Immune dysregulation in Glycogen Storage Disease 1b - a CyTOF approach

Glycogen Storage Disease type 1b (GSD1b) is a rare disease manifesting as hypoglycemia, recurrent infections and neutropenia, resulting from deleterious mutations in the SLC37A4 gene encoding the glucose-6-phosphate transporter. The susceptibility to infections is thought to be attributed not only to the neutrophil defect, though extensive immunophenotyping characterization is currently missing. Here we apply a systems immunology approach utilizing Cytometry by Time Of Flight (CyTOF) to map the peripheral immune landscape of 6 GSD1b patients. When compared to control subjects, those with GSD1b had a significant reduction in anti-inflammatory macrophages, CD16+ macrophages, and Natural Killer cells. Additionally, there was a preference towards a central versus an effector memory phenotype in multiple T cell populations, which may suggest that these changes stem from an inability of activated immune cell populations to undergo the appropriate switch to glycolytic metabolism in the hypoglycemic conditions associated with GSD1b. Furthermore, we identified a global reduction of CD123, CD14, CCR4, CD24 and CD11b across several populations and a multi-cluster upregulation of CXCR3, hinting at a potential role of impaired immune cell trafficking in the context of GSD1b. Taken together, our data indicates that that the immune impairment observed in GSD1b patients extends far beyond neutropenia and encompasses innate and adaptive compartments, which may provide novel insights into the pathogenesis of this disorder.


Introduction
Glycogen Storage Disease type 1 (GSD1), also known as Von Gierke disease, is an autosomal recessive [1] disease, characterized by repeated episodes of hypoglycemia due to inability to break down glycogen storage. The prevalence of GSD1 ranges from 1 in 20,000 [2] to 1 in 400,000 [3] depending on study and the ethnic background of subjects. GSD1 can be further divided into two categories, GSD1a, resulting from a mutation in the glucose-6-phosphatase catalytic subunit 1 (G6PC1) [4], leading to a de ciency of glucose-6-phosphatase catalytic activity, and GSD1b, resulting from mutations in the glucose-6phosphate (G6P) exchanger SLC37A4 [4], blocking the transport of G6P into the endoplasmic reticulum. Both mutations lead to an inadequate conversion of G6P into glucose, impacting glycogenolysis and gluconeogenesis pathways, with subsequent hypoglycemia and buildup of glycogen and adipose tissue in liver and kidneys [4]. Clinical manifestations are typically noticeable after 3 to 4 months of age and include hepatomegaly, nephromegaly, hypoglycemic seizures, and failure to thrive [1], [4]. Patients with GSD1 also suffer from short stature with thin extremities, delayed puberty and a protuberant abdomen caused by the massive hepatomegaly [4]. Interestingly, a signi cant number of patients with GSD1b suffer from in ammatory bowel disease (IBD), typically with a Crohn's disease phenotype [5], [6].
A unique phenotype in patients with GSD1b is reoccurring infections, that traditionally have been attributed to either associated neutropenia or dysfunctional chemotaxis and intracellular bacterial killing of neutrophils. The cause of neutropenia in GSD1b patients remains unclear. Evidence suggests that the decreased neutrophil abundance and activity is a consequence of increased apoptosis of neutrophils in affected individuals [7]. However, the neutropenia alone cannot solely explain the immunode ciency state in this disorder, since GCSF therapy doesn't necessarily eliminate the risk of infections. Therefore, it is likely that the susceptibility to infections is likely attributed to additional effects on other immune subsets. As such, Melis and colleauges[8] were able to observe lymphopenia in patients with GSD1b, but not in GSD1a patients, with a reduction in CD4 + T cells, CD8 + T cells, and natural killer (NK) cells, compared to controls.
We hypothesized that patients with GSD1b exhibit a broad range of immune dysregulation, beside in the neutrophil compartment, leading to an immunode ciency phenotype with in ammatory features. In the present study, we utilized Cytometry by Time of Flight (CyTOF) to perform deep immunophenotyping of peripheral blood mononuclear cells (PBMCs) obtained from 6 patients with GSD1b and 3 control subjects to show dysregulation in several immune populations in patients with GSD1b.

