Gaucher Disease: Basic and Translational Science Needs for More Complete Therapy and Management

“In the middle of difficulty lies opportunity.”

Albert Einstein

1. Introduction

Since the approval of enzyme therapy (ET) for Gaucher Disease (GD) type 1 (GD1) 30 years ago, the medical community has focused on the transformative improvements in the lives of affected individuals and their families. Although these achievements cannot be overstated, the focus of ET, substrate synthesis reduction therapy (SRT), and the associated clinical science has been directed primarily to manifestations of non-neuronopathic, N370S mutation-related GD1’s and GD3’s hematovisceral effects. Over these three decades, the authors have experienced, collectively and individually, the broad spectrum of clinical and genotypic variants of GD in the “N370S Western world” as well as in the much larger populations of North Africa, the Middle East, India, and China. Indeed, there are a great many more GD patients globally than reflected in mostly European-centric literature. In these other populations, the predominant disease mutations are mostly L444P or other similar ethno-specific variants that predispose to neuronopathic GD (nGD). Our collective longitudinal experiences with several thousand individuals afflicted with GD variants, as well as the authors’ and others’ basic mechanistic and translational research, underscore the need to more fully address the totality of GBA1 mutation-related disease(s). The transformative therapeutic effects in GD1 have not yet been experienced by those afflicted with central nervous system (CNS) involvement, i.e., GD2 and GD3, or GD1 patients and heterozygote carriers with increased lifetime risk of Parkinson disease (PD) and Lewy body dementia (LBD).

Here, we attempt to crystallize some aspects of three broad areas of unmet needs and questions. The basic science of the lysosomal glucocerebrosidase (EC 3.2.1.45, lysosomal acid β-glucosidase (LBG)) enzymology and the heterogeneous tissue and cellular involvements, within and among the GD types. LBG will be used here instead of glucocerebrosidase (GCase), since this enzyme has at least two causal pathogenic substrates, glucosylceramides (GCs) and glucosylsphingosines (GSs), and may have extralysosomal (1) and non-enzymatic functions. Translational and clinical sciences related to finding tools to facilitate decisions on when, how, and why to initiate treatment. These would necessitate understanding the determinants of the great phenotypic variation within and among the GD types and the reversibility of specific tissue responses to therapies. Clinical and ethical considerations related to prognostic difficulties in individual GD patients, the impact on the families, and the skill sets and resources needed by the treating physicians, genetic counselors, and other medical professionals involved in the care of affected individuals and families.

2. What are the Impacts of Gaps in the Basic Science of GD on Understanding Pathogenesis and Therapeutic Developments?

2.1. Background:

For the last decade, the visceral organ and CNS cellular heterogeneities have been detailed at molecular and histological levels. Such heterogeneity may be fundamental to the pathophysiology and, potentially, the therapeutic efficacy in GD. This section will focus on several aspects of GD in need of additional research, including: A) for wild-type (WT) and mutant LBGs, the nature of their molecular microenvironment(s), substrate preferences, and the functions of associated proteins, B) the heterogeneous origins and biologies of the major affected visceral and CNS myeloid cells, C) regional neuronal and microglial interactions contributing to GD CNS disease, and D) the pathobiology of the GD bone involvement. In particular, what are the temporal and spatial evolutions of molecular events that create and change the GD cellular and regional environment? Importantly, the initiating disease cause differs from the emergent propagation mechanisms arising during disease progression and are mechanistically distinct from the LBG deficiency.

2.2. LBG Structure and Function:

About 30 high-resolution structures of WT LBG have been solved with and without occupancy of its single active site (https://www.ncbi.nlm.nih.gov/structure/?term=GBA1)[e.g., (2–5)]. Such details are known for only one mutant LBG, p.N370S (a.k.a., p.N409S). The p.N370S and WT structures showed little difference, except for the former having a more rigid loop near the active site (2). Many other point mutant LBGs have decreased stability, thereby impeding isolation of mutant enzymes in sufficient purity for in vitro structural work.

2.2.1. Approaches, Needs, and Impacts:

Comparative LBG WT and point mutant structures could provide insights for development of pharmacologic chaperones that are not active site-directed inhibitors, i.e., allosteric sites that would not inhibit catalytic activity and could improve intracellular stability and activity (3). These approaches likely will require structures with active site-bound substrate analogues to “reveal” allosteric sites as well as more detailed structures of the active site. These would include accommodation of the sphingosyl and fatty acid acyl chains (FAACs) of GC. Curiously, in vitro GS exhibits non-competitive inhibition kinetics, suggesting that although GC and GS are hydrolyzed by LBG, different parts of the catalytic site are needed to accommodate the hydrophobic chains uniquely present on GCs or GSs (6–8). Also, such structures could provide insight into substrate preferences of LBGs, particularly accounting for the ~100-200-fold decreased turnover number (k_cat) for GSs vs. GCs (6). Importantly, GSs are more soluble compounds than GCs, and they induce inflammation and cellular toxicity (9–13), as well as being an antigen in ~25% of sporadic and in GD-associated multiple myeloma patients (14).

Mouse models showed the in vivo substrate preferences of several Gba1 point mutations with tissue-specific accumulation of GCs with variable FAACs (14). This implies that differential tissue involvement may be modulated by the catalytic properties of the mutant LBG (e.g., (15, 16)). This has import for therapeutic decisions based on the mutant LBGs’ preference for specific substrates. For example, a mutant LBG with preference for GCs with shorter FAACs would lead to greater involvement of organs expressing GCs with long-chain FAAC. The therapeutic import of this preference would include GD with manifestations in organs with decreased accessibility to ET via intravenous (IV) or gene therapy (17), i.e., lung alveolar-macrophages. In GD-like mice, lung alveolar-macrophages accumulate long-chain FAACs. If lung alveolar-macrophage-accessible pharmacological chaperones were used to increase this mutant LBG activity, they would be less likely to be successful compared to an SRT with similar accessibility. The chemical heterogeneity of GCs and/or GSs could also provide insight into such regional pathology and biomarkers of tissue disease status, as suggested by studies of differing ceramides from various cell types (18).

2.3. Functional Consequences of Higher-Order Structures:

The intracellular structures of the WT and mutant LBGs during transport to and residence in the lysosome are not rigorously defined and have important pathophysiologic implications. There is a potential for differing effects if the WT and mutant LBGs are monomers, dimers, or higher-order structures. Nearly all disease-causing missense mutations in GBA1 produce LBGs with some residual catalytic activity (7), and, in vivo, heterozygote carriers would be expected to have at least ~50% of WT activity. In heterodimers of WT and mutant LBGs, the latter could have dominant-negative effects leading to <50% activity of WT homodimers. Such effects could be exacerbated by a pharmacologic chaperone with preferential effects, e.g., greater stabilization, on the mutant vs. WT LBG. Such dominant-negative and/or preferential effects could have consequences for heterozygote carriers, e.g., increased risk of PD (see below).

LIMP2 is the receptor for trafficking WT and most point mutant LBGs (19, 20) to the lysosome in many cells (19–21). The 1:1 stoichiometry of LIMP2 and WT LBG suggests a monomeric form of WT LBG during targeting from the ER to the lysosome. After the lysosomal pH-induced dissociation, LBG’s structure is not established. LIMP2 may not deliver LBGs to the lysosome in all cell types (22, 23). Indeed, loss-of-function mutations in LIMP2 do not cause GD (24), but rather a progressive myoclonic epilepsy (25, 26) or its variant, action myoclonus-renal failure syndrome (27), implying other lysosomal trafficking mechanisms for LBGs. A more complete biology of LIMP2 remains to be defined, e.g., LIMP2 dimerization when binding phospholipids, including phosphatidylserine (an activator of LBG), for delivery to lysosomes (28).

2.3.1. Approaches and Impacts:

The higher-order structures of LBG may be accessible by cryogenic electron microscopy or other imaging approaches within the cellular and lysosomal environments. Furthermore, delineation of LIMP2’s capacity to deliver WT and mutant LBGs to the lysosome has implications for therapies. Most LBGs are not secreted in any significant amounts from cells expressing LIMP2, and at least a 5- to 10-fold LBG overexpression may be needed to saturate LIMP2, leading to secretion into cellular media or plasma (17). This also impacts hematopoietic stem cell transplantation (HSCT), leading to differentiation into cells for macrophage repopulation that express WT levels of LBG. The lack of secreted enzyme from such cells would preclude cross-correction of neighboring cells.

