Author response: Galectin-9 regulates the threshold of B cell activation and autoimmunity

Article Figures and data Abstract Introduction Results Discussion Materials and methods Appendix 1 Data availability References Decision letter Author response Article and author information Metrics Abstract Despite the mechanisms of central and peripheral tolerance, the mature B cell compartment contains cells reactive for self-antigen. How these cells are poised not to respond and the mechanisms that restrain B cell responses to low-affinity endogenous antigens are not fully understood. Here, we demonstrate a critical role for the glycan-binding protein galectin-9 in setting the threshold of B cell activation and that loss of this regulatory network is sufficient to drive spontaneous autoimmunity. We further demonstrate a critical role for galectin-9 in restraining not only conventional B-2 B cells, but also innate-like B-1a cells. We show that galectin-9-deficient mice have an expanded population of B-1a cells and increased titers of B-1a-derived autoantibodies. Mechanistically, we demonstrate that galectin-9 regulates BCR and distinct TLR responses in B-1a cells, but not B-1b cells, by regulating the interaction between BCR and TLRs with the regulatory molecules CD5 and CD180, respectively. In the absence of galectin-9, B-1a cells are more readily activated and secrete increased titers of autoantibodies that facilitate autoantigen delivery to the spleen, driving autoimmune responses. Introduction B cells contribute to host immunity through their potent capacity to activate T cell responses, modulate inflammatory profiles, produce high-affinity antibodies, and provide lasting immunological memory (Hoffman et al., 2016). Many of these functions are initiated by binding antigen to the B cell receptor (BCR), triggering downstream signaling and ultimately shaping cell fate (Hasler and Zouali, 2001). BCR signaling, however, must be tightly regulated to prevent aberrant B cell activation to low-affinity self-antigens and development of autoimmunity. Indeed, since the specificity of the BCR is stochastically generated, B cells with specificity for self-antigen arise (Kurosaki et al., 2010). Receptor editing, clonal deletion, and anergy limit the generation and activation of autoreactive B cells; yet despite these mechanisms of central and peripheral tolerance, the mature B cell compartment contains cells reactive for self-antigen (Liu et al., 2010). How these cells are restrained from responding to low-affinity endogenous antigen, and the mechanisms that regulate antigen affinity discrimination are not fully understood. BCR signal transduction is regulated by the nanoscale distribution of the BCR with respect to co-stimulatory molecules that antagonize or enhance signal transduction (Treanor, 2012). Loss or engagement of co-stimulatory molecules such as CD45, CD19, and CD22 dramatically alter the intensity of BCR signaling and ultimately shape B cell activation (Hasler and Zouali, 2001). These regulatory networks play a key role in tuning B cell responses in secondary lymphoid organs and help to restrain inappropriate activation to innocuous antigens (Goodnow et al., 1990). In fact, co-stimulatory molecules such as CD19 and CD45 have been shown to alter the threshold of B cell activation, and perturbations in their function can lead to autoimmunity (Goodnow et al., 1989). However, the mechanisms that establish the molecular interaction between BCR and positive and negative co-stimulatory molecules are not fully clear. Galectin-9 (Gal9) is a soluble, bivalent glycan-binding protein that binds directly to IgM-BCR and CD45 (Cao et al., 2018; Giovannone et al., 2018). Binding of Gal9 alters the organization of the BCR, enhancing interactions with inhibitory co-receptors CD45 and CD22, thereby reducing BCR signal transduction (Cao et al., 2018). These findings raise the question of whether Gal9 acts as a rheostat for antigen affinity discrimination, ultimately restraining B cells from responding to low-affinity antigens and therefore limiting autoimmune responses. Moreover, many B cell subsets have unique regulatory networks and play distinct roles in host immunity. For example, B-1a cells, which reside in the peritoneal and pleural cavities, are poised to rapidly respond to foreign antigen by producing natural IgM; yet their BCR polyspecificity and cross-reactivity for self has also implicated them in the pathogenesis of autoimmune diseases (Wollenberg, 2011). Whether Gal9 exerts subset-specific roles on B cell function and the importance of these on the pathogenesis of autoimmunity is