1. Immunology and Inflammation
Download icon

B cells suppress medullary granulopoiesis by an extracellular glycosylation-dependent mechanism

  1. Eric E Irons
  2. Melissa M Lee-Sundlov
  3. Yuqi Zhu
  4. Sriram Neelamegham
  5. Karin M Hoffmeister
  6. Joseph TY Lau  Is a corresponding author
  1. Roswell Park Comprehensive Cancer Center, United States
  2. Blood Research Institute Versiti, United States
  3. University at Buffalo, United States
Research Article
  • Cited 0
  • Views 736
  • Annotations
Cite this article as: eLife 2019;8:e47328 doi: 10.7554/eLife.47328

Abstract

The immune response relies on the integration of cell-intrinsic processes with cell-extrinsic cues. During infection, B cells vacate the marrow during emergency granulopoiesis but return upon restoration of homeostasis. Here we report a novel glycosylation-mediated crosstalk between marrow B cells and hematopoietic progenitors. Human B cells secrete active ST6GAL1 sialyltransferase that remodels progenitor cell surface glycans to suppress granulopoiesis. In mouse models, ST6GAL1 from B cells alters the sialylation profile of bone marrow populations, and mature IgD+ B cells were enriched in sialylated bone marrow niches. In clinical multiple myeloma, ST6GAL1 abundance in the multiple myeloma cells negatively correlated with neutrophil abundance. These observations highlight not only the ability of medullary B cells to influence blood cell production, but also the disruption to normal granulopoiesis by excessive ST6GAL1 in malignancy.

https://doi.org/10.7554/eLife.47328.001

Introduction

Hematopoiesis generates the blood cells necessary for gas exchange, hemostasis, and immune defense. Cell-intrinsic developmental programs orchestrate these lineage decisions, but they are guided by systemic signals to convey the dynamically changing needs for specific cell types (Lee et al., 2017). Dysregulated communication of such extrinsic cues results in imbalanced blood cell production and can trigger pathologic processes including anemia, thrombocytopenia, inflammation, and autoimmunity (Chovatiya and Medzhitov, 2014; Calvi and Link, 2015). In malignancy, the disruption of normal differentiation within hematopoietic stem and progenitor cells by tumors can lead to insufficiencies in one or more blood cell lineages, a common complication (Gabrilovich, 2017).

The sialyltransferase ST6GAL1 is a glycan-modifying enzyme mediating the attachment of α2,6-sialic acids. Canonically, it resides within the intracellular ER-Golgi secretory apparatus, but there is also an extracellular blood-borne form (Kim et al., 1971; Bernacki and Kim, 1977; Ip, 1980). In addition to ST6GAL1, a number of other terminal glycosyltransferases are also present in systemic circulation (Lee-Sundlov et al., 2017). Fluctuating levels of blood-borne ST6GAL1 are associated with a wide array of conditions, especially metastatic cancers, where high levels of blood enzyme have been associated with poor patient outcomes (Ip and Dao, 1978; Evans et al., 1980; Weiser et al., 1981; Berge et al., 1982; Dao et al., 1986; Cohen et al., 1989a; Magalhães et al., 2017; Rodrigues et al., 2018). Early reports also associated elevated blood ST6GAL1 with systemic inflammation (Kaplan et al., 1983; Jamieson et al., 1993), atherosclerosis (Gracheva et al., 1999), Alzheimer’s disease (Maguire et al., 1994), and alcohol-induced liver disease (Malagolini et al., 1989; Garige et al., 2006; Gong et al., 2007; Gong et al., 2008). The physiologic contributions of extracellular ST6GAL1 in these diseases, however, remain poorly understood. We have hypothesized that secreted ST6GAL1 can access distant sites to modify circulating plasma components and surfaces of target cells that do not express ST6GAL1 (Nasirikenari et al., 2014; Manhardt et al., 2017). Previously, we observed that blood-borne ST6GAL1 can profoundly modify leukocyte differentiation by attenuating G-CSF dependent granulocyte production (Dougher et al., 2017) while promoting BAFF-dependent survival in B cells (Irons and Lau, 2018). In mouse models, circulatory ST6GAL1 insufficiency results in an exuberant granulocytic inflammatory response (Nasirikenari et al., 2010; Nasirikenari et al., 2006) that can be therapeutically ameliorated by intravenous infusion of recombinant ST6GAL1 (Nasirikenari et al., 2019). Extracellular, liver-derived ST6GAL1 is also a major determinant of serum immunoglobulin G sialylation, which activates anti-inflammatory pathways within innate immune cells through Fc receptors (Jones et al., 2016; Jones et al., 2012; Pagan et al., 2018). In the periphery, activated platelets can supply the necessary sugar donor-substrate to support such extrinsic glycosylation reactions (Lee-Sundlov et al., 2017; Manhardt et al., 2017; Wandall et al., 2012; Lee et al., 2014).

The principal source of extracellular ST6GAL1 in circulation is believed to be the liver (Jamieson et al., 1993; Appenheimer et al., 2003; Lammers and Jamieson, 1986), where ST6GAL1 expression is activated by glucocorticoids and IL-6 (Jamieson et al., 1993; Jamieson et al., 1987; Wang et al., 1990; Dalziel et al., 1999). However, B cells also robustly express ST6GAL1, which synthesizes the ligands for sialic acid-binding receptors CD22 and Siglec-G (Wuensch et al., 2000; Müller and Nitschke, 2014). Here, we report that hematopoietic lineage cells, particularly B cells, contribute to the extracellular pool of functional ST6GAL1 and the sialylation of non-self cells both in vitro and in vivo. B cells secrete functionally active ST6GAL1 that sialylates hematopoietic progenitors in co-culture to suppress granulopoietic differentiation. In mouse models, we observed a positive correlation between IgD+ B cells and richly α2,6-sialylated niches of the bone marrow. In bone marrow specimens of treatment-naïve human multiple myeloma, there was a striking negative association between marrow plasma cell ST6GAL1 expression and the prevalence of bone marrow neutrophils. Our study is the first to demonstrate that the liver is not the sole source of ST6GAL1 responsible for extracellular sialylation, and underscores a novel potential relationship between B lymphocytes and hematopoietic progenitors influencing neutrophil production in the marrow.

Results

Human B lymphoblastoid cells secrete enzymatically active ST6GAL1

B cell expression of ST6GAL1 is critical for B cell development and function secondary to engagement of the lineage-specific lectin CD22 with α2,6-sialic acid (Hennet et al., 1998). Although ST6GAL1 is expressed in multiple tissue and cell types, it is thought that secreted, extracellular ST6GAL1 is exclusively derived from the liver, as a hepatocyte conditional knockout of St6gal1 results in vastly reduced serum α2,6-sialyltransferase activity (Appenheimer et al., 2003). In addition to hepatocytes, mature B cells strongly express ST6GAL1 (Wuensch et al., 2000), and numerous other cell types also express ST6GAL1 to varying degrees (Dalziel et al., 2001). The ability of non-hepatic cells to secrete functional ST6GAL1 to drive extrinsic sialylation has not been formerly studied.

We have recently analyzed the expression of ST6GAL1 within bone marrow and splenic B cell populations in mice and found that maximal ST6GAL1 expression occurred in early transitional and mature stages of development (Irons and Lau, 2018). However, it is unclear if the ST6GAL1 expressed in B cells is also actively released into the environment. In order to assess if human B cells are capable of secreting ST6GAL1, we analyzed four B lymphoblastoid cell lines derived from multiple stages of differentiation. ST6GAL1 mRNA expression was detectable in all cell lines except myeloma line RPMI 8226, with highest expression observed in the Burkitt lymphoma line Louckes (Figure 1a). Since Louckes is a germinal center B cell derivative, this observation is consistent with our previous observations that BCR activation induces ST6GAL1 expression (Wang et al., 1993). Expression of the β-site amyloid precursor protein-cleaving enzyme 1 (BACE1), thought to be required to liberate ST6GAL1 from its N-terminal membrane anchor prior to secretion (Kitazume et al., 2001; Deng et al., 2017), was detected within all lines except the myeloma cell MM1.S (Figure 1a). The cellular content of ST6GAL1 protein, assessed by western blot of total cell lysates, essentially followed ST6GAL1 mRNA levels. A possible exception was MM1.S cells, which expressed more ST6GAL1 protein than expected from ST6GAL1 transcript levels. All cells examined expressed, as expected, the cellular full-length ST6GAL1 form of 50 kDa. (Figure 1b). To assess the ability of the B lymphoblastoid cell lines to secrete functional ST6GAL1, the cells were seeded in serum-free medium for 3 days, and ST6GAL1 released into the medium was analyzed by western blot and assayed for sialyltransferase activity. All cell lines, except RPMI-8226, released measurable ST6GAL1 protein in a time-dependent manner into the medium (Figure 1c). B cells (NALM-6, Louckes) expressing BACE1 secreted the expected 42 kDa soluble form of ST6GAL1, consistent with the proteolytic liberation of the soluble catalytic active domain from the full-length protein by BACE1. Cells also released a 50 kDa form, consistent in size with the full-length ST6GAL1. MM1.S cells, which do not express BACE1, released predominately the 50 kDa form. To the best of our knowledge, the release of unprocessed, full-length ST6GAL1 has never been reported. The putative identity of the large 50 kDa form and its potential biologic significance are not explored further here. Enzymatic assay confirmed that all released ST6GAL1 was catalytically active, regardless of the larger size observed particularly in MM1.S (Figure 1d). Together, these results demonstrate that human B cell lines can release ST6GAL1 in vitro.

Human B lymphoblastoid cells secrete ST6GAL1.

Human lymphoblastoid cell lines derived from the pre-B (NALM-6), germinal center (Louckes) and plasma cell (RPMI 8226 and MM1.S) stages were profiled for ST6GAL1 expression and secretion. (A) RT-qPCR analysis of ST6GAL1 and beta-secretase BACE1 mRNA (n = 3 replicates) (B) Total ST6GAL1 protein analyzed by western blot (left) and quantified (right, n = 3). (C) Protein levels of ST6GAL1 in the serum-free conditioned medium of cell cultures 1–3 days after plating 10^6 cells/ml, analyzed by western blot (top) and quantified for 50 kD and 42 kD sizes (bottom). (D) Sialyltransferase activity in conditioned medium, relative to media only control, was determined by incorporation of [3H]NeuAc onto Gal(β4)GlcNAc-O-Bn acceptor substrate. [3H]-Labeled products were separated by SNA-agarose chromatography into [3H]NeuAc-α2,6-Gal(β4)GlcNAc-O-Bn α2,6- (SNA binding) and [3H]NeuAc-α2,3-Gal(β4)GlcNAc-O-Bn (SNA non-binding) fractions. The data shown are representative of multiple experiments with similar results.

https://doi.org/10.7554/eLife.47328.002

Theoretically, extracellular ST6GAL1 can enzymatically reconstruct sialic acid on cell surfaces in the presence of a sialic acid donor substrate (Lee et al., 2014). In order to determine if B cell secreted ST6GAL1 is capable of extrinsically remodeling cell surfaces, we applied concentrated conditioned medium from Louckes cells (concentrated ~20X) to sialidase-pretreated human hepatoma HepG2 cells (Figure 2a). The target HepG2 cells were rendered metabolically inert by formalin fixation to disable cell endogenous capacity to regenerate the cell surface sialyl glycans. Louckes conditioned media alone was insufficient to restore HepG2 cell surface SNA reactivity, but required the presence of the sugar donor substrate, CMP-Sia. The ability of Louckes conditioned media to restore HepG2 surface SNA reactivity in the presence of CMP-Sia could be reproduced in HepG2 cells in suspension, as analysed by quantitative flow cytometry (Figure 2b and c). These results indicate that the sialyltransferase secreted by B cells is enzymatically active and capable of extrinsic sialylation of cell surface glycans when supplemented with the sugar donor.

B cell conditioned medium extrinsically restores SNA reactivity of target cells.

(A) HepG2 human liver cells were grown on glass cover slides and fixed (10 min in 5% formalin) to disable endogenous metabolism. Cells were treated with C. perfringens sialidase C (Roche) for 1 hr at 37C to remove cell surface sialic acid, then incubated with concentrated (~20X) B cell conditioned medium (CM) from Louckes grown in serum-free medium, in the presence or absence of 0.05 mM CMP-sialic acid. Representative images of cell surface sialylation, as indicated by SNA lectin stain, are shown. HepG2 cells in suspension were subjected to the same treatments and analyzed by flow cytometry for SNA reactivity. SNA reactivity of nucleated cells is shown (B) by representative histogram and (C) as average mean fluorescence intensity of biological replicates.

https://doi.org/10.7554/eLife.47328.003

B cell ST6GAL1 sialylates hematopoietic progenitors to suppress granulopoiesis in co-culture

Mice deficient in circulatory ST6GAL1 have exaggerated neutrophilia inducible by various inflammatory stimuli (Nasirikenari et al., 2010; Nasirikenari et al., 2006; Appenheimer et al., 2003). Supplementation of recombinant ST6GAL1 is sufficient to blunt development of G-CSF and IL-5 dependent granulocytic colonies from whole bone marrow cells in vitro (Jones et al., 2010). In vivo, elevation of blood ST6GAL1 reduces bone marrow neutrophil abundance and neutrophilic inflammation secondary to sialylation of the multipotent granulocyte/monocyte progenitor (GMP). Ex vivo sialylation of GMPs potently inhibits G-CSF induced STAT3 signaling to reduce expression of myeloperoxidase, C/EBP-α, Gr-1, and ultimately, neutrophil production (Dougher et al., 2017). The data presented here suggest that B cells secrete enzymatically active ST6GAL1 capable of sialylating cell surfaces. Early B cells, mature B cells, and antibody-producing plasma cells represent a significant fraction of bone marrow cells. Therefore, we hypothesized that B cell-derived extrinsic ST6GAL1 participates in the modification of neighboring hematopoietic stem and progenitor cells (HSPC) to control granulopoiesis.

