope reduces the zeta potential by approx. 50%.

Full size image
Figure 3
figure 3

Transmission electron microscopy (TEM) images of PPSGG and the anti-HNK-1 IgM antibody with the lower panels displaying the squared areas of the images. The size and shape of PPSGG (A), anti-HNK-1 IgM (B), and the PPSGG/anti-HNK-1 IgM complex (C) were characterized by TEM. PPSGG and anti-HNK-1 IgM samples were incubated on a Copper grid 400 mesh coated with Parlodion and carbon, blotted and washed with water, and stained with 2% uranylacetate. Images were taken with a Tecnai G2 Spirit instrument (120 kV TEM Microscope with EMSIS Morada camera). PPSGG has a size of approximately 100 nm, IgM is pentameric with a diameter of 40–50 nm, and the formed complexes are of similar size as the antibody. No crosslinking or large aggregate formation was observed.

Full size image

An alternative method for the assessment of the binding stoichiometry of complexes is based on the determination of the sedimentation coefficient by analytical ultracentrifugation (AUC), allowing the calculation of the molecular mass. PPSGG and the fluorescein isothiocyanate (FITC)-labeled anti-HNK-1 IgM were pre-incubated and the sedimentation velocity was monitored by fluorescence detection. The data were analyzed using a sedimentation coefficient distribution (c(s)) analysis. Based on the frictional ratio (f/f0) and the sedimentation coefficient, the masses were calculated by two different approaches. Firstly, a fixed frictional ratio for IgM of 1.82 was used34, indicating that IgMs have a fairly extended structure compared to a perfect sphere with the frictional coefficient of 135. Secondly, to account for the presence of differently sized species, a bimodal f/f0 fitted distribution was applied. Both approximations led to similar results showing two distinct peaks for the sedimentation coefficient at 22 S and 33 S, corresponding to a molecular weight of 1.3 MDa (1:1-binding stoichiometry) and 2.3 MDa (1:2-binding stoichiometry), respectively. Furthermore, the peak intensities were dependent on the ratio of the two binding partners, meaning that higher antibody concentration lead to an increase of 1:2 binding (Fig. 2B). In summary, we could confirm the 1:1- and 1:2-binding stoichiometry of the PPSGG/anti-HNK-1 IgM complex as suggested by a previous in vivo dose titration study in mice1.

An additional parameter of interest in polymer formulations is the colloidal stability described by the zeta potential, a measure of the charge density on the particle surface, which for homopolymers correlates with the propensity for flocculation. Thus, the zeta potential of 1 mg/ml PPSGG in 10 mM NaCl over a pH range from 4 to 10 amounted to an average of − 46 mV (± 3.7 mV SD) over the whole pH range indicating a low probability for aggregate formation (Fig. 2C). The surface charge of the PPSGG results from the negative charges of the carboxylate and sulfate group present on the HNK-1 epitope mimetic (Fig. 1). The negative surface charge is unaffected over the whole pH range because at pH 4 the carboxylate (pKa = ≤ 4) is only partially protonated and the sulphate (pKa = ≤ 2) fully deprotonated. The negative zeta potential explains the observed short half-life of PPSGG in vivo (17 min) due to the high level of negative charge enabling internalization through scavenger receptors (discussed in more detail below), and reduces some toxicity concerns related to positively charged nanoparticles, such as sequestration in the lungs11,36,37. Compared to a reference glycopolymer (undisclosed) with similar carbohydrate mimetic density but only one negative charge per glycoepitope, the contribution of the negatively charged groups on the zeta potential becomes evident (Fig. 2C).