2.3.2. Intracellular Complexes of LBGs and Other Proteins:

LBGs interact with additional proteins for their lysosomal delivery, stability, and catalytic activity, including saposin C (29–33), progranulin (34), and HSP70 (35). Saposin C, an activity and stability enhancer of WT and some variant LBGs (7), is an 80 amino acid protein that is proteolytically cleaved in the lysosome from its precursor, prosaposin, that contains three other saposins (e.g., (29, 36–38)). Each of these saposins has specificity for and different mechanisms of enhancing lysosomal sphingolipid hydrolase functions (39, 40). Saposin C deficiency causes a nGD variant (29) indicating in vivo specificity for LBG. Except for the requirements of negatively-charged phospholipid interfaces (e.g., (32)), saposin C’s mechanism(s) of affecting LBGs’ catalytic activities are not well understood. To date, direct interactions, i.e., co-crystallization, cross-linking, or binding, of saposin C and WT LBG have been suggested only in silico (41). Saposin C has fusogenic properties for negatively-charged phospholipid liposomes (32) and has similar effects on T-cell membranes for lipid antigen presentation (42). Saposin C is trafficked to the lysosome within prosaposin via the mannose 6-phosphate receptor in a heterodimer with progranulin, a protein with multiple small subdomains (43). Progranulin loss-of-function mutations impair prosaposin processing and result in decreased LBG activity (44). Progranulin has multiple growth-promoting and immunosuppressive effects (45). Along with HSP70, a chaperone for WT and mutant LBG ex vivo (35), progranulin may form a cytoplasmic complex or aggregate with LBG and LIMP2 under stress conditions (35).

2.3.3. Approaches and Solutions:

Further rigorous characterization is needed for the effects and interactions of these, and potentially other proteins on the stability, trafficking, and activity of LBG variants. These results should define therapeutic targets of direct modifiers of LBG function and biomarkers for GD and provide additional understanding of their functions, as well as insight into the heterogeneity of GD.

3. Diversity of Myeloid Cell Responses to LBG Abnormalities

3.1. Background:

GD pathology and pathogenesis have been dominated by a focus on monocytes and macrophages in most visceral tissues without significant attention to their properties and diversity beyond the simplistic M1 and M2 classifications (see (46, 47)). The inflammatory aspects of GD have been catalogued; some of the serum and bodily fluid proinflammatory cytokines and chemokines are biomarkers for disease activity. More recently, with efforts directed to treating and elucidating the CNS and bony manifestations, focus has turned to the ontogeny of involved cells and the immune system’s critical role as components of disease propagation and reversibility (48, 49). The findings in mouse models and human cells and samples implicate the innate (autoinflammatory) and adaptive (autoimmune) immune systems as integral to GD visceral and CNS disease propagation (50, 51). Furthermore, the “healing phases” of the inflammatory responses can lead to irreversible end-stage “repair,” e.g., fibrosis and gliosis. If uncontrolled, these reparative phases may progress to organ failure or damage even with elimination of the causal initiation, i.e., substrate accumulations.

3.2. Incomplete Cellular and Molecular Understanding of Therapies:

ET of GD1 has produced remarkable degrees of clinical recovery in many cellular renewing visceral soft tissues (52–58). Advanced bone diseases (fractures, infarctions, and severe osteoporosis) appear to be mostly irreversible, but prevention has been successful before signs of significant clinical involvement (53). Few detailed data on the tissue effects of ET from humans are available; these are mostly from genetic mouse models of GD. Imiglucerase or velaglucerase had identical biochemical and histologic effects on soft tissues (59). Whole genome transcriptome sequencing revealed differing splenic levels of GATA1 (erythropoietic lineage specification, velaglucerase) and PU.1 (myelopoietic lineage specification, imiglucerase) pathway genes (60). These interesting findings may reflect underlying myeloid bias in GD pathophysiology and differential mannose receptor-mediated uptake by various cells; additional studies are needed. For example, osteoblast defect of GD is reversed by imiglucerase, although this cell type does not exhibit mannose receptors (61). The greatest therapeutic challenge of all could be viewed as getting the therapeutics targeting root cause to cells and tissues in sufficient amounts to prevent irreversible complications.

3.2.1. Approaches and Solutions:

Although the clinical import of these basic findings, if translatable to humans, is unknown, these findings do highlight the need for characterization of the spatial-temporal molecular events and histopathologic or imaging analyses for disease staging. These are a major challenge for translational and clinical sciences as well as clinical trial design and the evaluation of current and future therapies. Once the downstream pathways beyond the primary metabolic defect come into play, the use of ET and SRT (via various proteins, genetics, iRNAs) and small-molecule approaches may not be sufficient to address these disease processes in their entirety. Understanding the timing and causes of the cellular variability in involved organs and their accessibility is essential, but poorly understood. Importantly, the nature of myeloid cell responses to insults across regions of organs makes rational development of primary (directed at reducing pathogenic lipids) and adjunctive therapeutic modalities essential for personalized and more complete disease management.

3.3. Complexity of Tissue or Regional Specific Disease Effects:

The details of the initiating, propagation, and reparative effects of the involved myeloid cells depend upon their intrinsic or acquired responses within the tissue regions (e.g., (62, 63)). Single-cell RNA (scRNA) and single-nucleus RNA (snRNA) sequencing (62, 64–66), together with detailed morphologic and organelle composition data (e.g., lysosomes (67, 68)), show heterogeneity of macrophages between visceral organs and brain perivascular macrophages (PVMs), as well as resident microglia within regions of brain parenchyma (e.g., (46, 69–76)). In the CNS, resident microglia and PVMs are derived from embryonic yolk-sac-derived precursors. In contrast,. tissue macrophages in other organs that are derived from bone marrow hematopoietic stem cells (HSCs) (74, 77) (Fig. 1). Therefore, caution should be exercised in applying the concept of HSC-targeted therapies to correcting brain immune dysregulation. Self-repopulating resident macrophage-like cells include those of the lung alveoli (78), liver Kupffer cells (70, 71), those in specific splenic regions (76), partially for osteoclasts (79, 80), and brain parenchymal microglia (81) (Fig. 1). These cells display significant morphological and metabolic heterogeneity between and within tissues (82). This heterogeneity implies topographically different tissue responses to pathological insults, including pathogens or metabolic insults, e.g., GD variants. Yolk-cell-derived brain microglia and PVMs provide additional heterogeneity of these myeloid cells, with major functions in maintenance of neuronal homeostasis and detecting foreign agents. Furthermore, these myeloid cells modulate CNS vascular permeability and, therefore, permeability of the blood brain barrier (BBB) (69, 73). Indeed, such BBB leakage can lead to chronic microglial-mediated inflammation of CNS parenchyma (83). The spatial-temporal structural, functional, and GC FAAC differences and downstream bioactive lipids (84) in these various myeloid populations are critical to delineating the pathophysiology, propagation, and therapeutic reversibility of regional tissue contributions to GD.

Figure 1:

Schematic of myeloid cell ontogeny and development from the yolk-sac or bone marrow stem cells. Through complex lineage specification, several visceral organs derive resident myeloid cells from precursors originating in the fetal liver or yolk-sac, e.g., microglia and PVM of Virchow-Robinow spaces of blood vessels in the brain. Each of these has unique properties and gene expression profiles that differ from other tissue macrophages. Similarly, bone marrow-derived myeloid cells undergo complex differentiation with major roles in the innate and adaptive immune systems, including elicitation of anti-GC antibodies in GD with activation of the C5a/C5aR1 complement axis. Osteoclasts have primary functions in bone resorption and remodeling.

3.3.1. Approaches and Solutions:

Elucidation of such disease-related events may provide new therapeutic targets and potentially tissue regional-specific biomarkers that could address the significant challenges in improving therapeutic outcomes. Such data provided insights into the evolution of inflammatory processes in mouse models of Alzheimer disease and other neurodegenerative diseases, particularly the role of microglia (62, 85–93). Unique disease-associated microglia (DAMs) showed specific spatial identity markers, localization, and metabolic pathways that may limit neurodegeneration via modulation of independent (activation) and dependent (down-regulation) of TREM2 DAM checkpoints (62, 91, 93).

Scant data are available for nGD, including the types of microglia (88), but whole-genome regional mRNA and miRNA sequencing from genetic GD-like mice has shown similarities, albeit not identity, to patterns in such DAM expression profiles (94). These single timepoint end-stage studies had partial concordance with the DAM Stage 1 TREM2-independent phase of microglial activation (62, 91). Across the cerebellum (CB), cerebral cortex (CO), midbrain (MID), and brainstem (BS), there was variable expression of several genes, e.g., Axl (increased in CO, CB, and BS only), Lpl (increased in BS and CB only), ApoB (increased in CO, MID, and BS only) (95), and RIPK3 (94, 96) (increased only in BS). However, very elevated pyrin, Tyrop2, and Tmem119 mRNA levels were found in all four regions (94), indicating generalized microglial activation. TREM2 is an important modulator during neurodegeneration (92, 95, 97–99). The modest elevation of TREM2 in all regions suggests that the microglia were in transition between DAM stage 1 and 2 signatures (91), even in the presence of significant microglial activation and neuronal death (necroptosis) (94). Many other proinflammatory genes were highly expressed, including many complement receptor subunits, e.g., C3aR and C5aR1, indicating a generalized innate immune activation (100, 101). These GD-like mice were also treated with a BBB-penetrant, active site-directed, pharmacologic chaperone (isofagomine) that in cultured cells increased the lysosomal activity and protein of the mutant LBG present in these mice (94). Curiously, with large doses of isofagomine, no changes were detected in CO, CB, BS and MID LBG activity or CNS GC or GS levels. In comparison, there were large and broad decreases in proinflammatory gene expression, as well as a ~33% increase in lifespan, with significant behavioral improvements. Such a decoupling of substrate accumulations and off-target decreases in the total CNS proinflammatory state provides proof-of-concept that other BBB-penetrant potent anti-proinflammatory agents could positively affect, but not eliminate, the CNS involvement in nGD without affecting the root cause (LBG deficiency). Additional work is needed to further validate this concept using pure mouse models of nGD, as the above data were obtained in a homozygous GBA1 mutation knock-in model in saposin C-deficient background. Together with therapies targeting the root cause of pathogenic bioactive lipid accumulation, e.g., enzyme enhancement or SRT, adjuvant therapies targeting decreases in propagator pathway activities in the proinflammatory environment, e.g., RIPK1/3 (96) or C5aR1/C5a axis (51) inhibitors, may further mitigate propagation of nGD. In these GD mouse models, amplification of C5aR1/C5a axis drives increased glucosylceramide synthase (GCS) expression. As a result, this pathway could reasonably be targeted via substrate reduction approaches (inhibitors or iRNAs) or pharmacologic C5aR1 blockade, thereby addressing a root cause of bioactive lipid accumulation (51). Future studies should carefully examine targeting root-cause of GC and GS accumulation vs. combination therapies involving targeting of downstream pathways (Fig. 2).

