ABSTRACT
In addition to their cell autonomous activities, a number of homeoprotein transcription factors transfer between cells and regulate gene transcription, protein translation and chromatin organization in a non-cell autonomous way. ENGRAILED-1 homeoprotein is endowed with the latter properties and is expressed in spinal cord V1 inhibitory interneurons that synapse on large spinal cord α-motoneurons. Based on the neuroprotective effects of several homeoproteins in the adult central nervous system, we have analyzed the motor phenotype of mice lacking one functional Engrailed-1 allele. Engrailed-1 heterozygote mice with Engrailed-1 expression reduced by half in the spinal cord start to show muscle weakness, abnormal spinal reflex and partial neuromuscular junction denervation between 2 and 3 months of age. Alpha-motoneuron degeneration is first observed at 4.5 months and progresses with time, reaching 50 per cent in 15.5-month-old mice. A single intrathecal injection of exogenous recombinant human ENGRAILED-1 at 3 months in the lumbar enlargement allows for an addressing of the protein to lumbar α-motoneurons and, to a lesser degree, to cervical ones. This results in full restoration of strength and extensor reflex for at least 2 months, in parallel with a preservation of the number of lumbar α-motoneurons, and blocks endplate denervation. The similarities of the Engrailed-1 heterozygote phenotype to motor neuron disease symptoms and the long-lasting effects of a single ENGRAILED-1 injection suggests that this approach may be of interest in the search for therapies alleviating the consequences of α-motoneuron degeneration.
INTRODUCTION
Homeoprotein transcription factors are encoded by a class of genes discovered on the basis of their developmental functions 1. More recently, it was shown that several of them are also expressed in the adult, suggesting that they participate in physiological homeostasis 2,3. Similarly to what has been observed during development, adult homeoprotein (HP) functions involve both cell autonomous and non-cell autonomous properties 4,5. Non-cell autonomous properties arise from the ability of several HPs to transfer between cells due to two highly conserved sequences present in their DNA-binding domain or homeodomain 6–9. Examples of non-cell autonomous activities in the central nervous system, include the regulation of cerebral cortex plasticity and retinal ganglion cell (RGC) survival by OTX2 10–12, the control of oligodendrocyte precursor cell and Cajal-Retzius cell migration by PAX6 13,14 and the axon guidance activities of ENGRAILED-2 and VAX-1 15–17.
An important ENGRAILED-1 (EN1) function is to ensure the survival of mesencephalic dopaminergic (mDA) neurons of the Substantia Nigra pars compacta (SNpc). In En1 heterozygote (En1-Het) mice, mDA neurons die progressively starting at post-natal week 6 and only approximately 60% of them remain in one-year-old mice 18. Infusion or injection of EN1 or EN2 proteins at the level of the SNpc and their internalization by mDA neurons rescues these neurons from degeneration in several Parkinson Disease mouse models 18–20 and in a primate model 21. Following a single EN1 injection in mice, mDA rescue lasts for several weeks, a long-lasting effect reflecting that EN1, in addition to directly regulating transcription and translation 5,22,23, also works as an epigenetic chromatin remodeler 20,24.
In the ventral spinal cord, En-1, but not En2, is expressed in V1 inhibitory interneurons 25–29 which form inhibitory GABAergic and glycinergic synapses on large α-motoneurons (αMNs) 30,31 and regulate alternating flexor-extensor activity 28,32. They contribute to the excitatory/inhibitory balance with excitatory V0 interneurons and the maintenance of this balance may be important for αMN survival 30.
In light of our studies demonstrating that EN1, EN2 and OTX2 have pro-survival activities 5, we investigated whether EN1 expressed by V1 interneurons may be important for αMN physiology and survival with which they are in synaptic contact. We demonstrate that En1 expression is maintained in adulthood and that En1-Het mice have early muscle weakness and abnormal spinal reflex associated with partial denervation of the neuromuscular junction (NMJ) and a later loss of large αMNs. These degenerative changes are progressive, increasing between 2 and 15.5 months of age. A single intrathecal injection of human recombinant EN1 (hEN1) at 3 months restores muscle strength, normalizes spinal reflex and NMJ innervation and promotes survival of αMNs. This restorative and protective activity lasts for at least 2 months.
