The accumulation in the cortical plaque of different protein types was consistent with the involvement of one or more scaffolding protein(s) that would be recruited by a putative meningococcus-activated receptor. Two such scaffolding proteins, β-arrestin 1 and 2 (βarr1 and βarr2), which are principally involved in G protein coupled receptor (GPCR) regulation and signalling, appeared as plausible candidates. Previous studies had shown that βarrs were essential for the co-localization of stimulated GPCRs, the actin-binding protein filamin-A and active ERK in membrane ruffles (21). Moreover, βarrs were shown to participate in the activation of CDC42 (22), PI3K (23), and Src (24).

The hypothesis of a role for βarrs in Nm-induced signalling in host endothelial cells was validated. In monolayers of hCMEC/D3 cells, a human brain endothelial cell line, which stably maintains phenotypic features of BBB in culture (25, 26), capsulated (pathogenic) piliated meningococci induced the accumulation in cortical plaques of both exogenous βarr2-GFP and endogenous βarr2, together with the proteins that were commonly found under bacterial colonies (27). Inhibition of βarrs with specific siRNAs inhibited the formation of cortical plaques (visualized by ezrin staining), demonstrating that βarr recruitment to the putative unknown Nm receptor was actually an upstream signalling event. Interestingly, siRNA-mediated inhibition of GRK2, a kinase which specifically phosphorylates GPCRs thus providing docking sites for βarrs, prevented βarr translocation under bacterial colonies and cortical plaque formation. This finding represented a strong experimental argument for the involvement of a GPCR in meningococcus signalling. Multiple studies combining loss and gain of function assays demonstrated that the missing signalling receptor in hCMEC/D3 cells was the β₂-adrenergic receptor (β₂AR) and provided a comprehensive scenario of the signalling events elicited by Nm following its adhesion to human endothelial cells (27). After binding onto endothelial cells, Nm TFP interact with the host cell β₂AR, leading to its βarr-biased activation, independently of the activation of cognate heterotrimeric Gα_s protein and of its downstream adenylyl cyclase-cAMP pathway. The pilus components PilE and PilV were identified as the specific “bacterial ligands” involved in β₂AR activation. Once translocated to β₂ARs, βarrs drive the 2 major events involved in Nm tissue infection: docking and activation of Src, leading to the formation of actin-rich cellular protrusions, which stabilizes bacterial colonies under the blood flow; accumulation under bacterial colonies of proteins, such as VE-cadherin and p120-catenin that are normally located in intracellular junctions, which, once depleted, become permeable to bacteria. However, from the studies above it also emerged that the β₂AR was not competent for the initial adhesion of the pathogen, corroborated by the fact that β₂AR-depleted cells still support initial pilus-mediated Nm adhesion. These observations were confirming previous reports indicating that TFP-mediated adhesion and signalling are two independent events (14).

A differential, quantitative, large-scale analysis of gene expression was conducted in a cell line, which only become permissive for the attachment of piliated capsulated meningococci after phorbol ester 12-O-Tetradecanoylphorbol-13-acetate treatment. This strategy led to the identification of the immunoglobulin superfamily member CD147 (also known as basigin or emmprin) as a brain microvasculature early adhesion receptor for meningococcus (28). PilE and PilV pilins specifically mediate the interaction of the pathogen with the C-terminal part of the CD147 Ig domain, showing that the same bacterial ligands are involved in the interaction with the signalling and the adhesion receptor. These findings indicated that CD147 and β₂AR might cooperate to promote firm Nm adhesion and subsequent activation of signalling events, and suggested a potential functional connection between these receptors that might pre-exist pathogen infection ( Figure 1 ).

The hypothesis of a spatio-temporal coordination between adhesion to CD147 and β₂AR stimulation by Nm TFP was specifically addressed in a subsequent study (29). CD147 and β2AR were found in pre-existing complexes at the cell surface of endothelial cells independently of bacterial infection. This interaction was stabilized by the cytoskeletal protein actinin-4, which in addition promoted the assembly of highly ordered clusters of receptor complexes in response to meningococcus adhesion. In vivo, under blood flow, this multi-molecular assembly process likely provides sufficient binding strength for the initial interaction of TFP with CD147 and then the subsequent rapid activation of the β₂AR, which ultimately enhances and stabilizes bacterial adhesion.