Uptake of PPSGG by liver and spleen resident macrophages

Similar to observations reported for nanoparticles27,38, we expected that distribution and clearance of glycopolymers are also mediated by their chemical composition, size, shape, and surface charge. In a previous study, we demonstrated the plasma half-life of PPSGG in naïve BALB/c mice to be approximately 17 min11. Therefore, we investigated the tissue distribution of PPSGG, its metabolic fate, and the possibility of the formation of aggregates and tissue accumulation in vivo. To monitor the tissue distribution, naïve mice were intravenously injected with the PPSGG (10 mg/kg) labeled with the fluorescent dye sulforhodamine 101 (PPSGG-sulforhodamine, Fig. 1). The animals were perfused, their organs harvested, and large sections of the organs embedded in activated agarose. The solidified agarose blocks were then processed with a microtome and images were recorded with a two-photon system; in red to detect the PPSGG-sulforhodamine and in green to show the naturally occurring autofluorescence of the organs, enabling a label-free visualization of the organ structure. Analysis of the serial two-photon tomography showed extensive uptake of PPSGG-sulforhodamine by liver and spleen (Fig. 4A). While liver uptake is distributed more broadly, the uptake by the spleen is localized in the marginal zone that surrounds the B and T cell rich follicular zone. Since sulforhodamine 101 is widely used to fluorescently stain mouse brain astrocytes in vivo39, the uptake into the brain of the experimental animals was studied with PPSGG-sulforhodamine. No staining of the brain could be observed, and extracellular degradation of PPSGG-sulforhodamine, leading to a loss of the fluorescent dye, can be excluded. Thus, the crossing of the blood–brain barrier is unlikely (Supporting Information, Fig. S2). The tissue distribution further highlights the short half-life of PPSGG (approx. 17 min), emphasizing the quick uptake in the cells of the mononuclear phagocyte system (MPS) after 10 min and 1 h (Fig. 4A). The uptake, however, does not lead to an excessive accumulation, since the fluorescence signal diminishes quickly and completely disappears after 6 h and is sustained over 72 h post injection (Supporting Information Fig. S3).

Figure 4
figure 4

Tissue distribution of PPSGG-sulphorhodamine and anti-HNK-1 IgM in mice after intravenous injection. (A) The tissue distribution of 300 µg fluorescence labeled and IV injected PPSGG-sulforhodamine (red) in BALB/c mice after 10 min and 60 min, in comparison to control samples. The perfused organs were harvested and embedded in agarose for imaging with a TissueVision by serial two-photon tomography. PPSGG-sulforhodamine did not form aggregates in the kidney and was quickly and extensively taken up by the liver and cells of the marginal zone of the spleen. (B) The lower left panel shows a confocal microscopy image of the liver uptake of PPSGG-sulforhodamine (red) after 15 min, with tissue resident macrophages (F4/80) stained in green and colocalization depicted in yellow. (C) The image on the lower right shows the uptake of 60 µg anti-MAG IgM (blue) when co-injected with 300 µg PPSGG-sulforhodamine (red) after 60 min. The anti-HNK-1 IgM colocalized (purple) with PPSGG-sulforhodamine indicating a targeted degradation through the MPS.

Full size image

The liver and marginal zone in the spleen are rich in macrophages that are involved in the clearance of endogenous debris as well as pathogens and therapeutics40,41. We therefore raised the hypothesis that the uptake of PPSGG is mediated by the MPS. To test this hypothesis, we analyzed liver sections by staining liver resident macrophages with anti-F4/80 antibodies, a marker for Kupffer cells42,43. When confocal microscopy images of PPSGG-sulforhodamine (red) and anti-F4/80 antibodies (in green) were recorded, colocalization of PPSGG-sulforhodamine and macrophages was depicted in yellow, confirming the specific uptake of PPSGG by cells of the MPS (Fig. 4B). This distribution in the liver is very similar to the pattern described by Rothkopf et al., showing uptake of liposomal nanoparticles not only in Kupffer cells but additionally in sinusoidal endothelial cells of the liver. A contribution of these cells to the uptake of PPSGG is likely, considering the scavenger receptor mediated uptake of anionic nanoparticles reported by Campbell et al.37,44.

In addition, naïve mice were intravenously injected with Pacific Blue™ labeled anti-HNK-IgM followed by an injection of PPSGG-sulforhodamine. The liver sections showed uptake of both PPSGG-sulforhodamine (red) and anti-HNK-1 IgM (blue) by cells of the MPS, leading to purple colocalization (Fig. 4C). Given the excess of labeled PPSGG-sulforhodamine to anti-HNK-1 IgM, an increase of PPSGG-sulforhodamine uptake in the liver was detected. However, the uptake of the PPSGG-sulforhodamine/anti-HNK-1 IgM complex was delayed by one hour compared to the rapid clearance of PPSGG-sulforhodamine within the first 10 min. This is illustrated by the lack of uptake after 10 min (injected separately with PPSGG-sulforhodamine and anti-HNK-1 IgM, Supporting Information Fig. S4) and the delayed signal one hour after administration in liver and spleen sections of mice (1 h PPSGG + IgM, Fig. 3A) compared to the much more pronounced fluorescent signal in the marginal zone of spleen and liver of animals treated with PPSGG-sulforhodamine alone (Fig. 4A). These results point out the importance of the negative charges of PPSGG and their accessibility for receptor recognition. For steric reasons, the surface charges of PPSGG/anti-HNK-IgM complexes are less accessible leading to delayed uptake into the MPS. Nevertheless, as shown in previous work, the in vivo elimination of anti-HNK-1 IgM after complex formation with PPSGG was highly selective and efficient. The reduction of anti-HNK-1 IgM levels upon treatment was sustained, even during the continuous production of anti-HNK-1 IgM in the immunological mouse model11.