MATERIALS AND METHODS
Animal Management
All animals were treated in accordance with the guideline for the care and use of laboratory animals (US National Institutes of Health), the European Directive number 86/609 (EEC Council for Animal Protection in Experimental Research and Other Scientific Utilization) and French authorizations n° 00703.01 “Therapeutic homeoproteins in Parkinson Disease” and n° APAFIS #6034-2016071110167703 v2, “Spinal cord motoneurons neuroprotection” delivered by the Minister of higher education, research and innovation.
Adult mice were housed two to five per cage and maintained with ad libitum food and water, under a 12h light/dark cycle. Transgenic mouse strain En1-Het was bred by the Rodent Breeding Services provided by the Animal Care Services at College de France. Colony maintenance and genotyping were performed as previously described 33. Both, females and males were included in all the studies.
Assessment of disease progression and humane end-point
Mice were considered for euthanasia if they displayed 15% loss of bodyweight or showed signs of paralysis, whichever was reached first. None of the mice were euthanized due to these reasons.
Behavioral analyses
The tests used to evaluate muscle strength and spinal reflexes are forepaw grip strength test, inverted grid test and hindlimb extension reflex. Mice were habituated to the behavioral room and experimenter 24 hours before the day of testing and again 1 hour before each behavioral test. All tests were performed the same day and behavioral assessment was carried by evaluators blind to genotype and treatment. Mice were weighed before each session.
Forepaw Grip Strength
The Transducer (IITC Life Science Grip Strength Meter, ALMEMO 2450 AHLBORN, World Precision Instruments) was calibrated and the scale of the values set to grams. During the test, each mouse was lifted by the tail to position the front paws at the height as the bar (about 15 cm) and moved towards the bar. Symmetric firm grip with both paws was verified and the mouse was pulled backward at a constant speed until the grasp was broken and the maximal value was recorded. The test was repeated 5 times per animal with a minimal resting time of 5 minutes between tests and the mean of all values was normalized to the weight of each animal. At least 4 mice per group per experiment were used.
Inverted Grid Test
The mouse was placed on a wire grid (15×10 cm) and let to freely explore it. After 3-5 minutes, the grid was raised 30 cm above a soft surface and gently turned upside down. Latency to falling was recorded three times per mouse with a minimum resting time of 5 minutes between trials. The longest latency was used for further analysis.
Hindlimb Extensor Reflex
Mice were gently suspended by the tail at a constant height (30 cm) and scored for hindlimb extension reflex. The scores were assigned from 0 to 3 as follows: 3 normal symmetric extension in both hind limbs without visible tremors; 2.5 normal extension in both hind limbs with tremor in one or both paws; 2.0 unequal extension of the hind limbs without visible tremors; 1.5 unequal extension in the hind limbs with tremors in one or both paws, 1.0 extension reflex in only one hindlimb, 0.5 minimum extension of both hindlimbs, 0 absence of any hindlimb extension.
Tissue preparation
Spinal cord
Adult mice were euthanized by an overdose of Dolethal (Pentobarbital, Vétoquinol) (1μl/g body weight). The spinal cord was dissected after laminectomy and placed in PBS (20mM PO4, 155mM NaCl) to remove the meninges and surrounding connective tissue. Cervical and Lumbar enlargements were separated and placed in Paraformaldehyde 4% (PFA, Thermo Scientific) for 1 hour at room temperature with mild mixing. The tissue was then washed in PBS, three times for 30 minutes at room temperature and placed in PBS, 20% sucrose overnight at 4°C. After cryoprotection, the spinal cords were embedded in Tissue Freezing Medium (TFM, Microm Microtech), frozen on dry ice and 30μM sections prepared on an HM 560 Microm cryostat (Thermo Scientific).
Muscle
The extraocular (EOM), tongue and lumbrical muscles from the plantar surface of the paw were dissected into cold PBS and fixed at room temperature in 4% PFA for 10 minutes for extraocular and lumbrical muscles or 1 hour for the tongue. The muscles were then washed in PBS and cryoprotected. Extraocular and lumbrical muscles whole-mounts were stained to allow for visualization of the entire innervation pattern and detailed analysis of the neuromuscular junctions 34. Tongue muscles were embedded and sectioned at 30μm.