Figure 2:

Schematic of the stress responses of cells to untreated Gaucher disease. The specific cellular response varies with cell type, but innate immune cells (macrophages of various ontogenies, microglia, and dendritic cells) evoke many of the indicated pathways leading to multiple different types of programmed cell death. The major features result from lysosomal stresses due to excess GCs ad GSs leading to disruption of the autophagy/lysosomal system from destabilization of the lysosomal and autophagosome membranes. Such disruption leads to dissociation of mTORC1 from the lysosomal membrane making it inactivate for the phosphorylation of TFEB/TFE3. These factors then enter the nucleus and activate the transcription of genes involved in autophagic, lysosomal biogenesis and enzymes, peroxisomal biogenesis and enzymes while the pathways for lipogenesis and adipogenesis are blocked by in inactivation of Lipin 1. Components of the mitochondria are also disrupted with resultant disruption of the oxidative pathways (not shown). The overall effects lead directly or indirectly to proinflammatory activation through cell dependent cell death pathways. The damage to various organellar membranes is shown as dashed circles and the resultant effects on substrates and precursors are shown in dashed lines. The diminished or altered function of the various organellar pathways indicate the degrees of metabolic disturbance. Dissection of these cellular-specific pathway abnormalities should provide for identification of potential modifiers of GD expression within and among phenotypes.

4. Impact of Regional Microglial and Neuronal Effects in GD

4.1. Therapeutic Implications of Tissue and Regional Cellular Involvement in GD:

The similar paucity of detailed CNS spatial-temporal histopathology, cell type evolution, cell surface receptors, substrate effects, and organ-crosstalk hampers targeted development of treatments, clinical trial design, and more accurate prognostication. Across studies, the most consistent findings are lipid-laden macrophages, “Gaucher cells,” in the brains of GD1, GD2, and GD3 on the abluminal surface of brain blood vessels, particularly arterioles, i.e., the Virchow-Robinow spaces (102–104). These are derived from resident PVMs, not from bone marrow myeloid cells, and normally function to modulate inflammation (including systemic inflammation); when activated, they express the mannose receptor (69, 73). Understanding of PVM-aberrant functions in GD is limited. Before the advent of ET, increased BBB permeability was shown by blood plasma proteins within PVMs, microglia, and CNS parenchyma of Norbottnian GD3 patients (103, 104). Additional molecular and functional changes are unknown. Also unknown are potential changes in replacement and repair of in situ PVMs and microglia as well as the potential role of bone marrow-derived myeloid cells with different properties than the resident myeloid cells, as shown in other disease states. In one study of a chemically induced GD mouse model, such CNS-infiltrating myeloid cells were not present (63). This is an important topic for further investigation.

The integrity of the BBB has implications for IV-administered agents. In untreated Norbottnian GD3 patients, the PVMs were filled with GCs whose longer FAAC contents indicated visceral origins. In comparison, increased parenchymal cell GCs had shorter FAACs, indicative of CNS ganglioside origins (12, 104, 105). The PVM GC levels were greatest in splenectomized patients and partially correlated with the rapidity of dementia, although this was more directly related to greater parenchymal cell GS levels and neuronal death (104). These data imply that the excess GCs supplied to the brain exceeded a threshold of substrate hydrolytic ability, producing more rapid progression. In addition, autopsy data from a single GD3 patient treated for 11 years with mannose-terminated WT LBG showed Gaucher cell clearance of liver, spleen, and bone marrow GCs, but not in lung alveoli or CNS PVMs (106). Macrophages in all visceral organs, including lung alveolar “Gaucher cells,” expressed the mannose receptor, but the CNS PVMs and microglia (106) did not express this marker of activation (73). The presence of lung alveolar “Gaucher cells” after many years of ET likely resulted from this compartmenťs inaccessibility via a vascular-alveolar barrier. However, PVMs are directly exposed to the vascular compartment, and their persistent pathology was, at least in part, related to the lack of the mannose receptor for sufficient enzyme uptake. Since CNS PVM “Gaucher cells” are universal in GD variants, their progressive involvement by GC storage, even with visceral improvements on ET, could lead to chronically altered BBB permeability and activation of neighboring microglia with progressive gliosis (102), as well as luminal narrowing of involved blood vessels and potentially distal ischemia.

In human GD2 and GD3 CNS, regional and specific layers exhibit non-uniform neuronal losses via necroptosis and neuronophagia (102, 104), as highlighted by losses of hippocampal CA2-CA4 neurons, but not in CA1 (102). These findings were not found in all autopsy reports (106, 107) or series (104). Neuronal losses correlate to some degree with the presence of activated microglia and CNS GS, but not GC, levels (104). Whether this or similar selectivity of other neuronal layers of the cerebral cortex are related to differences or similarities in types of locally activated microglia has not been studied in human tissues, e.g., layers 3 and 5 and/or the partial (104) or total (106) losses of cerebellar Purkinje cells. Data are also absent on the critical aspect of GC and GS flux in individual brain parenchymal and immune cell types.

4.1.1. Approaches and Solutions:

Confirmatory studies require detailed autopsy analyses of the morphology, receptor expression, activation status, and molecular characteristics of CNS PVMs and microglia in additional GD patients treated with IV ET, SRT, or IV or intra-CSF gene therapy. The lack of sufficient and appropriate human brains or more human-like mouse model brains inhibits the needed gene expression and detailed cellular subtype analyses within and across the CNS. The use of unbiased scRNA and snRNA total transcriptome sequencing or proteomic analyses could provide critical data, even in autopsy material from end-stage GD patients. The potential molecular analyses of organoids containing induced pluripotent stem cell (iPSC)-derived astrocytes and neurons (108) as well as microglial cells or the transplantation of normal or GD human neurons into mouse GD-like brains (109) may provide critical insights into these complex neurodegenerative pathways and their potential for mitigation.

4.2. The Impact of Systemic Inflammation on CNS Pathology and Function:

An extensive literature on neurodegenerative diseases documents the influence of systemic inflammation on microglial activation and exacerbation of existing cognitive deficits and, if chronic, on neurodegeneration (110–112). Classically, GD variants exhibit variable, but at times massive, macrophage expansion as well as proinflammatory involvement in visceral organs. The “Gaucher cells” within these soft tissues can secrete large amounts of cytokines and chemokines, an “autoinflammatory storm,” arising from activation of the innate immune system (50, 51, 94). An “autoimmune” response, triggered by autoantibodies to GCs and potentially to GSs, and activation of the C5a/C5aR1 complement axis perpetuated this immune storm by increased synthesis of GC via upregulation of GCS (51); this pathway is also activated in CNS microglia of nGD-like mice (51, 94).

Microglia are “primed” by neurodegenerative disease processes to respond more rapidly and severely than those not exposed to such influences (110, 112). This priming is less specific than, but akin to, the anamnestic response of the adaptive immune system. The microglia within the CNS of patients with GD2 and GD3 are likely to be primed by intra-CNS innate immune activation associated with progressive GC and GS accumulations. This leads to over-response to systemic proinflammatory cytokines and chemokines and induction of additional CNS neuroinflammation. Primary early-onset progressive CNS disease is not a component of GD1. Patients not receiving ET, however, can experience severe fatigue, depression, poor school performance, and, potentially, some cognitive impairment. The symptoms and sign decrease or disappear within months of initiating ET, with concordant effects on the visceral cytokine and chemokine levels (113). Treatment of the systemic disease by ET in patients with GD3 leads to a lessening of the proinflammatory state. The CNS improved function correlates with these therapeutic effects. Such effects were indicated in a patient with GD2 who received IV ET, which does not cross the BBB, from 4 days after birth to her death at 15 months. She showed great improvement in her visceral organ involvement, and she walked, was interactive, recognized her mother, and understood a few words, albeit with progressive CNS disease (114). In comparison, her untreated brother, who died at 10 months, had a classical GD2 disease course without achieving any significant developmental milestones. Their post-mortem CNS pathology was indistinguishable (107). Such CNS effects were evident from the authors’ experiences with a large population of patients with GD3 receiving ET, who exhibited improvement in saccadic initiation defects or cognition, concordant with major decreases in visceral disease (Egyptian Expert Medical Committee, personal communication). These defects recurred during a prolonged period when these patients received suboptimal doses of ET due to production shortages. These dose reductions were concordant with return of visceromegaly and the clinical proinflammatory state.

