Background
Currently, over 8% of the global adult population live with diabetes mellitus [
1]. Diabetes negatively impacts cardiovascular health and as a result, is a risk factor for the development of microvascular diseases which impact multiple vascular beds. Two of the most common microvascular complications of diabetes are diabetic retinopathy (DR) and diabetic kidney disease (DKD). DR is the leading cause of vision impairment and vision loss in the working age population, and affects over 34% (approximately 93 million people) of the diabetic population globally [
2]. DKD impacts up to 40% of diabetic patients (> 116 million people), and is the leading cause of end stage renal failure [
3].
Both DR and DKD are characterized by microvascular barrier dysfunction, resulting in increased vascular permeability. The blood-retina-barrier is formed of pericytes which wrap around a basement membrane, and a monolayer of non-fenestrated endothelial cells with junctions, forming a relatively impermeable barrier. In DR, a breakdown in this barrier results in abnormal vascular leak of fluid and circulating proteins, leading to diabetic macular oedema (DME) [
4]. In the kidney, the glomerular filtration barrier is composed of specialised epithelial cells called podocytes with interdigitating podocyte foot processes, which wrap around the glomerular basement membrane and fenestrated endothelial cells. It is highly permeable to water and small solutes but not to macromolecules. DKD is characterised by increased urinary albumin excretion (albuminuria), resulting from damage to the glomerular filtration barrier causing increased permeability [
5]. Prospective and epidemiological studies in people with diabetes have shown that development of microalbuminuria,
i.e. small quantities of albumin in the urine, is an independent predictor of progressive renal disease [
6,
7].
As both DR and DKD have microvascular barrier complications, it is perhaps unsurprising that the incidence of one complication significantly increases the likelihood of developing the other [
8‐
10]. Studies suggest DR and DKD have common mechanisms in their pathogenesis, as improved glycaemic and blood pressure control reduces the incidence of both [
11,
12]. There are no preventive treatments for DR or DKD. Retinal laser therapy and anti-VEGFA treatments remain the most effective treatments for sight-threatening DR [
13,
14], as they reduce DME, the most common cause of blindness in DR [
14]. However, laser therapy is inherently destructive and 40–50% of patients with DME do not fully respond to anti-VEGFA treatments [
4]. The incidence of DR increases with duration of diabetes, with a prevalence of 80% in patients who have had diabetes for > 15 years [
15], suggesting that early intervention may reduce DR development. Similarly, development of DKD usually occurs more than 10 years after the onset of diabetes, providing an opportunity for early intervention. Current DKD treatments such as sodium-glucose cotransporter-2 inhibitors (SGLT2i), renin-angiotensin system (RAS) inhibitors, and mineralocorticoid receptor antagonists (MRA) have renoprotective effects beyond those attributable to blood pressure lowering alone [
16]. However, despite available treatments, DKD still progresses towards kidney failure in a proportion of patients. As both DR and DKD involve microvascular barrier dysfunction, there is potential for identification of therapeutic targets which are suitable for systemic prevention/treatment of diabetes associated microvascular barrier dysfunction in general. As glomerular and retinal vessel architecture is distinct from one another, these vessel beds allow for the opportunity to explore common mechanisms of widespread microvascular dysfunction.
The endothelial glycocalyx (eGlx) lines the luminal surface of all vascular endothelial cells. It has emerged as a key determinant of endothelial and vascular health, including through its role in barrier function [
17,
18]. The eGlx is a multifunctional structure serving as; a protective layer on endothelial cells, a regulator of immune cell and molecule adhesion, as well as a barrier to macromolecules in circulating blood [
17,
19,
20]. The eGlx limits endothelial cell activation by cytokines, and damage to this protective layer has been linked to endothelial dysfunction and increased permeability [
21,
22], a key characteristic of both DR and DKD. The eGlx is a carbohydrate rich layer comprised primarily proteoglycans and glycosaminoglycans (GAGs) including heparan sulphate (HS), chondroitin sulphate, and hyaluronic acid [
18]. HS chain elongation is a key step in HS synthesis and is carried out by HS-polymerase, a heterodimer made up of two glycosyltransferases, Exotosin-1 (Ext1) and Exotosin-2 (Ext2) [
23]. Upon completion of HS synthesis in the Golgi apparatus, membrane HS proteoglycans are transported to the cell surface.
