Abstract
The neurotrophin growth factors bind and activate two types of cell surface receptors: the tropomyosin receptor kinase (Trk) family and p75. TrkA, TrkB, and TrkC are bound preferentially by nerve growth factor, brain-derived neurotrophic factor, and neurotrophin 3 (NT3), respectively, to activate neuroprotective signals. The p75 receptors are activated by all neurotrophins, and paradoxically in neurodegenerative disease p75 is upregulated and mediates neurotoxic signals. To test neuroprotection strategies, we engineered NT3 to broadly activate Trk receptors (mutant D) or to reduce p75 binding (mutant RK). We also combined these features in a molecule that activates TrkA, TrkB, and TrkC but has reduced p75 binding (mutant DRK). In neurodegenerative disease mouse models in vivo, the DRK protein is a superior therapeutic agent compared with mutant D, mutant RK, and wild-type neurotrophins and protects a broader range of stressed neurons. This work rationalizes a therapeutic strategy based on the biology of each type of receptor, avoiding activation of p75 toxicity while broadly activating neuroprotection in stressed neuronal populations expressing different Trk receptors.
SIGNIFICANCE STATEMENT The neurotrophins nerve growth factor, brain-derived neurotrophic factor, and neurotrophin 3 each can activate a tropomyosin receptor kinase (Trk) A, TrkB, or TrkC receptor, respectively, and all can activate a p75 receptor. Trks and p75 mediate opposite signals. We report the engineering of a protein that activates all Trks, combined with low p75 binding, as an effective therapeutic agent in vivo.
Introduction
Neurotrophins (NFs) are growth factors with key roles in the embryonic and the adult nervous systems (Chao et al., 2006). Mature NFs include nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin 3 (NT3). NFs bind to two classes of receptors, tropomyosin receptor kinase (Trk) and p75, which play opposite and often paradoxical roles in vivo.
Trk receptors are bound by mature NFs with relative selectivity. NGF is the preferred ligand for TrkA, BDNF for TrkB, and NT3 for TrkC. In some cellular contexts, NT3 can activate TrkA and TrkB with lower efficiency (Ivanisevic et al., 2003; Ivanisevic et al., 2007). Activation of Trk receptors preserves the phenotype and function of neurons and promotes neuronal survival in neurodegenerative diseases (Saragovi et al., 2009; Josephy-Hernandez et al., 2017; Saragovi et al., 2019).
Regarding p75 receptors, they are promiscuous and are bound by all NFs and the NF precursors (Hempstead 2002; Hempstead 2006; Ibáñez and Simi, 2012). The p75 receptors are upregulated in acute and chronic neurodegenerative diseases and mediate multiple actions, such as elevation of proinflammatory tumor necrosis factor α (TNFα) (Lebrun-Julien, et al., 2009a; Bai et al., 2010a; Lebrun-Julien et al., 2010; Barcelona et al., 2016; Mossa et al., 2020), resulting in neuronal dysfunction and death.
The NFs have been investigated as therapeutic agents for neurodegeneration (Josephy-Hernandez et al., 2017; Saragovi et al., 2019), but clinical trials with mature NFs have not been successful to date. Many reasons have been postulated for the lack of clinical efficacy of NFs, such as poor pharmacokinetics due to their short half-lives, the difficulty of delivery to target neuronal compartments, the challenges of stable expression/manufacturing, and lack of receptor selectivity leading to upregulated p75 activity prevailing over the desired neuroprotective signals of Trk receptors.
Several strategies have been explored to develop neurotrophic-targeted therapies (Saragovi et al., 2009; Longo and Massa, 2013; Josephy-Hernandez et al., 2017; Saragovi et al., 2019). Here we discuss combining two of them. Selectively activating a single type of Trk receptor used Trk-agonistic small molecules or monoclonal antibodies (mAbs) that do not bind to p75 (Bruno et al., 2004; Zaccaro et al., 2005; Bai et al., 2010a; Bai et al., 2010b; Aboulkassim et al., 2011; Simmons et al., 2013; Brahimi et al., 2016; Szobota et al., 2019; Brahimi et al., 2020). A variation of this was to mutate NGF or NT3 at their p75-binding domain to abolish interactions between p75 and NF p75less mutant but leave TrkA or TrkC activation intact (Urfer et al., 1994; Guo et al., 1996; Mahapatra et al., 2009; Bai et al., 2010a; Enomoto et al., 2013). These agents have been evaluated therapeutically in vivo.
A separate approach was to activate all Trk receptors simultaneously using engineered proteins with sequences from NGF, BDNF, and NT3 (pan-NF) (Urfer et al., 1994; Rydén and Ibáñez, 1996; Rydén and Ibáñez, 1997; Urfer, et al., 1997). Since neuronal subpopulations express different Trk receptors (e.g., TrkA cholinergic, TrkB dopaminergic), a pan-NF would promote survival simultaneously in different injured neuronal phenotypes. However, purified proteins have not been evaluated in vivo in models of disease, and moreover in these proteins there remains the potential for unintended activation of p75, which can be toxic.