4.2.1. Approaches and Solutions:

Such findings make clear the need to control the visceral disease prior to documenting the primary CNS effects of a therapy. In addition, the direct involvement of the C5/C5aR1 axis and RIPK1/3 in the CNS GD process or propagation (51, 96) provides targets for additional systemic and potentially CNS therapeutic interventions. Investigations to determine whether targeting C5/C5aR1 axis is needed require further validation, i.e., exploring C5a serum levels in GD1 and response to ET/SRT (see Translational and Clinical Science).

5. Challenges in the Pathobiology of GD Bony Involvement

5.1. Background:

The mechanistic bases for the imbalances between bone absorption (osteoclastic) and deposition (osteoblastic) leading to GD bone disease are poorly understood (115). Much of the bone disease can be prevented by early initiation of ET, which decreases bone marrow GC and, potentially, GS (53, 116, 117). In children, before peak bone mass develops, there is significant reversal of osteopenia. Evidence suggests that this reversal may be mediated via decreasing the GS impairment of osteoblast function (49, 61). Although ET is macrophage mannose receptor-targeted, the osteoblast defect in GD iPSC-derived osteoblasts was reversed by imiglucerase ET, consistent with bone density responses in younger patients (116). In older adults, many of those with GD1 continue to develop osteopenia or osteoporosis even after long-term ET. This may reflect susceptibilities to multifactorial etiologies of osteopenia, as occurs in the general population. The myeloid ontogeny of osteoclasts (79, 80) suggested that these cells could have a greater rate of bone resorption in GD, possibly related to sphingolipid abnormalities. A controlled, 2-year clinical trial tested whether bisphosphonates would inhibit this osteoclastic resorptive activity in stably ET-treated men and women with GD1 (ages 18-50 years) (118). Because osteoclasts were thought to be activated in GD1 bone, the expected result was a significant improvement in bone mineral density (BMD). However, the same BMD increases and plateau of effect were found in similarly treated non-GD post-menopausal women with osteoporosis. Other bony abnormities present in the GD bisphosphonate group remained unchanged (118). These data indicate that the osteoclastic resorptive activity in GD was not the major cause of decreased BMD or other bony lesions. These data are consistent with studies in GD-like mice (49), GD-like zebrafish, and ex vivo (61, 119) that found an imbalance between deposition and resorption with major involvement of osteoblast dysfunction.

5.2. Disruptions of Bone Metabolism in GD:

Histologic and imaging analyses of GD bone marrow show a complex pathology involving macrophages, monocytes, and adipocytes as well as osteoblasts and osteoclasts (115) and primary effects of GD on HSC and hematopoiesis (49, 120). Indeed, bone marrow “Gaucher macrophages” are a hallmark of the disease. These GC-laden inflammatory macrophages can, without treatment, replace bone marrow hematopoietic cells in a proximal-to-distal progression (121). With this advancement of storage cells, adipocytes, an important component of bone marrow, can be completely replaced (122, 123). Osteoclast precursors mature into multinucleated cells that attach to bone and create lacunae for bone resorption that contain “exteriorized” lysosomal contents (79, 124, 125). These compartments may account for the lack of visible lysosomal storage in these cells. Interestingly, the commonly described radiological feature of bone manifestations of GD, endosteal scalloping, invites speculation of association of these lesions with endosteal hematopoietic niches (120, 126, 127); indeed, impaired hematopoiesis contributes to cytopenia in GD (48). However, inhibition of GCS decreased mCSF-mediated osteoclastogenesis via plasma membrane-raft mediated alteration of the RANKL receptor interaction (128, 129). Similarly, decreased osteoclastic activation via GCS inhibition was shown to ameliorate myeloma bone disease in a mouse model (129). Important data quantifying the density (number/unit volume) of osteoclasts in bone from humans or mice with GD, as well as their respective membrane glucosylsphingolipid compositions, should be a focus of research. Osteoblasts, osteocytes, and adipocytes derive from bone marrow mesenchymal stem cells, and their fate specifications balance production of one cell type vs. the other (130). For example, the specification preference of adipocytes over osteoblasts with aging may lead to osteoporosis (131). The molecular events directing the fate specifications of these cells (127) provide fertile areas for GD research.

The pathology of GD bone involvement is more complex than a paucity of osteoblasts and adipocytes or a resulting imbalance of resorption and deposition of bone matrix. This pathology also includes heterogeneous bone marrow infarction and myelofibrosis, as well as differences between trabecular and cortical bone (115).

5.2.1. Approaches and Solutions:

Characterizations are needed of the inflammatory status of the GD bone marrow, the density of important cell types, and the molecular status and interactions of the major bone cell types (127, 132–134). Indeed, the normal cytokine networks and differentiation pathways have been elucidated for osteoclastogenesis and osteoblastogenesis, as well as the osteocyte immunologic roles (79, 124, 125, 127, 130, 131, 135–137). These provide targets for analyses in human GD for their assessments and prediction of the bone resorptive-deposition balance as well as their roles in osteoimmunological responses in GD. The role of osteoblasts in regulating hematopoietic niches, and therefore blood and immune cells, suggests in GD1, in addition to osteoporosis, a decrease in their ability to suppress hematological malignancies (136). For clinical research, it will be important to advance assessment from an operational definition of bone density and the extrapolated fracture risk in the general population to bone microarchitecture, using high-resolution peripheral quantitative computed tomography (HRpQCT) (138).

6. What Challenges in GD Impact Improvement in Disease Understanding and Therapeutic Developments?

6.1. Background and Unsolved Questions are addressed in the following sections:

Do GD1, GD2, and GD3 represent a continuum of disease manifestations from the same underlying spatial-temporal molecular and cellular mechanisms? What are the bases for the highly variable phenotypes within “types” and between some siblings? How would knowing these impact the identification of patients with GD for treatment and the choice of therapy? What are the criteria and tests for defining reversibility and irreversibility of disease manifestations, and how might these impact the outcome expectations?

6.2. The Continuum Hypothesis and Its Implications:

A continuum of GD, in contrast to a categorical delineation of GD “types,” has become vogue (139–141). A continuum of a common pathophysiology was postulated, in particular, to explain the GD CNS involvement that differed by degree, not kind; it is based on the concept of levels of “residual enzyme activity” (142, 143). This implies that the degree of involvement inversely relates to the amount of enzyme activity present in specific cells, tissues, or regions to meet their metabolic demands. Also, an inherent inverse dependency is implied for “adaptability,” and therefore phenotype, of involved tissues and organisms to various environmental, genetic, and other external insults. This implies that lower “residual enzyme activity” will lead to less ability to respond to internal or external insults, thereby leading to phenotypes being more severe and stereotyped. Superimposed on this consideration is the emerging and unexpected evidence of increased GC synthesis in a setting of inflammation that amplifies substrate accumulation in a setting of LBG deficiency (49, 51). Further evidence of the central role of GC as an immunostimulatory molecule comes from its property as the endogenous ligand for macrophage-inducible C-type lectin (144). Moreover, basic clinical considerations suggest that as splenomegaly progresses, accompanying hypersplenism will add increasing GC load from a rapid turnover of blood cell membranes.

Support for basic concepts regarding residual enzyme activity derives from ex vivo (140) and in silico models (Grabowski, unpublished) as well as analyses of correlations between genotype and gross clinical parameters from hundreds of untreated patients with GD1 (145). GD physiology was simulated in murine macrophage-like J774 cells treated with increasing doses of conduritol B-epoxide (CBE), a covalent inhibitor of LBG, and “fed” damaged RBCs as a source of GC (142). GC accumulated only at LBG activities ≤11-15% of the uninhibited WT levels; GS was not assessed. In vitro K_cat values of LBGs expressed from various alleles showed p.N370S ~14-18% WT and p.L444P ~0.1-1% WT; K_ms were equal to WT, and 84GG produced no LBG (7). Using these parameters, and assuming no dominant negative effects of LBG pathogenic variants, intracellular GC and GS metabolisms were simulated in silico. GC accumulated when LBG k_cat ≤9-14% of WT values. GS accumulated at LBG k_cat ~30% of WT (Grabowski, unpublished). This simulation assumed WT LBG’s substrate preference of 100-200-fold for GC vs. GS (6). The ex vivo and in silico analyses suggested increasing GC or GS accumulations as k_cat was progressively decreased to estimated levels at or below the k_cat of p.N370S and p.L444P. The in silico simulations assumed the same acid ceramidase k_cat for cleavage of ceramide to sphingosine and GC to GS, as in vitro data were not available. Also, the in silico studies are complicated, since GS has in vitro non-competitive inhibitor kinetics (6, 7), whereas GC is a partial inhibitor of GS cleavage (6). Overall, the ex vivo and in silico studies were concordant and indicated thresholds of LBG activity for GC and GS accumulations.