Macrophage uptake of PPSGG mediated through scavenger receptors

After clarification of binding stoichiometry and metabolic pathway in mice, the uptake mechanism in human macrophages was studied. For this purpose, THP-1 monocytes were differentiated with phorbol 12-myristate 13-acetate (PMA, 100 nM), for 72 h into CD14+ mature macrophages45,46. These cells were then incubated with 0.1 μM PPSGG labeled with the pH-sensitive dye pHAb (PPSGG-pHAb, Fig. 1) or with 0.1 μM non-glycosylated poly-l-lysine-pHAb and the fluorescence signal was recorded by fluorescence spectroscopy and point scanning confocal microscopy. Since the fluorescence intensity of the pHAb dye is drastically increased in an acidic environment47, the internalization of PPSGG-pHAb by macrophages via endosomes and lysosomes leading to the degradation of poly-l-lysine and the carbohydrate components, can be detected48. The active uptake mechanism was confirmed by incubating the cells at 4 °C49, where no signal could be detected in contrast to incubation at 37 °C (Supporting Information, Fig. S5). Because PPSGG-pHAb was internalized to a much higher degree than even a ten-fold higher concentration of poly-l-lysine-pHAb, the major contribution of the HNK-1 mimetics to endocytosis (Fig. 5A1) could be confirmed.

As described by Rothkopf et al.37 and Campbell et al.44 polyanionic nanoparticles are readily taken up through scavenger receptors by cells of the MPS and liver sinusoidal endothelial cells. For investigating the contribution of this endocytic pathway to the uptake of PPSGG-pHAb, we treated THP-1 derived macrophages with dextran sulfate (Fig. 5A2), a negatively charged polysaccharide internalized by scavenger receptors50,51. Incubation with increasing concentrations of dextran sulfate inhibited the uptake of PPSGG-pHAb in a concentration-dependent manner, but had no effect on the uptake of poly-l-lysine-pHAb. This suggests that the uptake is considerably mediated by the two negative charges present on each carbohydrate moiety of the glycopolymer (see highly negative zeta potential, Fig. 2C) and to a lesser extent by non-concentration-dependent mechanisms such as fluid-phase micropinocytosis52,53. The main members of the scavenger class A family, SR-A1 and SR-A6, formerly known as MARCO, are primarily expressed on tissue macrophages such as Kupffer cells54. To further narrow in on the receptors responsible for the endocytosis of PPSGG, we incubated the cells with AcLDL (human acetylated low density lipoprotein, Fig. 5A3), a specific ligand for SR-A1, but not for SR-A655,56. The resulting inhibition of PPSGG uptake was similar to the result obtained by the treatment with the unspecific inhibitor dextran sulphate, indicating that SR-A1 is the main scavenger for the endocytosis of PPSGG.

Figure 5
figure 5

Scavenger receptor mediated uptake of PPSGG, anti-MAG IgM and complexes thereof into human macrophages. A) By measuring the fluorescence after 9 h of incubation, the uptake efficiency of PPSGG-pHAb and poly-l-lysine-pHAb by THP-1 derived macrophages is compared. The fluorescence values are normalized to uptake of 0.1 µM PPSGG-pHAb being 100% in A1. It shows that an efficient uptake depends on the negatively charged HNK-1 mimetics meanwhile only a fraction of the positively charged poly-l-lysine-pHAb is taken up after 9 h of incubation, even at an elevated concentration of 1 µM. Co-incubation with the scavenger receptor ligands dextran sulfate (A2) and AcLDL (A3) shows the concentration dependent inhibition of PPSGG-pHAb uptake, while having no effect on the uptake of the uncharged poly-l-lysine-pHAb. B) Pre-incubation of anti-HNK-1-IgM-FITC (green, B1) with PPSGG-pHAb (red, B2) and subsequent incubation with THP-1 derived macrophages leads to an uptake of the two individual components, as well as the complexes (yellow, B3) in acidic compartments of the cells.