Here, we report work combining the two strategies of mutating NFs to enhance activation of all Trk receptors and avoid p75 activation. We generated a pan-NF protein that can activate all Trk receptors, combined with mutations that reduce binding to p75 (hereafter “DRK” based on the engineered mutations). In experimental models of acute (optic nerve axotomy) or chronic (diabetic retinopathy) retinal neurodegeneration in vivo, DRK has superior therapeutic efficacy compared with NF mutants that only reduce binding to p75, or to NF mutants that only activate all Trk receptors, or to control wild-type NFs. Efficacy was significant with DRK administered as a single dose, after injury, with ongoing degeneration, and at low concentrations. This work helps in the rationalization of therapeutic strategies by accounting for the pattern of receptor expression and the biology of each targeted receptor.
Materials and Methods
Synthesis of Mutants
Mutant forms of NT3 (D, RK, DRK) were codon-optimized and expressed in Escherichia coli. Proteins were purified and refolded from the collected inclusion bodies. Appropriate refolding and purity were confirmed by SDS-PAGE. Mutant D15A (D) was designed to enhance TrkB binding, and mutant R114A/K115A (RK) was designed to reduce p75 binding (Urfer et al., 1994; Rydén and Ibáñez, 1996; Rydén and Ibáñez, 1997; Urfer et al., 1997). Mutant D15A/R114A/K115A (DRK) was designed to combine these features to generate a molecule that would potentially activate TrkA, TrkB, and TrkC with reduced p75 binding. DRK is a p75less pan-NF. Wild-type NFs were used as controls. As an additional control, a previously published mutant of NGF, termed NGF-C, does not bind to p75 but activates TrkA fully (without activating TrkB or TrkC) (Bai et al., 2010a; Aboulkassim et al., 2011).
Cell Lines
Cells have been described (Maliartchouk and Saragovi, 1997; Guillemard et al., 2010; Barcelona and Saragovi, 2015; Szobota et al., 2019; Brahimi et al., 2020). HEK293 or NIH3T3 cells were stably transfected to express either TrkC, TrkB, TrkA, or p75. The nnr5 cells are a variant of the rat PC12 cells (TrkA+ p75+) that lost TrkA expression. Nnr5 cells were stably transfected with human TrkC cDNA (nnr5-TrkC) and are TrkC+ p75+. Receptor expression was routinely verified by flow cytometry, using selective monoclonal antibodies directed to the ectodomain. All cells were monitored routinely to exclude mycoplasma contamination (M208001, ZM Tech Scientific).
Mice
The animal protocols and endpoints were approved by the Lady Davis Institute Animal Care Committee and by McGill University Institutional Animal Care and Use Committee. Experiments were done according to the guidelines of the Canadian Council on Animal Care, and the 3R principle of replacement, reduction, and refinement was applied. Healthy wild-type male C57BL/6N mice, 8–10 weeks of age, 18–21 g (Charles River Laboratories) were used for the optic nerve axotomy model. C57BL/6J (Jackson Laboratories) were used for the mouse model of diabetic retinopathy. A maximum of five mice per cage were kept in a 12-hour dark-light cycle with food and water ad libitum.
Direct Binding Assay
Wild-type NT3 or mutant DRK proteins were biotinylated according to the manufacturer’s instructions (EZ-Link Sulfo-NHS-LC-biotin; 21335, Thermo Fisher Scientific). The stoichiometry of biotin:protein was verified to be ∼4:1 for both wild-type NT3 and DRK mutant. Not all mutant NFs could be biotinylated due to lack of pure protein. Direct binding of the biotinylated ligands (a range of 0, 2, 10, and 50 nM concentrations) was evaluated by quantitative flow cytometry with detection by fluorescein-conjugated avidin [avidin–fluorescein isothiocyanate (FITC); A 2050, Sigma]. Binding and washes were carried out in binding buffer (PBS/0.2% bovine serum albumin/0.1% Na Azide pH 7.3) at 4°C, on cells stably transfected to express TrkA, TrkB, TrkC, or p75. Receptor expression was routinely verified using specific anti-receptor mAbs.
Binding Competition to Biotinylated NT3
The binding competition assay was performed as described (Maliartchouk and Saragovi, 1997; Guillemard et al., 2010; Barcelona and Saragovi, 2015; Brahimi et al., 2020). Cells were preincubated for 15 minutes in binding buffer at 4°C with negative control vehicle (no competition), positive control competition wild-type NT3 (450-03, Pepro Tech), or test proteins (1 µg). Then, biotinylated NT3 (NT3-bio, 40 nM) was added for 15 minutes. After washes in binding buffer, cells were incubated with fluorescein-conjugated avidin (avidin-FITC, Sigma) for 15 minutes at 4°C, washed at 4°C, and analyzed immediately by flowcytometry. Mean channel fluorescence (MCF) values of bell-shaped histograms were standardized to control vehicle = 100% binding, and avidin-FITC without NT3-bio = 0% binding. Shown are % MCF, mean ± S.D. of three independent experiments.