unknown. Here, we demonstrate that loss of Gal9 in mice results in a breakdown of peripheral tolerance and leads to spontaneous autoimmunity with age, marked by splenomegaly, spontaneous germinal center formation, autoantibody production, and nephritis. Mechanistically, we demonstrate that Gal9 regulates the threshold of follicular B cell activation, and loss of this regulatory network permits activation to low-affinity and low-density antigens. Additionally, we observe an expansion in the B-1a compartment in the absence of Gal9 and demonstrate that these B-1a cells, but not B-1b cells, are more sensitive to BCR and distinct toll-like receptor (TLR) stimuli. We identify that Gal9 directly regulates BCR and distinct TLRs by binding to IgM-BCR and CD5 on the surface of B-1a cells as well as TLR4 and the regulatory molecule CD180, dampening signal transduction by altering their nano-scale co-distribution. We demonstrate that enhanced activation of B-1a cells in the absence of Gal9 is detrimental and exacerbates autoimmunity by facilitating transfer of autoantigens to secondary lymphoid organs where they drive autoimmune responses. Results Gal9 regulates the threshold of B cell activation BCR signal strength upon antigen binding is defined by a host of co-receptors that fine-tune and co-ordinate signal transduction. Increased signal strength leads to productive B cell activation when antigen affinity or concentrations are limiting (Zikherman and Lowell, 2017). Therefore, BCR co-receptors modulate BCR signaling and define the threshold of antigen required for B cell activation (Tedder et al., 1997). We have shown that Gal9 regulates BCR signaling by binding directly to IgM-BCR, altering its membrane organization with respect to the inhibitory co-receptors CD45 and CD22, thereby dampening BCR signal transduction (Cao et al., 2018). To investigate whether Gal9 impacts the threshold of B cell activation, we first asked if Gal9 regulates the intensity of signal transduction downstream of BCR ligation. To address this, we stimulated primary naïve murine wild-type (WT) and Lgals9−/− (Gal9KO) B cells with defined concentrations of anti-IgM F(ab’)2 as surrogate antigen and measured total tyrosine phosphorylation at 5 min by flow cytometry. Modulating BCR signal strength through limiting the concentration of stimulatory antigen leads to a decrease in the magnitude of the signaling response at 5 min, as observed by the decrease in geometric mean fluorescence intensity (gMFI) of total-phosphotyrosine (Figure 1A,B). In the absence of Gal9, however, BCR signaling is enhanced at limiting antigen concentrations compared to WT B cells (Figure 1A,B). Figure 1 with 2 supplements see all Download asset Open asset Gal9KO B cells respond to limiting concentrations of antigen. (A) Representative histograms of total tyrosine phosphorylation (p-Tyr) in WT (black) and Gal9KO (red) B cells at 5 min post-stimulation with anti-IgM F(ab’)2, as indicated. FMO (gray shaded). (B) Summary geometric mean fluorescence intensity (gMFI) of p-Tyr shown in (A). (C) IgM internalization over 20 min for WT (black) and Gal9KO (red) B cells following stimulation with anti-IgM F(ab’)2 as indicated. (D) Rate of IgM internalization (k) of data shown in (C). (E) Proportion of total IgM internalized at 20 min for data shown in (C). (F) Representative histograms of CD86 expression on WT (black) and Gal9KO (red) B cells stimulated with increasing concentrations of anti-IgM F(ab’)2 (left). Summary statistic, proportion of CD86 expressing B cells (right) as a function of F(ab’)2 concentration. (G) EC50 of F(ab’)2 titration shown in (F). Data show mean ± SEM and are representative of nine biological replicates over three independent experiments. Statistical significance was assessed by Mann–Whitney **p≤0.01, ****p<0.0001. Following antigen binding and initiation of the BCR signaling cascade, the BCR-antigen complex becomes internalized and brought to major histocompatibility complex class II (MHCII) containing lysosomes, which facilitate antigen processing and subsequent presentation to T cells, a crucial step to achieve full B cell activation (Avalos and Ploegh, 2014). To assess whether altered BCR signaling in Gal9-deficient B cells impacts BCR internalization, we stimulated WT and Gal9KO primary murine B cells with defined concentrations of anti-IgM F(ab’)2 and measured surface expression of IgM-BCR over 20 min. Following stimulation with high concentrations of anti-IgM F(ab’)2, we see a sharp decrease in surface IgM that plateaus by 20 min, whereas stimulation with decreasing concentrations leads to less IgM internalized (Figure 