Full size image

When cells were treated with pre-incubated anti-HNK-1 IgM-FITC and PPSGG-pHAb, the uptake in acidic compartments of the macrophages could be demonstrated (Fig. 5B) for the individual components (green: anti-HNK-1 IgM-FITC, Fig. 5B1 and red: PPSGG-pHAb, Fig. 5B2), as well as the complexes (yellow: PPSGG-pHAb/anti-HNK-1 IgM-FITC, Fig. 5B3). We therefore conclude that the uptake of the complexes is mainly driven by the negatively charged PPSGG, since we have observed, that as a consequence of the formation of the spherical structures depicted in the TEM images (Fig. 3C), the antibody is shielded by the glycopolymer, thus withdrawing the Fc part from recognition (Supporting Information, Fig. S6).

Uptake and elimination of PPSGG by the MPS did not induce hepatotoxicity (ex vivo)

To investigate if the extensive uptake by the liver could possibly lead to drug-induced liver injury, we tested the hepatotoxic potential of PPSGG on primary human-derived liver microtissue57. The cell clusters comprised of hepatocytes and Kupffer cells were incubated with PPSGG concentrations exceeding the anticipated human dose range by more than a factor of 101. After incubation for 48 h, the viability of the liver cell was assessed by measuring free ATP with the CellTiter Glo Kit specifically designed to assess microtissue viability. Whereas the cell viability was not affected by high concentrations of PPSGG (6 mM epitope concentration, Fig. 6B), chlorpromazine, used as a positive control, showed cytotoxicity (EC50 = 14.3 μM, Fig. 6C). The uptake of PPSGG in liver microtissue was confirmed by fluorescence microscopy (PPSGG-sulforhodamine, Fig. 6A).

Figure 6
figure 6

Hepatotoxicity evaluation of PPSGG in a primary human liver microtissue model. (A) The uptake of PPSGG-sulforhodamine by primary human-derived liver microtissue. (B) PPSGG showed no cytotoxic activity over a 48 h incubation period, (C) while chlorpromazine was cytotoxic with an EC50 of 14.3 µM.

Full size image

Conclusion

With our studies we have gained insight into the mode of action of the glycopolymer PPSGG (Fig. 7) and the suitability of its PK/PD profile for the therapeutic use in peripheral neuropathy mediated by anti-MAG autoantibodies. We could show that tissue distribution and the elimination route of PPSGG through the scavenger receptors on cells of the MPS lead to quick endocytosis and subsequent degradation of its amino acid and carbohydrate components via acidic endosomes and lysosomes in liver and spleen48. These data explain the previously observed short half-life of PPSGG and the fast and efficient removal of anti-HNK-1 IgM antibodies in mice11. The fast uptake of large amounts of PPSGG did not cause cytotoxicity in primary human liver microtissue and showed favorable colloidal properties, i.e. no formation of large aggregates in solution or further mitigating toxicity concerns in vivo. With anti-HNK-1 IgM antibody, PPSGG forms spherical complexes with a binding stoichiometries of 1:1 or 1:2 (PPSGG/anti-HNK-1 IgM), confirming the suspected binding stoichiometry based on a dose titration study in mice1. PPSGG/anti-HNK-1 IgM is of comparable hydrodynamic size as PPSGG and the anti-HNK-1 IgM antibody. Moreover, in spite of the multivalent nature of both the glycopolymer and the anti-HNK-1 IgM, large aggregates through crosslinking are not formed. The active uptake of PPSGG/anti-HNK-1 IgM complexes was confirmed in vitro in a human macrophage cell line and in vivo in mice. Finally, we could show that PPSGG can selectively and quickly pull an anti-carbohydrate autoantibody (anti-HNK-1 IgM) into the MPS and then makes use of the natural degradation system of the immune system for the removal of a disease-causing autoantibody. The approach was tested in a Phase 1 clinical trial and aborted after occurrence of a serious adverse event in the sentinel patient61. The patient experienced a “feeling of warmth though body” with subsequent increase in heart rate and blood pressure. After stopping the infusion and administration of IV antihistamine therapy the patient fully recovered and left the clinical unit in the following morning. Follow up investigation indicated changes in complement levels that are associated with complement activation-related pseudoallergy, that was triggered by the release of C5a. To prevent in vivo complement activation, we propose an extracorporeal application using the anti-MAG IgM specific epitopes as a scaffold on beads for immunoadsorption. With this the highly efficient and specific binding to pathogenic anti-MAG IgM could be retained, while mitigating adverse events62.