Although the matrix metalloprotease (MMP) enzyme family can cleave HS proteoglycan core proteins found in the eGlx such as Syndecan-4 [
24,
25], heparanase-1 is the only known mammalian endoglycosidase with the capability to directly cleave HS, and is systemically upregulated in diabetes. Heparanase is elevated in the vitreous and epiretinal membranes of DR patients as well as the kidneys of DKD patients [
26‐
28]. We hypothesised that eGlx HS plays a key role in the barrier properties of retinal and glomerular capillaries and that inhibiting heparanase to protect eGlx HS, could serve as a novel systemic approach to protect against microvascular complications in diabetes.
Here, we show that HS plays a key role in microvascular barrier function of the eGlx in vivo
. We demonstrate for the first time that the eGlx can be therapeutically targeted with the novel heparanase inhibitor OVZ/HS-1638, a unique tetravalent dendrimer heparanase inhibitor with no-off target anticoagulant activity [
29], to prevent microvascular dysfunction in diabetes in multiple vessel beds,
i.e. eye and kidney, in a mouse model of type-2 diabetes.
Research design and methods
Animals
Animals were kept according to the Guidelines on the Use of Animals in Research and studies performed in compliance with the University of Bristol’s guidelines and the Animal (Scientific Procedure) Act 1986 under licences approved by UK Home Office.
C57BL/6J and FVB mice (mean weight 25 g) were obtained from Charles Rivers (Charles Rivers Laboratories, Wilmington, MA). BKS.Cg-+
Leprdb/+
Leprdb/OlaHsd (
db/db) and lean
db/ + littermate controls (lean) were obtained from Envigo (Huntindon, England). Endothelial specific Ext1 conditional knockout mice (
Ext1(ECKO)) and control mice were generated on a C57BL/6 background by crossing
Tie2-Tet-O-
Cre mice with
Ext1-floxed (
Ext1fl/fl) [
30] mice and bred in house by the Animal Service Unit at the University of Bristol.
Heparinase III enzyme treatments
Six-week-old FVB mice were used for transmission electron microscopy (TEM) kidney studies and C57BL/6J mice were used for fluorescein angiography experiments due to incompatibility of albino FVB mice with fluorescein angiography analysis. Mice were given retro orbital injections (under anaesthesia) of heparinase III derived from Flavobacterium (H8891, Sigma-Aldrich, St. Louis, MO) reconstituted in PBS at 8.2 units/kg or heat-inactivated enzyme as negative control. After 30-min, mice were prepared for cardiac perfusions or sodium fluorescein angiography as described below.
Targeted knockout of Exotosin-1 in endothelial cells
Tie2-rtTa+/-, Tet-Cre+/-, Ext1fl/fl (Ext1(ECKO)) and age/sex-matched littermate controls (LMC) which lacked at least one transgene, were given 2 mg/ml of doxycycline with 5% sucrose in drinking water for three weeks to induce knockout.
Urine collection and urine albumin creatinine ratio
Urine was collected and urine albumin creatinine ratios (uACR) were quantified as previously described [
31]. Briefly, albumin concentration was measured using Mouse Albumin ELISA kit (E90-134, Bethyl Laboratories, Inc) and creatinine concentration measured at Langford Vets Diagnostic Laboratories (The University of Bristol). Two urine samples in the diabetes study (one
db/db sample and one
db/db + HI sampled) were invalidated due to bacterial contamination and therefore omitted.
l-lysine treatments
l-lysine was injected via intraperitoneal (i.p.) route, with 2 g/kg of sterile
L-lysine (L5751, Sigma-Aldrich) dissolved in PBS, a dose previously shown to be effective at inhibiting tubular reabsorption in mice [
32]. ‘Post-treatment’ urine was then collected after
L-lysine treatment over the course of 2-h in metabolic cages.