Cresyl violet staining
Slides with 30μM spinal cord sections were washed in PBS (1X) 3 times, cleared in O-Xylene (CARLO-HERBA) for 5 minutes, then hydrated in graded alcohols with increasing water and placed in Cresyl Violet acetate (MERCK). Sections where then dehydrated in increasing alcohols and mounted.
RT-qPCR
Spinal cords were removed as above and cervical and lumbar enlargements were separated and rapidly frozen on dry ice. Total RNA was extracted (RNeasy Mini kit, Qiagen) and reverse transcribed using the QuantiTect Reverse Transcription kit (Qiagen). RT-qPCR were done using SYBR-Green (Roche Applied Science) and Light Cycler 480 (Roche Applied Science). Data were analyzed using the « 2-ddCt » method and values were normalized to Glyceraldehyde 3-phosphate dehydrogenase (Gapdh). The following primers were used: Engrailed-1 sense: CCTGGGTCTACTGCACACG, antisense: CGCTTGTTTTGGAACCAGAT; Gapdh sense: TGACGTGCCGCCTGGAGAAAC, antisense: CCGGCATCGAAGGTGGAAGAG
Spinal cord and muscle immunofluorescent labeling
Slides with 30μm spinal cord or muscle sections were washed in PBS and permeabilized with 2% Triton. After 30 minutes at RT in 100μM glycine buffer, sections were blocked in 10% Normal Goat Serum (NGS, Invitrogen) or Fetal Bovine Serum (FBS, Gibco) in the presence of 1% Triton and incubated with primary antibodies (Sheep anti-Choline Acetyltransferase -ABCAM, 1:1000− or goat anti-Choline Acetyltransferase -Millipore, 1:500-, rabbit anti-Engrailed1 −1:300-, mouse anti-neurofilament 165kDa -DSHB, 1:50− and mouse anti-synaptic vesicle glycoprotein 2A -DSHB, 1:100-) overnight at 4°C, washed and further incubated with corresponding secondary antibodies for 2 hours at room temperature. For muscle staining, α-bungarotoxin (Alexa fluor 488 conjugate) was used at the same time as the secondary antibodies. Slides were washed and mounted with DAPI Fluoromount-G® (Southern Biotech). Controls without primary antibodies were systematically included.
Protein production
Human EN1 (hEN1) was produced as described 11 and endotoxins were removed by Triton X-144 phase separation 35. In brief, pre-condensed 1% Triton X-144 (Sigma) was added to the protein preparation. The solution was incubated 30 minutes at 4°C with constant stirring, transferred to 37°C for 10 minutes and centrifuged at 4000 rpm for 10 minutes at 25°C. The endotoxin-free protein was aliquoted and kept at −20°C.
Intrathecal injections
Mice were anesthetized with ketamine (Imalgene 1000) and xylazine (Rompur 2%) in NaCl 50mM and placed on the injection platform. The tail of the animal was taken between two fingers of one hand and its back was gently flattened with the other hand. The L3 vertebral spine was identified by palpation and a 23G x 1” needle (0.6×25mm Terumo tip) was placed at the L1 and T13 groove and inserted through the skin at an angle of 20° (Hylden and Wilkox, 1980). The needle was slowly advanced forward to the intervertebral space until it reached the injection point, provoking a strong tail-flick reflex. Five μl were injected at a rate of 1μl/min with or without 1μg recombinant protein. The needle was left in place for two minutes after injection and then slowly removed. Animals were placed in a warmed recovery chamber until awakening. Extensor reflex and gait analysis were examined 2 and 24 hours after injection to ascertain the absence of spinal cord damage.