Initial investigations showed that the βarr-biased activation of the β₂AR by PilE and PilV components of TFP was allosteric, since a pre-incubated orthosteric antagonist of the receptor did not block it, and likely involved the extracellular N-terminal region of the receptor (27). Homogeneous time resolved FRET experiments with purified PilE and PilV confirmed their specific direct interaction with this region of the receptor (30). Intriguingly, the amino-acid composition of β₂AR N-terminus is conserved in mammals and almost identical for some species to that of human β₂AR, contrasting with the strict human specificity of meningococcus. Moreover, in a fully reconstituted cellular model of meningococcus signalling, the mouse β₂AR was activated by the pathogen in cells of human origin, whereas in a symmetrical experiment the human β₂AR was not stimulated in mouse cells. These findings suggested that host factors, independent of the amino-acid composition, were involved in meningococcus interaction/activation of β₂AR. The N-terminus of the β₂AR also contains two asparagine-branched glycan chains, which were then investigated as potential alternative pilin binding sites. In the β₂AR sequence the asparagine residues from which the glycan chains arise are separated by 9 amino-acid positions. When the two consensus N-glycosylation sites of the β₂AR were introduced in the sequence of the angiotensin AT1R receptor, which is also expressed in endothelial cells but cannot be stimulated by meningococcus, the resulting chimera became activatable by the pathogen. Interestingly, not only the number but also the distance of the glycan chains in the N-terminus sequence was critical for receptor activation (30). A panel of lectins (i.e. proteins that exhibit high avidity for glycoprotein- and/or glycolipid-associated carbohydrates) were then pre-incubated with endothelial cells to block the specific glycan(s) involved in the interaction between the β₂AR and meningococcus. Nm signalling was exclusively inhibited by lectins with the capacity of binding to terminal sialic acid (or N-acetyl neuraminic acid). Supporting a role of sialic acid in the interaction of N-terminal glycans with meningococcus, biochemical studies on purified N-terminal β₂AR domain confirmed that both glycan chains contain sialic acid, whereas the inhibition of sialyl transferases (the enzymes involved in the addition of sialic acid to the glycan chain) prevented Nm signalling. These findings were particularly interesting in the context of meningococcus species selectivity. Two principal forms of sialic acids are found in mammals, N-Acetylneuraminic acid (Neu5Ac) and N-Glycolylneuraminic acid (Neu5Gc), which differ by a single oxygen atom (31). In mammals, Neu5Gc is predominant, whereas humans mainly express Neu5Ac (32). Neu5Gc is actually synthesised from Neu5Ac by the cytidine monophosphate-N-acetylneuraminic acid hydrolase (CMAH), which is absent in humans, due to a mutation in the CMAH gene (33). The hypothesis that the difference in sialic acids might contribute to the species selectivity was confirmed experimentally. Indeed, the deletion of CMAH in mouse endothelial cells restored meningococcal signalling via the β₂AR (30). The nature of the human N-glycan recognized by Nm was further investigated using recombinant CD147 receptors, confirming the role of a complex sialilated N-glycan devoid of fucose residues (34).

Although it had been established that host-pathogen interactions can involve host cell glycans (35) and that the signalling properties of some receptors can be modulated by glycosylation (36, 37), the activation of a GPCR via the interaction between a ligand and N-glycan chains had not been reported previously. On the other hand, in addition to their usual role of chemosensors, some GPCRs can be activated by mechanical cues that are transduced by the receptor into intracellular chemical signals (reviewed in (38)), raising the hypothesis that mechanical forces applied via meningococcal TFP to β₂ARs might similarly produce their activation ( Figure 1 ). Indeed, bacteria growing at the cell surface of endothelial cells are permanently submitted to forces exerted by blood flow. In addition, TFP retraction, powered by the PilT ATPase can generate traction forces independently of hemodynamic flow (13, 39). Wild type and mutant meningococci deficient in PilT activity [ΔpillT mutants (40)] were compared for their capacity to induce signalling in hCMEC/D3 cells. Under basal conditions ezrin recruitment (as a marker of the cortical plaque formation) was significantly impaired under ΔpillT mutants compared to wild type Nm. The application of centrifugal forces on bacteria enhanced ezrin accumulation under both wild type and ΔpillT colonies, although the effect was significantly larger for wild type bacteria. Maximal signalling was thus obtained with the additive effect of PilT-induced pilus retraction and exogenous forces applied to meningcocci (30).