1C). Notably, in the context of reduced antigen concentrations, IgM internalization is increased in Gal9KO compared to WT B cells (Figure 1C). To quantify the rate of IgM internalization, we plotted the rate of change (k) of the fitted decay curve, which revealed Gal9KO B cells internalize IgM at a faster rate than WT B cells (Figure 1D). Furthermore, this results in an increase in total BCR internalization at 20 min to limiting antigen concentrations in Gal9KO B cells (Figure 1E). Taken together, these data suggest that increased BCR signaling in Gal9-deficient B cells leads to an increase in internalization of IgM-BCR. To assess the relationship between BCR signal strength and BCR internalization, we used a Nur77-eGFP reporter system, in which GFP expression is driven by Nur77 and is rapidly upregulated upon BCR signaling (Zikherman et al., 2012). Following stimulation for 20 min with defined concentrations of anti-IgM F(ab’)2, WT and Gal9KO Nur77-eGFP B cells were washed of excess antigen and allowed to incubate for 16 hr to express GFP proportional to the extent of BCR signal transduction. Consistent with our previous findings, BCR signal transduction to low-dose stimuli, as measured by the gMFI of the GFP reporter, is enhanced in the absence of Gal9 (Figure 1—figure supplement 1A,B). Furthermore, the extent of BCR signaling (gMFI of GFP) correlates with the rate of BCR internalization over the same interval (k, over 20 min) (Figure 1—figure supplement 1C). Taken together, these data demonstrate that Gal9 regulates BCR internalization through modulating BCR signal strength. BCR signal transduction leads to a B cell activation program, marked by upregulation of activation markers and a change in transcriptional and metabolic profiles. We then asked if enhanced BCR signaling in response to limiting antigen concentrations in Gal9KO B cells affects the threshold of B cell activation as assessed by the upregulation of the co-receptor CD86. To address this, we stimulated primary naïve B cells from WT or Gal9 deficient mice with titrated concentrations of anti-IgM F(ab’)2 for 16 hr and assessed upregulation of CD86 by flow cytometry. Following 16 hr of stimulation we see increased B cell activation proportional to increasing concentration of anti-IgM; however, in the absence of Gal9 B cells respond to lower concentrations of antigen (Figure 1F), quantified by the decreased EC50 value (Figure 1G). Taken together, these data demonstrate that Gal9 restrains B cell activation when antigen concentrations are limiting. Gal9 is a pleiotropic protein with many described roles in a variety of immune cells. Because of its secreted nature, Gal9 can be produced by one cell and act on another. This enhanced layer of complexity prompted us to ask which cell types within the spleen can produce Gal9 and contribute to extrinsic sources of Gal9 in the tissue? To address this, we compared Gal9 expression on the surface of a variety of splenic immune cell populations by flow cytometry (Yu et al., 2016; Figure 1—figure supplement 2A). Additionally, to delineate cells which may be excreting Gal9, we quenched the surface Gal9 with an unlabeled monoclonal antibody, then permeabilized cells, and stained for intracellular Gal9 using a fluorescent conjugation of the same monoclonal antibody (Figure 1—figure supplement 2B). Importantly, following quenching with unlabeled antibody, unpermeabilized cells did not stain with fluorescent antibody, demonstrating effective saturation of the quenching antibody in this assay (data not shown). Interestingly, we see high levels of intracellular and extracellular Gal9 expression from many myeloid cell populations, with little detectable Gal9 from the T cell compartment. The contribution of subset-derived Gal9 is a factor of both the expression of Gal9 by a given subset and the overall abundance of that subset within the splenic environment. As such, we normalized the total Gal9 expression (intra- and extracellular) against the relative abundance of each cell subset in the spleen, providing a metric of Gal9 expression that considers the size of specific populations (Figure 1—figure supplement 2C). We see that when we factor in subset abundance, B cells appear to have the highest Gal9 expression relative to abundance within the spleen. We next asked which cell types within the spleen contribute to the Gal9 detected on the surface of B cells. To investigate this, we performed adoptive transfer experiments using Gal9KO B cells transferred into WT hosts, µMT hosts (which lack B cells), and clondronate liposome-treated WT hosts (which results in selective depletion of phagocytic