6.2.1. Approaches and considerations:

The J744 results may not apply to other cell types for which the sources of GC are not dependent on phagocytosis, e.g., neurons. The in silico simulations may not reflect actual intracellular flux, since in vitro the kinetic parameters (K_ms, k_cats, and K_i;GS) were obtained in non-physiologic systems, although their relative values, e.g., k_cat,Gs/k_cat,Gc, in the simulations may be more reflective of in vivo metabolic balances. Additional ex vivo substrate(s) flux studies are needed in a variety of cell types whose metabolisms of GCs and GSs may differ from that of cultured macrophages, e.g., non-phagocytic cells with internal sources of these lipids or different types of GC or GS internalization and production. Furthermore, in vitro kinetic analyses show differences in the WT and mutant LBGs kinetic parameters. These include varying preferences for cell, tissue, and region-specific GCs of varying lengths of FAACs, as well as K_i or IC₅₀ values for GS, N-C_n-deoxynojirimycin, CBE, and other inhibitors (7, 8, 146). Some of these differences are mutation-specific, as are the intracellular stabilities and the delivery to lysosomes. Currently there are no reliable and reproducible LBG assays in tissue homogenates or blood-formed elements that directly correlate LBG activity with phenotype. With these significant caveats, such data provide some support for the “residual enzyme activity”-continuum hypothesis. Beyond cellular GC and GS metabolism, it will be important to delineate whole-body metabolism of these lipids, especially GS and plasma flux in serum lipoproteins (147). Early studies showed marked lipoprotein dysregulation that may arise through altered lipid composition related to GC and GS transport and HDL-mediated reverse-GSs transport for biliary secretion (148). Lipoprotein-mediated GC and GS may be surprisingly high, such that it majorly impacts overall phenotype expression.

6.3. Phenotype-Genotype Correlations and the Continuum Hypothesis:

Analyses of clinical assessments provide support for a continuum within, but not between, GD types, i.e., GD1 and GD2 (145, 149, 150). This is evident in patients with GD1 who do not develop primary early-onset CNS disease and have distinct neuropathology, i.e., lack of neuronal involvement, compared to patients with GD2 and GD3 (102).

6.3.1. GD1:

Genotype (N370S/N370S=A, N370S/L444P=B, N370S/84GG=C, N370S/undetermined or rare =D)-phenotype correlations were determined in several hundred untreated individuals with clinical signs of GD1: median age of onset or diagnosis → A>B>C, hepatic and splenic volumes → C>B>A, and percent with any bone manifestations → A~B~C (Fig. 3). The D group showed a wide variation in all parameters, highlighted by the median age of onset or diagnosis of 13 years., similar to the B group. The median for age of onset or diagnosis was ~2 and 2.5 decades later in A than in B and C, respectively. The spread of individual values in A was from 2 to 85 years. In comparison, the values in B and C were more tightly clustered, and D was nearly bimodal. These results are consistent with a genotype rank order of increasing disease severity of C>B>A that was also consistent with larger hepatic and splenic volumes, as well as more significant hematologic abnormalities at earlier ages (145). Within GD1, this ranking order inversely correlated with the degree of disease involvement and the predicted in vivo levels of residual activity, i.e., k_cat of p.N370S>p.L444P>84GG implies p.N370S/p.N370S would be predicted to have twice the activity of p.N370S/84GG, as 84GG is null. p.N370S/p.L444P would be in between, consistent with p.L444P LBG having some activity. A priori, the severity of skeletal disease due its devastating impact would be correlated with the severity of visceral and hematological disease. The most surprising finding is that there is no correlation such that advanced disabling bone complications can occur in patients with minimal visceral or hematologic disease (151). Such a dichotomy of phenotypic expression underscores the need to understand the molecular phenotype of storage cells in the bone marrow vs the viscera.

Figure 3:

Age at diagnosis or detection (years) of GD 1 individuals stratified by genotype based on allele specific analyses. The numbers of patients and mean/median ages are shown on the right for each genotype. The RED box indicates the potential mis-genotyping due to the Δ55 deletion in Exon 9 that would prevent appropriate binding of the primer specific for the N370S allele. The correct genotype in such a case would be N370S/Δ55 corresponding to a protein type of p.N370S/Null, similar to 84GG. The dots represent individual patients. The 1^st and 3^rd quartiles are upper and lower bounds of the boxes, respectively. The horizontal lines within the boxes are the median values.

6.3.1.1. Caveats:

The results are based on allele-specific PCR genotyping, not whole-GBA1 sequencing. In the group testing as N370S/N370S, the mean and median values may have been shifted to greater severity of clinical parameters. This results from using PCR primers that bind in exon 9 of GBA1, the presence of the N370S mutation at the 5’ end of exon 9, and a pseudogene-derived deletion, termed Δ55, within exon 9. This deletion would prevent the binding of the 3’ primer of exon 9 because of the Δ55 allele, and it would only recognize the N370S allele; this would give a false genotype of N370S homozygosity. Like 84GG, the Δ55 allele is null. In analogy to N370S/84GG, these N370S/Δ55 genotypes would be clustered in the youngest members of the “N370S/N370S” cohort, which, if removed, would shift the true N370S/N370S, assessed by whole GBA1 sequencing, to later-onset or diagnosis with smaller livers and spleens. Recently, follow-up chart reviews of hematovisceral findings in 103/440 (treated/untreated) Ashkenazi Jews with GD1 (~80% were N370S/N370S) were reported as stable (152). However, short-term (5-year) direct follow-up of patients focused only on hematovisceral parameters should be viewed in the context of the known dissociation of these parameters from indicators of skeletal disease (153, 154). In these studies, N370S/N370S patients diagnosed symptomatically as well as through screening showed minimal disease activity when viewed in the context of hematovisceral parameters, but with a high burden of skeletal disease, including osteonecrosis. The challenge in the clinic is how to identify N370S homozygous patients with mild, stable hematovisceral disease who are at risk of significant bone manifestations.

In the 1990s, Ashkenazi Jewish heterozygote prevalence data were compared to large clinic populations (Scripps and MSSM). The comparison suggested that that 40-50% of N370S/N370S patients do not come to medical attention due to the paucity of obvious clinical findings (155, 156). During this period, the patients with GD1 in clinics were almost always those with more advanced disease, which suggests a potential for significant overestimation of individuals with N370S/N370S who had subclinical disease or less. Nevertheless, including in these analyses patients who do not exhibit any signs of the disease would further decrease the overall disease severity in this genotype. It could also cause the spread of values in these clinical parameters to become very broad and thus increase the difficulties of determining, a priori, who should be treated. Additional, more complete follow-up data on untreated patients with GD1 are needed even if criteria are met for enzyme therapy. Such data would facilitate assessment of rates of progression in relationship to the genotype. Although insufficient details, e.g., molecular, histological, or detailed imaging analyses, are available to reach rigorous conclusions, overall data are consistent with great phenotypic variation in the N370S/N370S genotype. The bases accounting for such stability, or lack thereof, within and among those patients with GD1 with at least one N370S allele are unknown; they have major import for disease management and implicate genetic and environmental factors beyond in vivo LBG residual activity in disease expression and progression. Identifying potential modifier genes of these phenotypes is both timely and important for disease management. Large cohorts with appropriate replication and control groups are required for such studies, as well as comprehensive timeline assessments, including osteonecrosis or hematovisceral findings, as these may be disassociated in many patients.

6.3.2. GD2:

In GD2, genotype-phenotype correlations are derived from case reports and small data sets, but at least two types have been identified: as the “classical” variant with overt age at onset of ~2-4 months, lack of developmental progress, and, depending on supportive care, death from neurological involvement by 1-2 years, and a “neonatal” (a.k.a., “collodion-baby”) variant with severe CNS disease and a poor skin permeability barrier leading to death at 1-3 weeks (149, 150, 157). Most individuals with “classical” GD2 have an L444P/recombinant genotype; the recombinant alleles are highly variable but contain multiple point mutations, deletions, and/or insertions that derive from the highly homologous GBA1 pseudogene and are null alleles (150). The genotypes of the patients with “neonatal” GD2 are generally null/null, with various combinations of recombinant alleles or other null alleles. These limited data support a correlation within GD2 of genotype and thus residual activity and disease severity.

6.3.3. GD3:

GD3 presents an evolving clinical management and scientific conundrum. Of the multiple different, potentially ethno-specific GBA1 alleles found in individuals with GD3, L444P/L444P is the most common genotype. Unlike the N370S allele that arose as a founder mutation in the Ashkenazi Jewish population (158, 159), the L444P alleles derive from recurrent gene-conversion events between GBA1 and its pseudogene (150). Whole-GSA1 gene-sequencing is needed to exclude recombinant alleles containing L444P or other cis mutations (see (160, 161)), as well as to explore the potential effects of intronic variations.