Blood glucose measurements
Weekly blood glucose measurements were taken using Accu-Check standard glucometer (range 0–40 mmol/L) with Accu-Check AVIVA test strips (Roche Diagnostics Limited, West Sussex,England).
OVZ/HS-1638 heparanase inhibitor treatments
Seven-week-old BKS.Cg- +
Leprdb/ +
Leprdb/OlaHsd (
db/db) (n = 10, mean weight 45 g) were randomly assigned to experimental groups (vehicle or OVZ/HS-1638) and caged accordingly. Heparanase inhibitor OVZ/HS-1638 (Additional file
1: Fig. S2, described previously as compound 29) [
29] was reconstituted in PBS at 20 mg/kg and administered to nine-week-old
db/db mice (n = 5) for 2 weeks via i.p. injection.
Db/db vehicle control mice (n = 5) and non-diabetic lean mice (n = 5) were administered PBS only.
Cardiac perfusions for eGlx staining and transmission electron microscopy
Mice under anaesthesia were cardiac perfsued with Ringer’s solution to remove red blood cells from the circulation, followed by Alcian blue in fixative solution (0.15 M sodium cacodylate pH 7.4, 2.5% glutaraldehyde, 0.01 g/mL of Alcian Blue 8GX (75,881–23-1, Santa Cruz Biotechnology, Dallas, TX) in water) to stain the eGlx in vivo, as previously described [
31]. Mouse retina and kidney samples were removed, and tissue was processed for TEM at the University of Bristol Wolfson Bioimaging Facility. See Additional file
1: Methods S1 ‘Processing of samples for transmission electron microscopy’ for full details.
Researchers were blinded for imaging and analysis. High power images (49,000 × magnification) were acquired on a FEI Tecnai-12 BioTwin Spirit transmission electron microscope (ThermoFisher Scientific, Hillsboro, OR). To standardise eGlx depth measurements, images were overlayed with a grid on ImageJ, and measurements taken where intersecting grid lines also intersected lipid bilayers (Additional file
1: Fig. S1). See Additional file
1: Method S1 ‘Electron microscopic glomerular filtration barrier measurements’ for more details.
Immunofluorescence for HS on mouse kidney and retinal tissue
Mouse kidney tissue was fixed in 4% formaldehyde (PFA), paraffin embedded, and cut into 3 μm sections. Slides were deparaffinized in Xylenes and rehydrated. A panel of antibodies was initially tested to determine which antibody bound to the eGlx HS (data not shown). The vesicular stomatitis virus G (VSV-G) tagged phage display antibody HS3A8V derived from bovine kidney HS [
33,
34] (gift from Jerry Turnbull, Keele University) detected HS in the eGlx in this model. HS3A8V was diluted 1:5 and incubated at 4 °C overnight. All washes were with PBS, pH 7.2. Secondary rabbit anti-VSVG (A190-131A, Bethyl Laboratories, Inc., Montgomery, TX) was incubated for 1 h at 1:50 at room temperature, washed, and tertiary anti-rabbit 488 (A-11001, ThermoFisher Scientific, Waltham, MA) incubated for 1 h at 1:250. Tissue was counterstained in 300 nM 4’,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI) (D1306, ThermoFisher Scientific) for 5 min followed by a 15-min membrane stain with Octadecyl Rhodamine B Chloride (R18) (O246, ThermoFisher Scientific) diluted 1:1000 in PBS. Slides were washed and mounted with Pro long gold (P10144, ThermoFisher Scientific) and imaged on a Multi-Laser CLSM Leica SP5 confocal microscope (Leica Microsystems, Wetzlar, Germany).
Mouse eyes were enucleated and retinas isolated. Retinas were flat mounted and fixed in 4% PFA and stained following the same procedure as above, with the VSV-G tagged phage display antibody RB4Ea12V derived from mouse skeletal muscle (gift from Jerry Turnbull, Keele University). A different HS antibody was used in the retina than in the kidney due to the wide variation of HS epitopes which vary between different tissue structures, as has been previously demonstrated [
35]. Retinas were coverslip mounted with Pro long gold (P10144, Thermo-Fisher) and imaged on a Carl Zeiss 700 confocal microscope provided by the Kansas State University Confocal Core.