Image analyses
Cresyl violet stained spinal cord section images were acquired with a Nikon-i90 microscope under bright field conditions at 10x with a Z-stack step of 0.5μm. Immunofluorescence stained spinal cord sections images were acquired with a Leica SP5 inverted confocal microscope at 20x (Leica DMI6000) and acquisitions of 3D z-stack (0.5μM) were made using the UV (405 nm, 50mW), Argon (488 nm, 200mW) and DPSS (561 nm, 15mW) lasers. For MN quantification, at least five spinal cord sections separated by ≥ 900μm were analyzed for each animal. ChAT+ cells and Cresyl stained cells with a cell soma area greater than 100μm2 were manually outlined using ImageJ software (NIH, Bethesda, MD) and their area determined. Analyses were carried out on Z-stacks through the entire 30μm thickness of the section. Cells were classified as small (100-199μm2 cross sectional area), intermediate (between 200-299 μm2) and large (greater than 300μm2).
Lumbrical, EOM and tongue muscles were imaged with a Leica SP5 inverted confocal microscope (Leica DMI6000) with a motorized XY stage. 3D z-stack (0.5μM) acquisitions were made as indicated above (UV, Argon and DPSS) and images analyzed using ImageJ software. Analyses were performed blind to the genotype and treatment. Endplates where categorized as fully innervated (neurofilament overlying more than 80% of the endplate), partially innervated (neurofilament overlying up to 80% of the endplate) or denervated (no neurofilament overlying the endplate) 36. Endplate morphology was evaluated by counting the number of endplates with perforations (areas lacking α-bungarotoxin staining). For postsynaptic analysis, each endplate was manually outlined using ImageJ software and area calculated. All analyses were done on the entire Z-stacks through the NMJ.
Statistical analyses
Results are based on seven independent experiments. Muscle weakness and abnormal reflex score in 3-month-old En1-Het mice was observed in 6-7 of the 7 experiments, NMJ denervation in 5 of 5 experiments and αMN loss in 4 of 4. The restoration of muscle strength and normalization of the spinal reflex after hEN1 injection was observed in 5 of 5 experiments and a significant increase in αMN survival in 4 of 4 experiments.
Data are expressed as mean ± SD unless otherwise indicated. Statistical significance was determined as indicated. For RT-qPCR, WT and En1-Het mice were compared by Unpaired T-test with equal SD. For behavioral and NMJ analyses and MN counting in the time-course study, groups were compared by Unpaired T-test with equal SD comparing WT with En1-Het for each time point. For the intrathecal injections, behavioral and NMJ analyses and MN counting, experimental data were compared by one-way ANOVA followed by a post hoc Dunnett’s test for comparisons to WT. For behavioral analysis in the time-course protection of injected hEN1, groups were compared by Unpaired T-test with equal variances comparing WT with En1-Het injected at each time point. For behavioral analysis in the time-course protection experiment following the injection of hEN1 in 9-month-old mice and of the hEN1 mutant hEN1Q50A, groups were compared by One-Way-ANOVA with Tukey’s multiple comparisons test comparing the groups for each time point.
RESULTS
En1-Het mice show early and progressive motor deficits
Previous studies reported En1 expression in V1 interneurons in the early postnatal mouse spinal cord 29. We verified that this expression is maintained at the same levels throughout life and is decreased by twofold in the En1-Het mouse both at 4.5 and 16 months, thus during the time frame of all experiments (Fig. 1a, b). This indicates that V1 interneurons survival is preserved in the En1-Het mice, in contrast with the situation for mDA neurons of the SNpc 18. En1-Het mice display the same average weight gain as their WT littermates at all ages studied (Fig. 1c). For the three tests used to assess muscle strength, WT and En1-Het animals evolved differently, the latter showing signs of loss of strength compared to their WT littermates (Fig. 1d-f). At one and two months of age forepaw grip strength was similar in En1-Het and WT mice, but a significant reduction was first measurable at 3 months that persisted through 15.5 months of age (Fig. 1d). The inverted grid test assesses the ability of the four limbs to resist gravity. Two-month-old En1-Het mice resisted significantly less than WT mice and holding time further diminished between two months and 15.5 months, revealing a progressive loss of strength (Fig. 1e). Finally, the hind limb extensor reflex was taken to monitor purely spinal function. WT mice had an extensor score of about three at all ages but the score of the En1-Het animals started to deteriorate overtime between 2 and 15.5 months (Fig. 1f). Taken together, these results show that En1-Het mice have a progressive loss of muscle strength and a deteriorating spinal reflex compared to WT siblings.