Cell Metabolism/Survival Assays
The growth/survival profile of the cells were quantified in 96-well plates using the tetrazolium salt reagent 4-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT; M2128, Sigma), 48 hours after plating, by reading the optical density (OD). Cells cultured in serum-free medium (SFM) die by apoptosis, but they can be rescued if they express TrkA, TrkC, or TrkB and are supplemented with NGF (450-01, Pepro Tech), NT3, or BDNF (450-02, Pepro Tech), respectively. Vehicle is the negative control and serum the positive control. Under these conditions MTT assay reflects cell survival and is comparable with live cell counting (Maliartchouk and Saragovi, 1997; Guillemard et al., 2010; Barcelona and Saragovi, 2015; Brahimi et al., 2020). All assays were in quadruplicate and were repeated three independent times. MTT data are standardized to optimal concentration (2 nM) of wild-type NF = 100% survival, and SFM = 0% survival, using the formula [(OD test – OD SFM) × 100/(OD optimal NF – OD SFM)].
Western Blot Analysis
The activation of signaling proteins [Trk, serine/threonine Protein kinase B (Akt), and mitogen-activated protein kinase (MAPK)] was studied after treatment of starved cells with wild-type NT3 or mutants (2 nM) for 15 minutes. Detergent lysates were analyzed by Western blotting with anti–phospho-Tyr mAb 4G10 (generic anti-phosphotyrosine; 05-321, Millipore), anti–phospho-MAPK [phosphorylated extracellular signal-regulated protein kinase (ERK) 1/2; 4377S, Cell Signaling], or anti–phospho-Akt (4060S, Cell Signaling). After stripping, membranes were reprobed with anti-actin (A5316, Sigma) to standardize loading.
AlphaLISA
HEK293 or NIH3T3 cells that overexpressed either human TrkB or human TrkC were grown in 96-well plates. Cells were serum-starved for 1–4 hours, then incubated with test agonists for up to 20 minutes. Cells were lysed, and a portion of the cell lysate was incubated with AlphaLISA reagents (PerkinElmer) in 384-well plates. Phosphorylated ERK (pErk) 1/2 (MAPK) was quantified using an EnSpire plate reader (PerkinElmer) as described by Szobota et al. (2019).
Differentiation Assays in Cells
nnr5 cells stably transfected with human TrkC cDNA (nnr5-TrkC) were plated on cover slips with full medium in 24-well plates. Twenty-four hours after plating, cells were untreated (control) or were treated for 72 hours with wild-type NT3 or mutants at 25 pM, 0.1 nM, and 2 nM. Cells on the coverslip were immunostained with rabbit anti –microtubule-associated protein 2 (ab183830, Abcam) followed by anti-rabbit Alexa fluor 488 (A11034, Thermo Fisher Scientific), then washed and further stained with 4′,6-diamidino-2-phenylindole. Pictures were taken by microscopy (40×). Differentiation was scored morphologically as percentages of cells with neurites (more than two cell bodies long). Quantification was done by image analysis (Galan et al., 2017; Galan et al., 2017) from at least three independent assays.
Upregulation of TNFα by Activated p75
HEK293-p75 cells (stably transfected and expressing p75) were treated with the indicated compounds for 6 hours. Quantification of TNFα mRNA was performed by real-time quantitative PCR and primers for TNFα and RNAs18 (Barcelona and Saragovi, 2015; Brahimi et al., 2020). Data are expressed as the means ± S.D. relative to the untreated (2–3 independent experiments, each in triplicate).
Optic Nerve Axotomy
The optic nerve axotomy model was described by Bai et al. (2010a) and Galan et al. (2014). The optic nerve of one eye was exposed and was completely transected 0.75–1.0 mm posterior to the eyeball with the use of micro tweezers, sparing vessels. The contralateral eye was mock surgery without damaging the optic nerve. Normal blood circulation in the retina was ascertained. This model axotomizes the retinal ganglion cell (RGC) axonal fibers, and the RGC cell body in the retina degenerates with fast kinetics (∼50% RGC loss at day 7 postaxotomy, ∼90% RGC loss at day 14 postaxotomy). In the optic nerve axotomy model animals received one treatment immediately after injury (under the same anesthesia), and RGC survival was quantified at the day 14 endpoint. RGC survival was quantified as described below.
The eyes subjected to optic nerve axotomy were injected with test agent, control vehicle, or control wild-type NFs. The uninjured contralateral eyes (mock surgery) served as naive normal controls for each individual mouse, n = 8 per group (each group is the indicated treatment of the injured eye). Surviving RGCs were quantified by counting brain-specific homeobox/POU domain protein 3 cells in flat-mounted retinas (Bai et al., 2010a; Barcelona et al., 2016; Galan et al., 2017; Galan et al., 2017). RGC counts for each individual mouse were analyzed as the ratio of the injured, treated eye to the uninjured contralateral eye (set to 100% RGCs) to derive a percentage of surviving RGCs for the injured eye of that individual mouse. The percentage of surviving RGCs data for n = 8 mice within a group was then averaged ± S.D. This experimental design accounts for possible mouse-to-mouse individual variations in RGC numbers.