Lectin staining for eGlx measurements in mouse retinas
Retinas were fixed in 4% PFA, paraffin embedded, and sectioned at 3 µm or flat mounted. Sections were deparaffinised in Xylenes and rehydrated and blocked in 1% BSA. We used FITC-conjugated Lycopersicon esculentum lectin (LEL) (L0401, Merk Life Science, Dorset, UK) to measure changes in eGlx depth. LEL was incubated at 1:200 in 1% BSA overnight and washed in 0.1% Tween-20 PBS. Sections were counterstained with DAPI as above, followed by a 15-min cell membrane stain with R18 (ThermoFisher Scientific) diluted 1:1000 in PBS. Slides were washed in PBS and coverslip mounted. Retinal flat mount (Additional file
1: Fig. S3) was stained in a similar manner excluding R18, and coverslip mounted.
Researchers were blinded for imaging and analysis. Glycocalyx depth was measured as previously described [
25,
36,
37]. Briefly, confocal images were taken using an 100X oil immersion objective on a Multi-Laser CLSM Leica SP5 (Leica Microsystems). Using ImageJ, fluorescence peak measurements were obtained using the line tool to draw a line perpendicular to the vessel wall. Fluorescence profiles were plotted for both green filter (LEL staining) and red filter (R18 staining) images (Fig.
2d). EGlx staining was evident by staining on luminal surface of endothelial cells. The distance between the two fluorescence peaks was measured (Fig.
2d), providing a representation of eGlx depth [
37]. A minimum of five vessels per animals were analysed. EGlx depth measurements were averaged per mouse.
Fluorescein angiography for solute flux measurements
A Micron IV imaging system (Phoenix Research Labs, Pleasanton, CA) was used for fluorescein angiography. The following method of imaging was adapted from Allen et al
. [
38]. Briefly, 30-min prior to imaging, mouse pupils were dilated with Tropicamide 1%w/v (Bausch and Lomb, Rochester, NY). Mice were anesthetized and positioned on a stage for imaging. The centre of the image was focused on the optic nerve of the left eye using the brightfield fundus setting. Images were scanned for abnormalities and optical coherence tomography images were taken. The camera was then switched to the green filter and mice were given 50 µl of 0.22 μm sterile filtered 10% Sodium Fluorescein diluted in water by i.p. injection. Recordings at 15 frames per second commenced immediately until the fundus image was saturated with fluorescein (approximately 4- minutes). All camera settings remained constant for the duration of the experiment.
Researchers were blinded for video analysis. The ratio of fluorescence intensity measured in tissue next to the main vessel (V
2) and within the main vessel (V
1) (Fig.
2f) was plotted against time to give the change in fluorescence intensity over time (∆V
2/1/∆t) = (∆V
R/∆t). To minimize the effect of differential vessel filling time from animal to animal, which may impact solute flux calculations, only data points after the main vessel had reached a steady state were used (
i.e. when the curve began to plateau but before saturation). This exact point was defined as the point at which the R
2 of the curve was closest to 1. Similar to Allen et al. [
38], we called this time = 0. Solute flux was then calculated using ∆V
R/∆t from the slope of the linear change in fluorescence intensity during a 200-s window after t = 0. We developed an automated program for this analysis to allow for a larger area of the retina to be analysed and a copy of this program is available upon request.