Subject Recruitment
Page 3/12 The study was approved by the local IRB committees. After informed consent was obtained blood samples were collected from participants.

Collection and isolation of PBMCs
PBMCs were isolated by density gradient with Lymphoprep™ (Stemcell™ Technologies) per vendor instructions [9].

CyTOF Staining
Cells were stained with a cocktail of antibodies per our previously published protocol [10]. In short, cells were stained with Rh103 for viability, washed, blocked with Fc-Block, and incubated with the cocktail of metal-coupled surface antibodies (Supplementary Table 1) for 30 minutes at room temperature. Cells were then xed in 1.6% formaldehyde and stained with Ir-DNA intercalator solution. Cells were resuspended in water containing 1:10 dilution of EQ beads and run on a Helios CyTOF machine (Fluidigm, South San Francisco, Calif) at the Yale School of Medicine CyTOF Core.

Data analysis and Dimensionality reduction
Raw Flow Cytometry Standard (FCS) le were normalized using the Helios2 software and imported into Cytobank (Cytobank Inc.), followed by manual gating of cells positive for both DNA dyes (191Ir_DNA1, 193Ir_DNA2) of the Intercalator solution applied after xing (cf. 3.2). From these double positive events signals with a 140Ce intensity > 1x10 3 were excluded as bead signals. Single viable cells were identi ed by two consecutive gating steps excluding all events with a relative event length > 40 and an intensity > 3x10 1 for the 103Rh intercalating dye applied before xation (cf. 3.2). CD45 + events were selected as all events with an intensity for 89Y > 1x10 1 and exported as a separate FCS le for each GSD1b patient and each control. The obtained les were imported into the R 4.1.1 based shiny plugin cytofkit2 [11], [12],. The data loading settings were chosen as follows: Merge Method ceil, Fixed number of cells from each FCS le 50,000, Transformation Method cytofAsinh. Dimensionality reduction was performed following t-Distributed Stochastic Neighbor Embedding (t-SNE) algorithm with a perplexity of 30, with maximal 1000 iterations and 42 was utilized as Seed. Clustering was carried out with Rphenograph with a k value of 30: Cellular Progression was set to NULL. Data was graphed in GraphPadPrism 9.2.0 (GraphPad Software). A twotailed unpaired t-test was used for the comparisons between groups.

Mean metal intensity calculations and exclusions
The mean metal intensity (MMI) values of individual cells calculated from the FCS les and output by cytofkit2 were averaged for each marker for each patient, with surface markers showing an expression value below 0.5 MMI in both groups being excluded from further analysis, as they are within the range of background signal. For the remaining expressed surface markers, a fold change was calculated as MMI of GSD1b divided by MMI of controls followed by a two-tailed unpaired t-test to identify markers signi cantly changed in individual clusters. CD45 and CD326 (EpCAM) were included in the marker analysis to observe any potential batch dependent effects, as expected none of the annotated clusters showed CD326 expression; for all but one cluster no signi cant change in CD45 expression was observed.

Patient cohort
We enrolled 6 patients with GSD1b and 3 controls for this study. Gender, age, in ammatory condition, current medications, absolute neutrophil count (ANC), and c-reactive protein (CRP) levels are provided in Table 1. A deleterious mutation in SLC37A4 gene was identi ed in all GSD1b patients (Table 1), and all had an ANC below 1000 cells/ mm 3 , indicative of neutropenia. The average age for the control group was 16.3 years compared to 8.2 years for the GSD1b patients (p = 0.039). Two GSD1b patients also had concurrent IBD, and two patients had signs of bronchiectasis.