Diabetic Retinopathy
Hyperglycemia was induced as described by Barcelona et al. (2016). Male 10-week-old mice received an intraperitoneal injection of streptozotocin (STZ ; 60 mg/kg; Sigma-Aldrich) dissolved in sodium citrate buffer (0.01 m, pH 4.5) on five consecutive days. Age-matched, nondiabetic C57BL/6 mice injected with sodium citrate buffer were used as controls. Blood glucose was measured 7 days after completion of STZ dosing (e.g., the induction of diabetes). Fasting blood glucose levels >17 mmol/L (300 mg/dl) were considered to be diabetic. In this model there is loss of RGC nerve fibers of ∼20% at week 6 of diabetes (quantified in vivo by tomography; see below) reflecting the progressive injury of RGCs (Barcelona et al., 2016). In the diabetic retinopathy model, animals received one treatment after 2.5 weeks of hyperglycemia, at a time of ongoing degeneration, and the endpoint was at the 6-week time point.
For intravitreal injections mice were anesthetized in 3% isoflurane, and 2 μl containing a total of 2 μg wild-type NF, mutant NFs, or vehicle (PBS) was slowly delivered into the vitreous chamber using a Hamilton syringe and confirmed microscopically. After the injection, the syringe was left in place for 30 seconds and slowly withdrawn to prevent efflux.
Quantification of the Nerve Fiber Layer by Tomography
A noninvasive spectrometer-based Fourier domain optical coherence tomography (OCT) system was used to acquire retinal images. Fourier domain OCT is a noninvasive method that allows time-kinetic studies in the same animal, with consistent and reproducible axial resolution in tissue nominally better than 2.2 μm (Bai et al., 2010a; Jian et al., 2013; Barcelona et al., 2016). In each B-scan, the thickness of the nerve fiber layer, ganglion cell layer, and inner plexiform layer, hereafter referred to as NGI, was measured at three adjacent points, using ImageJ software (http://imagej.nih.gov/ij) as described by Barcelona et al. (2016), Galan et al. (2017), and Galan et al. (2017). The outer nuclear layer (ONL) is a layer not affected in the time frames of the disease models and was also measured at the same locations as internal control for potential geographic differences in retinal thickness. Representative data are shown and quantified as average ± S.E.M. absolute thickness in μm in control versus diabetic injected with either vehicle (PBS) or with the indicated treatment (n = 3 independent experiments, 3–4 mice per group) and as percentage of nerve fiber loss ± S.E.M. by setting the NGI damage in vehicle-treated diabetic eyes as maximal (100%).
Statistical Analysis
In all experiments, one-way ANOVA with significance α = 0.05 or lower followed by Bonferroni post hoc analysis was used for calculating significance between groups. Intergroup comparisons were specified before data were viewed. All comparisons are reported. The present study is exploratory, and the P values are considered only descriptive.
Results
The mutant D15A (D) reportedly enhances TrkB binding, and the mutant R114A/K115A (RK) reportedly reduces p75 binding. Mutant D15A/R114A/K115A (DRK) was designed to combine these features to generate a molecule that would potentially have the properties of NT3 plus enhanced TrkA and TrkB binding and reduced p75 binding. These agents were compared with each other and with control wild-type NFs in binding and in biologic assays.
Receptor Binding Profile of the Mutant NFs
We evaluated ligand interactions with TrkC, TrkB, or p75 in a binding competition assay. The binding competition assay tests the ability of mutants to competitively inhibit the binding of biotinylated wild-type NT3 (NT3-biotin). Inhibition by unlabeled wild-type NT3 (“cold competition”) is the positive control (Table 1).
In TrkC-expressing cells, all three mutants significantly competed the binding of NT3-biotin to a similar degree and were comparable to cold competition by wild-type NT3. In TrkB-expressing cells all three mutants significantly competed the binding of NT3-biotin, comparable to cold competition by wild-type NT3 or by wild-type BDNF. In p75-expressing cells, mutant D significantly competed the binding of NT3-biotin, comparable to cold competition by wild-type NT3. In contrast, mutants RK and DRK did not compete NT3-biotin binding to p75 significantly (Table 1). The mutant proteins were not evaluated for competition of NT3-biotin binding to TrkA because detectable NT3-biotin binding to TrkA requires very high concentrations greater than 75 nM (Ivanisevic et al., 2003; Ivanisevic et al., 2007).
The binding competition data indicate that 1) all mutants D, RK, and DRK bind TrkC; 2) all mutants bind TrkB; and 3) mutant D binds to p75 like wild-type NT3, but mutants RK and DRK have reduced affinity for p75.
This conclusion about the binding profile of mutant DRK was confirmed in direct binding assays. Quantitative flowcytometry assays compared the direct binding of NT3-biotin and DRK-biotin to cells expressing TrkA, TrkC, TrkB, or p75 (Fig. 1). Ligand binding is concentration-dependent. Compared with wild-type NT3, the DRK mutant maintains equal or enhanced binding to TrkC, TrkA, and TrkB but has significantly reduced binding to p75.