To test this, human B lymphoblastoid cell lines were co-cultured with c-kit+ St6gal1KO murine HSPCs and supplemented with SCF, IL-3, TPO, Flt-3, G-CSF, and CMP-sialic acid for 3 days. In order to resolve the two populations by flow cytometry, murine cells were labeled with the plasma membrane dye CellTrace Violet (Figure 3a). Co-cultures were seeded at 1:1 and 4:1 ratio of B cells to HSPCs. Co-culture of HSPC with B cells resulted in modifications to their cell surface sialic acid levels, as measured by reactivity towards the lectins Sambucus nigra (SNA) and Maackia amurensis II (MAL-II), which recognize α2,6- and α2,3-sialic acids, respectively (Figure 3b). The ST6GAL1 secreting B cell lines, NALM-6, Louckes, and MM1.S, increased cell-surface SNA reactivity of HSPC in a dose-dependent manner. RPMI-8226 cells, which do not release ST6GAL1, did not raise SNA reactivity on the co-cultured HSPCs. In contrast, cell surface MAL-II reactivity had no apparent correlation with ST6GAL1 status in the co-cultured B cells.

Figure 3 with 1 supplement see all
B cells modify HSPC SNA reactivity and Gr-1 expression in co-culture.

Human B lymphoblastoid cell lines were co-cultured with CellTrace Violet-labeled murine St6gal1KO c-kit+ bone marrow cells for 3 days with SCF, IL-3, G-CSF, TPO, and Flt-3 at indicated ratios of 1:1, 2:1, or 4:1. (A) Resolution of B cells and murine HSPCs by flow cytometry, with HSPCs staining positive for CellTrace Violet. (B) Levels of SNA and MAL-II reactivity on the HSPCs in monoculture or co-culture with indicated B cells. (C) Flow cytometric separation of CD11b+ and Gr1+ cells from murine HSPCs, after 3 days of monoculture (left) or co-culture (right) with MM1.S myeloma cells at 4:1 ratio. (D) Correlation between murine cell SNA reactivity and frequency of Gr-1+ or CD11b+ cells (expressed as % of total CellTrace Violet+ cells). (E) Frequency of CD11b-/Gr-1- undifferentiated murine cells after co-culture. (F) Frequency of SNA+ and Gr-1+ murine cells after co-culture with genetically modified MM1.S cell lines (Cr3 and Cr4) with targeted ST6GAL1 knockout by CRISPR/Cas9. Data are from a single experiment representative of three individual experiments, with n = 4 technical replicates per condition *p<0.05, **p<0.01, ***p<0.001 by student’s T-test.

https://doi.org/10.7554/eLife.47328.004

We assessed expression of CD11b and Gr-1, markers of granulocytic differentiation, on the murine HSPC (Figure 3c). Although CD11b is expressed before Gr-1 during granulocyte development in vivo, Gr-1 expression can precede CD11b during in vitro culture. We analyzed total expression of Gr-1 after co-culture and observed that ST6GAL1 expressing B cells were able to modify expression of Gr-1 on the murine HSPCs. Co-cultured murine HSPCs exhibited a significant negative correlation between Gr-1 expression and increasing cell surface SNA reactivity (r2 = 0.66, p<0.0001) (Figure 3d). A similar correlation was not observed between pan-myeloid marker CD11b and SNA reactivity (Figure 3d), suggesting a specific effect on granulocytes, in contrast to total myeloid cells. We also analyzed the frequency of CD11b-/Gr-1- murine HSPCs, indicative of less-differentiated cells that had yet to commit to the myeloid or granulocyte lineage (Figure 3e). Consistently, there was a significant, dose-dependent increase in this undifferentiated population in co-cultures with ST6GAL1 expressing B cell lines, suggesting that the presence of ST6GAL1 is correlated with maintenance of a less differentiated state. Since neither the Ly6C or Ly6G epitope of Gr-1 is predicted to contain any N-linked glycans, this relationship is unlikely a direct result of altered antibody binding due to epitope sialylation. Collectively, our data are consistent with a role of B cell-derived ST6GAL1 in the sialylation and differentiation of co-cultured St6gal1KO HSPC. However, patterns in HSPC SNA reactivity do not correlate perfectly with B cell ST6GAL1 expression in our data.

To establish definitively that the altered HSPC differentiation was due specifically to ST6GAL1 expression, two independent ST6GAL1KO genetically-modified MM1.S cell lines were generated using a CRISPR/Cas9 approach. These MM1.S modified lines, Cr3 and Cr4, express and secrete vastly reduced amounts of ST6GAL1 (see Figure 3—figure supplement 1). In co-cultures with murine St6gal1KO HSPC, ST6GAL1 deletion in MM1.S cells abrogated the dose-dependent increase in SNA reactivity on murine HSPCs (Figure 3f, left). Furthermore, loss of ST6GAL1 also attenuated the dose-dependent suppression of Gr-1 expression by co-cultured MM1.S cells (Figure 3f, right). Together, these observations implicate B cell secreted ST6GAL1 in the α2,6-sialylation and suppression of granulocyte differentiation in co-cultured HSPCs.

B cells secrete ST6GAL1 to modify non-self hematopoietic cells in vivo.

Our data demonstrate that human B lymphoblastoid cells can release functional ST6GAL1 to modify the glycosylation and granulopoietic differentiation of co-cultured hematopoietic progenitors in vitro. After neutrophils, B cells are the second most abundant lineage of hematopoietic cells in the marrow, and our early data indicated that they express significant ST6GAL1 (Wuensch et al., 2000). To assess if hematopoietic cells are a significant source of extracellular ST6GAL1 in vivo, wild type (C57BL/6) or St6gal1KO whole bone marrow was transplanted into irradiated recipients. The recipients were St6gal1KO/μMT and thus deficient in ST6GAL1 and B cells due to loss of the heavy chain of IgM (Ighm-/-). Thus, all ST6GAL1 present in extracellular compartments (e.g. blood) must originate from donor hematopoietic cells (Figure 4a). Blood α2,3-sialyltransferase activity, which did not depend on ST6GAL1, was unchanged and remained within the limits of resting WT (native C57BL/6) animals (Figure 4b). In contrast, blood accumulation of α2,6-sialyltransferase activity against Gal(β4)GlcNAc-o-Bn acceptor was dependent on re-establishment of ST6GAL1-competent hematopoietic cells, and increased and even surpassed the baseline for resting WT mice by week 8 (Figure 4c).

Figure 4 with 1 supplement see all
Hematopoietic Cells Supply Extracellular ST6GAL1 for Extrinsic Sialylation in vivo.

(A) CD45.1 + St6gal1 sufficient (WT) or deficient (KO) whole bone marrow was used to reconstitute irradiated CD45.2+/St6gal1KO/μMT mice. (B) α2,3 and (C) α2,6-sialyltransferase activity was quantified in serum of bone marrow chimeras at indicated time points. (D) Representative histograms of SNA-reactivity are shown for Lin-/c-kit+/Sca-1- (LK) and Lin-/c-kit+/Sca-1+ (LSK) progenitor compartments in the bone marrow, 8 weeks post-transplant. (E) Mean fluorescence intensity of SNA in CD45.2+ recipient bone marrow cell subsets was quantified by flow cytometry (n = 5). All cell types were significantly (p<0.01) different between WT and KO donor chimeras by student’s T-test. (F) CD45.2+ WT, μMT, St6gal1KO, or St6gal1KO/μMT bone marrow was used to reconstitute irradiated CD45.1+ St6gal1KO recipients. (G) α2,6-sialyltransferase activity was quantified in serum of bone marrow chimeras at indicated time points. Statistical significance is indicated in comparisons of WT and μMT cohorts for student’s T-test (**p<0.01, ****p<0.0001).

https://doi.org/10.7554/eLife.47328.006

In order to distinguish between donor and recipient hematopoietic cells, we utilized CD45.1+ mice as donors, and CD45.2+/St6gal1KO /μMT mice as recipients. Presence of CD45.2+ host derived cells was comparable between experimental groups (see Figure 4—figure supplement 1). Amongst the CD45.2+ residual host-derived CD11b+ myeloid, c-kit+ hematopoietic progenitors, and CD41+ megakaryocyte lineage cells, all of them unable to express their own ST6GAL1, increases in SNA reactivity were noted. Particularly, progenitor Lin-/c-kit+/Sca-1- (LK) and Lin-/c-kit+/Sca-1+ (LSK) populations, as well as CD41+ megakaryocyte lineage progenitors, were dramatically modified by cell non-autonomous extrinsic ST6GAL1, in comparison to CD11b+ myeloid cell subtypes and Ter119+ erythrocyte progenitors. Representative data depicting LK and LSK cell SNA reactivity are shown in Figure 4d. Quantitative data depicting all analyzed bone marrow cell types are shown in Figure 4e. These obeservations demonstrate that hematopoietic-derived ST6GAL1 alters the sialylation of non-self cells, and further suggest target-dependent bias of the extracellular ST6GAL1 towards CD11b-neg cells. These data demonstrate that hematopoietic cells alone, without contribution from the liver the canonical source of circulating ST6GAL1, were sufficient not only to maintain baseline blood ST6GAL1, but also to confer cell surface SNA-reactivity onto other cells that were unable to express their own ST6GAL1.

In order to assess definitively the contribution of B cells to extrinsic ST6GAL1 in vivo, the bone marrow transplantation approach was expanded to include comparisons between donors capable or incapable of reconstituting the B lineage (i.e WT or μMT). Comparisons were also extended to donors that were ST6GAL1-deficient and either B cell-intact or B cell-absent (i.e. St6gal1KO or St6gal1KO/μMT). All recipients were St6gal1KO and subjected to full body irradiation before transplantation (Figure 4f). The data show that ST6GAL1 accumulated in the blood of the chimeras reconstituted with WT marrow, but not in in chimeras reconstituted with μMT marrow. Strikingly, almost all (roughly 75%) of the increase in blood ST6GAL1 activity could be attributed to B cells. The parallel comparison, between St6gal1KO and St6gal1KO/μMT donors, did not show any detectable increase in serum ST6GAL1 activity (Figure 4g). Importantly, since both comparisons differ only in the presence or absence of B cells, and only differ from each other in the expression of St6gal1, the difference in blood ST6GAL1 activity can be attributed to ST6GAL1 expression in B cells. Thus, the data demonstrate that B cells directly express and release ST6GAL1 into the blood to restore extracellular levels in the absence of host ST6GAL1 expression.

We visualized SNA reactivity within the chimeric marrow of B cell deficient, St6gal1KO/μMT recipients reconstituted with WT marrow cells, in which both ST6GAL1 and mature, IgD+ B cells can only come from the donor (as in Figure 4a). The formalin-fixed and frozen whole femurs from these chimeras had strikingly patchy areas of SNA reactivity across the bone marrow (Figure 5a). In contrast, uniform SNAreactivity was observed in chimeras where the recipients were ST6GAL1-competent, for example μMT mice. This observation suggested that the spatially punctate distribution of marrow α2,6-sialylation results only when hematopoietic cells are the sole source of ST6GAL1. The distribution of α2,6-sialyl structures, as measured by SNA, was analyzed in the St6gal1KO/μMT recipients after engraftment with wild-type hematopoietic cells. Areas representing high, medium, and low reactivity towards SNA, designated as Regions of Interest (ROIs), were selected for analysis. ROIs were selected based on the relative homogeneity of SNA staining, rather than on region size or number of cells present. Four such ROIs are outlined in white in a single femur, with observed SNA staining intensity of Region 1 > Region 4 > Region 3 > Region 2 (Figure 5b, top). The whole femur was further stained for IgD to identify donor-derived mature recirculating B cells (Figure 5b, bottom). Quantitative analysis within four recipients demonstrated that the number of IgD+ B cells was modestly positively correlated with the SNA reactivity of ROIs (Figure 5c, R2 = 0.2448, p<0.05). Distinctly, SNA+ cells were not limited to the IgD+ B cells, but often encompassed groups of cells in their vicinity, consistent with their extrinsic modification. Collectively, this suggests that some of the variation in bone marrow niche α2,6-sialylation can be attributed to the presence of mature IgD+ B cells.

IgD+ B cells are Enriched in Regions of High α2,6-Sialylation within the Bone Marrow.

(A) Wild-type or μMT bone marrow was allowed to reconstitute irradiated μMT or St6gal1KO/ μMT mice for 8 weeks. Upon sacrifice, whole femurs were fixed, frozen, and sectioned for immunofluorescence staining. (B) Heterogeneous SNA reactivity was observed and indicated regions of interest (ROI) were created with differing SNA reactivity (top). Mature B cells (IgD+) were identified within ROIs (bottom panels). (C) Correlation of overall SNA reactivity with abundance of IgD+ B cells in ROIs from chimeras. Data are derived from four biological replicates.

https://doi.org/10.7554/eLife.47328.008

ST6GAL1 expression in human multiple myeloma negatively correlates with bone marrow neutrophil abundance

B cells exist within the bone marrow medullary spaces at several stages during development. Whereas early B cells occupy the niche until successful arrangement of a functional BCR, mature B cells freely recirculate between the marrow, blood, and lymphoid tissues. In contrast, plasma cells can remain for years within the bone marrow to maintain systemic titers of protective antibodies, and are thought to occupy a specialized niche in proximity to megakaryocytes, eosinophils, and soluble pro-survival factors (Wilmore and Allman, 2017). In order to understand if B cell-derived ST6GAL1 is able to perturb normal HSPC development into mature leukocytes in a clinically relevant setting, we performed histological analyses of human bone marrow samples from freshly diagnosed, treatment-naïve patients with a plasma cell dyscrasia, multiple myeloma. Samples from only a limited number of patients (n = 15) were available, and these were examined. Because of the clonal origin of multiple myeloma, the ST6GAL1 expression of the neoplastic plasma cell could be assessed. Moreover, the high occupancy of the bone marrow by the neoplasm, which in some cases approached 60%, allowed for an assessment of the consequences of pathologically elevated ST6GAL1 within the bone marrow microenvironment.