GD3 presents a broad spectrum of visceral and CNS disease ranging from early-onset severe visceral and CNS disease, of later onset than in “classical” GD2, to phenotypes with less severe visceral or CNS disease. In later-onset GD3, the signs may be delayed to absent into the 4^th decade or later (162). Such variation occurs even in L444P/L444P GD3 patients who can be traced to cousins in the 17^th century (163, 164). Great visceral and CNS variation occurs in same-sex siblings with or without ET (authors’ observations, (165)). ET in GD3 has created additional heterogeneity in the degree and onset of the CNS disease, due to the altered visceral cytokine and chemokine effects on the CNS. However, within the GD3 population, sibling differences in overall disease burden can be stark, as can the degrees, progression, and types of CNS disease (authors’ observations). Indeed, two brothers having same GBA1 mutations had very discordant GD phenotypes. A heterozygous LIMP2 mutation (p.E471G) was associated with myoclonic epilepsy and dementia in one brother and was not present in the neurologically normal brother and controls, including 13 other patients with GD3 and myoclonic epilepsy (160). Longitudinal assessments (up to 20 years) of ET in individuals with GD3, mostly L444P/L444P, showed stable verbal, performance, and full-scale IQs (166). Individual patients’ IQs remained stable regardless of initial IQ. The lack of correlation between ET duration and IQ suggests specific developmental effects of GD3-related CNS disease processes. Alternative measures of cognitive function, e.g., attentive testing, may provide more relevant quantitative data. Other L444P/L444P GD3 patients had variable rates of CNS disease progression, including IQ deterioration (167, 168). Other measures of CNS involvement and progression may facilitate quantitative measures of disease progression; they include saccadometry and optico-coherence tomography of preretinal and retinal lesions (169). Investigations into detailed CNS imaging (diffusion-tensor imaging) and histopathologic or molecular analyses may provide additional tools for disease staging.

This great phenotypic variation within and among families and populations, as well as the deterioration of CNS function with splenectomy (104), implicate yet to be defined additional genetic and environmental factors that have significant impact on GD3’s clinical course. Moreover, these additional factors greatly influence not only the degree but the kind of CNS involvement, which clearly transcend a continuum hypothesis. Identifying specific genetic, metabolic, and/or environmental factors that impact variations in phenotypes within GD type possesses has major import for newborn screening, prognostication, clinical trial design, and therapy.

Within and among GD1 and GD3, the variations in hepatic (170), lung (157, 171–173), and bone (115, 118) transcend GBA1 genotype alone. These may relate to the factors influencing tissue-specific levels of LBG activity toward GCs and GSs that vary in chemical compositions specific to cells, regions, or their ontogenies, as suggested in mouse studies (15) as well as environmental and infectious insults. These would fall outside of a mechanistic continuum for some visceral phenotypes. The mechanistic bases accounting for these differences remain to be elucidated.

GD-like mouse models using LBG inhibition with CBE (51, 96, 174, 175) or Gba1 genetic modification (17, 176–180) have been useful in GD research, but such models are approximations to the human disease; similar human studies are needed. Currently, other than histopathology, no comparable human GD data are available from any tissue, especially the brain. These paucities of information are major challenges in the era of GDs, non-neuropathic and neuropathic, transitioning to chronic diseases with long-term emergent disease complications. Such emergent or inadequately treated manifestations are already evolving in patients receiving long-term ET (e.g., (53, 115, 181)).

7. Some Major Clinical Challenges and Potential Approaches and Solutions

7.1. Major GD Clinical Challenges:

Future-directed practical challenges in GD relate to treatment in GD being based on staging of the disease together with the prediction of disease progression and outcome. These include assessments of the reversibility or irreversibility of disease-specific tissue processes, the increased risk of PD in GD patients and GBA1 heterozygote carriers, challenges in managing patients identified through newborn screening, transitioning of medical care, and counselling for reproductive planning.

How can GD be staged and assessed for reversibility, and how are the degrees of reversal determined in each organ? In soft visceral tissues, the reversal of hepatosplenomegaly and improvements in blood-formed elements are obvious assessments, as are improvements in disease activity markers, e.g., LFTs, chitotriosidase, and lyso-Gbl. In addition to physical exams, various imaging studies provide more detailed assessments of initial and residual disease, including various MRI or CT techniques for lungs, liver, spleen, bone, and bone marrow (115). Hepatic elastography is becoming common, and it could expand to other soft tissues for estimations of fibrosis (182). Tissue histopathology by hepatic biopsy, pulmonary lavage, and bone and bone marrow biopsy is rarely used to quantify fibrosis, cellular compositions, replacement, or degree of mineralization. In addition, there is little if any insight into the causation or treatment of progressive, irreversible kyphoscoliosis in some patients with GD3, which initially seems independent of gross vertebral structural changes. How can this be staged, treated, or stabilized without knowing its basic disease mechanism? An additional emerging feature in a subset of patients with GD3 on long-term ET is massive intraabdominal lymphadenopathy that can result in profound malnutrition, intestinal obstruction, and proteinlosing enteropathy, as well as the proliferation of “Gaucher cells,” i.e., “Gaucheromas” (181, 183). The bases of these profound phenotypes are not understood.

Imaging analyses provide better overall assessments of the heterogenous organ involvement by GD. Since the advent of ET for GD1, these clinical tools have been pivotal in assessing disease improvements. However, more detailed studies are needed in some patients to evaluate less-than-expected ET responses or progression of disease, persistent abnormalities of disease markers, recurrence or resistance of specific disease manifestations, and the development of hematologic or other malignancies (154, 181). Refractory or recurrent anemia or thrombocytopenia resulting from bone marrow myelofibrosis develop in a subset of patients, necessitating a biopsy for diagnosis. Do chronically elevated lyso-Gb1, chitotriosidase, or other “biomarkers” predict ongoing disease activity requiring additional interventions? Elevation of serum C5a in GD is striking, and this topic would advance significantly with formal studies of biomarker validation. Also, other than gross destruction of bone, what degree of disease involvement of this and other visceral tissues signals irreversibility? Could such tools impact the choice(s) of therapeutic approach, including ET or SRT, and eventually genetic or cellular replacement therapy? For the long-term chronic management of treated or untreated GD1, we currently lack biomarkers and tools to assess risks and specific organ involvement for these components of the disease.

The above questions also apply to GD2 and particularly to GD3, but they would need to include development of quantitative tools for staging the CNS disease. Available CNS imaging currently plays little role in the evaluation of these diseases, except for GD3 homozygotes with D409H for the evaluation of developing hydrocephalus (184–186). Detailed neurological, developmental, and appropriate cognitive assessments currently can provide insights into the degree and progression of disease involvement. Biomarkers for assessment of CNS disease activity and neuroinflammation, e.g., lyso-Gb1 or GPNMB, are under investigation for use in clinical trials (187–189). Such biomarkers that correlate with the degree and kind of CNS disease and progression will be essential for the chronic management and treatment of GD3 (and perhaps GD2). Specifically, these are needed for the prediction of reversibility or stabilization and, critically, age of onset of progressive CNS disease.

7.1.1. Approaches and Solutions:

Based on the continuing expansion of the phenotypic variation in GD1 and GD3, additional classifications may be needed based on genotype and biomarkers that correlate with overall disease activity or tissue-specific involvement. In patients with GD3 the evolution of this “neuropathic/neurodegenerative” disease to one with variable visceral and progressive or static (or no) CNS disease necessitates revisiting the “classical” clinical phenotyping (Table 1), with significant impact on lifetime management of the disease. Indeed, if predictable, the phenotype of the longer-lived patients with GD3 without major CNS involvement would have a massive impact on all aspects of their lives. For GD1, except potentially for N370S/N370S, the predictions of the degree of disease involvement and progression can be reasonably based on genotype. Accurate biomarkers of these disease components are needed to facilitate clinical decisioning in the N370S/N370S cohort, particularly in the era of newborn screening. In addition, human brain banks are needed to facilitate exploring the spectra of histopathology and regional and cellular involvement in all variants of GD. These should be supplemented by detailed whole genome transcriptomic and proteomic analyses to dissect the full complexity of the differences in kind of their CNS involvement.

Table 1:

Phenotypes of Untreated Gaucher Disease Subtypes

¹The purpose of the proposed classification is an initial staging of Gaucher disease patients prior to or in the absence of any specific therapeutic intervention(s) and to provide a classification useful in disease prognosis for the affecteds and for genetic counseling for families. Genotyping based on whole GBA1 sequence analyses can supplement the clinical signs noted in the Table. See the text for details on genotyping prognostications for Type 1 variants.

²Excludes Parkinson/Lewy Body disease;

³Not done or not detected;

⁴Dependent on supportive care;

⁵Does not include kyphoscoliosis, which is likely of neuromuscular origins initially and occurs only in a subset of Gaucher disease type 3 patients;

⁶Includes progressive hydrocephalus;

⁷The solid black square represents the WT LBG activity in any tissue, i.e., 100%. In the subsequent table cells show gradients of mutant LBG activity that are postulated to be less (lighter color) in more severe phenotypes and greater (darker color) in less severe phenotypes. The thickness of the arrows represents a total LBG activity and is postulated overall to be rank ordered with GD1>GD3>GD2. For GD3C, the rectangle color and size indicates a similar postulated LBG activity to GD3B.