Figure 7
figure 7

Mode of action of PPSGG: the polyanionic surface of PPSGG is recognized by scavenger receptors (SR-A1) on the surface of cells of the mononuclear phagocyte system (MPS). The binding of the PPSGG/anti-HNK-1 IgM complexes triggers a receptor mediated endocytosis, forming endocytic vesicles that are further trafficked via the progressively increasing acidic environment of the endosome/lysosome pathway, resulting in the metabolism of the complexes58.

Full size image

Materials and methods

Dynamic light scattering (DLS) and zeta potential

DLS (Fig. 2A) and zeta potential (Fig. 2C) of PPSGG (Supporting Information) was measured using a Zetasizer Nano ZS (Malvern Instruments, UK). DLS measurements of 0.2 mg/ml PPSGG and 0.2 mg/ml purified anti-HNK-1 IgM (3H5, DSHB, USA) in PBS were recorded in five runs for 50 s each at 25 °C with a dispersant refractive index (RI) of 1.33 (water) and a sample RI of 1.45 (protein) in a quartz cuvette. A titration series of anti-HNK-1 IgM (C6680, Sigma-Aldrich, Switzerland) to PPSGG was recorded accordingly with 0.13 mg/ml (A3), 0.2 mg/ml (A4), 0.3 mg/ml (A5), and 0.33 mg/ml (A6) anti-HNK-1 IgM and 0.267 mg/ml (A3), 0.2 mg/ml (A4), 0.1 mg/ml (A5), and 0.067 mg/ml (A6) PPSGG (Fig. 2A). The hydrodynamic diameter was plotted, and polydispersity index (PdI) calculated using Malvern Zetasizer Software. Zeta potential measurements were performed with 1 mg/ml PPSGG in 10 mM NaCl at 25 °C by electrophoretic light scattering in disposable folded capillary cells (DTS1070, Malvern Instruments, United Kingdom) at different pH in the range from pH 4 to pH 10. The voltage (150 V), viscosity (0.8894 cP), and dielectric constant (78.6) was kept constant over five measurements per sample. Data were converted using the Smoluchowski model (Malvern Zetasizer Software)59.

Transmission electron microscopy (TEM)

The particle size and morphology of PPSGG, anti-HNK-1 IgM (C6680, Sigma-Aldrich, Switzerland), and their complexes were analyzed with TEM (Fig. 3). Aliquots of 5 μl (0.025 mg/ml PPSGG, 0.04 mg/ml anti-HNK-1 IgM) were adsorbed onto a glow-discharged carbon film-coated copper grid 400 mesh (G400-Cu, EMS, USA), washed with four droplets of pure water and subsequently stained with 2% uranlyacetate (22400, EMS, USA). Images were recorded using Tecnai G2 Spirit Electron Microscope (FEI, Netherlands) operating at 120 kV on an Morada CCD camera (EMSIS, Germany). The individual samples were blotted and stained immediately and the mixture of PPSGG and anti-HNK-1 IgM were incubated at room temperature for 30 min.

Analytical ultracentrifugation (AUC) with fluorescence detection

Sedimentation velocity experiments (Fig. 2B) were performed for samples comprising PPSGG and anti-HNK-1 IgM FITC (TB01, Invitrogen, Thermo Fisher Scientific, Switzerland) in PBS containing 0.05% Tween20 (93773, Sigma-Aldrich, Switzerland) to suppress adsorption of proteins to cell components. The buffer density (1.0052 g/ml) and viscosity (0.01022 Poise) were measured at 20 °C using an DMA 4500 M density meter (Anton Paar, Austria) and AMVn viscometer (Anton Paar, Austria). The 80 µL samples were housed in Spin Analytical AU-FDS 3 mm double sector charcoal-epon centrepieces, with sapphire windows. Centrifugation was performed at 30,000 rpm and 20 °C using a Beckman XL-I AUC with a Beckman An-60 Ti rotor. The rotor was incubated for three hours prior to acceleration to ensure homogeneous temperature. Sedimentation was monitored using an AU-FDS detector (Aviv Biomedical, USA) with continuous scanning for ten hours. Sedimentation velocity data were fitted to a diffusion-deconvoluted sedimentation coefficient distribution, c(s), using the software Sedfit60. A sedimentation coefficient range of 1-100 s was used with a resolution of 200 points. To obtain mass estimates for species of the size of IgM in the presence of a background of smaller fluorescent species, two approaches were used. In the first approach, the value of the frictional ratio (f/f0) was fixed to a value of 1.82, calculated from literature values for IgM (sedimentation coefficient 18.5, MW 950 kDa)34. In the second approach, a bimodal frictional ratio distribution was used allowing separate free-fitted values of f/f0 for species sedimenting with a coefficient above or below 10 s. As expected, the c(s) distribution was very similar for these two approaches.