Paraffin-embedded bone marrow sections were stained for ST6GAL1 using a DAB reagent. Plasma cell-specific expression of ST6GAL1 was assessed according to intensity (1-5) and frequency (0–100%) by counting five groups of ten cells each in at least five fields of view per patient. The product of intensity and frequency is referred to as ‘ST6GAL1 score’. ST6GAL1 expression was highly heterogeneous between patients, and varied from nearly completely absent to intense expression in 100% of examined cells (Figure 6a). The expression of ST6GAL1 in tumor cells was not associated with altered patient survival (Figure 6—figure supplement 1a). However, whereas the plasma cell burden and abundance of segmented neutrophils varied completely independently (Figure 6—figure supplement 1b, r2 = 0.001, p=0.88), we observed a striking relationship between plasma cell ST6GAL1 expression and segmented neutrophils. When ST6GAL1 score was compared to the frequency of granulocyte lineage cells, as assessed by a trained clinical pathologist, we identified a strong negative correlation between ST6GAL1 score and presence of segmented neutrophils (r2 = 0.42, p=0.0083) (Figure 6b). Patients with low ST6GAL1 scores (<100) had higher frequencies of mature neutrophils (22.32 ± 4.11%), whereas those with high ST6GAL1 scores (>100) had markedly lower frequency (9.1 ± 2.6%) (Figure 6b inset). Qualitatively, patients with low ST6GAL1 expression in plasma cells had evidence of abundant granulocytes on H and E staining, whereas high ST6GAL1 expressing patients had far fewer visible granulocytes (Plasma cells indicated by arrowheads on DAB stains, PMNs indicated by arrows and arrowheads in H and E of Figure 6c).

Figure 6 with 1 supplement see all
ST6GAL1 Expression in Human Multiple Myeloma Cells Correlates Negatively with Bone Marrow Neutrophil Abundance.

(A) Quantification of ST6GAL1 expression in bone marrow histological specimens from treatment-naive multiple myeloma patients (n = 15). (B) Negative correlation between ST6GAL1 expression and frequency of bands and segmented PMNs (p<0.01). Stratification of patients into low and high ST6GAL1 expression was predictive of abundance of segmented neutrophils (**p<0.01, student’s T-test). (C) Representative data from one patient with high and one patient with low myeloma ST6GAL1 expression, showing H and E with neutrophils indicated (left, arrows or arrowheads), and ST6GAL1 stain with myeloma cells indicated (right, arrowheads).

https://doi.org/10.7554/eLife.47328.009

Discussion

Effective cross-talk between components of the medullary environment is central to the demand-driven production of different lineages of blood cells. During systemic inflammation, B cells vacate the bone marrow due to the release of TNF-α and downregulation of CXCL12, salvaging space in preparation for the emergency generation of granulocytes (Ueda et al., 2004). However, it remains unclear if the departure of marrow B cells merely creates physical space for granulopoiesis or also disinhibits intrinsic granulopoietic processes. Our data support the existence of a paracrine signaling relationship, mediated by B cells via a novel extracellular glycosylation pathway, to influence the generation of granulocytes. This molecular pathway involves the release of catalytically active ST6GAL1 sialyltransferase from B cells, and may contribute to the reciprocal inhibition of granulopoiesis during homeostasis. Our observations add to the body of literature documenting the emerging regulatory relationship between B cells and neutrophils. The newly-defined ‘B-helper’ subset of neutrophils (NBH) can prime marginal zone B cells for antibody production during inflammation by the release of BAFF (Scapini et al., 2003; Puga et al., 2012). Conversely, evidence suggests that B cells inhibit neutrophil functions in a number of ways, including blocking chemotaxis and initiating β2 integrin-mediated apoptosis (Kondratieva et al., 2010; Kim et al., 2018). B cell suppression of neutrophilic influx into the liver and spleen is necessary for the resolution of infections with a number of intracellular organisms, where neutrophils can cause damaging local and systemic inflammation (Bosio and Elkins, 2001; Smelt et al., 2000; Buendía et al., 2002). Here, our observations suggest that B cells influence the generation of new neutrophils by regulating glycosylation within the bone marrow, consistent with a separate body of literature documenting the development of excessive neutrophilic inflammation due to ST6GAL1 insufficiency (Dougher et al., 2017; Nasirikenari et al., 2010; Nasirikenari et al., 2006; Jones et al., 2012; Jones et al., 2010). These include demonstrations of a profound role for extracellular, distally produced ST6GAL1 in muting the transition of granulocyte-monocyte progenitors (GMP) to granulocyte progenitors (GP) (Dougher et al., 2017), and that intravenously infused recombinant ST6GAL1 can attenuate demand granulopoiesis in a mouse model of airway inflammation (Nasirikenari et al., 2019).

The existence of sialyltransferases within the extracellular milieau, particularly the blood, has been known for quite some time. Upregulation of serum ST6GAL1 during inflammation was attributed to the induction of hepatic expression, leading to the designation of ST6GAL1 as an acute phase reactant (Kaplan et al., 1983; Jamieson et al., 1993; Jamieson et al., 1987; Jamieson, 1988; van Dijk et al., 1986). However, it was also recognized that in addition to hepatocytes, B cells are sophisticated expressers of ST6GAL1, utilizing multiple tissue-specific transcripts during development (Wuensch et al., 2000; Wang et al., 1993; Lo and Lau, 1996). Given the widespread distribution of ST6GAL1-expressing mature B cells within secondary lymphoid tissues, blood, and bone marrow, we hypothesized that this population could be contributing to the extracellular pool of ST6GAL1, thus regulating the sialylation and development of other hematopoietic cells. The observations in this study demonstrate that human B cells can release functional ST6GAL1 and are capable of modifying hematopoietic stem and progenitor cells (HSPC) in co-culture conditions to suppress granulocytic differentiation. Importantly, this effect is attributable directly to the expression of ST6GAL1 in the B cells, as demonstrated by the CRSPR/Cas9 targeted gene knockouts. In vivo, hematopoietic cell-derived ST6GAL1 is a significant modifier of the sialylation of diverse bone marrow cells. Indeed, the limited endogenous expression of ST6GAL1 in hematopoietic stem and progenitor populations may be compensated for by secreted, extracellular enzyme, making the bone marrow microenvironment a distinct niche space for extrinsic sialylation (Nasirikenari et al., 2014). In contrast to untreated wild-type mice, we observed striking regional heterogeneity in bone marrow sialylation among St6gal1KO chimeras reconstituted with wild-type bone marrow, with higher sialylation generally at the epiphysis and metaphysis of long bones. Interestingly, these sites of high sialylation also contained high frequencies of IgD+/ST6GAL1+ donor derived mature B cells. In our experiments, hematopoietic cells were capable of independently reconstituting blood levels of ST6GAL1 in St6gal1KO mice to baseline WT levels over 8 weeks, and the vast majority of this could be attributed to ST6GAL1 expression within B cells. These findings clearly implicate cells of the B lineage as major extra-hepatic determinants of blood ST6GAL1, and argue in favor of a tissue-agnostic model of extrinsic sialylation wherein multiple cell types actively secrete enzyme into the extracellular space. Future studies may uncover whether B cells contribute to the elevation of blood ST6GAL1 during inflammation or are regulated by an entirely different set of stimuli.

In the immortalized human B lymphoblastoid lines we examined, aside from the expected 42 kDa soluble form, we also observed a larger 50 kDa form similar in size to the full-length, unclipped ST6GAL1 (Weinstein et al., 1987). The 50 kDa form, which appeared to be catalytically active, was the predominant ST6GAL1 form released by the BACE1-negative MM1.S cells. While the nature and mechanism of release of the 50kDa ST6GAL1 form remain to be investigated, its size is consistent with an intact transmembrane domain, raising the possibility of the release of ST6GAL1 associated vesicles. If true, this mechanism may facilitate shuttling of enzyme between specific cell types to mediate extrinsic sialylation (McLellan, 2009; Zech et al., 2012). Alternately, the 50 kDa form could represent a form of ST6GAL1 processed by signal peptide peptidases, as reported for other glycosyltransferases, solubilizing the enzyme without significantly reducing its molecular weight (Voss et al., 2014; Kuhn et al., 2015).

ST6GAL1 has been implicated in a variety of biological processes relevant to the development of disease, particularly in systemic inflammation (Jamieson et al., 1993) and metastatic cancers (Lu and Gu, 2015). The sialylation of IgG by ST6GAL1 is necessary for the anti-inflammatory effects of IVIG therapy in autoimmune disease, and variations in serum IgG sialylation have been widely associated with inflammatory diseases (Pagan et al., 2018; Nimmerjahn and Ravetch, 2008; Biermann et al., 2016). Human GWAS studies have also associated genetic variation in ST6GAL1 with IgA nephropathy and flucloxacillin-induced liver damage (Li et al., 2015; Daly et al., 2009). In epithelial carcinomas, ST6GAL1 expression confers increased resistance to chemotherapy, hypoxia, and nutrient deprivation by promoting a stem-like phenotype, bolstering signaling through pro-survival and pro-proliferative EGFR, HIF-1α, and NF-κB pathways (Schultz et al., 2016; Britain et al., 2017; Chakraborty et al., 2018; Holdbrooks et al., 2018; Britain et al., 2018; Jones et al., 2018). Numerous early studies in both rodent models and humans have also documented concurrent increases in serum protein sialylation and sialyltransferase activity during malignancy, including in multiple myeloma, implying that ST6GAL1 expressing tumor cells are capable of secreting enzyme into the extracellular pool (Bernacki and Kim, 1977; Cohen et al., 1989b; Cohen et al., 1993; Chelibonova-Lorer et al., 1986; Dairaku et al., 1983; Gessner et al., 1993; Poon et al., 2005). The functional consequences of fluctuations in blood ST6GAL1 are yet unexplored, especially in malignancy. This current work is the first to examine how cancer-derived ST6GAL1 can perturb the generation of granulocytes, the best documented biologic role associated with extrinsic ST6GAL1 sialylation (Nasirikenari et al., 2014; Nasirikenari et al., 2006; Nasirikenari et al., 2019; Jones et al., 2010). Multiple myeloma (MM) was examined because of the natural localization of tumor cells in the marrow, in proximity to nearby healthy HSPCs. ST6GAL1 expression in MM varied dramatically from patient to patient. We observed that the abundance of segmented granulocytes within the marrow was strikingly and negatively associated with the level of ST6GAL1 expression, but not with the overall abundance of multiple myeloma plasma cells. While localized gradients of cytokines, chemokines, and growth factors are already understood to create functional and developmental marrow niche spaces (Birbrair and Frenette, 2016), our data now underscore a novel role for extracellular glycan-modifying enzymes such as ST6GAL1 in the marrow hematopoietic environment.

Recent work in our group suggests that extrinsic ST6GAL1 may have a broad ability to coordinate the development and function of multiple immune cell types (Dougher et al., 2017; Irons and Lau, 2018; Nasirikenari et al., 2019). Given the well-documented role of immune cells in cancer, further investigation into the role of extrinsic glycosylation in cancer is merited. The strong negative association between human multiple myeloma ST6GAL1 expression and neutrophil prevalence indicates that tumor-derived ST6GAL1 may dysregulate the development of bystander immune cells in the tumor microenvironment, with a variety of potential implications. At the very least, the ability of cancer-derived ST6GAL1 to disrupt granulopoiesis predicts a diminished capacity for the patient to combat bacterial infections. Furthermore, the correlation between ST6GAL1 levels and worse patient outcomes in a number of other cancers may be in part due to the extrinsic modification of mature tumor-associated leukocytes. This is consistent with reports that myeloid cell surface α2,6-sialylation diminishes maturation, activation, antigen cross-presentation and anti-tumor immune responses in dendritic cells, for example (Silva et al., 2016; Crespo et al., 2013; Cabral et al., 2013). Collectively, our data hint at biologic effects of ST6GAL1 in cancer that extend beyond the cell-intrinsic modulation of oncogenic signaling pathways, instead being mediated by a novel, extrinsic axis of glycosylation, and galvanized by the growing importance of immune cells in malignancy.

Materials and methods

Animal models

Request a detailed protocol

The St6gal1KO strain has been backcrossed 15 generations onto a C57BL/6J background and maintained at Roswell Park’s Laboratory Animal Shared Resource (LASR) facility. The B cell deficient B6.129S2 – Ighmtm1Cgn/J mouse μMT (The Jackson Laboratory) was used as a donor and recipient in bone marrow transplantation. The reference CD45.1 expressing wild-type strain used was B6.SJL-Ptprca Pepcb/BoyJ, in order to distinguish donor cells from recipient mice, which express the CD45.2 allele of the Ptprc locus. For transplantations, mice received 6 Gy whole body gamma-radiation and were rescued with 4.0 × 106 whole bone marrow cells from a single donor or two donors equally. Mice were euthanized after 8–10 weeks for analysis. Unless otherwise indicated, mice between 7–10 weeks of age were used, and both sexes were equally represented. Roswell Park Institute of Animal Care and Use Committee approved maintenance of animals and all procedures used.