cells, predominantly macrophages/monocytes). We then stained for Gal9 expression on transferred cells to assess the cell type-specific contribution to surface Gal9 expression on B cells. Following treatment with clondronate liposomes, we see approximately 50% reduction in splenic macrophage and monocyte populations compared to control liposome-treated mice (Figure 1—figure supplement 2D). Interestingly, when we compare the expression of Gal9 on transferred Gal9KO B cells with endogenous WT B cells in the same host, we see full reconstitution of Gal9 expression on follicular B cells when transferred into WT hosts (Figure 1—figure supplement 2E). Notably, however, Gal9 expression is only partially restored when gating on marginal zone B cells (Figure 1—figure supplement 2F), suggesting that B cells may have subset-specific requirements for intrinsic vs extrinsic sources of Gal9. Additionally, Gal9 expression was only partially restored when Gal9KO B cells were transferred into macrophage depleted or B cell-deficient hosts (Figure 1—figure supplement 2E,F), demonstrating that both B cell-derived and macrophage-derived Gal9 contribute to extrinsic sources of Gal9 within the splenic environment. Lastly, we asked if these perturbations in Gal9 expression affect the threshold of B cell activation. To address this, we stimulated cells with titrated concentrations of anti-IgM F(ab’)2 for 16 hr and assessed upregulation of CD86 by flow cytometry. With reconstitution of Gal9 expression by transfer into WT hosts, we see that Gal9KO B cells have similar threshold of activation compared to endogenous WT cells in the same host (Figure 1—figure supplement 2G,H). In contrast, we see a decrease in the threshold of activation of Gal9KO B cells transferred into macrophage-depleted or B cell-deficient hosts, which is proportional to Gal9 expression. Taken together, these data demonstrate that macrophage populations and B cells act as a source of Gal9 within the spleen and that these sources of extrinsic Gal9 are sufficient to alter the threshold of B cell activation. Gal9 regulates antigen affinity discrimination B cell activation is defined by factors such as the abundance of antigen as well as the affinity of that antigen for cognate BCR (Zikherman and Lowell, 2017). To assess the role of Gal9 in modulating B cell activation to antigens with reduced affinity, we used the transgenic MD4 system. In this model, B cells express an IgM-BCR specific for model antigen hen egg lysozyme, HEL (HyHEL10, Ka = 4.5 × 1010 M−1). MD4 B cells bind to lysozymes from other avian species with reduced affinity, such as Bobwhite Quail (QEL, Ka = 3 × 108 M−1) and Duck (DEL, Ka = 1.7 × 107 M−1) (Lavoie et al., 1992). To investigate a role for Gal9 in antigen affinity discrimination, we first settled WT or Gal9KO MD4 B cells on artificial planar lipid bilayers displaying HEL, QEL, or DEL antigens, allowing us to simulate antigen presentation in a similar fashion to how B cells interact with antigen in vivo (Harwood and Batista, 2008). At 90 s post-interaction, the point of maximal B cell spreading, cells were fixed and imaged using total internal reflection fluorescence (TIRF) microscopy (Figure 2A). Gal9KO B cells appear to spread more and accumulate more antigen in response to lower affinity antigens compared to WT B cells. To quantify this, we measured the contact area and total amount of antigen accumulated (Figure 2B,C). Indeed, Gal9KO B cells spread more and accumulate more antigen in response to low-affinity antigens presented on lipid bilayers compared to WT B cells. It is well established that the B cell spreading response, and thus the amount of antigen accumulated is regulated by BCR signaling (Fleire et al., 2006; Weber et al., 2008). Thus, to assess the impact of Gal9 on BCR signaling in the context of an antigen presenting cell, we labeled OP9 stromal cells and deposited lysozyme containing immune complexes onto their surface at three different densities, covering a fourfold range, corresponding to low, medium, and high density. OP9 stromal cells were incubated with WT or Gal9KO MD4 B cells for 5 min, and cell conjugates were then permeabilized, stained for total tyrosine phosphorylation, and analyzed by flow cytometry. In the absence of Gal9, BCR signaling is increased in response to low-affinity and low-density membrane-bound antigens compared to WT B cells (Figure 2D,E). Figure 2 with 1 supplement see all Download asset Open asset Gal9KO B cells respond more readily to low-affinity antigens. (A) Representative images of primary WT and Gal9KO B cells fixed on planar lipid bilayers containing