Mutants Promote Survival in Cells Expressing TrkC or TrkB
Survival assays evaluated whether the mutants have agonistic activity at Trk receptors. In a serum-deprivation cell culture model, the NFs promote survival. The bioactivity of the mutants was compared versus the corresponding wild-type NFs as positive controls (NT3 for TrkC-expressing cells, BDNF for TrkB-expressing cells, and NGF for TrkA-expressing cells). Concentrations of the NFs at low (0.2 nM, suboptimal) or high (2 nM, optimal) molarity were tested versus untreated control. The 0.2 nM and 2.0 nM ligand concentrations were predefined, respectively, as suboptimal and optimal for these cells in these assays (Ivanisevic et al., 2003; Zaccaro et al., 2005; Ivanisevic et al., 2007; Guillemard et al., 2010; Brahimi et al., 2014).
In TrkC-expressing cells, mutant RK promoted survival significantly better than wild-type NT3 (P < 0.01) (Fig. 2A). The D and DRK mutants promoted survival better than NT3 but without reaching statistical differences. These data indicate that mutations expected to enhance TrkB interactions could also have a positive impact on TrkC-mediated trophic survival because these cells do not express TrkB or p75.
In TrkB-expressing cells the mutants D and DRK promoted survival above untreated control cells. Mutant DRK was significantly better than mutant D (P < 0.05) indicating enhanced TrkB activation. The mutant RK and control wild-type NT3 did not induce significant survival at 0.2 nM or 2 nM (Fig. 2B), or even at the higher 10 nM concentration (10 nM; not shown).
In TrkA-expressing cells, neither the mutants nor wild-type NT3 promote survival at the concentrations tested (Fig. 2B). This is consistent with previous reports that wild-type NT3 promotes significant survival via TrkA only at concentrations greater than 75 nM (Ivanisevic et al., 2003; Ivanisevic et al., 2007).
The survival data (Fig. 2) are generally consistent with the binding data (Table 1; Fig. 1) and validate the concept that mutants D and DRK retain full or enhanced TrkC activation and also have enhanced TrkB activation.
However, it is noteworthy that binding and bioactivity assays are not always aligned; this is a recurrent theme in neurotrophin biology (Ivanisevic et al., 2003; Ivanisevic et al., 2007). For example, at the concentrations tested, both wild-type NT3 and DRK bind to TrkA but do not promote survival in biologic assays. Also, compared with wild-type NT3, mutant RK promotes enhanced TrkC survival without enhanced binding to TrkC (or enhanced biochemical signals; see below). Perhaps this is due to the RK mutant being more stable at 37°C, a feature that could be evident in the long-term biologic assays (48 hours) at cell culture temperatures.
Mutants Induce Differentiation via TrkC
In addition to promoting survival signals, NT3 promotes neurite growth and synapse formation. Hence, we tested TrkC-mediated induction of neurite outgrowth in nnr5-TrkC cells, which differentiate in response to NT3.
Given that nnr5-TrkC cells express TrkC and p75 and that p75 can impact negatively or positively on Trk agonists (Ivanisevic et al., 2003; Zaccaro et al., 2005; Ivanisevic et al., 2007; Guillemard et al., 2010; Brahimi et al., 2014) we first evaluated function on these cells using survival assays.
The nnr5-TrkC cells survive equally well in response to wild-type NT3 and to the mutants D, RK, and DRK (Fig. 3A). However, we note that in cells lacking p75 the mutant RK promoted TrkC-mediated survival significantly better than wild-type NT3 (Fig. 2A), and this observation remains intriguing.
In differentiation assays the mutants and wild-type NT3 all increased differentiation, defined as the percent of cells bearing more than two axons with axonal length greater than two cell bodies (Fig. 3B). In negative controls, treatment with mouse IgG or 10 nM NGF did not induce differentiation. Quantification demonstrates that the mutants and wild-type NT3 have similar neuritogenic activity (Fig. 3C). These data indicate that expression of p75 does not impact on the TrkC-mediated neurogenic differentiation induced by RK and DRK although these mutants have reduced affinity for p75. The lack of impact by p75 on the ability of mutants to induce differentiation (Fig. 3, B and C) is consistent with lack of impact by p75 on the ability of mutants to promote equivalent survival (Fig. 3A).
Mutants Induce Signal Transduction via TrkC and TrkB
As a correlate of the biologic endpoints of cell survival and cell differentiation, we evaluated biochemical endpoints of signal transduction in Western blot studies using specific antibodies to quantify signaling pathways.
The phosphorylation of TrkC (pTrkC) or TrkB (pTrkB), and the phosphorylation of downstream effectors AKT (pAKT) and pERK1/2 were quantified (Fig. 4). For quantification, densitometry data were standardized relative to actin and to total AKT, and the bands corresponding to pERK1/2, pAKT, and pTrks were quantified as a function of wild-type NT3 positive control. As a positive internal control for each cell type, the corresponding wild-type NFs was used (NT3 for TrkC, BDNF for TrkB, and NGF for TrkA).