Antibodies

For immunoblots and immunohistochemistry, anti-ST6GAL1 (R and D Biosystems), anti-β-tubulin (Cell Signaling Technology), anti-PF4 (Peprotech 500-P05), anti-IgD (eBioscience 11–26 c), SNA-FITC (Vector labs) were used. For flow cytometry, SNA-FITC (Vector Labs), biotinylated MAL-II (Vector Labs), anti-Gr1-PE/Cy5 (RB6-8C5), anti-CD11b-BV711 (M1/70), anti-CD45.2-PE/Cy7 (104), anti-CD45.1-PerCP/Cy5.5 (A20), anti-Ly6G-APC (1A8), anti-Ter119-BV510 (TER-119), anti-CD41-BV421 (MWReg30), anti-c-kit-APC/Cy7 (2B8), and anti-Sca-1-PE (D7) (all Biolegend) were used.

Analysis of cell lines

Request a detailed protocol

Human B lymphoblastoid cell lines were grown in RPMI base medium supplemented with 10% heat-inactivated fetal bovine serum. All analyses were performed during logarithmic growth phase, and cell lines were kept in passage for no more than 6 weeks. All cell lines are from Roswell Park Comprehensive Cancer Center repository. Lines MM1.S and HepG2, which are central to the conclusions here, are certified in our laboratory to be mycoplasma-free.

For RNA analysis, cells were washed, pelleted, and resuspended in TRI Reagent (MRC Inc) and RNA extracted according to manufacturer’s instructions. 1.0 μg RNA was converted to cDNA (iSCRIPT kit, Bio-rad), and then amplified by qPCR (iTaq Universal SYBR Green, Bio-rad) with intron-spanning primers towards human ST6GAL1 and BACE1. Relative expression (2dCt) was calculated in reference to B2-microglobulin. Primer sequences are as follows: B2M: F 5’-GTGCTCGCGCTACTCTCTCT–3’, R 5’-TCAATGTCGGATGGATGAAACCC–3’; ST6GAL1: F 5’-CCTTGGGAGCTATGGGACATTC–3’, R 5’-TATCCACCTGGTCACACAGC–3’; BACE1: F 5’-TCTTCTCCCTGCAGCTTTGT–3’, R 5’-CAGCGAGTGGTCGATACCT–3’.

For total protein, cells were washed, pelleted, and resuspended in RIPA cell lysis buffer with protease inhibitors, and 5–10 μg of total protein resolved on 10% SDS gels, transferred onto activated PVDF membranes, and blocked in 5% fat-free milk for 1 hr. Blots were probed with primary antibody overnight at 4C, then washed and incubated with HRP-conjugated secondary for 1 hr. Membranes were developed using Pierce ECL WB Substrate (Thermo Scientific) and imaged using ChemiDoc Touch (Bio-rad). For analysis of secreted protein, cells were seeded at a density of 106 cells in 1 ml of serum-free RPMI in 12-well plates. Cell-free conditioned medium was collected after 24, 48, and 72 hr by pipetting and centrifugation at 1,000 rpm to separate cells from supernatant. In order to control for secreted protein per cell, an equal volume (10 μl, 1%) of conditioned media was resolved by 10% SDS-PAGE. In quantification of enzymatic activity, 1.5 μl (0.15%) of total volume was used. Densitometric quantification of adjusted band intensity was performed separately for 50kD and 42kD forms of ST6GAL1 using ImageJ software.

Sialyltransferase enzymatic activities were quantified by following transfer of 3[H]NeuNAc from CMP-3[H]NeuNAc onto the artificial acceptor Gal(β1,4)GlcNAc-o-Bn and separation of the 3[H]-trisaccharide products from unreacted 3[H]NeuNAc by Sep-Pak C18 reverse phase chromatography. The 3[H]NeuNAc- Gal(β1,4)GlcNAc-o-Bn products were further subjected to SNA-agarose chromatography to separate the α2,6-3[H]NeuNAc- (ST6Gal1 product and SNA binding) from the α2,3-3[H]NeuNAc- (SNA flow through) products. This procedure has been described, validated, and utilized previously (Lee-Sundlov et al., 2017; Nasirikenari et al., 2006; Nasirikenari et al., 2019; Jones et al., 2012).

Extrinsic sialylation of fixed hepatocytes 

Request a detailed protocol

HepG2 cells (ATCC) were seeded at 2 × 105 cells/ml onto sterile glass cover slips in 6-well dishes for 3 days. Wells were washed with PBS and fixed for 5 min in 5% formalin solution. Cover slips were carefully removed from wells, and subjected to 1 hr treatment with 20 μl/ml bacterial sialidase C (Roche) at 37°C, followed by incubation with ~20X concentrated Louckes conditioned medium at 10% total volume for 2 hr at 37°C, in the presence or absence of 100 μM CMP-sialic acid charged sugar donor (EMD Millipore). Cover slips were blocked in 5% BSA for 1 hr, stained with SNA-FITC lectin overnight, washed with DAPI, then mounted onto charged microscope slides in 10% glycerol. During all steps, cells were kept moist by incubation within a water-containing chamber. Fluorescence was visualized immediately using a Nikon Eclipse E600 microscope with EXFO X-cite 120 light source. Spot RT3 camera and Spot Software were used to capture images.

LK (Linneg cKitpos) cell co-culture

Request a detailed protocol

St6gal1KO mouse bone marrow mononuclear cells were obtained and enriched for c-Kit+ cells using MACS columns (Miltenyi Biotechnology). Resulting Lin-neg:cKit+ (LK) hematopoietic progenitors (HSPCs) were stained for 20 min at 37C with CellTrace Violet (Thermo Fisher), as per manufacturer’s instructions. Stained cells were quenched with media before quantification, and 10,000 cells cultured in 96-well round-bottom plates with either 10,000 or 40,000 human B lymphoblastoid cells at logarithmic growth phase, supplemented with 0.05 mM CMP-Sia (EMD Millipore). To induce differentiation and proliferation, cultures were supplemented with recombinant SCF (50 ng/ml; BioVision), G-CSF (20 ng/ml; Peprotech), IL-3 (5 ng/ml; BioVision), TPO (25 ng/ml; Peprotech), and FLT-3 (30 ng/ml; Peprotech) in a total volume of 200 μl RPMI medium supplemented with 10% FBS. After three days, cells were analyzed by flow cytometry for CellTrace Violet to discriminate between HSPCs and B cells, and murine cells further analyzed for cell surface glycans and expression of granulocyte markers. Flow cytometry data were acquired with BD LSR II flow cytometer and analyzed with FlowJo software.

Knockout of ST6GAL1 by CRISPR-Cas9

Request a detailed protocol

sgRNA was purchased from Genscript (Cr3: CATTCGCCTGATGAACTCTC and Cr4: CAGATGGGTCCCATACAATT). Neon system (Thermo Fisher) was used for electroporation. 1 μg Cas9 (New England Biolabs Inc) was incubated with 1 μg sgRNA for 20 ~ 25 min at room temperature to form the RNP (ribonucleoprotein) complex. 0.1 ~ 0.25 million MM1.S cells was used in each electroporation. Cells were washed twice with PBS and resuspended in buffer R (10 μl), supplied in the Neon transfection kit. Four electroporation conditions were tested (E1: 1600V, 10 ms, three pulses; E2: 1700V, 10 ms, three pulses; E3: 1600V, 20 ms, three pulses and E4: 1600V. 20 ms, two pulses). The editing efficiency was checked 2 ~ 4 days after electroporation by measuring SNA in flow cytometry. Editing was observed under all electroporation conditions, with significant SNA reduction under E2 conditions. SNA-low transfected cells were sorted by fluorescence activated cell sorting (FACS), then analyzed for intrinsic expression of ST6GAL1 at the RNA and protein level to confirm reduction in expression.

Analysis of bone marrow chimeras

Request a detailed protocol

Femurs of indicated bone marrow transplantation chimeras were flushed extensively to obtain cells. Peripheral blood was collected from the retro-orbital venous plexus in citrate-containing anticoagulant. All samples were subjected to ammonium-chloride-potassium (ACK) lysis buffer in order to remove anucleated cells, then stained with the appropriate combination of antibodies for 20 min, washed, and analyzed by BD LSR II flow cytometer. Data were analyzed with FlowJo software, and donor status of individual cells was distinguished by CD45.1 and CD45.2 staining.

Histological analysis of whole murine femurs

Request a detailed protocol

Femurs were fixed in a paraformaldehyde–lysine–periodate fixative overnight (0.01 M Sodium-M-Periodate, 0.075M L-Lysine, 1% PFA), rehydrated in 30% sucrose in a phosphate buffer solution for 48 hr, embedded in OCT (TissueTek, Sakura), and snap frozen in an isopentane/dry ice mixture (Nombela-Arrieta et al., 2013). Whole longitudinal sections (7 μm) sections were obtained using a Leica Cryostat and the Cryojane tape transfer system. Tissue sections were thawed, rehydrated and permeabilized in Tris-buffered saline with 0.1% Tween (T-TBS), blocked with 5% BSA, then incubated with FITC-conjugated SNA lectin (VectorLabs) for 1 hr, then washed prior to incubation with the appropriate primary antibodies. These were followed by the corresponding Alexa Fluor secondary antibodies (1:500, Invitrogen). Fluorescence whole slide imaging was performed on a Nikon Eclipse Ti2. Quantification of SNA density staining and IgD+ cell localization and numeration analysis was executed using Imaris (Bitplane) and Matlab (MathWorks) software. Using consistent parameters for every bone based on their staining intensity, IgD+ B cells were identified and counted. SNA density staining of the marked regions of interest (ROIs) was normalized from different femurs by accounting for total SNA intensity in the whole bone scan. ROIs were selected at random and varied in sizes and SNA staining intensity. ROIs were located in various parts of the bone marrow but consistent with hematopoietic cell localization, and not covering any major vasculature or bony areas. However, SNA intensity within each individual ROI exhibited low variability.

Histological analysis of human bone marrow

Request a detailed protocol

All experiments involving human samples were evaluated and approved by the Institutional Review Board (IRB) prior to their initiation. Banked human biospecimens were provided by the Disease Bank and BioRepository at Roswell Park Comprehensive Cancer Center under protocol BDR 082017. Human multiple myeloma samples collected from treatment-naïve patients were included in the analysis. Samples from 15 treatment-naïve patients were available from the Repository, de-identified prior to transfer, and associated survival and demographic information and pathology reports were provided via an honest broker. Paraffin-embedded sections of bone marrow biopsies were melted at 55C for 1 hr, twice dehydrated in xylene-containing HistoClear (National Diagnostics), then rehydrated in successive ethanol solutions, and heated in Antigen Unmasking Solution (Vector Labs) for 30 min. Slides were blocked in 5% BSA for 1 hr, incubated overnight with anti-ST6GAL1 antibody, then with anti-goat-HRP secondary (R and D Biosystems) for 1 hr. Tissues were then immersed in Impact DAB stain (Vector Labs) for 120 s and rinsed in water for 3 min. Slides were counterstained for 60 s with hematoxylin. Plasma cells were identified under pathologist guidance as cells with radial distribution of heterochromatin and perinuclear clearing corresponding to the lipophilic Golgi apparatus. The presence of malignancy was confirmed by identification of high density sheets or clumps of morphologically similar plasma cells. ST6GAL1 expression was quantified by evaluation of frequency of positively-staining plasma cells and intensity of staining in at least 5 fields of view per specimen typically containing 50 cells each. The frequency of ST6GAL1+ plasma cells was quantified in five random selections of ten cells from five independent images per patient. The average intensity of staining was graded as follows from 0 to 5: 0 – no evidence of staining, 1 – faint staining, 2 – nut brown perinuclear staining occupying <25% of cell, 3 – nut brown perinuclear staining occupying >25% of cell, 4 – dark brown perinuclear staining, 5 – dark brown or black staining occupying the perinuclear and nuclear regions. Bone marrow plasma cells and neutrophils were quantified by pathologist evaluation at time of diagnosis and provided by the Pathology Shared Resource Network (PSRN).