Reduction in total NK cells and macrophage/monocyte populations characterizes GSD1b
To characterize the immune landscape of the peripheral blood of patients with GSD1b vs. controls, we subjected all leukocytes (CD45 + cells) to unsupervised Rphenograph clustering. Automated clustering of the CD45 + populations identi ed 33 unique populations. The clusters were manually annotated based on average expression values of each marker in each cluster of combined patient samples displayed in the expression heatmap generated by cytofkit2 (Fig. 1A). Utilizing this method, we were able to identify multiple clusters of naïve T cells including: CD4 + T helper (T h ) cells, CD25 hi CD127 lo regulatory T cells (T reg ),

cells, CD8 + cytotoxic T cells (T c ), a population of activated and non-activated T cells and memory T cells. The B cell compartment was represented by naïve B cells, pro-B cell, activated B cells, plasma cells (PC) and
anergic B cells. The innate compartment accessible with our methodology consisted of classical pro-and anti-in ammatory macrophages (MΦ), CD16 high and CD16 low monocytes, monocyte precursors, CD15 high and CD15 low neutrophils, CD16 high and CD16 low NK cells, a small granulocyte population (Basophil, Neutrophil, Neutrophil -myeloid derived suppressor cells (MDSC)), and innate lymphoid cells (ILC) (Fig. 1A-C). Cluster 32 could not be assigned to a known immune population because of its atypical high expression of CD11c, CD3 and CD19 and likely represented an immune aggregate or an artifact of unsupervised clustering. Given the low overall abundance of this cluster (0.095% in GSD1b cases and 0.13% in controls), it was labeled as unidenti ed and excluded from further analysis.
To get an overview of the obtained data and determine the differences in the composition of major immune subtypes, the identi ed clusters were then arranged into superclusters based on their marker similarities ( Fig. 1B-C (3), granulocytes (13, 18, 28, and 19) and ILC (29 and 31) (Fig. 1B).
NK and myeloid cell percentage changes in GSD1b are population speci c We then determined if there were signi cant changes at the cluster level. The analysis of innate cell demonstrated a signi cant reduction in clusters 6, consisting of anti-in ammatory MΦ, and cluster 17, CD16 high monocytes ( Fig. 2A). The large decrease in anti-in ammatory MΦ observed may explain the overall decrease in the MΦ & monocyte population described above (Fig. 1B). For the NK populations, cluster 7, consisting of CD16 high NK cells, was signi cantly reduced (fold change GSD1b/control = 0.39, p-value = 0.038, resulting in a global decrease in the number of NK cells, Fig. 1B, Fig. 2A, Fig. 2C). The loss of CD16 expression on NK cells is usually associated with their activation [13] and cluster 7 CD16 high NK showed signi cantly lower CD16 levels in GSD1b patients (fold change GSD1b/control = 0.496, p-value = 0.026), indicative of increased activation of NK cells. However, no relevant or signi cant increase was observed in CD16 low NK clusters. While the overall neutrophil populations were similar between patients and controls, cluster 28, Neutrophil-MDSC, was signi cantly reduced in GSD1b patients (fold change GSD1b/control = 0.359, p-value = 0.0283, Fig. 2B). Nevertheless, assessment of the neutrophilic compartment after a gradient preparation is of limited value and may not provide an accurate estimation of cell composition.

GSD1b patients exhibit high central memory and low effector memory T cells levels
On the adaptive immunity side, two T cell clusters were altered between GSD1b subjects and controls. Cluster 11, representing central memory T helper 1 cells (cmT h 1), was increased (fold change GSD1b/control = 3.54, p-value = 0.015, Fig. 2A [14] in healthy controls and GSD1b was 3.4 times higher in patients with GSD1b, with a con dence interval of 94.61% (p = 0.054, Fig. 2B). Additionally, we observed a trend in the overall reduction of memory T cells and an increase in naïve T cells in the GGD1b subjects with an increase in the ratio of naïve to memory T cells in GSD1b subjects ( Fig. 2A, Fig. 2B).