In addition, detailed analyses of GD visceral organ and CNS involvement should also include either targeted or unbiased lipidomics. The chemical structures of GCs and GSs from visceral tissues and regions of the CNS could provide markers for the degree of CNS region (18, 84) involvement and enable tracking of their response to therapies. Unbiased lipidomics of plasma membranes from GD fibroblasts reveal lipid patterns that were predictive of specific TLR innate immune responses (190). Once determined in tissue, their analyses in plasma or serum or CSF could provide powerful tools for monitoring overall and tissue- or cell-specific involvement. Assessments of the state of proinflammation/neuroinflammation may also provide more general staging of disease activity, as has been suggested for chitotriosidase, MIP-α, and gene networks (125, 191). These reflect the activation status of macrophages and other immune cells, but they are not part of the pathophysiology of GD. Potentially, components of the autoinflammatory and autoimmune systems, e.g., C5aR1/C5a, may provide improved assessments of disease activity, as they appear integral to the visceral and CNS pathophysiology of GD. As GD1 and particularly GD3 evolve as chronic disorders, tissue-specific biomarkers are needed for tailoring and personalizing disease management; such specific biomarkers will be essential to addressing long-term consequences and their management tailored to individual patients.

8. What Are the Mechanistic Bases of Increased Risks of PD and GBA1 Variants?

8.1. Background of PD and GBA1 mutations:

Epidemiologic studies have demonstrated that GBA1 mutations are a major lifetime-risk factor in developing PD. Patients with GD1 patients who are homozygous or biallelic for GBA1 mutations or heterozygote carriers of a GBA1 mutation are at increased lifetime-risk of developing earlier-onset and more rapidly progressive PD compared to individuals with PD without GfiAl-mutant alleles (192–196). Among Ashkenazi Jews, patients with GD1 patients and GBA1-mutant heterozygote carriers are both at 4- to 5-fold increased risk of PD (197). The epidemiology supporting these risks is buttressed by additional studies of increased PD risk with other lysosomal gene variants (198).

8.1.1. Mechanisms of GBA1 Effects on PD Risk:

As part of the “continuum hypothesis,” two mechanisms have been suggested for increased PD risks resulting from GBA1 mutations. First, at a tissue and biochemical level, the loss of substantia nigra compacta (SNPC) dopaminergic neurons results from similar processes in GBA1-variant heterozygote carriers and patients with GD1, i.e., accumulations of excess GCs or GSs mediate “toxicities” that directly or indirectly lead to pathogenic α-synuclein and other protein accumulations. Second, at a cell biologic level, there is a disruption, potentially subtle, of the “autophagy/lysosomal/mTORC1 system (199–201)” and the SNARE ykt6 α-synuclein interactions (202) resulting in abnormal processing, trafficking, and/or stability of mutant LBGs (203) and other proteins. The finding that ykt6 enhanced effects on lysosomal protein trafficking and folding are modulated by inhibitors of its farnesylation led directly to a potential novel therapeutic, i.e., farnesyltransferase inhibitors (202). Which mechanisms are dominant is unsettled, since supporting data for them are indirect; they are under intense investigation and will require analyses of human tissues.

8.1.2. The Continuum vs. Cell Biological Disruption:

The former postulates excess GC and GS in the SNPC and other PD-involved areas (e.g., hippocampus) in patients with GD1 and GBA1-variant heterozygous carriers. The latter suggests a disruption of the “greater lysosomal system” involving potentially subtle changes in oxidative stress, free radical generation, low-level chronic inflammatory effects, or an intrinsic disruptive effect of the mutant LBG in the SNPC dopaminergic and other neurons. Providing basic support for these proposals, in vitro and ex vivo studies show a clear relationship between GC and GS levels and α-synuclein pathogenic aggregation, as well as impairment of LBG processing or trafficking by interaction with α-synuclein monomers (202, 204–208). In vivo studies used genetic mouse models of GD (Gba1 D409V homozygotes, 9V/9V) or of PD over-expressing the A53T mutant α-synuclein in Gba1 WT homozygotes (209–211). Overexpression of AAV9 WT GBA1 in hippocampus of D409V/D409V (abbreviated here -- 9V/9V) and A53T α-synuclein homozygotes showed significant resolution of the PD manifestations (e.g., decreased novel object recognition) and the increased GS accumulation. Also, abnormal α-synuclein deposition improved. Importantly, WT/9V had normal levels of GC and GS and WT novel object recognition but increased α-synuclein, albeit to a lesser degree than 9V/9V. In comparison, Gba1^+/+/Gba1^−/− mice did not have abnormalities in any of these assessments (210). Thus, the excess α-synuclein was found in 9V/9V and Gba1 WT/9V mice and not in Gba1 WT/null mice, no genotypes accumulated excess GC, and only 9V/9V had increased GS in the CNS. These results suggest that some intrinsic property of the mutant 9V LBG, e.g., misfolding, led to a dose-dependent increase in α-synuclein (210). The increased GS, but not GC, in the untreated 9V/9V mice likely results from the LBG substrate preference for GC vs. GS.

Such studies suggest that parts of both proposals may be operative: In GD1-associated PD, the accumulation of substrates would lead to gradual excesses of α-synuclein and increased negative impact on mutant LBG activity (208). In the GBA1 heterozygous carriers, lesser α-synuclein excesses would not depend on substrate accumulation (212–214), but more on disruptive effects of the mutant LBG, e.g., UPR or misfolding, producing subtle defects in the “autophagy/lysosomal system.” Implicit in this proposal would be a difference in the onset and rates of progression, with PD in GD1 having earlier-onset and GBA1 heterozygote carriers having later-onset, slower progression of PD. However, some clinical data suggest that in the Ashkenazi Jewish population, PD in GD1 or GBA1 carriers have similar onset and progression (197), whereas PD carriers of GBA1 variants associated with GD2 or GD3 may have a more rapid disease course compared to the less disruptive N370S-associated mutation (215–217). The mechanistic bases for the GBA1-associated PD remain unsettled.

8.1.3. Findings in Human PD studies:

The results from several human studies have not shown consistent increases in GC or GS in the brains of individuals with GBA1 (heterozygote)-PD (e.g., (213, 214)). Human studies have supported an inverse relationship between age and WT LBG activity levels in brains of individuals with idiopathic PD and of those without (214), as well as sex-related PD SN changes in ganglioside content (218, 219). CNS glycosphingolipid content also displays regional changes during prenatal development, as well as during aging into the 9^th decade (214, 220–222). The differences in CNS regions analyzed for glycosphingolipid and enzyme contents, as well as a paucity of histological characterization in these mostly end-stage samples, likely contribute to inconsistencies of results (220). Analyses conducted in PD SNPC that showed GC levels in GBA1 mutant heterozygotes (n=4) were comparable to the non-PD controls and appeared lower than in idiopathic PD SNPC (214). Although GS was increased in this whole PD cohort, the age effects or relationships to GBA1 genotype were not detailed. Histology was not presented, but such correlations could assist in interpreting the lipid characteristics (i.e., structures) and enzyme levels. Histopathology of SN from patients with PD, who are mostly end-stage, shows a predominance of microglia in varying stages of “activation” and a paucity of dopaminergic neurons (87). These cellular differences might have significantly affected the findings and those in other such studies, which may be dissociated from the earlier PD disease processes.

From the authors’ experiences and neurological data in GD3 (reviewed in (157)), patients up to 40 years of age receiving ET have variable cognitive and other CNS involvement, and some have shown signs that occur in PD, but not all individuals with GD3 (162, 166, 168, 223). Dissecting the mechanistic bases for the clear epidemiologic risk data for GBA1-associated PD, as well as any excess risks for treated nGD patients with prolonged survival, is a major challenge in GD and PD.

9. Significant Clinical Management and Ethical Challenges

9.1. Background:

The clinical and ethical challenges are grounded in accurate and predictive data and tools that are currently unavailable for lifetime decisioning by families, patients, and their physicians. Such data and tools are essential for significantly improving the provision of accurate, scientifically-based information for prognostication and assessing the lifetime health of those impacted by GD and GBA1 mutations.

Up-to-date, unbiased assessments of prognosis, GD phenotype, stage of GD, and availability of specific and adjunctive therapies must be provided to individuals and families with existing diagnoses and to those who are newly diagnosed or detected (via newborn screening) as affected individuals, as well as the PD risk. These are massive disruptions that require large adjustments in life and life planning. Such assessments are best provided by those with significant expertise and experience in GD variants. These experts also greatly value the collaboration with primary care physicians for routine care and with subspecialists and multidisciplinary teams who can provide specialized care for GD complications and co-morbidities.

For example, there are challenges for patients with GD1 and GD3 who have received ET since early childhood and who are now transitioning through adolescence and into adulthood. Some are rebelling, as with patients having DM1, against therapy. These require pre-emptive intervention by appropriate mental health professionals to avoid recurrence of disease manifestations, as might be expected in this population of early-onset progressive GD. Many patients with GD3 have sufficient cognitive abilities to understand the potential for progression of their CNS disease, but specialized interventions may be needed to facilitate and assess their comprehension. Adequate and appropriate intervention will be needed to address their perceptions of disease burden and require significant culturally sensitive medical and family support.