Statistical analyses

Request a detailed protocol

Experiments were conducted with a minimum sample size calculated for appropriate power to detect changes of at least 2-fold (α = 0.05, β = 0.80, SD = 0.5). Raw data are presented in all figures as mean ± SD.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
    Increase in serum sialyltransferase in tumor bearing rats: the origin and nature of the increased enzyme
    1. K Dairaku
    2. T Miyagi
    3. A Wakui
    4. S Tsuiki
    (1983)
    Gan 74:656.
  20. 20
  21. 21
  22. 22
  23. 23
    Serum and tumor sialyltransferase activities in women with breast Cancer
    1. TL Dao
    2. C Ip
    3. JK Patel
    4. GS Kishore
    (1986)
    Progress in Clinical and Biological Research 204:31.
  24. 24
  25. 25
  26. 26
    Correlation of seurm, tumor, and liver serum glycoprotein: n-acetylneuraminic acid transferase activity with growth of the R3230AC mammary tumor in rats and relationship of the serum activity
    1. IM Evans
    2. R Hilf
    3. M Murphy
    4. HB Bosmann
    (1980)
    Cancer Research 40:3103.
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
    Plasma sialyltransferase activity in healthy subjects and atherosclerotic patients
    1. EV Gracheva
    2. NK Golovanova
    3. MV Ezhov
    4. PP Malyshev
    5. VV Kukharchuk
    6. NV Prokazova
    (1999)
    Biochemistry. Biokhimiia 64:1315.
  33. 33
  34. 34
  35. 35
  36. 36
    Alterations in serum glycosyltransferases and 5'-nucleotidase in breast Cancer patients
    1. C Ip
    2. T Dao
    (1978)
    Cancer Research 38:723–728.
  37. 37
  38. 38
  39. 39
  40. 40
    Sialyltransferase: a novel acute-phase reactant
    1. JC Jamieson
    2. G McCaffrey
    3. PG Harder
    (1993)
    Comparative Biochemistry and Physiology Part B: Comparative Biochemistry 105:29–33.
    https://doi.org/10.1016/0305-0491(93)90165-2
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
    Studies on the effect of inflammation on rat liver and serum sialyltransferase: evidence that inflammation causes release of Galgal beta1-4 GlcNAc alpha2-6 sialyltransferase from liver
    1. HA Kaplan
    2. BM Woloski
    3. M Hellman
    4. JC Jamieson
    (1983)
    Journal of Biological Chemistry 258:11505.
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
    Transcriptional regulation of the liver beta-galactoside alpha 2,6-sialyltransferase by glucocorticoids
    1. XC Wang
    2. TJ Smith
    3. JT Lau
    (1990)
    Journal of Biological Chemistry 265:17849–17853.
  83. 83
    Chromosome mapping and organization of the human beta-galactoside alpha 2,6-sialyltransferase gene. Differential and cell-type specific usage of upstream exon sequences in B-lymphoblastoid cells
    1. X Wang
    2. A Vertino
    3. RL Eddy
    4. MG Byers
    5. SN Jani-Sait
    6. TB Shows
    7. JT Lau
    (1993)
    Journal of Biological Chemistry 268:4355–4361.
  84. 84
    Primary structure of beta-galactoside alpha2,6-sialyltransferase. Conversion of membrane-bound enzyme to soluble forms by cleavage of the NH 2 -terminal signal anchor
    1. J Weinstein
    2. EU Lee
    3. K McEntee
    4. PH Lai
    5. JC Paulson
    (1987)
    Journal of Biological Chemistry 262:17735–17743.
  85. 85
  86. 86
  87. 87
  88. 88

Decision letter

  1. Jamey Marth
    Reviewing Editor; University of California, Santa Barbara, United States
  2. Satyajit Rath
    Senior Editor; Indian Institute of Science Education and Research (IISER), India
  3. Jamey Marth
    Reviewer; University of California, Santa Barbara, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "B cells suppress medullary granulopoiesis by an extracellular glycosylation-dependent mechanism" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Jamey Marth as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Satyajit Rath as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

Overall the reviewers were in agreement of the general importance of the findings presented, and the potential for a revision of this manuscript to become appropriate for publication in eLife. However there were many specific concerns raised by all three reviewers. These comments indicate the need for a significant revision that will also require additional data. It is recommended that the authors address each comment of the reviewers carefully in composing a revised manuscript.

Essential revisions:

Multiple datasets identified in the reviews need additional data points to achieve a more robust statistical significance. In addition, some data appears insufficient and of poor quality, and thus inadequate for review and publication. Various methods used in the manuscript are missing. Data analyzed in the co-culture experiments of Figure 3 should include one or more ST6 KO B cells to determine the B cell effect and considering poor correlations with ST6 RNA and SNA binding (Figure 1/Figure 3). Data presented in the studies of Figure 5 should be further elaborated upon regarding choosing and analyzing ROIs as suggested in the reviews, and the inclusion of µMT donors would be of value. The impact of B cells as presented from data in Figure 6 needs further analysis including cell subsets top gauge the degree of B cell effect versus other cell types present.. The human patient data of Figure 7 needs careful explanation of methods used, as indicated, and better quality histological sections for review and publication.

The individual reviews are provided below so as to provide the authors access to the context and logic for the revisions needed.

Reviewer #1:

Irons et al. report ST6Gal1-dependent sialylation of bone marrow progenitor cells results in reductions of Gr1+ granulopoeisis in mice and in corollary cases of human patients with Multiple Myeloma. The authors follow up on their past publications reporting the presence of extracellular sialylation by secreted forms of the ST6Gal-1 sialyltransferase, which is made in a number of tissues and cell types including the liver and B cells, and its role in modulating leukocyte differentiation and inflammatory responses. Cell type specific sources of extracellular ST6Gal1 and its donor substrate among specific studies published have remained mostly unclarified as this enzyme can be found at cell surfaces including among platelets and hepatocytes as well as being secreted into the blood. With the current manuscript, a role of ST6Gal1 in modulating granulopoiesis is the focus. The mechanism involved regarding which proteins are being sialylated and how sialylation modulates granulopoiesis in the bone marrow is not addressed, which is a weakness of the manuscript. At various places the manuscript is difficult to understand and key experimental methods are sometimes lacking. However, there are a number of important findings that with some additional data would provide a strong rationale for publication. The findings of potential involvement of bone marrow megakaryocytes are important, and the authors demonstrate strong linkages of ST6Gal-1 function vis SNA lectin binding with regulation of bone marrow hematopoiesis. The text can be pared down especially Introduction and Discussion sections to achieve a more focused presentation. I have listed other critiques for revision below, including the use of data sets larger than n=2, which would provide me with even greater enthusiasm for publication.

1) The secretion of the 'full length' version of ST6Gal1 may reflect the presence of exosomes in the assays of Figure 1, and is something that might in the future be investigated as exosomes have the potential to transport not just ST6Gal1 but pre-sialylated proteins and donor substrates, in trans. The authors do not comment or reference this possibility. In that regard, the reference to the use of "intracellular proteins" in the Figure 1 legend and in the Materials and methods is incorrect. Total cellular proteins appear to have been analyzed (B and C), including cell surface proteins, and which may have included exosomes from the supernatant, depending on the method of protein isolation, including the centrifugal forces used. The method of protein isolation is absent from the Materials and methods.

2) The studies of HSPCs as presented in Figure 3B could be strengthened by providing more than 2 data points per condition. Key data presented in Figure 3C cannot be read as the text is blurry and too small. So it is not clear if this particular data point is represented in Figure 3D. I also suggest the inclusion of 0:1 co-culture condition data for Figure 3C, along with a clearer depiction and presentation of the gating of cell populations and their frequencies in these examples, and some description of the double-positive population (Gr1+/CD11b+) in the text, which is likely also affected by changes to abundance of Gr1+/CD11b- cells.

3) Partial reconstitution of the bone marrow compartment in chimeras was used to assess the trans sialylation of multiple cell types in vivo. This approach is effective as presented but has only 2 data points (Figure 4B and 4D). The use of only two data points at various places in the manuscript, while technically allowable as a bare minimum, becomes much less effective and compelling when including data involving animal models with individual variation.

4) The authors present an in-situ bone marrow analysis of chimeras to acquire data linking higher levels of ST6Gal1 function (detected by SNA binding), to regions of the bone marrow that are also high in abundance of donor-derived B cells and megakaryocytes. These data are presented in Figure 5. It is not clear if the same cell numbers are being analyzed in each chosen ROI. That data should be provided. It would also be valuable to have a normal correlate panel of wild-type marrow, which could be used to determine whether megakaryocyte lineage cells are normally approximated with highest SNA binding levels in the bone marrow, which if so would further strengthen the relevance of the findings. The potential for sialylation without bone marrow B cells, as is suggested, can be determined using µMT mutant donors as well as the same double mutant recipient used in the current study. The fraction of non B cell ST6Gal1 sialylation in the bone marrow could then be determined. The Figure 5 term "Full Stain" is undefined and unclear. This study also presents very small data sets from small numbers of ROIs and as indicated in the Figure 5 legend from 2 biological replicates.

5) To determine the contribution of B cells, the authors do a study using bone marrow chimeras, which now includes µMT mutant donors, looking specifically at circulating blood cells 8 weeks following partial reconstitution. The data provide evidence of B cell involvement in providing a small degree of ST6Gal1 sialylation measured by SNA binding among circulating donor-derived blood cells. Controls in this study would be helpful to make this determination more clear and robust, including relevant donors without ST6Gal1. It would also be valuable to have more information regarding sialylation among specific cell subsets in circulation, as perhaps some cell types in this model are more of less affected in their sialylation by B cells. In Figure 6C, it is difficult to determine if the difference presented in the cytometry scans reflect the same number of events/cell numbers, as it looks as if there may be more cells being analyzed in the case of the µMT donor set, while the range of SNA binding is not elevated beyond normal.

6) The data provided by the human patients diagnosed with Multiple Myeloma needs more clarification. It is unclear how the data are being analyzed in Figure 7A, there is nothing in the Materials and methods or legend that addresses this. As with all the histology results in Figure 5, the numbers of sections in area and numbers of cells analyzed in total in Figure 7 can be provided by current histological techniques. The tissue panels in Figure 7D are of insufficient quality to identify neutrophils (arrows) and other basic cellular histological features. Regions of interest including higher magnifications are also needed. In studying human bone marrow, the authors have the opportunity to follow up on the megakaryocyte findings, and the SNA reactivity measurements, which might further tie in their findings in Figure 5 with in-situ human studies.

Reviewer #2:

This manuscript provides evidence that ST6GAL1 is secreted from B cells to contribute to serum levels and that B cells in bone marrow, perhaps in collaboration with megakaryocytes, are responsible for adding sialic acid to hemopoietic cell progenitors leading to suppression of granulopoiesis. The data are generally of good quality but numerous specifics are missing. The following points should be addressed:

1) In the Introduction, second paragraph: "terminal glycosyltransfearases" – spelling as well as jargon that would not be comprehended by scientists in general.

2) Subsection” Human B lymphoblastoid cells secrete enzymatically active ST6Gal-1”, first paragraph, "formally studied." Do the authors mean to say formerly?

3) Subsection “Human B lymphoblastoid cells secrete enzymatically active ST6Gal-1”, second paragraph. Single examples of human B cell lines at different stages of development were examined. The data would be more robust if independent cell lines at those same stages gave the same results.

4) "In conditioned medium from day 3 of culture, α2,6-sialyltransferase activity generally agreed with protein analyses, whereas α2,3-sialyltransferase activity varied independently (Figure 1D)." This statement is meaningless unless the authors can explain "protein analyses"

5) Figure 1 Are the secreted 50 and 42 kD forms of ST6Gal1 equivalently active? It may not be a completely full length form that is present in the medium. Do SPPL peptidases play any role in cleaving either form? How was the Western blot in Figure 1C imaged – it has a non-specific black blob covering the ST6Gal-1 band in the last lane.

The histogram in Figure 1C appears to represent only the blot in Figure 1C. How reproducible are these data? A histogram including results of several blots should be shown.

What amount of conditioned medium was assayed to obtain the data in Figure 1D ? How was the experiment performed in terms of samples collected at different times, replicates etc.? The description and methods are too brief to enable replication.

It is not acceptable to infer that the remaining activity was due to ST3GAL activity and to label the figure as though that fact was established.

6) Figure 2. Details of methods, amounts, timing and replication are needed in the legend and Materials and methods.

The cellular localization of the SNA signal does not seem consistent with sialidase acting on the surface of fixed cells, nor with SNA apparently inside fixed cells.

7) Details of experimental methods to set up co-cultures and to analyze by flow cytometry, numbers of cells co-cultured, culture conditions, concentrations of cytokines etc. are needed.

Why were ckit+ bone marrow cells used? Progenitors ae usually selected by a more elaborate protocol involving lineage depletion and Sca1.

8) Figure 4 Show a2,3sialyltransferase (SiaT) activity analysis and describe how this specific activity was assayed. Do the levels of the a2,3 enzyme also rise to those found in serum from WT mice? How many chimeras were assayed?

What was the efficiency of chimerism? Were the numbers of hemopoietic cells of each subtype similarly represented? These data should be included in the manuscript, preferably in Figure 4.

What does n=5 relate to in Figure 4D? Is each symbol representative of a mouse?

9) Figure 5. The methods for quantitating femur co-localization data need to be more completely described. Were the ROI data obtained from each femur of each mouse? The correlations would be more robust with more data, especially of high SNA ROIs.

10) Figure 6B cells are clearly not the only source of serum a2,6SiaT. This should be discussed.

How many mice were used to generate Figure 6B versus 6C?

11) Figure 7. The bone marrow H&E from the SNA-low patient is poor. Another example should be shown.

12) General.

Complete information on each Ab used should be given, including clone numbers.

How was non-mitogenic FCS prepared?

Define ACK lysis.

Nomenclature should conform with HGNC convention of ST6GAL1 for protein, St6gal1 (italicized) for the mouse gene, and ST6GAL1 (italicized) for the human gene.

Reviewer #3:

In this manuscript, the authors describe the ability of B cell-lineage cells to release the sialyltransferase ST6Gal1, which may contribute to the 'extracellular glycosylation' reported by the corresponding author and others previously. While there are a number of tantalizing observations herein, the issues with the data shown and the conclusions provided reduce enthusiasm to this contribution to a very important field of study. For example, it is not clear that the data in Figure 7 has anything to do with the rest of the manuscript. Also, in several cases, the number of biological replicates was 2, with only 1 being shown, which is inadequate. Further detailed comments are below:

- In Figure 1, the authors are using cancer/immortal cell lines to assess ST6Gal1 and BACE1 expression, but the relationship of these values to normal primary cells in vitro and in vivo is unclear. As pointed out by the authors, cancer is associated with robust changes in glycoforms. Moreover, culturing in standard media contains more glucose than what is normal in vivo, which is also associated with changes in glycosylation. These data have little value in understanding the biology of ST6Gal1 in B cells in the absence of studies using primary cells (human or mouse) for confirmation.

- In Figure 1, the authors also make generalizations about the relative expression levels in B cells from different anatomical locations/developmental stages. But again, these are cancer/immortalized cell lines. Such connections require much more data and multiple cell lines, not just one, for each stage being represented.