Experimental animals

The wild type BALB/cJRj (Ref. C-BJ-56-M/F, Janvier Labs, France) mice (5–10 weeks old male and female) were bred at the University Hospital Basel and kept under specific pathogen-free conditions in a controlled environment (12-h light–dark cycle at 20 °C) with food and water ad libitum. All animal experiments were performed in accordance with the relevant guidelines and regulations and approved by Animal Research Authorities of the Canton of Basel-Stadt, Switzerland.

Perfusion of mice

BALB/c mice were IV injected with 300 µg PPSGG-sulforhodamine (4%) only or with 60 µg Pacific Blue™-labeled anti-HNK-1 IgM (3H5, DSHB, USA) followed by 300 µg PPSGG-sulforhodamine after 1 h. In case of PPSGG-sulforhodamine only, organs were isolated 10 min, 1 h, 6 h, 24 h, 48 h, 72 h post-injection, while for the combination of anti-HNK-1 IgM and PPSGG-sulforhodamine, organs were isolated 10 min, 1 h, and 6 h after PPSGG-sulforhodamine injection. Mice were sedated with a ketamine/xylazine (5 mg/kg of 2% Rompun, Bayer HealthCare, Germany and 100 mg/kg of 10% Ketasol, Graeub, Switzerland) and transcardially perfused with PBS, followed by 4% PFA (paraformaldehyde, P6148, Sigma-Aldrich, Switzerland). Organs were isolated and post-fixed overnight at 4 °C in 4% PFA. The next day, organs were cut in half and transferred into 50 mM of phosphate buffer (pH 7.4) for two-photon tomography or into 30% sucrose (84100, Sigma-Aldrich, Switzerland) for confocal microscopy imaging.

Confocal microscopy and F4/80 staining

Organs were embedded in cryomolds with OCT matrix (81-0771-00, Biosystems, Switzerland) and frozen on dry ice. Then 5 µm cryosections were cut. The macrophages were visualized by performing an immunofluorescent staining against the macrophage marker F4/80 (Fig. 4B). For this, sections were thawed, and antigen retrieval was performed by boiling the tissue sections in citrate buffer (10 mM, pH 6, 5949-29-1, Sigma-Aldrich, Switzerland). After cooling to room temperature, sections were incubated with blocking buffer containing 3% normal goat serum (10000C, Invitrogen, Thermo Fisher Scientific, Switzerland) and 0.5% Triton X (9002-93-1, Sigma-Aldrich, Switzerland) for 1 h at room temperature. Next, tissues were incubated with primary rat anti-mouse F4/80 antibody (1/1000 in blocking buffer, MCA497RT, Bio-Rad, Switzerland) overnight at 4 °C. For detection, the tissues were incubated with AF488 goat anti-rat IgG (1/1000 in blocking buffer, 112-545-003, Jackson laboratories, USA) for 2 h at room temperature and sections were mounted with Fluoromount-G mounting medium (BML-AP402, Invitrogen, Thermo Fisher Scientific, Switzerland) containing DAPI (nuclear staining).