Ex vivo glomerular albumin permeability assay
The following assay was previously described in depth by Desideri et al
. [
39]. In brief, mice were anaesthetised and cardiac perfused with Ringer’s solution as above. Kidneys were placed into cold Ringer’s solution with 4% BSA, pH 7.4. Glomeruli were isolated by graded sieving and resuspended in 4% BSA ringer containing 36.5 µg/mL R18 (ThermoFisher Scientific) for 15 min at 4 ºC, to label cell membranes. Glomeruli were washed with 4% BSA ringer and incubated in 4% BSA ringer containing 30 µg/mL AlexaFluor 488 BSA (A13100, ThermoFisher Scientific) for 15 min at 4 ºC to allow fluorescently labelled BSA to diffuse into capillaries. Researchers were blinded for analysis. Albumin permeability across single glomerular capillaries was measured as previously described [
39]. Briefly, an individual glomerulus was trapped using a glass capillary pipette on a thin glass bottom petri dish and washed with unlabelled 4% BSA. Videos recorded on Nikon Ti-E inverted confocal microscope (Nikon instruments Inc., Melville, NY). The rate of decline of fluorescence intensity in individual capillary loops was used to calculate apparent glomerular albumin permeability.
Albumin staining and analysis for retinal extravascular albumin measurements.
Mouse retina sections were deparaffinised and rehydrated as described above. Tissue was permeabilised and blocked with 0.3% Triton-X, 3% BSA in PBS solution for 30-min. Goat anti-mouse albumin (A90-134A-17, Bethyl Laboratories, Inc., Montgomery, TX) was incubated at 1:200 in blocking buffer overnight at 4 °C in a humidity chamber. Sections were washed in 0.3% Triton-X in PBS. Donkey anti-goat Alexa Fluor 594 (A-11058, ThermoFisher Scientific) was incubated at 1:200 in blocking buffer for 1-h at room temperature. Sections were washed, stained with DAPI, and coverslip mounted.
Researchers were blinded for imaging and analysis. Sections were imaged using a Multi-Laser CLSM Leica SP5 microscope (Leica Microsystems). In one retina, a minimum of three fields of view were taken per animal using the 40 × objective. To analyse extravascular albumin staining, brightfield images were used to trace around the retina and exclude choroidal and vitreous staining, using ImageJ. In the red channel, retinal area, Integrated density of the retina (IntDenretina), and four measurements of background mean fluorescence were taken to calculate corrected total retinal albumin fluorescence (CTFretina). Retinal vessel albumin staining appeared as bright circles (vessels in cross section) or long tubes (longitudinal sections of vessels). Vessels were traced around, and the area and IntDenvascular were measured to calculate vascular albumin staining (CTFvascular). Extravascular albumin (CTFEXV) was defined as CTFvascular subtracted from CTFretina. The CTFEXV for each field of view was averaged for each animal.
Statistics
Quantitative data are expressed as means ± SEM. GraphPad Prism version 5.00 used for statistical analyses (GraphPad Software, La Jolla California USA). Unpaired Student’s t-test were performed when comparing two groups. For normally distributed data, one-way ANOVA and Tukey’s multiple comparisons test was performed when comparing more than two groups and for non-normally distributed data, a Kruskal–Wallis test was performed. Statistical significance was reached when P < 0.05.
Discussion
Our results demonstrate that microvascular complications in diabetes can be treated systemically, by targeting HS in the eGlx to restore vascular barrier properties. Microvascular permeability changes occur early in DR and DKD, and are predictors of disease progression [
4,
51]. We show that HS is a key component of the microvascular barrier, using the retina and glomerulus as examples. In this study, we used two independent methods to deplete HS from the eGlx in vivo: in the first, mice were treated with heparinase III, and in the second, the HS polymerase component Exotosin-1 was conditionally knocked out from endothelial cells in
Ext1(ECKO) mice. Our results showed that HS depletion by either method resulted in reduced retinal and glomerular eGlx depth. Furthermore, both methods resulted in greater retinal solute flux and glomerular permeability, indicating the importance of HS to eGlx barrier function in two different vascular beds. Importantly, our data provided further evidence that protection of the eGlx using a novel class of heparanase inhibitors OVZ/HS-1638, is an effective systemic approach to protect against the microvascular complications in diabetes. OVZ/HS-1638 holds competitive advantages over commercially available inhibitors, including lack of anticoagulant activity, increased stability, and cheaper/less complex synthesis [
29], making this class of heparanase inhibitor an excellent candidate for clinical use.