- Also in Figure 1, the authors state that the Western blot "confirms" the mRNA levels, but this is a mischaracterization. The protein level (Figure 1B) does not mirror the mRNA level (Figure 1A). Moreover, the protein level was seemingly measured only once, hence no error bars in the graph (Figure 1B). The Western blots are also missing visible size standards.

- Figure 1C is missing repeat experiments and statistics.

- Figure 1D is missing controls with 2,3 and 2,6 recombinant sialyltransferases to demonstrate the rigor of the assay, and the completeness of the SNA capture. The authors also describe the SNA-agarose precipitation as "column chromatography", which is somewhat of a mischaracterization.

- In Figure 2, the images are nice, but a measure of fluorescence intensity of the cells under the varied conditions (and with biological replicates) using flow cytometry would provide statistical power and quantitation.

- Figure 3 is confusing. First, the amount of change in SNA staining does not correlate with the amount of ST6Gal1 being produced by each B cell line (compare changes in Figure 3B with the protein levels in Figure 1B). This calls into question whether the effect of the B cells is truly because they produce ST6Gal1, or if their presence in the culture alters the metabolic state of the HSPCs such that their glycosylation changes. Indeed, one can see a decrease in a2,3-Sia with the RPMI8226 cells, even though they are releasing an a2,3-Sia transferase (Figure 1D). Similarly, the MM1 cells seem to produce a lot of a2,3-Sia transferase activity, and yet no change in glycosylation is seen in the HSPCs. Does this mean that these STs cannot add Sia to the cells? The authors should demonstrate a dependence on ST6Gal1 in these assays by knocking out ST6Gal1 in each line and repeating the experiment.

- In Figure 3D, is it possible that the anti-GR1 antibody binding if impacted by sialylation? Does GR-1 transcription (mRNA) change as well?

- Figure 4B is lacking a comparative control of WT mice not subjected to any bone marrow transfer/chimerism. This is important to provide the relative contribution of the hematopoietic compartment to the total ST6Gal1 activity in the extracellular space (marrow and plasma).

- Figure 5B is lacking statistical power (2 biological replicates). There are too few data points to make a meaningful correlation. Additional repeats should solidify the findings easily. Moreover, and perhaps more importantly, the term "co-localization" is not being applied in a clear way. For example, how close do the cells have to be to be called "co-localized"? Simply within the region drawn on the image? If so, there are cells in Region 1 (panel A) that are closer to Region 2 than they are to the left side of Region 1. Thus, the quantitation here is largely arbitrary and without a specific definition (i.e. within X µm). Finally, it would help if these images were done with Z-axis detail.

- For Figure 6, what are the differences in cellularity between the WT and µMT bone marrow, other than the lack of B cells? Could changes in the cell population that accompany the lack of B cells be playing a role? Also, why would it take 6 weeks for the ST6Gal1 activity to become different, and what happens after 10 weeks, since the curves seem to be converging as they were in earlier weeks?

- In Figure 6, the authors state that the Ighm-/- hematopoietic cells within the blood and bone marrow show increase SNA, but the graph in Figure 6C for the blood indicates a lack of statistical significance in this change. This also highlights another problem. Why is data from only one experiment shown? Multiple biological replicates plotted/averaged together should be shown.

- The connection of the data in Figure 7 to the rest of the manuscript is not clear. There is no functional connection made between ST6Gal1 and the granulocyte observation. In addition, the data in Figure 7D are difficult to see (the H&E). If the number of granulocytes from patients was the readout, flow cytometry or automated hematological cell counting should be performed to provide a more quantitative measure of cellularity.

https://doi.org/10.7554/eLife.47328.013

Author response

Essential revisions:

Multiple datasets identified in the reviews need additional data points to achieve a more robust statistical significance. In addition, some data appears insufficient and of poor quality, and thus inadequate for review and publication. Various methods used in the manuscript are missing. Data analyzed in the co-culture experiments of Figure 3 should include one or more ST6 KO B cells to determine the B cell effect and considering poor correlations with ST6 RNA and SNA binding (Figure 1/Figure 3). Data presented in the studies of Figure 5 should be further elaborated upon regarding choosing and analyzing ROIs as suggested in the reviews, and the inclusion of µMT donors would be of value. The impact of B cells as presented from data in Figure 6 needs further analysis including cell subsets top gauge the degree of B cell effect versus other cell types present.. The human patient data of Figure 7 needs careful explanation of methods used, as indicated, and better quality histological sections for review and publication.

Essential changes to the manuscript:

The authors thank the many insightful suggestions from the editor and reviewers. In the revision, we have introduced 2 major new experiments to strengthen the manuscript:

1) To validate further that ST6GAL1 released by the B cells was responsible for modification of HSPC cell surface sialylation and muted Gr-1+ cell generation, 2 CRISR/Cas9 mutants with knocked-out ST6GAL1 were generated and analyzed (new Figure 3F). The CRISPR/Cas mutants lost the ability to suppress Gr-1+ cell generation, relative to the original MM1.S cells.

2) New chimeras were constructed to validate the hypothesis that B cells are the major hematopoietic cells supplying extracellular ST6GAL1 in vivo (Figure 4F and G). ST6GAL1-normal donor hematopoietic cells can restore resting circulating ST6GAL1 activities, but donors without B cells (μMT) are unable to elevate circulating ST6GAL1 activities.

3) Construction and analysis of the new cell lines (part 1) involved new collaborators Yuqi Zhu and Sriram Neelamegham. Hence these individuals are now added to the authorship as authors 3 and 4.

Data sets have also been expanded and descriptions edited. (Please see response to individual reviewers’ comments).

The individual reviews are provided below so as to provide the authors access to the context and logic for the revisions needed.

Reviewer #1:

[…] 1) The secretion of the 'full length' version of ST6Gal1 may reflect the presence of exosomes in the assays of Figure 1, and is something that might in the future be investigated as exosomes have the potential to transport not just ST6Gal1 but pre-sialylated proteins and donor substrates, in trans. The authors do not comment or reference this possibility. In that regard, the reference to the use of "intracellular proteins" in the Figure 1 legend and in the Materials and methods is incorrect. Total cellular proteins appear to have been analyzed (B and C), including cell surface proteins, and which may have included exosomes from the supernatant, depending on the method of protein isolation, including the centrifugal forces used. The method of protein isolation is absent from the Materials and methods.

We thank the reviewer for bringing this salient point to our attention. We agree that the presence of the full-length form implies that the protein is still membrane-embedded and is consistent with its localization in a vesicle. This might suggest exciting new possibilities in ST6Gal-1 transfer and extrinsic sialylation between different cells. Our observations are consistent with both the 50kDa and the 42kDa forms being catalytically active, but we feel it is beyond the scope of this manuscript for further mechanistic dissection into the origin and nature of the 50kDa form. Results and Discussion have been expanded to address this.

The reference to “intracellular proteins” is replaced with “total cellular proteins”.

2) The studies of HSPCs as presented in Figure 3B could be strengthened by providing more than 2 data points per condition. Key data presented in Figure 3C cannot be read as the text is blurry and too small. So it is not clear if this particular data point is represented in Figure 3D. I also suggest the inclusion of 0:1 co-culture condition data for Figure 3C, along with a clearer depiction and presentation of the gating of cell populations and their frequencies in these examples, and some description of the double-positive population (Gr1+/CD11b+) in the text, which is likely also affected by changes to abundance of Gr1+/CD11b- cells.

We apologize for the low-resolution figure quality in the submitted review draft. All figures have been provided as higher-resolution JPEGs in this submission, and various instances of blurry print were replaced with clearer text for your evaluation. The highest resolution images suitable for publication will be included separately. The data in Figure 3C is from one of the single points shown in Figure 3D. Also, Figure 3C has been updated to include the 0:1 monoculture condition as comparison, wherein it can be appreciated that there is a maximal number of Gr-1+ and CD11b+ cells. In addition, a more detailed explanation of CD11b and Gr1 expression in this experiment and in general is provided in the corresponding Results section for context. We have also included data regarding CD11b-/Gr-1- murine cells, which represent cells that have not differentiated to the extent required for expression of these markers. Here, ST6Gal-1 expressing cell lines are seen to enlarge this population.

3) Partial reconstitution of the bone marrow compartment in chimeras was used to assess the trans sialylation of multiple cell types in vivo. This approach is effective as presented but has only 2 data points (Figure 4B and D). The use of only two data points at various places in the manuscript, while technically allowable as a bare minimum, becomes much less effective and compelling when including data involving animal models with individual variation.

Similar experiments to Figure 4B and D have been performed by our group in the past and published (Nasirikenari et al., 2010, Jones et al., JBC), and these have clearly demonstrated sialylation by extracellular ST6Gal1 in multiple cell types in vivo. Collectively, our data have demonstrated the importance of non-self ST6Gal-1 expression in cell surface sialylation in a number of contexts. The purpose of Figure 4 was to show that extracellular ST6Gal1 can also come from hematopoietic cells. Figure 4 has been modified to show clearly that the collected data had more than 2 points, specifically in reference to Figure 4B, C and E.

4) The authors present an in-situ bone marrow analysis of chimeras to acquire data linking higher levels of ST6Gal1 function (detected by SNA binding), to regions of the bone marrow that are also high in abundance of donor-derived B cells and megakaryocytes. These data are presented in Figure 5. It is not clear if the same cell numbers are being analyzed in each chosen ROI. That data should be provided. It would also be valuable to have a normal correlate panel of wild-type marrow, which could be used to determine whether megakaryocyte lineage cells are normally approximated with highest SNA binding levels in the bone marrow, which if so would further strengthen the relevance of the findings. The potential for sialylation without bone marrow B cells, as is suggested, can be determined using µMT mutant donors as well as the same double mutant recipient used in the current study. The fraction of non B cell ST6Gal1 sialylation in the bone marrow could then be determined. The Figure 5 term "Full Stain" is undefined and unclear. This study also presents very small data sets from small numbers of ROIs and as indicated in the Figure 5 legend from 2 biological replicates.

The reviewer raised important points. The careful design and description of methods in relatively new analytical techniques are critical to accurate interpretation of results. The Materials and methods section for this experiment has been modified to address the points raised here and by other reviewers, including cell numbers per ROI. As a comparison to our original analysis of WT transplanted marrow into St6gal1-KO recipients, which showed patchwork restoration of marrow SNA staining, we did stain a wild-type femur and noted the absolutely homogenous SNA staining. This indicates the sialylation patchiness depends on having a St6gal1-deficient recipient animal; WT and µMT mice both have native ST6Gal-1, and all showed diffuse and even SNA staining. These are now included in the figure.

The term ‘full stain’ is removed. The number of biological replicates is 4 animals (Figure 5C, ROIs from each replicate in a different colour). Our data indicate a modest positive relationship between IgD+ B cells and the SNA reactivity of the surrounding cells. To maintain focus of this manuscript, which is on the contribution of B cells to marrow sialylation, we are electing to strike the data associating megakaryocyte from this manuscript. The contribution of megakaryocytes to marrow sialylation will be taken up separately at a later time. Including megakaryocytes will bring up the natural mechanistic question of how megakaryocytes contribute to sialylation, which lies beyond the scope of this manuscript.

5) To determine the contribution of B cells, the authors do a study using bone marrow chimeras, which now includes µMT mutant donors, looking specifically at circulating blood cells 8 weeks following partial reconstitution. The data provide evidence of B cell involvement in providing a small degree of ST6Gal1 sialylation measured by SNA binding among circulating donor-derived blood cells. Controls in this study would be helpful to make this determination more clear and robust, including relevant donors without ST6Gal1. It would also be valuable to have more information regarding sialylation among specific cell subsets in circulation, as perhaps some cell types in this model are more of less affected in their sialylation by B cells. In Figure 6C, it is difficult to determine if the difference presented in the cytometry scans reflect the same number of events/cell numbers, as it looks as if there may be more cells being analyzed in the case of the µMT donor set, while the range of SNA binding is not elevated beyond normal.

In addition to the data presented in the original Figure 6, we have performed a follow-up transplantation using a combination of donors: St6gal1-KO, St6gal1-KO/μMT, μMT, and WT (C57BL/6) that were used to reconstitute the hematopoietic compartment of St6gal1-KO recipients. Under this scheme, the specific contribution of ST6Gal-1 originating from B lineage cells to extrinsic sialylation can be assessed with less ambiguity. The new data involving also donors without ST6Gal-1 and/or without B cells clearly show that reconstitution of hematopoietic compartment with ST6Gal-1-competent B lineage cells is absolutely necessary to raise circulating ST6Gal-1 levels. Lack of B cells but otherwise ST6Gal-1 normal (i.e. μMT donor) is clearly insufficient to restore circulating ST6Gal-1 levels. This new data are now appended to the new Figure 4. The original data, presented as the old Figure 6, used a ST6Gal-1 competent recipient, which allows liver-expressed ST6Gal-1 to contribute to the overall extrinsic ST6Gal-1 pool. The new data, moreover, make the old Figure 6 redundant. Therefore we have removed the old Figure 6 since it adds nothing new.

6) The data provided by the human patients diagnosed with Multiple Myeloma needs more clarification. It is unclear how the data are being analyzed in Figure 7A, there is nothing in the Materials and methods or legend that addresses this. As with all the histology results in Figure 5, the numbers of sections in area and numbers of cells analyzed in total in Figure 7 can be provided by current histological techniques. The tissue panels in Figure 7D are of insufficient quality to identify neutrophils (arrows) and other basic cellular histological features. Regions of interest including higher magnifications are also needed. In studying human bone marrow, the authors have the opportunity to follow up on the megakaryocyte findings, and the SNA reactivity measurements, which might further tie in their findings in Figure 5 with in-situ human studies.