Presynaptic pathology in the neuromuscular junction of the lumbrical muscles in En1-Het mice
Motor deficits are often accompanied by early synaptic changes at the NMJ, a specialized synapse between αMN axon terminals and skeletal muscle fibers. Since NMJs are highly vulnerable from very early stages of motor disorders, we analyzed the NMJ morphology and synaptic characteristics of three muscle groups with different degrees of vulnerability in neuromuscular diseases 36. Data corresponding to lumbrical endplates are described in Figure 2 and those corresponding to extraocular and tongue muscles in Supplemental Figure 1.
Alpha-bungarotoxin and antibodies against neurofilament and synaptic vesicle protein 2A were used in combination to visualize acetylcholine receptor (AChR) clusters of the motor endplate and motor axon terminals, respectively (Fig. 2a). The typical endplate morphology well centered on the lumbrical muscle fibers was observed in all mice examined 34. No changes were found in the average number of AChR clusters between WT and En1-Het mice (Fig. 2b). The area of the motor endplate did not differ between WT and En1-Het mice throughout 9 months of age, and is significantly reduced in En1-Het mice at 15.5 months of age (Fig. 2c). Endplate perforations (post-junctional folds) are a region of the endplate devoid of AChR. Their number reflects the state of maturation of the NMJ and normal healthy muscles show a complex morphology with multiple perforations 37. The percentage of endplates with perforations shows no difference between WT and En1-Het mice throughout 9 months of age (Fig. 2d) but is significantly reduced in the En1-Het mice at 15.5 months of age. Finally, the lumbrical NMJ innervation was evaluated. Motor endplates where categorized as fully innervated if neurofilament immunoreactivity overlaid more than 80% of the endplate. Figure 2e shows the significantly reduced NMJ innervation in En1-Het mice, starting at 3 months and worsening with age.
In contrast with lumbrical endplates, no differences between the two genotypes were observed in any of the NMJ morphological characteristics for extraocular and tongue muscles, at any age (Supp. Fig 1).
Loss of large αMN in the En1-Het mouse spinal cord
The progressive muscle weakness of the En1-Het is paralleled with a partial denervation of the muscle fibers possibly reflecting αMN loss. We thus analyzed the number of neurons present in the cervical and lumbar enlargements of the spinal cord in the En1-Het mice and their WT littermates. At 15.5 months, the total number of Cresyl-stained neurons in the ventral spinal cord was similar for the cervical (Supp. Fig 2) and lumbar enlargements at 15.5 months (Fig. 3a,b). In contrast, a decrease in the number of large densely stained cells was clearly apparent in the En1-Het mouse spinal cord. To understand this, the neurons were classified according to their surface: small neurons of 100-199μm2, intermediate size neurons of 200-299μm2, and large neurons 300μm2 or more. The number of small and intermediate size neurons remained constant at all ages, with no difference between En1-Het and WT mice (Fig. 3c, d) while the number of large neurons decreased progressively between months 4.5 (≈ 20% loss) and 15.5 (≈ 50% loss) in En1-Het animals (Fig. 3e).
Choline Acetyltransferase (ChAT) is selectively expressed in γ and αMNs that can be distinguished on the basis of size 38. In order to better identify the neurons that are lost in the ventral spinal cord of En1-Het animals we used ChAT immunoreactivity in 9-month-old mice. Compared to WT, Figure 4a clearly shows a loss of large ChAT+ cells in the En1-Het spinal cord and Figure 4b-c which compares Cresyl violet and ChAT staining for medium size and large neurons, confirms that the cells that disappear in the En1-Het mutant are αMNs. It is of note that the loss of large αMNs in the En1-Het mouse spinal cord is not accompanied by a corresponding increase in the intermediate size class of MNs, precluding αMN shrinkage.
Taken together, these results show that, compared to WT, En1-Het mice present a loss of large αMNs in the ventral spinal cord that is first apparent at 4.5 months and progresses with time, reaching about 50% at 15.5 months of age.