In TrkC-expressing cells (Fig. 4A), wild-type NT3 and all the mutants afforded significant pTrkC and activated the downstream pERK1/2 and pAKT. In TrkB-expressing cells (Fig. 4B), D and DRK were the most potent, followed by wild-type NT3 and to a lesser degree RK. These assays are consistent with the survival data for TrkC- or TrkB-expressing cells.
In TrkA-expressing cells (Fig. 4C), wild-type NT3 and mutants D and DRK activated biochemical signals that, although statistically significant, represent a small fraction of the positive control NGF-activated signals. This may explain why the agents did not promote detectable survival of TrkA-expressing cells in biologic assays (see Fig. 2B).
As independent confirmation of the biochemical endpoints, Trk receptor activity was assessed by AlphaLISA, a commercial assay that quantifies Trk-dependent MAPK (pErk1/2) activation after stimulation with the various ligands or controls for 15 minutes.
In cells expressing human TrkB, the mutants showed a trend toward lower EC50 values when compared with NT3, with similar percent maximal stimulation values (Fig. 5A; Table 2). In cells expressing human TrkC, all three mutants produced concentration-response curves for ERK1/2 activation that were similar to NT3, with equivalent EC50 and percent maximal stimulation values (Fig. 5B; Table 2) and a trend toward lower EC50 values with respect to BDNF.
p75 Activation by RK and DRK Mutants
In cells expressing p75 in the absence of Trks, ligand-dependent p75 activation leads to an increase in TNFα mRNA and protein. We studied the mutants for their ability to promote TNFα production in HEK293 cells expressing p75. At 10 nM all mutants induced significant TNFα expression compared with untreated control, and not different from the positive control lipopolysaccharide (1 µg/ml). The TNFα expression induced by mutants RK and DRK was lower but not significantly different from wild-type NT3 (P = 0.09, one-way ANOVA) (Fig. 6). This is somewhat surprising considering that RK and DRK mutants have lower p75 binding than wild-type NT3.
Unfortunately, this in vitro assay requires high ligand concentrations (10 nM NF) and is not robust to evaluate lower agonist concentrations (e.g., at 1 nM ligands do not induce TNFα expression significantly over background control). At high ligand concentrations the differences in p75 activation may not be evident (Barcelona and Saragovi, 2015). Nonetheless, the relatively lower induction of TNFα expression, and the enhanced Trk activation profiles were encouraging for evaluating the agents in vivo.
Neuroprotective Efficacy of Mutants in In Vivo Models of Retinal Neurodegeneration
We evaluated the therapeutic action of the mutant NFs in two in vivo mouse models of retinal neurodegeneration: the optic nerve axotomy causing rapid and acute neuronal death (Fig. 7), and the diabetic retinopathy model causing slower and chronic neuronal death (Fig. 8). In these in vivo models, activation of TrkA or TrkB (expressed in neurons) is neuroprotective, whereas activation of p75 (upregulated in glia) causes neuronal death through elevation of proinflammatory TNFα (Bai et al., 2010a; Barcelona and Saragovi, 2015).
DRK Mutant Prevents Retinal Ganglion Cell Death in Optic Nerve Axotomy
Optic nerve axotomy is a model of acute injury in which the optic nerve is completely severed, causing the RGCs to die and their cell bodies in the retina to degenerate rapidly (Fig. 7). Equal protein concentrations of test agents were injected intravitreally (1.6 μg/2 μL), and RGC survival was quantified at the 14-day postaxotomy endpoint.
In these experiments the right eye of each mouse was injured, whereas the uninjured left eye represents 100% RGCs in that individual mouse. Randomized groups of mice received injections of treatments or controls (n = 8 per group). The percent RGC survival data (right eye/left eye ratio) of each single mouse was averaged by group.
In the PBS-treated (control) axotomy group, only ∼10% of the RGC cell bodies remained detectable. Injection of PBS control had no effect (9.2 ± 1%). Mutant RK (12.9 ± 1.0%) and control wild-type NT3 (11.6 ± 1.0%) did not preserve RGCs compared with controls, likely due to the poor TrkC expression in RGCs.
Significant preservation of RGCs was promoted by mutant D (19 ± 1.3%) (P ≤ 0.0001 versus PBS control group) and even more significantly by mutant DRK (27.7 ± 1.8%) (P ≤ 0.0000061 versus PBS control group). Mutant DRK was significantly better than D, likely due to lower p75 binding by DRK, since is the only difference between these mutants. Indeed, mutant DRK is significantly better than wild-type BDNF (P ≤ 0.001) (Fig. 7), which is the gold standard growth factor for neuroprotection in this model.
Ligands Lacking p75 Binding Are More Efficacious at Preserving Retinal Ganglion Cells After Optic Nerve Axotomy
To further evaluate the consequence of ligands binding to p75, and as a benchmark comparison, we used mutant NGF-C. NGF-C is a previously reported mutant of NGF that activates TrkA but does not bind to p75. NGF-C is reported to be significantly protective of RGCs in the optic nerve axotomy model, whereas wild-type NGF is not protective (Bai et al., 2010a).