fluorescently conjugated antigen, as indicated, after 90 s of spreading and imaged by TIRF microscopy. Images mapped to a blue-orange ice 8-bit color scale (ImageJ). Scale bar 2 μm. (B) Quantification of cell contact area of WT (open) and Gal9KO (filled) in response to planar lipid bilayers containing HEL (black), QEL (orange), and DEL (purple) antigens. (C) Quantification of the amount of accumulated antigen as in (B). (D) Representative histograms of total tyrosine phosphorylation (p-Tyr) in WT (open) and Gal9KO (filled) B cells stimulated for 5 min with indicated antigen deposited on OP9 stromal cells, as indicated. FMO (gray shaded). (E) Summary gMFI of data shown in (D). (F) Representative histograms of CD86 expression on WT (open) and Gal9KO (filled) B cells stimulated with increasing concentrations of lysozymes, as indicated (left). Summary statistic, proportion of CD86 expressing B cells (middle). EC50 of lysozyme titration (right). (G) Internalization rate (k) of IgM; (H) Proportion of total IgM internalized; and (I) gMFI of intracellular lysozyme expression for WT (open) and Gal9KO (filled) B cells following 20 min stimulation with lysozyme, as indicated. Data show mean ± SEM and are representative of nine biological replicates over three independent experiments. Statistical significance was assessed by Mann–Whitney **p≤0.01, ***p≤0.001, **** p<0.0001. We then asked if this altered signaling in response to low-affinity and low-density membrane-bound antigens in Gal9KO B cells impacts the threshold of downstream B cell activation. To investigate this, we stimulated WT or Gal9KO MD4 B cells with titrated concentrations of these antigens and assessed B cell activation by upregulation of CD86. In the absence of Gal9, upregulation of CD86 is increased in response to lower concentrations of antigen in all three affinities of lysozyme (Figure 2F). Furthermore, in the absence of Gal9, B cells internalize IgM at a faster rate in response to low-affinity antigens compared to WT B cells (Figure 2G, Figure 2—figure supplement 1A). This results in a greater amount of IgM and antigen internalized in Gal9KO B cells when stimulated with low-affinity antigens (Figure 2H,I, Figure 2—figure supplement 1B). Taken together, these data demonstrate that Gal9 regulates B cell responsiveness to low-affinity and low-density antigens. Gal9 regulates autoimmunity BCR signaling is tightly regulated to mitigate inappropriate activation to otherwise innocuous autoantigens. Many central and peripheral tolerance mechanisms are in place to eliminate B cells with high affinity for autoantigens from the B cell repertoire, resulting in a pool of B cells that should have little to no affinity toward autoantigens (Basten and Silveira, 2010). Given our observation of enhanced B cell activation to low-affinity antigens in Gal9-deficient B cells, we then asked, does loss of Gal9 result in spontaneous autoreactivity and the development of autoimmunity in mice? To investigate this, we examined aged (>8 months) WT and Gal9KO mice for evidence of spontaneous autoimmunity. We find that aged Gal9KO mice have enlarged spleens (Figure 3A), with detectable germinal center (GC) B cells in the absence of immunization, suggesting they are driven toward autoantigens (Figure 3B,C). Furthermore, we see enhanced antibody secreting cell (ASC; defined as FSChi, surface IgMneg, and CD138hi expressing cells) (Pracht et al., 2017) development in the spleen of Gal9KO mice compared with WT littermate controls (Figure 3D,E). Further characterization of the ASC compartment revealed a clear skewing toward B220-negative long-lived ASCs in the absence of Gal9 (Pracht et al., 2017; Figure 3F). Consistent with this, we observe increased frequency of B220- ASCs in the bone marrow the absence of Gal9 (Figure 3—figure supplement 1A,B). We next asked if these spontaneous germinal centers and increased ASCs are driving autoimmunity. To address this, we looked for circulating autoantibodies in the serum of aged Gal9KO We see increased autoreactivity to cells with serum from Gal9KO mice compared to WT circulating that are and circulating IgM that are reactive (Figure Furthermore, and membrane autoantibodies are increased in Gal9KO mice (Figure In autoantigen containing immune complexes are a source of as they in the and other organs and and Thus, we next assessed circulating immune complexes and function in aged WT and Gal9KO We observe an increase in circulating in the serum (Figure as well as increased (Figure of in aged Gal9KO mice compared to WT Taken together, these data that Gal9KO mice spontaneous autoimmunity with Figure 3 with 1 supplement