The methods for this experiment have been rewritten to include more detail, particularly in regard to the quantification of ST6Gal-1 expression in myeloma cells and the number of fields of view/cells used to make this determination. Changes have been made to allow the reader to more clearly assess the histological features in H&E stains and appreciate the noted difference in neutrophil abundance. These include an overall increase in the size of pictures, inclusion of an insert wherein individual cellular structures can be observed, and clear indications of both neutrophils and plasma cells for the reader to qualitatively assess the correlation. Although we appreciate the suggestion of furthering the investigations into megakaryocytes in the context of the human samples, we believe that this approach would be unlikely to yield meaningful results for several reasons. Firstly, megakaryocytes are well-documented to participate in niche maintenance for bone marrow plasma cells and have a pro-tumorigenic role in multiple myeloma (Yaccoby et al. Blood 2005, Takagi et al. Blood 2015). Thus, any correlative analysis of megakaryocyte abundance or localization should have to consider the confounding effects of megakaryocytes on the tumor itself. Secondly, the DAB staining process only allows analysis of one histological parameter at a time, making the co-identification of megakaryocytes and sialylation difficult.

Reviewer #2:

[…]) In the Introduction, second paragraph: "terminal glycosyltransfearases" – spelling as well as jargon that would not be comprehended by scientists in general.

This has been corrected. Thank you.

2) Subsection” Human B lymphoblastoid cells secrete enzymatically active ST6Gal-1”, first paragraph, "formally studied." Do the authors mean to say formerly?

This has been corrected. Thank you.

3) Subsection “Human B lymphoblastoid cells secrete enzymatically active ST6Gal-1”, second paragraph. Single examples of human B cell lines at different stages of development were examined. The data would be more robust if independent cell lines at those same stages gave the same results.

The purpose of our survey of human B cell lines was not to establish a pattern of ST6Gal-1 expression between different stages of development. Such an analysis in primary cells was recently done by us in a separate publication (Irons et al., 2018), and in the B lymphoblastoids much earlier (Wuensch et al., 2000). Our purpose here was to select B-lineage cells that releases differing levels of ST6Gal-1 in order to assess how B-cell derived ST6Gal-1 can influence extrinsic sialylation of nearby cells. While the panel of B cell lines happened to represent different developmental stages, we cannot and do not infer correlations to stages of development based on this very limited sampling.

4) "In conditioned medium from day 3 of culture, α2,6-sialyltransferase activity generally agreed with protein analyses, whereas α2,3-sialyltransferase activity varied independently (Figure 1D)." This statement is meaningless unless the authors can explain "protein analyses"

We have modified this statement to specifically indicate that the reference is to the quantitation of secreted ST6Gal-1 protein from immunoblot analysis of conditioned medium.

5) Figure 1 Are the secreted 50 and 42 kD forms of ST6Gal1 equivalently active? It may not be a completely full length form that is present in the medium.

Do SPPL peptidases play any role in cleaving either form?

How was the Western blot in Figure 1C imaged – it has a non-specific black blob covering the ST6Gal-1 band in the last lane.

The histogram in Figure 1C appears to represent only the blot in Figure 1C. How reproducible are these data? A histogram including results of several blots should be shown.

What amount of conditioned medium was assayed to obtain the data in Figure 1D ? How was the experiment performed in terms of samples collected at different times, replicates etc.? The description and methods are too brief to enable replication.

It is not acceptable to infer that the remaining activity was due to ST3GAL activity and to label the figure as though that fact was established.

The sialylated Gal(β4)GlcNAc-o-Bn products have been extensively characterized in the past. We have only seen Sia(α2,6)- and Sia(α2,3)- trimers synthesized from blood-borne enzymes. The characterization includes HPLC and MS verification. While the former is retained by SNA-agarose, the latter is not. Thus based on our prior published characterization, we feel comfortable in stating that the SNA-agarose non-binding fraction represents only Sia(α2,3)-trimers. The Materials and methods section now includes bibliographic references to the analysis, particularly Lee-Sundlov et al., 2017.

To the best of our knowledge, both 50kD and 42kD forms are enzymatically active, as evidenced by the detectable sialyltransferase activity in the conditioned medium of MM1.S cells, which express minimal BACE-1 and largely secrete the 50kD form. Although the ability of these various cell lines to extrinsically sialylation HSPC in co-culture conditions varied, we cannot currently attribute such differences to associated changes in protein size.

We appreciate the insightful reference to aminopeptidases, as previous reports have indicated their involvement in the proteolytic processing of ST6Gal-1. Along with the comment regarding exosomes of reviewer 1, we have included an expanded discussion of the possibility for an aminopeptidase-processed ~50kD soluble form in the Discussion.

All immunoblots were imaged using a Biotek ChemiDoc Touch machine for between 30-120 seconds. We believe the “blob” in the last lane of Figure 1C may be partially due to a distortion in that well during gel loading. However, we have also noted in repeat analyses that the 50kD form of the secreted MM1.S ST6Gal-1 often appears as several bands in close proximity, which we have interpreted as multiple forms of similar size with differences in proteolytic or post-translational modification. Please see the included image of the MM1.S cell line secretions with other multiple myeloma cell lines for comparison, where this can be appreciated (see Figure 2—figure supplement 1B and C). We have repeated the experiment to generate histograms with the averages and variation in these samples.

In Figure 1D, parallel cultures of 10^6 cells were seeded in 1ml of serum-free medium in 12 well plates. At 72 hours post-seeding, conditioned medium was collected by pipetting and centrifuged to separate cells. A fixed volume of 1.5ul (0.15% of total) was used in the sialyltransferase activity assay. Details and clarification have been included into the Materials and methods section to allow for replication.

We have changed the figure, Materials and methods, and Results section to more accurately indicate that the data show SNA-reactive and SNA-unreactive sialyltransferase activity.

6) Figure 2. Details of methods, amounts, timing and replication are needed in the legend and Materials and methods. The cellular localization of the SNA signal does not seem consistent with sialidase acting on the surface of fixed cells, nor with SNA apparently inside fixed cells.

These details have now been included as suggested. Our interpretation of the images in Figure 2 is that low-intensity, ‘hazy’ SNA reactivity was due to cell surface sialic acid, which was removed by sialidase and restored in the final panel. In contrast, the bright, punctate points that persist throughout all treatments are intracellular, likely corresponding to the Golgi apparatus, wherein nascent glycoproteins are intrinsically sialylated. We have performed similar experiments with other cell types to validate observed trends in SNA reactivity on HepG2 cells. Also, we have demonstrated that Louckes-derived supernatant is able to raise SNA reactivity on beads conjugated to an asialylated protein substrate (asialofetuin). It is unclear how the reviewer is interpreting these images if sialic acid is not present on the cell surface or inside cells, but we would emphasize that HepG2 cells are generally large cells with an extended cytoplasm, consistent with the distribution of SNA staining.

To further address this point, we have performed flow cytometry analysis of HepG2 cells in a similar experiment, gating only on nucleated cells to remove residual debris and any acellular components. The results have been included in the figure and demonstrate that an increase in SNA reactivity on the cells can be detected.

7) Details of experimental methods to set up co-cultures and to analyze by flow cytometry, numbers of cells co-cultured, culture conditions, concentrations of cytokines etc. are needed. Why were ckit+ bone marrow cells used? Progenitors ae usually selected by a more elaborate protocol involving lineage depletion and Sca1.

We have added detail and clarification to the methods for this section. The culture of c-kit+ hematopoietic cells is routinely performed in the presence of recombinant ST6GAL1 to inhibit the production of granulocytes. We have previously tested and compared various conditions of cell isolation in this assay, including using lineage depleted bone marrow, c-kit+ bone marrow, and lineage-depleted, c-kit+ bone marrow. We have found that the purity of the LK isolation is comparable among these methods, and typically use only c-kit conjugated magnetic beads to isolate the relevant cells. Further, we have successfully reproduced the suppression of granulopoiesis with recombinant ST6GAL1 in cultures of either Lin-/c-kit+ and c-kit+ cells. We also would note that Sca-1+ cells would exclude the granulocyte/monocyte progenitor (GMP), which we have previously identified as the relevant cellular target of extrinsic sialylation to inhibit granulopoiesis.

8) Figure 4 Show a2,3SiaT activity analysis and describe how this specific activity was assayed. Do the levels of the a2,3 enzyme also rise to those found in serum from WT mice? How many chimeras were assayed?

What was the efficiency of chimerism? Were the numbers of hemopoietic cells of each subtype similarly represented? These data should be included in the manuscript, preferably in Figure 4.

What does n=5 relate to in Figure 4D? Is each symbol representative of a mouse?

The a2,3 SiaT activity has been added to the figure and clearly indicate that levels are steady in the blood after bone marrow transplantation. In addition, we have included resting wild-type average + SD values for both a2,6 and a2,3 sialyltransferase activity for reference (these are 0.1764 + 0.044 and 1.368 + 0.270, respectively). These values are derived from a total of ten mice from three independent experiments. Please note that a2,3 activity that was measured was restricted to the formation of Sia(α2,3)-Gal(β1,4)GlcNAc- from the SNA-unbound fraction of total sialyltransferase activity. The protocol using Gal(beta4)GlcNAc-o-Bn has been used for the past two or more decades, and the nature of the sialyl linkage, e.g. α2,3 or α2,6, with respect to SNA reactivity has been structurally validated previously (e.g. Lee et al., 2014). Blood-borne enzymes construct no other sialyl forms (aside from α2,3 or α2,6) has been found on Gal(β4)GlcNAc-o-Bn, based on HPLC and MS structural product analysis (Sundlov-Lee et al., 2017).

Data shown in Figure 4B were originally the mean representing 4-5 individual chimeras in each group, pooled into smaller subsets of 2-3 animals. Data from individual animals have now been generated and shown in the figure for clarity. The chimerism of host-derived CD45.2+ cells was comparable within the BM of both treatment groups. A complete tabulation of the frequency of the examined host-derived cell types as a fraction of total CD45.2+ cells has been created as well. These data are now included as Figure 4—figure supplement 1, and show that all parameters are comparable between experimental groups. Each symbol in Figure 4 is representative of an individual mouse.

9) Figure 5. The methods for quantitating femur co-localization data need to be more completely described. Were the ROI data obtained from each femur of each mouse? The correlations would be more robust with more data, especially of high SNA ROIs.

The methods have been improved for this section, particularly in regard to the selection of ROIs. In addition, multiple additional biological replicates have been added to this figure in order to increase rigor, including additional high SNA ROIs. As noted in response to reviewer 1, the strength of observed correlations between B cells and SNA reactivity decreased with additional data collection but remains statistically significant. Data for megakaryocytes have been removed from this manuscript as we deemed this lies outside of the focus of the present manuscript, which is about B-lineage cells as a source of extracellular ST6Gal-1 (see our response to reviewer 1)

10) Figure 6B cells are clearly not the only source of serum a2,6SiaT. This should be discussed.

How many mice were used to generate Figure 6B versus 6C?

The original data in Figure 6 was generated with n=3 mice per experimental group. The Discussion has been modified to consider the relative contribution of B cells and other cells to the blood level of ST6Gal-1. In response to other questions, we have performed a further marrow transplant experiment using ST6Gal-1 null recipients, and re-establishing the hematopoietic compartment with either B lineage competent/incompetent, and/or ST6Gal-1 competent/incompetent. This new data, presented as the second half of the new Figure 4, clearly established the role of B cells in releasing circulating ST6Gal-1. The old Figure 6 became redundant data, and therefore we have removed it from the present manuscript. Please see our response to reviewer 1 on this matter.

11) Figure 7. The bone marrow H&E from the SNA-low patient is poor. Another example should be shown.

The images have been modified to improve visibility of both the ST6Gal-1 DAB stain and neutrophils. Plasma cells and neutrophils have also been indicated within the images for clarity.

12) General.

Complete information on each Ab used should be given, including clone numbers.

Clone number and catalogue number have been added to the Materials and methods section where possible.

How was non-mitogenic FCS prepared?

This was mistakenly written and has been removed. Our FBS from Atlanta Biologicals is not necessarily free of mitogens. We thank the reviewer for catching this.

Define ACK lysis.

This has been rewritten as ammonium-chloride-potassium lysis buffer that selectively destroys red blood cells by osmolysis.

Nomenclature should conform with HGNC convention of ST6GAL1 for protein, St6gal1 (italicized) for the mouse gene, and ST6GAL1 (italicized) for the human gene.

This has been modified throughout the text where specifically appropriate. However, when generic references are made to ST6Gal-1 expression or function, the text has been left as “ST6Gal-1”.

Reviewer #3:

[…] While there are a number of tantalizing observations herein, the issues with the data shown and the conclusions provided reduce enthusiasm to this contribution to a very important field of study. For example, it is not clear that the data in Figure 7 has anything to do with the rest of the manuscript. Also, in several cases, the number of biological replicates was 2, with only 1 being shown, which is inadequate. Further detailed comments are below:

- In Figure 1, the authors are using cancer/immortal cell lines to assess ST6Gal1 and BACE1 expression, but the relationship of these values to normal primary cells in vitro and in vivo is unclear. As pointed out by the authors, cancer is associated with robust changes in glycoforms. Moreover, culturing in standard media contains more glucose than what is normal in vivo, which is also associated with changes in glycosylation. These data have little value in understanding the biology of ST6Gal1 in B cells in the absence of studies using primary cells (human or mouse) for confirmation.

We acknowledge the reviewer’s critique that the analysis of ST6GAL1 and BACE-1 expression levels does not necessarily extrapolate to primary B cells. However, the goal in this figure is not to characterize the natural expression of these genes in B cell populations (which was performed recently in another publication, Irons et al., 2018), but to determine if cells of the B cell lineage are capable of secreting ST6GAL1, given its expression. The relevance of this observation to in vivo, primary B cells is tested in Figure 4, wherein the presence or absence of B cells in donor bone marrow is seen to directly alter blood ST6GAL1 and non-self cell sialylation. Furthermore, although we agree that glucose in standard media can alter the production of glycan sugars, it is less clear to us how this would alter the expression of sialyltransferases.