The DRK mutant and NGF-C are significantly protective when compared with untreated or to vehicle controls. Together these data support the concept that after injury in vivo activation of a Trk receptor while circumventing p75 binding is beneficial, as demonstrated when comparing mutant DRK with mutant D and comparing wild-type NGF versus NGF-C.
Mutants Prevent Loss of RGC Nerve Fibers and Synapses in Diabetic Retinopathy
We then compared the neuroprotective effect of NFs or mutants in vivo in a chronic degenerative mouse model of diabetic neuropathy (Fig. 8). Retinopathy affects both eyes. Hence, mice received an intravitreal injection of the indicated agent in one eye and PBS vehicle (control) in the contralateral eye to represent maximal degeneration in that individual mouse. This design provides with the ability to internally standardize the data to the untreated diabetic eye (100% damage) in each animal. Then, the groups are averaged ± S.D. Drug delivery was at week 2.5 of diabetes, at a time of ongoing degeneration.
To evaluate long-term effects in this chronic disease model, endpoint and quantification was at the 6-week time point. This is a time of significant retinal degeneration, but the animals remain relatively healthy in spite of persistent hyperglycemia. Quantification of retinal structures was done by OCT, a technique that quantifies the thickness of the Nerve fiber layer (axons), the Ganglion cell layer (RGC cell bodies) and the Inner Plexiform layer (synapses), which we termed NGI. Quantification of the NGI thickness is validated as a marker of RGC and synaptic health (Bai et al., 2010a; Barcelona et al., 2016; Galan et al., 2017; Barcelona et al., 2018). In all eyes, the OCT measurements were taken at an equivalent geographic topology, guided by fundoscopy markers and confirmed by measuring the thickness of the ONL at the same locations. The ONL is not affected at 6 weeks of diabetes, and its thickness serves as internal control (Platon-Corchado et al., 2017).
Quantification of the neuronal structures of the retina at 6 weeks after onset of diabetes showed that mutant DRK achieved a significant ∼37% preservation (P < 0.001 versus untreated) (Fig. 8, A and B). Mutant D achieved a significant ∼17% protection (P < 0.05 versus untreated) (Fig. 8C). In the diabetes model the efficacy of the DRK and D mutants is not significantly different when compared with each other, but DRK is significantly better than D when they are compared with the untreated group. Moreover, in the diabetes model mutant DRK performs significantly better than BDNF control (P < 0.05).
Together, the data for the optic nerve injury and the diabetic retinopathy models indicate that DRK has overall better efficacy than the other mutants or than wild-type NFs and induces a prolonged physiologic response leading to the preservation of retinal neurons and retinal structures in acute and in chronic models of disease.
Discussion
Strategies to achieve neuroprotection in neurodegenerative states have evolved over the last four decades. Approaches based on treatments with NFs have to account for the biology of these systems. In the present studies, we explored the idea that it is possible to engineer desirable features into a neurotrophin molecule that would improve its ability to protect neuronal function in vivo.
This was based on the idea that a molecule possessing agonist activity at more than one Trk receptor would have the benefit of enhancing the survival and function of a wider range of neurons, or potential synergistic effects within a neuron expresses more than one Trk receptor subtype. These benefits would be further enhanced by lowering affinity for p75 and thereby reducing the deleterious consequences of activating this receptor, particularly under pathologic conditions when p75 is upregulated (Josephy-Hernandez et al., 2017; Saragovi et al., 2019).
The present work shows in models of retinal degeneration the improved therapeutic efficacy of DRK, a pan-NF that can activate multiple Trks while bypassing p75. In terms of mechanisms that explain a benefit, signal transduction data indicated that both the D and DRK mutants have a greater ability to activate TrkB in terms of autophosphorylation and stimulation of the downstream ERK and AKT pathways. This was evident for pERK1/2 in cells expressing human TrkB where in dose-response analyses DRK showed a trend toward a lower EC50 (i.e., greater potency) than NT3. Signal transduction was similar between the mutant NFs and NT3 in TrkC assays; however, in assays of TrkA signal transduction, again, D and DRK showed increased activity relative to NT3. The overall picture emerging from these studies is that the D and DRK mutants retain the ability to activate TrkC and have improved functional activity for TrkB and TrkA relative to the parent NT3 and are thus pan-Trk agonists. This is consistent with the original report that the D15A mutant of NT3 increases the affinity of NT3 for TrkB (Urfer et al., 1994).
With regards to p75, the RK and DRK mutants showed significantly less affinity than NT3 in ligand displacement experiments. However, all three mutant NFs stimulated TNFα production in p75-expressing cells. Although the NFs with the RK mutation did so to a lesser extent, the difference was not significant in this assay that requires high concentration of test agent (preventing detailed dose range studies). Consequently, the data from the binding and signal transduction assays are consistent with the D mutation providing improved agonist activity at TrkB and the RK mutation, potentially providing for reduced p75 binding and activation.