see all Download asset Open asset Gal9KO mice spontaneous autoimmunity. (A) of WT (black) or Gal9KO (red) mice aged (B) Representative of B cells in the spleen of aged as indicated. (C) Summary proportion of B cells, as in (B). (D) Representative of antibody secreting cells in the spleen of aged (E) Summary proportion of as in (D). (F) Proportion of ASC expression. (G) Representative images of cells stained with from WT or Gal9KO (H) Summary mean fluorescence intensity of IgM and (right) of as in (I) specific membrane specific immune complexes by protein into the Data are representative of biological replicates over at three independent experiments. Statistical significance was assessed by Mann–Whitney **p≤0.01, ****p<0.0001. T follicular cells are for the development of autoimmune as they are required for the and of et al., 2018). population of regulatory cells expressing cells) have been and these cells play a critical role in regulating B cell responses in peripheral Loss of cells leads to activation of autoreactive B cells and autoimmunity (Wollenberg, 2011). Thus, we measured the frequency of and cells in the spleen of aged Gal9KO mice compared to WT We see that aged Gal9KO mice have an increase in cell frequency (Figure 3—figure supplement and the increase in B cells (Figure 3B,C). Additionally, we observe a similar increase in the frequency of cells in the spleen of Gal9KO however, the of regulatory cells to cells is similar in both WT and Gal9KO mice (Figure 3—figure supplement These data suggest that loss of Gal9 does not lead to spontaneous autoimmunity through generation of regulatory cells. BCR signal strength and T cell help are key of tolerance in peripheral To assess whether is a breakdown in peripheral tolerance in Gal9KO B cells, we Gal9KO mice onto the model In this system, B cells express a transgenic BCR specific for HEL, as well as HEL secreted by cells. binding antigen in this B cells a of peripheral tolerance anergy where they to BCR surface and have a decreased (Goodnow et al., 1989). Therefore, in this model of is little to no of toward the We observed a distinct population of cells in the spleen of Gal9KO mice that high levels of surface IgM (Figure 3—figure supplement and Furthermore, following IgM this population signaling compared to both WT cells and cells in Gal9KO mice (Figure 3—figure supplement Additionally, we titers of in the serum of Gal9KO mice compared to WT controls (Figure 3—figure supplement These data suggest is a breakdown in peripheral tolerance in the absence of Gal9, to a population of cells that tolerance Gal9 regulates B-1a cell B-1a cells are a population of B cells from that a positive providing B-1a cells with an autoreactive It is that this is for in the of and cells and B-1a cells, defined by their expression of the T cell regulatory protein with B-1b cells are at such as the of the and peritoneal cells rapidly respond to both BCR and TLR stimulation and can readily secrete in the absence of T cell B-1a cells are expanded in of their specific role defined and We therefore asked if B-1a cells were expanded in aged Gal9KO Indeed, we find an increase in the proportion of B-1a cells in the of Gal9KO mice compared to WT mice (Figure Additionally, we see an increase in cell for B-1a cells in the of Gal9KO mice (data not shown). Furthermore, B-1a cells from Gal9KO mice have expression of activation markers at (Figure Figure supplement suggesting that Gal9 restrains B-1a activation. Figure with 1 supplement see all Download asset Open asset Loss of Gal9 leads to enhanced activation and of B-1a cells at (A) Representative of IgM expressing cells in the of aged WT and Gal9KO as indicated. (B) Summary proportion of CD5 expressing B-1a cells in the shown in (C) Representative histograms of CD86 expression at on cells shown in (A). (D) Summary gMFI of CD86 expression shown in (E) Representative of ASCs in the of aged (F) Summary proportion of ASCs shown in (G) specific IgM in of aged (H) Representative of cells of IgM expressing cells in the spleen of aged (I) Proportion of cells shown in Representative of B-1a and B-1b cells of Summary proportion of CD5 expressing B-1a cells in the spleen of aged mice shown in Data are representative of biological replicates over three independent experiments. Statistical significance was assessed by Mann–Whitney ****p<0.0001. of cells readily development into Consistent with our previous we find an expanded population of ASCs in the of aged Gal9KO mice compared to WT littermate controls (Figure We then asked if these ASCs were to circulating

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.