Serial-two-photon tomography

Tissue distribution (Fig. 4A) was assessed by whole organ imaging with a serial two-photon tomography system with included microtome (Tissue Cyte, TissueVision, Cambridge, USA). Liver, kidney, spleen, and brain were transferred into a 50 mM phosphate buffer 1 h before embedding in 4.5% activated agarose (A6013; NaIO4, S1878; Sigma-Aldrich, Switzerland). The agarose block was stored in borohydride/borate solution (pH 9, 19 mg/ml Na2B4O7·10H20 221732; 3 mg/ml H3BO3, B6768; 0.2 mg/ml NaBH4, 452882; Sigma-Aldrich, Switzerland) at 4 °C overnight. The block was mounted in a 10 mM phosphate buffer bath and 30 µm sections were cut and recorded at two different depths of the section (15 µm). Two laser pulses at 800 nm (Mai Tai eHP) coupled to a 20× magnification lens (Zeiss Plan Apo, NA 1.0) were used as illumination source. The signal was recorded in the blue, green, and red channel, each split with a SP 500, a LP 500, and a LP 560 dichroic mirror. Approximately 300 sections were cut, and the tiles were automatically stitched (StichIt, MATLAB, The MathWorks, USA).

Cell culture

THP-1 cells (ATCC, TIB-202) were maintained in RPMI 1640 medium (Roswell Park Memorial Institute medium, R8758, Sigma-Aldrich, Switzerland) supplemented with 10% FBS (fetal bovine serum, 10270-106, Gibco, Thermo Fisher Scientific, Switzerland), 1% antibiotic–antimycotic solution (15240062, Gibco, Thermo Fisher Scientific, Switzerland) 10 mM HEPES buffer (5-31F00-H, BioConcept, Switzerland), 1% sodium pyruvate (S8636, Sigma-Aldrich, Switzerland), and 0.05 mM mercaptoethanol (11528926, Gibco, Thermo Fisher Scientific, Switzerland). To induce differentiation of THP1 monocytes into macrophages, PMA (16561-29-8, Sigma-Aldrich, Switzerland) at a final concentration of 100 nM was added to seeded THP-1 cells 72 h prior to the uptake experiment. The cells were incubated with 1 μM poly-l-lysine-pHAb (1%) and 1 μM PPSGG-pHAb (1%) in cell culture medium for 1 h and washed with PBS (20012068, Gibco, Thermo Fisher Scientific, Switzerland). Before imaging the cells were treated with 5 µg/ml CellMask Deep Red Plasma membrane stain (C10046, Invitrogen, Thermo Fisher Scientific, Switzerland). Images were recorded (Fig. 5A) with the Leica SP8 confocal point scanning microscope using a HC PL Apo CS 20× (NA 0.75) and a HC PL Apo CS 40× (NA 1.1) objective (Leica Microsystems, Germany). Differentiated cells were also incubated with 0.1 μM PPSGG-pHAb and 1 μM poly-l-lysine-pHAb, pre-mixed with increasing concentrations of dextran sulfate (42867, Sigma-Aldrich, Switzerland) or AcLDL (L35354, Invitrogen, Thermo Fisher Scientific, Switzerland) and incubated for 9 h. The cells were washed twice with 100 µl PBS before measuring fluorescence with the Tecan Spark (Tecan, Germany) microplate plate reader at Ex535/Em595 nm.

InSphero liver microtissue toxicity

3D InSight™ Human Liver Microtissues (hLiMT280, IpHH_11, IPHN_10, InSphero AG, Switzerland) were maintained at 37 °C in a humidified 5% CO2 cell-culture incubator in BSA-free 3D InSight™ Human Liver Maintenance Medium TOX (CS-07-001-02, m307/17, InSphero AG, Switzerland). Chlorpromazine hydrochloride (C8138, Sigma-Aldrich, Switzerland) and PPSGG were dissolved and diluted in Human Liver Maintenance Medium (hLiMT/M)—TOX. Triplicates of Human Liver Microtissues were incubated for 48 h at final concentrations of chlorpromazine hydrochloride and PPSGG of 1 mM (1/5 dilution series) and 6 mM epitope concentration (1/10 dilution series), respectively. Human Liver Microtissues were transferred to a white, clear-bottom 96-well Cell Culture Microplate (655088, Greiner Bio One, Switzerland). The viability of Human Liver Microtissues was determined (Fig. 6) at the end of the experiment with the CellTiter-Glo® 3D Cell Viability Assay (G9681, Promega, Switzerland. In brief, Human Liver Microtissues were incubated with equilibrated Celltiter Glo reagent (G9681, Promega, Switzerland) for 30 min and the plate shaken vigorously (5 min at 1350 rpm) for cell lysis, then the luminescence was measured with the Synergy HT (integration 1 s, bottom measurement, gain 135, Biotek Instruments, Switzerland.