While HS is abundant throughout the retina [
52], to date, there have been no studies published demonstrating the role of eGlx HS in the retina. Previously, Witmer et al
. investigated HS in the vascular basement membrane of postmortem retina tissue from diabetic patients. The authors suggested that increased microvascular permeability, measured by PAL-E staining [
53], was not associated with changes in retinal basement membrane HS or its basement membrane proteoglycans agrin and perlecan [
54]. This is in line with studies on mice lacking perlecan and agrin in the glomerular basement membrane, which have no changes in GFB function or structure [
55]. These studies suggest that basement membrane HS does not contribute to microvascular barrier function. In contrast, our data shows clear changes in barrier function when HS within the eGlx is depleted in vivo, demonstrating an important role of HS within the eGlx in the retinal microvasculature, not previously shown.
The systemic role of HS within the eGlx was supported by our heparinase III model and
Ext1(ECKO) mouse models, in which depletion of HS resulted in higher glomerular albumin permeability [
43]. In health, small amounts of albumin which are filtered through the GFB are almost entirely reabsorbed by the tubules [
56]. It is only when the system is overwhelmed by increasing glomerular leakage of albumin in disease, that changes in albuminuria are observed. We did not observe significant changes in uACR in
Ext1(ECKO) mice. Yet, we found that when we inhibited tubular reabsorption with
L-Lysine, there was a large fold increase in the uACRs in
Ext1(ECKO) mice compared to LMCs, in agreement with our ex vivo glomerular albumin permeability results which measures permeability in the absence of tubular reabsorption. Contrary to previous literature which have used albumin clearance
i.e. uACR, and not glomerular albumin permeability, our data provides clear evidence that depleting HS from the glomerular eGlx is sufficient to negatively impact glomerular barrier function. Microalbuminuria is a predictor of progressive renal disease [
57], so factors that causes moderate increases in microalbuminuria are important to consider in the context of diabetes and the early management of DKD. We also showed that both enzymatic and genetic depletion of HS did not alter GFB structures other than the eGlx, including glomerular basement membrane thickness, podocyte foot process and slit diaphragm width, and endothelial cell fenestrations, indicating that the functional changes we observed were specifically caused by structural changes to the glycocalyx.
Current treatments for diabetic microvascular complications are designed to treat the specific local complication (i.e., DR or DKD and not both), they are not preventative, and they each have possible risks. For example, retinal laser photocoagulation therapy for DME can result in scotomas, reduced visual acuity and subretinal fibrosis [
14], and anti-VEGFA therapies may lead to proteinuria, compounding kidney injury in those already at risk for DKD [
58,
59]. Furthermore, DKD is still progressive, despite the potential renoprotective effects of treatments such as SGLT2i and RAS inhibitors. Therefore, there is a clinical need for treatments that can prevent the development and progression of DR and DKD.
Heparanase has gained interest as a druggable target in some diseases that are associated with chronic inflammation, such as cancer and diabetes. In line with our results presented here in diabetes, Gil. et al. demonstrated that the heparanase inhibitor SST0001 reduced albuminuria in T2D mice and further determined that heparanase upregulation in DKD results from activation of the heparanase promoter by the transcription factor early growth response-1 (Egr1) [
60]. As indicated by the authors, although protective effects of heparanase inhibition were shown in diabetic mice, a limitation of the study was that the findings were not sufficient to determine the mode of heparanase action in DKD pathogenesis [
60]. Using our novel heparanase inhibitor OVZ/HS-1638, we have now demonstrated that heparanase acts on the eGlx, as inhibiting heparanase protects the eGlx and prevents microvascular permeability changes. We showed that OVZ/HS-1638 treatment of
db/db mice protected the retinal eGlx and reduced albumin leak in the retina. In the same mice, we also showed OVZ/HS-1638 treatment prevented glomerular eGlx damage and reduced glomerular permeability. The clinical relevance to patients is that HS in the eGlx of different vascular beds can be therapeutically targeted with a systemic medication, thereby preventing the microvascular permeability changes associated with DR and DKD.