In cell survival assays, the consequence of the improved TrkB agonist activity of D and DRK translated into improved activity relative to NT3 in TrkB-expressing cells, and surprisingly all three mutants had superior survival activity relative to NT3 in TrkC-expressing cells. Although the signal transduction assays indicated improved activity in TrkA-expressing cells for D and DRK, this did not translate into improved survival of TrkA-expressing cells, likely because the level of activity remained below a critical threshold for survival under these conditions.
All three mutant NFs promoted survival and differentiation in nnr5-TrkC cells that express both TrkC and p75, indicating that this activity was retained. Indeed, neurite outgrowth and complexity were also shown to be improved with D, RK, and DRK in rat spiral ganglion neurons that express both TrkB and TrkC (Szobota et al., 2019; unpublished data), with mutant DRK showing robust activity.
In Vivo Therapeutic Paradigms
Taken together, these data predict that the mutant NFs and in particular DRK, which combines pan-Trk activity with reduced p75 affinity, have activity profiles that bode well for improved efficacy in vivo. The two in vivo models of retinal neurodegenerative disease (and the ex vivo model of hearing loss) suggest that these disorders may be indications suitable for treatment with mutant DRK.
In the in vivo experimental paradigms we apply the therapeutic agents once only, after injury. In the optic nerve axotomy model the agents are administered shortly after injury because this is an acute model with a short therapeutic window. In the diabetic retinopathy model the agents are administered after 2.5 weeks of continuous injury, as this is a chronic model. Drug dosing and timing were selected to better represent a real-life scenario of intervention in trauma versus a chronic disease that requires a diagnosis.
In optic nerve axotomy DRK is better than D. In the diabetic retinopathy model, where injury is chronic but modest and subtle, it would be difficult to see differences between D and DRK. However, a comparison of DRK or D versus controls (BDNF positive control or untreated negative control) shows that DRK provides a more significant benefit than D.
In addition, we aimed to avoid administration of a Trk-agonist able to fully activate p75, especially in chronic diseases that require chronic drug administration. For example, evaluation of wild-type NFs in optic nerve injury shows that higher frequency of administration results in higher toxicity unless p75 action if blocked. For example, wild-type NGF given two times (days 1 and 7 after optic nerve injury) is ineffective, even though more drug is administered. In contrast, wild-type NGF given once (immediately after injury) or twice (immediately after injury and at day 7 postinjury) given together with p75 antagonist (or in p75 knockdown or knockout mice) is effective and provides a more than additive benefit than when using each agent alone (Shi et al., 2007; Lebrun-Julien et al., 2009b; Bai et al., 2010a).
In a different approach (Enomoto et al., 2013) Schwann cells transfected with the cDNA of a mutant NFs similar to DRK were implanted in spinal cord at the time of spinal cord damage, aiming to achieve high local expression of the mutant NF concomitant with injury. That work reported more surviving Schwann cells and sensory fibers but did not result in significant functional locomotive improvement or recovery. In our work soluble proteins are used as the therapeutic agents, with the DRK mutant protein being efficacious at a relatively low dose and given at low frequency (single administration), even when the treatment was administered after neuronal injury and with ongoing degeneration. Moreover, in vivo the DRK mutant protein was more effective than wild-type NT3 or than the related mutants D and RK.
In summary, we present data supporting the concept that the administration of NF molecules that are capable of broad Trk activation while at least partially avoiding p75 activation is beneficial for neuronal survival, regeneration, and delay of disease progression. These concepts may help to rationalize therapeutic strategies for neurodegeneration.
Authorship Contributions
Participated in research design: Brahimi, Galan, Foster, Saragovi.
Conducted experiments: Brahimi, Galan, Siegel, Szobota.
Performed data analysis: Brahimi, Galan, Sarunic, Foster, Saragovi.
Wrote or contributed to the writing of the manuscript: Brahimi, Galan, Siegel, Szobota, Foster, Saragovi.
Footnotes
- Received April 14, 2021.
- Accepted August 12, 2021.
This work was supported by a grant from the Canadian Institutes of Health Research (Pharmacology), in part by the US Department of Defense [Optic Nerve Trauma W81XWH1910853], and in part by the Canadian Consortium on Neurodegeneration in Aging (CCNA) to H.U.S.
Abbreviations
- BDNF
- brain-derived neurotrophic factor
- ERK
- extracellular signal-regulated protein kinase
- FITC
- fluorescein isothiocyanate
- mAb
- monoclonal antibody
- MAPK
- mitogen-activated protein kinase
- MCF
- mean channel fluorescence
- MTT
- 4-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide
- NF
- neutrophin
- NGF
- nerve growth factor
- NGI
- nerve fiber layer,ganglion cell layer,and inner plexiform layer
- NT3
- neurotrophin 3
- OCT
- optical coherence tomography
- OD
- optical density
- ONL
- outer nuclear layer
- pAkt
- phosphorylated Akt
- pErk
- phosphorylated ERK
- pTrk
- phosphorylated Trk
- RGC
- retinal ganglion cell
- SFM
- serum-free medium
- STZ
- streptozocin
- TNFα
- tumor necrosis factor α
- Trk
- tropomyosin receptor kinase
- Copyright © 2021 by The American Society for Pharmacology and Experimental Therapeutics