Global downregulation of myeloid markers and upregulation of CXCR3 in GSD1b
To assess differential expression of surface markers between the groups, we evaluated the MMI in all the clusters and plotted changes found signi cant in Fig. 3A (Supplementary Figure S1B shows all marker fold changes). Although every cluster demonstrated some differences in marker expression between the two groups and almost all markers were altered in at least some of the clusters, we identi ed a few markers that were consistently downregulated in a majority of clusters among patients with GSD1b, including CD123 (IL-3Rα), CD14 (receptor for lipopolysaccharide-lipopolysaccharide [LPS] binding protein complex), CCR4 (receptor for CCL17 and CCL22), CD24 (ligand of P-selectin, SELP), CD11b (myeloid lineage marker, subunit of integrin CR3) and CD127 (IL-7R). Interestingly, CXCR3 (Receptor for CXCL9, 10, 11), which is involved in tra cking of effector T cells, was signi cantly upregulated across multiple clusters (Fig. 3B) in GSD1b patients.
In addition, cluster 6, consisting of anti-in ammatory MΦ, exhibited decreased expression of CD163 (fold change GSD1b/control = 0.54, p-value = 0.002, Fig. 3A). CD163 is a cysteine-rich scavenger receptor on monocyte lineage cells involved in immune regulation and tissue homeostasis and has been shown to be strongly downregulated on MΦ and Monocytes in response to proin ammatory signalling [15], which may be consistent with the ndings in the GSD1b patients, and may indicate impaired function of this anti-in ammatory macrophage subset. Cluster 27, activated emT c , showed higher expression of CXCR3 (fold change GSD1b/control = 1.78, p-value = 0.01, Fig. 3A), which is considered to represent an activation marker for CD8 + T cells; for this effector memory T cell population, it likely indicates activation and targeting to sites of infection [16]. The largest observed neutrophil population, cluster 18, CD15 high -expressing cells, had an increase in MMI for CD45RO (fold change GSD1b/control = 1.763=, p-value = 0.02, Fig. 3A) and a decrease for CD45RA (fold change GSD1b/control = 0.735, p-value = 0.03, Fig. 3A). CD45RO is usually found on non-activated, resting neutrophils, while CD45RA is considered a marker of activation [17], which may suggest that this neutrophil population is less activated in GSD1b patients.