In health, heparanase expression is tightly regulated; it is expressed only in a subset of tissues, mainly placenta and skin [
61,
62]. Heparanase expression and activity is largely modulated by pathophysiological conditions, such as pH, as heparanase is essentially inactive in neutral conditions [
63]. This increases the attractiveness of heparanase as a drug target. Heparanase inhibitors have been shown to slow disease progression in cancer studies; the most promising and potent inhibitor in clinical trials is PG545, a polysulfated oligosaccharide-cholestanyl aglycone [
60,
64,
65]. Despite this, no heparanase inhibitors have been approved for clinical use as they are limited by cost, production complexity, and off target effects such as anticoagulant activity and thrombocytopenia resulting from their structural similarity to heparin [
29,
66,
67]. Unlike other heparanase inhibitors currently in clinical trials, such as SST0001 and PG545 which are all oligo-/polysaccharide derivatives, OVZ/HS-1638 is a polyvalent dendrimer, allowing for multiple heparanase interactions at multiple points [
29,
66]. The heterogenous nature of oligo-/polysaccharide derivatives such as PG545 also results in batch variations, further complicating FDA approval. In contrast, dendrimer synthesis yields single entity compounds, which means OVS/HS-1638 preparations are homogenous. Previous in vitro experiments show that OVZ/HS-1638 inhibits heparanase at nanomolar concentrations (IC
50 11 nM) and displays no anticoagulant activity toward Factor Xa, even at concentrations > 100-fold than the IC
50 for heparanase inhibition [
29], indicating the highly specific nature of this inhibitor. Computer modelling of these polyvalent dendrimers suggest they bind to heparanase in the same region as HS with a greater affinity, indicating OVZ/HS-1638 may act as a competitive inhibitor [
68]. In addition to the forementioned competitive advantages, OVZ/HS-1638 inhibits angiogenesis [
29], and was recently shown to rescue blood–brain-integrity in a model of neuroinflammation [
69]. These results indicate OVZ/HS-1638 holds significant clinical promise. Further studies on pharmacokinetics, bioavailability, and safety of OVZ/HS-1638 are in progress, to support its potential for clinical evaluation [
29].
Limitations
To our knowledge, this study is the first to demonstrate systemic prevention of microvascular complications associated with diabetes, demonstrated in the eye and kidney. However, there are still limitations which remain. Firstly, our diabetic study lacked a group in which lean non-diabetic mice were treated with OVZ/HS-1638. At the time this study was designed, quantities of this inhibitor were limited due to small scale synthesis. However, this inhibitor has been used in other in vivo disease models such as myeloma, colorectal cancer, osteogenesis, as well as multiple sclerosis [
29,
68‐
70]. As in our study, treatment with OVZ/HS-1638 was well tolerated, with no apparent adverse effects observed. Additionally, we did not observe any increased negative effects in our treated
db/db mice in the parameters measured, but we cannot be certain that OVZ/HS-1638 does not have any effect in healthy mice. The forementioned pharmacokinetic, bioavailability, and safety studies using this inhibitor will be conducted in non-diseased models, and therefore any effect of OVZ/HS-1638 in healthy mice will be determined. Along with these studies, it will be informative to measure heparanase activity in
db/db mice treated with OVZ/HS-1638. In addition to the previously discussed heparanase activity studies by Zubkova et al
. 2018, our in vitro work in endothelial cells which confirmed inhibition of HS shedding by heparanase in OVZ/HS-1538 treated cells, gives us full confidence in OVZ/HS-1638’s ability to inhibit heparanase activity. Nevertheless, studies which confirm reduced heparanase activity in vivo will add value to our ongoing work as we progress this inhibitor toward clinical use. Another limitation of this study is the time of treatment in
db/db mice. Diabetic mice began treatment at 9-weeks-of-age, a relatively early stage of disease progression for DR and DKD in this model. It is likely that in a clinical setting, such treatments would be given at a wide range of disease stages (early and late). Although
db/db mice were significantly hyperglycemic and obese at the time of the treatment, we cannot conclude how this inhibitor would perform when given at a more advanced stage of disease. Ongoing research in our group will address this question and will be of great clinical relevance.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.