- In Figure 1, the authors also make generalizations about the relative expression levels in B cells from different anatomical locations/developmental stages. But again, these are cancer/immortalized cell lines. Such connections require much more data and multiple cell lines, not just one, for each stage being represented.

We have scaled back any claims regarding the relative expression levels in B cells at different developmental stages. Our goal is to evaluate ST6Gal-1 secretion in cells of the B lineage in general, and we selected multiple cell lines at multiple stages of origin in order to present an unbiased sampling. However, we appreciate the reviewer’s valid critique. Given the later focus on multiple myeloma and bone marrow plasma cells, we have analyzed two more plasma cell stage cell lines (U266 and ARH77), which both express and secrete ST6Gal1 enzyme. Thus, in our sampling, 3 of 4 plasma cell lines express and secrete enzymatically active ST6Gal1.

- Also in Figure 1, the authors state that the Western blot "confirms" the mRNA levels, but this is a mischaracterization. The protein level (Figure 1B) does not mirror the mRNA level (Figure 1A). Moreover, the protein level was seemingly measured only once, hence no error bars in the graph (Figure 1B). The Western blots are also missing visible size standards.

We have more accurately described the results and refrained from the language pointed out by the reviewer. We have also repeated the data in multiple immunoblots in order to include variation in the data, as suggested. The size standards have been enlarged and bolded for visibility.

- Figure 1C is missing repeat experiments and statistics.

We thank the reviewer for bringing this salient point to our attention. We have observed this on a number of occasions, and we showed a representative sample. Moreover, we feel that the activity of the released ST6GAL1 (Figure 1D) is more relevant to the central topic area of this report. While we agree that the presence of the full-length form implies that the protein is still membrane-embedded and is consistent with its localization in a vesicle. This might suggest exciting new possibilities in ST6Gal-1 transfer and extrinsic sialylation between different cells. Our observations are consistent with both the 50kDa and the 42kDa forms being catalytically active, but we feel it is beyond the scope of this manuscript for further mechanistic dissection into the origin and nature of the 50kDa form. Results and Discussion have been expanded to address this

- Figure 1D is missing controls with 2,3 and 2,6 recombinant sialyltransferases to demonstrate the rigor of the assay, and the completeness of the SNA capture. The authors also describe the SNA-agarose precipitation as "column chromatography", which is somewhat of a mischaracterization.

The reference to chromatography has been removed. For our sialyltransferase activity assays, we regularly use both blank saline and serum from ST6Gal-1 KO mice as negative controls, on top of which the values given are calculated. Recombinant ST6Gal-1 protein is also used to calibrate the calculation so that a2,3-sialyl product in this reaction is set at zero. In relation to the data presented, the KO serum control is thus set to 0 fmol/min*ul and a stock recombinant rat ST6Gal-1 enzyme gave an activity of 3800 fmol/min*uL at concentration of 2ug/ml. We do not utilize an a2,3-sialyltransferase to verify the specificity of the a2,3-sialyl product or completeness of SNA capture, since the focus of our work is on a2,6-sialylation and the structure captured by SNA lectin has previously been demonstrated by us to be highly specific for a2,6-sialylation (Nasirikenari et al., 2014).

- In Figure 2, the images are nice, but a measure of fluorescence intensity of the cells under the varied conditions (and with biological replicates) using flow cytometry would provide statistical power and quantitation.

We have repeated this experiment with HepG2 cells in suspension and analyzed SNA reactivity by flow cytometry. Average SNA values with biological replicates are now included in the figure.

- Figure 3 is confusing. First, the amount of change in SNA staining does not correlate with the amount of ST6Gal1 being produced by each B cell line (compare changes in Figure 3B with the protein levels in Figure 1B). This calls into question whether the effect of the B cells is truly because they produce ST6Gal1, or if their presence in the culture alters the metabolic state of the HSPCs such that their glycosylation changes. Indeed, one can see a decrease in a2,3-Sia with the RPMI8226 cells, even though they are releasing an a2,3-Sia transferase (Figure 1D). Similarly, the MM1 cells seem to produce a lot of a2,3-Sia transferase activity, and yet no change in glycosylation is seen in the HSPCs. Does this mean that these STs cannot add Sia to the cells? The authors should demonstrate a dependence on ST6Gal1 in these assays by knocking out ST6Gal1 in each line and repeating the experiment.

We acknowledge that there is an imperfect correlation between ST6Gal-1 expression within the B cell lines and the degree of SNA reactivity gained in the murine HSPCs. However, we would bring the attention of the reviewer to several points. Firstly, the increase in a2,6-sialylation only occurred within B cell lines that we have observed to be secreting ST6Gal-1. Secondly, the reactivity towards SNA has been previously reported to be among the most specific lectin-based methods of analyzing glycosylation, with high affinity to a2,6-sialyl-Gal-B1,4-GlcNAc structures, which are only known to be produced by ST6Gal-1 and ST6Gal-2. Thirdly, the ST6Gal-1 deficient HSPCs would be unable to self-sialylate, even if metabolic changes had occurred secondary to the co-culture with B cells. Therefore, we find it unlikely that endogenous sialyltransferase expression within the HSPCs is driving the observed changes in SNA reactivity. However, in order to more rigorously address this, we have generated 2 CRISPR-Cas9 versions of the MM1.S cell line, in which ST6Gal-1 expression was significantly reduced. The data supporting the creation of these modified cell lines is included in the supplementary figures. Data generated from the co-culture of these new cell lines with St6gal1-KO LK cells has been added to Figure 3, and demonstrate that dependence on ST6Gal-1 expression for both changes in SNA reactivity and Gr-1 expression.

- In Figure 3D, is it possible that the anti-GR1 antibody binding if impacted by sialylation? Does GR-1 transcription (mRNA) change as well?

We thank the reviewer for this salient question, and have addressed it both here and within the relevant section of the Results. The Gr-1 epitope consists of both Ly6C and Ly6G proteins, both of which are not predicted (by sequence) to contain any N-linked glycosylation sites. Both proteins contain GPI-linked anchors that likely contain glycolipid structures, which ST6Gal-1 is not known to modify. Further, there have been no publications reporting N-linked glycosylation of either protein. Sialylation of IgG can occur on a single N-linked glycan on the Fc portion, which affects Fc receptor binding, but not Fab antigen recognition. Furthermore, altered expression of signaling, gene expression, and acquisition of Gr-1 has been demonstrated in our previous publication detailing the effects of ST6Gal-1 extrinsic sialylation on G-CSF sensitivity in granulocyte progenitors (Dougher et al.).

- Figure 4B is lacking a comparative control of WT mice not subjected to any bone marrow transfer/chimerism. This is important to provide the relative contribution of the hematopoietic compartment to the total ST6Gal1 activity in the extracellular space (marrow and plasma).

We have included the average for 10 adult WT animals from 3 independent experiments in this panel, demonstrating that the hematopoietic compartment can elevate blood a2,6-sialyltransferase activity to levels comparable to (or even higher than) wild-type mice. These average values and standard deviation are shown in blue for reference.

- Figure 5B is lacking statistical power (2 biological replicates). There are too few data points to make a meaningful correlation. Additional repeats should solidify the findings easily. Moreover, and perhaps more importantly, the term "co-localization" is not being applied in a clear way. For example, how close do the cells have to be to be called "co-localized"? Simply within the region drawn on the image? If so, there are cells in Region 1 (panel A) that are closer to Region 2 than they are to the left side of Region 1. Thus, the quantitation here is largely arbitrary and without a specific definition (i.e. within X µm). Finally, it would help if these images were done with Z-axis detail.

We thank the reviewer for these salient comments, and have expanded the data in this figure to include more biological replicates. In addition, the specific methods involved in selecting ROIs, defining colocalization, and quantification. Unfortunately, we have not yet optimized the technology and methods necessary to visualize the Z-axis.

- For Figure 6, what are the differences in cellularity between the WT and µMT bone marrow, other than the lack of B cells? Could changes in the cell population that accompany the lack of B cells be playing a role? Also, why would it take 6 weeks for the ST6Gal1 activity to become different, and what happens after 10 weeks, since the curves seem to be converging as they were in earlier weeks?

The hematopoietic capacity of μMT mice, an established and frequently used line to address a wide range of questions relating to B cell function, is well-documented. Therefore we saw no reason to revisit this question here. Nevertheless, we have also characterized the general hematopoietic populations between WT and μMT marrow, blood, and splenic cells. In general, aside from the reduced B cells, we saw only an increase in T cells. However T cells are an extremely minor population in the bone marrow. Regarding the differences manifesting at 6 weeks, we have observed that reconstitution of the hematopoietic compartment to the stage of the mature B cell occurs approximately at 4 weeks post-reconstitution, so that the ST6GAL1 derived from this lineage may only significantly contribute to blood levels from this point onwards. Existing ST6GAL1 activity in the blood before then likely reflect traditional sources such as the liver. We have performed a repeat experiment with additional controls, as suggested by reviewer 1, and included it as the latter part of the new Figure 4. The old Figure 6 is now redundant data, and given the issues pointed out by the reviewer of ST6Gal1 from other tissues, is now removed.

- In Figure 6, the authors state that the Ighm-/- hematopoietic cells within the blood and bone marrow show increase SNA, but the graph in Figure 6C for the blood indicates a lack of statistical significance in this change. This also highlights another problem. Why is data from only one experiment shown? Multiple biological replicates plotted/averaged together should be shown.

We have performed a similar experiment in St6gal1-KO mice to address reviewer 1, as well as to strengthen the robustness of the original data. The old Figure 6 is removed, and the new data appended to end of as second part of the new Figure 4. Statistical significance issue is addressed.

- The connection of the data in Figure 7 to the rest of the manuscript is not clear. There is no functional connection made between ST6Gal1 and the granulocyte observation. In addition, the data in Figure 7D are difficult to see (the H&E). If the number of granulocytes from patients was the readout, flow cytometry or automated hematological cell counting should be performed to provide a more quantitative measure of cellularity.

We thank the reviewer for pointing out the lack of conceptual connection between Figure 7 and the rest of the work. The observation that extrinsic sialylation influences granulocyte production has been extensively studied in our group over a period of almost fifteen years. The data in this manuscript is not meant to define the mechanism of this phenomenon, but simply to demonstrate a novel context in which it occurs. The quality of the pictures in this figure have been improved. Unfortunately, the samples available in our Institute’s biorepository only contain 15 formalin-fixed pre-treatment samples and do not include banked specimens amenable to cell counting or flow cytometry. Aside from a lengthy and time-intensive prospective collection of patient samples, this kind of analysis would not be feasible. Furthermore, cytology analysis by trained pathologists is the original and gold standard for bone marrow hematologic cell quantification.

https://doi.org/10.7554/eLife.47328.014

Article and author information

Author details

  1. Eric E Irons

    Department of Molecular and Cellular Biology, Roswell Park Comprehensive Cancer Center, Buffalo, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  2. Melissa M Lee-Sundlov

    Blood Research Institute Versiti, Milwaukee, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8290-8586
  3. Yuqi Zhu

    Department of Chemical and Biological Engineering, University at Buffalo, Buffalo, United States
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  4. Sriram Neelamegham

    Department of Chemical and Biological Engineering, University at Buffalo, Buffalo, United States
    Contribution
    Resources, Supervision, Visualization, Methodology
    Competing interests
    No competing interests declared
  5. Karin M Hoffmeister

    Blood Research Institute Versiti, Milwaukee, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Joseph TY Lau

    Department of Molecular and Cellular Biology, Roswell Park Comprehensive Cancer Center, Buffalo, United States
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Methodology, Writing—original draft, Writing—review and editing
    For correspondence
    joseph.lau@roswellpark.org
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5128-2664

Funding

National Institutes of Health (R01AI140736)

  • Joseph TY Lau

National Institutes of Health (R01HL089224)

  • Karin M Hoffmeister

National Institutes of Health (K12HL141954)

  • Joseph TY Lau

National Institutes of Health (K12HK141954)

  • Karin M Hoffmeister

National Cancer Institute (CA076056)

  • Joseph TY Lau

BRI Director's Fellowship Award

  • Melissa M Lee-Sundlov

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by grant R01AI140736 (to JTYL), R01HL089224 (to KMH), K12HL141954 (to KMH and JTYL). The core facilities of Roswell Park Comprehensive Cancer Center used in this work were supported in part by NIH National Cancer Institute Cancer Center Support Grant CA076056. Additional support includes BRI Director’s Fellowship Award to MML-S. We would like to thank Jon Wieser for the computational analysis of bone marrow imaging.

Ethics

Animal experimentation: Roswell Park Institute of Animal Care and Use Committee approved maintenance of animals and all procedures used, under protocol 1071M.

Senior Editor

  1. Satyajit Rath, Indian Institute of Science Education and Research (IISER), India

Reviewing Editor

  1. Jamey Marth, University of California, Santa Barbara, United States

Reviewer

  1. Jamey Marth, University of California, Santa Barbara, United States

Publication history

  1. Received: April 2, 2019
  2. Accepted: August 10, 2019
  3. Accepted Manuscript published: August 13, 2019 (version 1)
  4. Version of Record published: August 28, 2019 (version 2)

Copyright

© 2019, Irons et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 736
    Page views
  • 147
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Immunology and Inflammation
    2. Microbiology and Infectious Disease
    Chung Park, John H Kehrl
    Research Article
    1. Human Biology and Medicine
    2. Immunology and Inflammation
    Allison M Owen et al.
    Research Article