To confirm that pulling forces transmitted via N-glycans are sufficient to activate β₂AR signalling agarose beads coated with lectins that specifically recognize Neu5Ac were used in a reconstituted cell system. After incubation on cell monolayers to allow lectin binding to sialic acid-terminated N-glycan chains of the β₂AR the beads were submitted to orbital rotation. Whereas under basal conditions, only background accumulation of ezrin (used as activation readout) was observed under the beads, rotation appliance significantly enhanced ezrin accumulation but only when exogenous β₂AR and βarrs were co-expressed in the cells (30).

In addition to their well-established role of chemosensors, which transduce catecholamine stimulation, β₂AR can also function as mechanosensors for the traction forces applied on their N-terminal glycan chains. The signalling output produced by mechanical stimulation appears biased toward βarr-dependent pathways, since the receptor cognate Gs protein is not activated. However, in the presence of particular ligands, GPGRs can switch their coupling to a non-cognate G protein, likely because these ligands stabilize a unique receptor conformation that is competent for this unusual coupling. For example, carvedilol, an inverse-agonist of the classically Gs-coupled β₁-adrenergic receptor (β₁AR), was reported to promote its coupling to the inhibitory G protein Gi, in addition to the recruitment of βarrs (41). Moreover, this study demonstrated the existence of interplay between Gi and βarrs in the activation of the ERK1/2 MAP kinase pathway. Cell pre-incubation with pertussis toxin (PTX), which prevents Gi from interacting with GPCRs, partially inhibited meningococcus-induced signalling ( Figure 2 ), suggesting that the mechanical traction forces on glycan chains similarly stabilize a unique β₂AR active conformation capable of synergistic Gi coupling and activation of βarr signalling. This or a close conformation is likely also obtained by applying traction forces via beads that are coated with lectins specific for the exposed sialic acid of receptor glycan chains. Finally, the effect of PTX is consistent with previous studies indicating that signalling events attributed to βarrs might actually require some G protein contribution (42).

Although there is no evidence at the moment that, in addition to the pathophysiological role in meningococcal infection, this mode of activation of the β₂AR is involved in physiological processes, this hypothesis merits further investigations. Indeed, many cell types express integral lectins as structural components of their plasma membrane. Among them, I-type lectins are glycan-binding proteins in which the binding domain is homologous to immunoglobulin superfamily proteins. The Siglec family is a subgroup of I-type lectins, which all recognizes terminal sialic acids with specificity for adjacent carbohydrates within the glycan chain. Most Siglecs have one or more tyrosine-based signalling motifs in their cytoplasmic tails, or associate with membrane adaptor proteins containing cytosolic tyrosine motifs. Following their tyrosine phosphorylation by Src family kinases, they recruit and activate SH2-domain-containing effectors (43). Siglecs are expressed by blood cells, microglia, osteoclasts, myelin forming cells, and placental trophoblasts. Macrophages, adhere to and roll over endothelial cells before stopping under blood flow and penetrate into tissues. They express at the cell surface Siglec-1 (35), which displays the same glycan specificity as Mal-I, one of the lectins that block meningococcal signalling. In this context Siglec-1 might bind to the β₂AR and induce signalling pathways contributing to macrophage diapedesis. Also, signalling promoted by cell-to-cell interactions are essential during development, during which sialic-acids are known to play a major role, as shown by early embryonic lethality in case of disruption of sialic acid synthesis (43).

More generally, GPCRs are almost ubiquitous and 80% of them display one or more N-glycan chains, which might similarly be involved in the interaction with surrounding cells through lectins. Although, it is difficult at the moment to evaluate how many GPCRs could actually be activated via a traction applied on their N-glycans, β₂AR might not be a unique example, since the angiotensin AT1R engineered to express a second glycan chain in its N-terminus can be activated by meningococcus.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

SM wrote the first version of the manuscript, which was reviewed and corrected by the co-authors. All authors contributed to the article and approved the submitted version.

This article was supported by ANR-14-IFEC-0006-01, ANR-15-CE15-0002 and ANR-19-CE14-0045 grants. SM and MC are also supported by the INSERM, the CNRS, and the Université de Paris.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.