9.2. Transitions of Care

With the advent of treatments, patients with GD1 and GD3 who might have died or developed serious chronic health issues are now surviving into adulthood, with the potential for age-related co-morbidities and disease complications. To address such concerns, providers, especially those caring for pediatric patients, need to address the additional challenges of adolescents and young adults, i.e., transition or transfer of care and reproduction.

9.2.1. Challenges:

As children develop, they should be incorporated into their own care and treatment to the degree that is developmentally appropriate. For normally developing children, the goal may be that they assume primary responsibility for their condition by adulthood. Adolescents and young adults with disabilities may require greater assistance from others. Transition into adolescence may disrupt treatment adherence as a result of adolescents’ emergent autonomy and immature executive functioning, as well as competing activities (224). As patients become adults, they should transition from a pediatric, parent-supervised to an independent, patient-centered health care. Such a transition to an adult model of care may entail a transfer to a new clinicians or clinicians, which can be traumatic (225).

9.2.1.1. Approaches:

Patients may nonetheless fail to take their medications, not keep appointments, neglect to seek care for medical problems, or engage in health-adverse behaviors. In addition to collaborating with patients to reduce the burdens of treatment regimes, providers should promote self-management tasks and processes, such as identifying goals of medical treatment and managing potential discrepancies between patients and providers (224).

Core elements of the transition include assessing transition readiness and developing a transition plan. These elements are typically performed between the ages of 14 and 18 years. It is also important during this time to incorporate one-on-one time between patient and provider into clinical encounters. Transfer elements also include preparing a medical summary, transferring care, and confirming the patient evaluation by the new clinician(s), potentially one at a time, if pediatric care has been multidisciplinary and needs to continue. Transfer to adult care providers may also negatively influence treatment adherence, as they generally provide less intensive ongoing support and monitoring. Care coordination support by nurses or social workers can increase the likelihood of success of the process (225). Few, if any, studies have assessed this transition for patients with GD, and particularly GD3, individuals, who may require decision supports, such as guardianship.

9.3. REPRODUCTIVE HEALTH

9.3.1. Challenges:

Individuals with GD1 and GD3 face actions, e.g., adolescent sexual activity, and intentional decisions related to pregnancy, marriage, partnerships, and long-term family planning.

9.3.1.1. Approaches:

Appropriate health and genetic counseling and testing should be made available for all individuals with GD 1 and GD3 who are sexually active. These would include the potential health effects of pregnancy or hormonal birth control on GD processes and treatments. Individuals with GD and their partners should have access to health and genetic counseling to discuss the status of their disease and therapy, the recurrence risk, and carrier screening. In addition, full GBA1 genotyping of both individuals would be needed, since the affected individual may be GBA1 heteroallelic. If the partner carries a GBA1-variant allele, the predicted genotypes of offspring and their expected phenotype require in-depth discussions. The ethical issues involved in having a child who may or will be affected, given the availability of treatments, are complex and beyond the scope of this article.

Similarly, if the partner of an individual with GD1 or GD3 is not heterozygous for a GBA1-variant allele, all offspring will be carriers of a mutant-GBA1 allele and at increased lifetime risk of developing GBA1-related PD and/or LBD. This increased risk varies from 4- to 10-fold vs. the general population of about 10% of PD by 80 years of age among heterozygous children. Critical research is evolving to solidify and to identify factors that modify this PD risk (226, 227), in addition to the nature of the mutant GBA1 allele.

During the reproductive years, specific GD therapy for women would be restricted to some form of enzyme therapy, e.g., IV ET or ET by preexisting gene therapy, as current SRT approaches would be contraindicated due to potential adverse transplacental fetal or embryonic effects. This potential change in therapy could require lifestyle changes, e.g., return to an infusion center for ET, making the pregnancy more complicated.

Also, pregnancies and the post-partum time for in women with GD should be considered high-risk, particularly in those who are not on established therapies. In 453 pregnancies, among 189 women participating in the Gaucher Outcome Study, complications, e.g., higher risk of bleeding, occurred in 74.2%, of women who did not receive GD-specific treatment (228). Pregnancy outcomes, live births delivered at term with no congenital abnormalities, and spontaneous abortions were the same with or without ET. Additional analyses are needed in order to improve counseling of women who are considering becoming pregnant and the care of women who are pregnant.

9.4. Costs of Therapy

The individual lifetime costs of ET for GD are very high and could be unaffordable without insurance or other health care coverage that includes preexisting conditions. There are also attendant costs of administration, regular follow-up testing, complications of GD, and co-morbidities. These costs and coverage impact newborn screening for GD, the lifetime burden of the disease, and reproductive decisioning. For many populations of GD patients, access to medical care and treatment are severely restricted due to location, transportation, or ability to pay. Developing systems to provide effective therapies present a major challenge for society and the medical communities.

9.4.1. Approaches:

To address drug costs, regulatory approaches include the US Orphan Drug Act and the European Union regulations on orphan medicinal products. They offer, in part, market exclusivity for a period of time following approval. Innovative approaches to funding include “megafund” financing using research-backed obligations securities. “Megafunds” pool a large number of drug development efforts into a single financial entity such as “research-backed obligations,” which are bonds collateralized by the portfolio of potential d rugs and their associated intellectual property. This mechanism has the potential benefit of diversification (229).

10. Conclusions:

Transformative changes bring great rewards and challenges for those who can envision the future. In GD, we as a community have experienced these beneficial changes over the past three decades, and some of the remaining challenges are evident, as delineated above. The introduction of effective visceral therapies for GD made evident the need to develop treatments for the remaining bone disease and CNS involvement. For the latter, the dearth of deep biologic understanding of the CNS disease processes related to GBA1 mutations provides great opportunity for scientific advancement and also great challenges in evaluating the efficacy of new therapies on a long-term basis. As with ET for GD1, emergent complications due to treatment limitations should provide little solace, but significant inspiration for filling the knowledge and clinical voids needed to effectively treat GD3, and perhaps GD2. That current and future basic and clinical sciences would have tentacles extending far beyond rare diseases to common neurodegenerative diseases, e.g., PD and LBD, could not have been anticipated three decades ago. This provides for joint opportunities for multidisciplinary bidirectional clinical and basic science explorations leading to mutual benefit for patients with these devastating diseases. Such interdigitations have already begun and should expand to mutual benefit. Here, we have outlined only some prospects. As a community of scientists, clinicians, and affected families, it is our responsibility and our privilege to meet the unmet needs, to seize this time to improve treatments while foreseeing the potential future.

Abbreviations

The following are conventional/common and recently revised nomenclatures

N370S, L444P, and D409H

The GBA1 genotypes of the encoded respective mature mutant proteins

P.D409H and P.L448H

The mature and immature proteins, respectively, derived from a GBA1 mutation in exon 9 that is 3’ to that causing p.N370S, a.k.a. p.N409S

P.L444P and P.L483P

The mature and immature proteins, respectively, derived from a GBA1 mutation in exon 10

p.N370S and P.N409S

The mature and immature proteins derived from a GBA1 mutation in exon 9. The immature proteins above contain a 39 amino acid N-terminal peptide leader sequence with two initiating methionines

Other Abbreviations in common usage

BBB

Blood Brain Barrier

BMD

Bone Mineral Density

BS

Brainstem

C5a/C5aR1

Complement C5a and its R1 receptor

CB

Cerebellum

CBE

Conduritol B-Epoxide

CNS

Central Nervous System

CO

Cerebral Cortex

DAM

Disease Associated Microglia

ET

Enzyme Therapy/Enzyme Replacement Therapy

FAAC

Fatty Acid Acyl Chain

GBA1/Gba1

The gene encoding human and mouse lysosomal glucocerebrosidase (EC 3.2.1.45), respectively

GCS

Glucosylceramide/Glucocerebroside synthase

GCs/GSs

Glucosylceramides/Glucosylsphingosines

GD1,2,3 and nGD

Gaucher disease types 1, 2 or 3, and neuronopathic GD, respectively

HRpQCT

High-Resolution Peripheral Quantitative Computed Tomography

HSC

Hematopoietic Stem Cell

HSCT

Hematopoietic Stem Cell Transplantation

HSP70

Heat Shock Protein 70

iPSC

Induced Pluripotent Stem Cell

IV

Intravenous

LBD

Lewy Body Disease

LBG

Lysosomal glucocerebrosidase protein, a.k.a. acid β-glucosidase, GCase

LIMP2

Lysosomal Integral Membrane Protein 2, a.k.a. SCARB2

MID

Midbrain

PD

Parkinson Disease

PVM

Yolk sac-derived perivascular macrophages in the CNS

SNPC

Substantia nigra pars compacta

SRT

Substrate Reduction Therapy, a.k.a. Substrate Synthesis Inhibition Therapy

TREM2

Triggering Receptor Expressed on Myeloid Cells 2

WT

Wild-type, i.e., normal protein or gene