Discussion
To date, the etiology of the immunode ciency state and hyper-in ammatory intestinal condition observed in GSD1b remain unclear. Here, we utilize CyTOF technology to provide an in-depth overview of the innate and adaptive immune landscape in patients with GSD1b. Although our sample size was small, analysis of cell proportions in GSD1b patients showed little variability, indicating a robust assessment of signi cant changes not driven by variation within our groups. We were able to detect and assess 32 immune populations in both the innate and adaptive compartment, illuminating changes in GSD1b at a previously unprecedented depth.
Neutropenia is a key characteristic in patients harboring mutations in patients harboring deleterious mutations in SLC37A4 leading to GSD1b. Although blood sample processing using lymphoprep gradient eliminates most circulating neutrophils, we still found small remaining neutrophil populations that showed upregulation of CD45RO and downregulation of CD45RA in patients with GSD1b, indicating an impairment in neutrophil function or lack of activation. A link to increased apoptosis in these neutrophils has been suggested in the past, along with a possible inability to switch to a glycolytic metabolism after activation in a hypoglycemic environment [18]. Furthermore, we show major changes in other immune cell populations. In line with the ndings of Melis et al.
[8] we observe a decrease in NK cells in GSD1b patients. The mechanism how GSD1b causes NK cell reduction is unclear, but might also be related to impaired glycolytic function in a hypoglycemic enviorment [19]. This is similar to the effect reported for lymphopenia and impairment of Warburg shift for T cells in GSD1b[8], although NK cells do not seem to undergo the switch to glycolytic energy generation as rapid as T cells [19]. As such, we observed a down-regulation in CD16 in NK cells in patients with GSD1b, likely due to the failure to sustain expansion and effector function in a hypoglycemic environment.
The strong reduction in anti-in ammatory MΦ (M2-like) is not explainable with the hypoglycemic environment being insu cient for sustaining a stable population after activation, because anti-in ammatory MΦ have been reported to rely mostly on oxidative phosphorylation (OXPHOS) to meet their needs while also maintaining some glycolytic activity [20], [21].
However, it is di cult to determine the association between the in ammatory state (M1-/M2-like) and effect of MΦ based on their surface marker expression (HLA-DR high vs. HLA-DR low in our case) alone. Some MΦ populations (e.g. M1-like) are known to undergo a switch to Warburg metabolism [20], [21]. Further characterization of MΦ populations, in ammatory status, and mode of activation in GSD1b patients might therefore be required.
We also observed a trend of increasing percentages of naïve T cells in GSD1b, consistent with the observation by Melis et al. of T cell dysregulation due to an impairment in glycolysis. However, we could not con rm their observation of a reduction in overall memory T cell levels. More striking is the observed transition from effector memory to central memory in T cells, a change that is predicted to result from the Warburg shift. Effector memory T cells, especially emT c (cluster 27), were signi cantly reduced in GSD1b and, in contrast to cm T cells, are reported to rely heavily on glycolysis to meet their energy needs [22]- [24]. On the contrary, central memory T cells (cluster 11) were signi cantly increased in GSD1b and are reported to preferably perform oxidative phosphorylation, mainly fueled by fatty acid oxidation. [22]- [24] Our data shows a shift away from effector memory and towards the central memory T cell populations in GSD1b, which can be explained by the hypoglycemic conditions and impaired glycolysis in GSD1b favoring central memory over effector memory T cell phenotype.
We observed no signi cant changes in the B cell compartment similar to Melis et al.[8]. Naïve B cells rely on OXPHOS with only a moderate increase of glycolysis upon activation [25], [26], explaining their ability to differentiate and form stable effector and memory populations, even in hypoglycemic conditions.
The reason for the global downregulation of myeloid markers in GSD1b remains unclear. The upregulation of CXCR3 in multiple clusters might indicate increased activation and targeting to in amed tissue in GSD1b, especially the lung [27].
Although one explanation of the altered immune landscape is the hypoglycemia environment, this by itself cannot solely explain the immune phenotype observed in patients with GSD1b. An important comparison should be made to patients with GSD1a, resulting from G6PC mutations, that also suffer from severe recurrent hypoglycemia, but do not develop neutropenia or IBD. Therefore, it is likely that deleterious SLC37A4 mutations impair immune function via additional mechanisms. Immunometabolomic Seahorse data on GSD1b T cells is sparse, but Melis et al. observed impaired glycolysis engagement in GSD1b patients [8]. Our results are in line with this conclusion, that glycolysis could explain the low levels of effector memory T cells and high levels of central memory T cells in GSD1b patients.

Conclusion
In conclusion, the study presented here is to our knowledge the rst to provide a detailed view of the immune landscape in GSD1b using a high-parameter panel. The alterations observed demonstrate that the underlying dysregulations of the immune system go much deeper than the commonly recognized neutropenia and affect multiple populations across all arms of the immune system, likely in uencing the clinical manifestations of GSD1b such as recurrent infections and IBD. We speculate that the mechanism of the immune dysregulation in GSD1b is partially mediated by a failure of anaerobic glycolysisdependent immune populations to differentiate and proliferate in the hypoglycemic environment commonly present in GSD1b patients, leading to immune de ciency associated with this rare disease. Further studies should address more speci cally whether these immune defects are the result of metabolic defects, and whether speci c interventions (e.g. SGLT2 inhibitors) ameliorate also these immune defects.

Competing interests
The authors have no relevant nancial or non-nancial interests to disclose.

Author Contributions
Dror Shouval supervised patient enrollment as well as collection, processing, and shipment of the obtained samples. Blake McCourt was responsible for performing the CyTOF stain. Arne Gehlhaar performed all steps of subsequent data analysis, literature research and composed the manuscript. Eduardo Gonzalez Santiago provided editing for the manuscript. Baruch Yerushalmi and Galina Ling provided patients' samples and clinical meta-data. Lael Werner assisted in sample processing. Liza Konnikova planned and supervised the entirety of the project and provided scienti c guidance.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics Approval
The study was approved by the local IRB committees.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.