Recombinant hEN1 reverses the En1-Het motor phenotype, restores NMJ morphology and reduces αMN loss
Previously, we showed that exogenous HP transcription factors, including EN1, EN2 and OTX2, rescue mDA neurons and RGCs from an in vivo stress 11,19,20 and that this survival effect holds for several weeks following a single HP injection 11,20. This led us to test whether a single intrathecal hEN1 injection into the spinal cord of En1-Het mice could have total or partial rescuing activities on any of the analyzed phenotypes. In the En1-Het mice muscle weakness and abnormal extensor reflex appear between 2 and 3 months (Fig. 1) while αMN loss is only visible at 4.5 months (Fig. 3e), suggesting that three months of age is a valuable “therapeutic” window. One μg of hEN1 was injected and 24h later strong EN1 immunofluorescence was detected in the perivascular space and in neurons of lumbar spinal and cervical cord (Fig. 5a, b). ChAT immunofluorescence showed that many of the cells that internalized hEN1 are large-sized αMNs and that the number of hEN1-stained cells was much lower in the cervical region probably due to its distance from the lumbar injection site (Fig. 5b).
Three-month-old En1-Het and WT mice were evaluated for muscle weakness and hindlimb extensor reflex and, 2 days later, En1-Het mice were separated into two groups. A first group received 5μl of vehicle and the other 1μg of hEN1 in 5μl of vehicle without incidence on weight (Fig. 5c). As already shown, before intrathecal injections, En1-Het mice (compared to WT siblings) had reduced forepaw grip strength, reduced time holding onto the inverted grid and had a slightly lower hind limb reflex score (Fig. 5d-f). One month and a half later, En1-Het mice treated with vehicle continued to do worse than WT mice in the three tests whereas the performance of hEN1 injected mice were undistinguishable from those of WT controls (Fig. 5d-f). Synapse morphology and αMN death were analyzed in the 4.5-old-mice injected at 3 months with vehicle or hEN1. The percentage of fully occupied endplates, reduced in En1-Het mice (Fig. 2), returns to normal in hEN1 treated animals (Fig. 6a). At 3 months of age, En1-Het mice have a normal number of αMNs characterized by size and ChAT staining, but some loss has already occurred 1.5 months later (Fig. 3e). Figure 6 (b, c) illustrates that the number of Cresyl-stained large cells and ChAT-stained large cells at the lumbar enlargement is identical at 4.5 months between WT mice and En1-Het siblings injected with hEN1 at month 3.
As a control, a mutant of hEN1 (hEN1Q50A) in which glutamine in position 50 of the homeodomain was mutated into alanine, thus allowing internalization but preventing high affinity binding to DNA 39 was also injected but showed no rescuing activity (Supp. Fig 3).
Taken together, these experiments demonstrate that a single hEN1 injection at month 3 and at the lumbar level restores muscle strength, normalizes the spinal reflex and restores a normal NMJ morphology. In addition, the treatment blocks lumbar, but not cervical αMN degeneration, for a period of at least 1.5 months (Supp. Fig 4). That En1-Het mouse muscle strength is restored in the forepaw grip strength and inverted grid tests suggests that the surviving αMNs in the cervical enlargement have benefited from hEN1 injection.
A single hEN1 injection at 3 months of has long-lasting effects on En1-Het mice and no effect at 9 months
To investigate if the rescue effect of a single hEN1 injection in three-month-old En1-Het mice could last more than 1.5 months, 1μg of hEN1 was injected in three-month-old En1-Het mice that were assessed weekly during the first month following injection and then monthly up to six months. Before injection, three-month-old En1-Het mice showed significant muscle weakness and an abnormal reflex compared to WT mice (Fig. 7a-c). One week after injection the forepaw grip strength and the extensor reflex in the En1-Het mice were normalized to WT values. This improvement was fully maintained for 3 months, after which performance declined progressively to reach pre-injection levels. The improvement in the inverted grid test was more modest but followed a similar time course. When compared to 9-month-old untreated En1-Het mice, mice injected with hEN1 at 3 months and kept for 6 months did not maintain wild type values but performed better in the 3 tests than untreated age-matched En1-Het mice. At the end of the study, muscles and spinal cord tissue were dissected and analyzed. Figure 8 shows that En1-Het mice treated with hEN1 show higher NMJ innervation at 9 months than untreated age-matched En1-Het mice, although less than WT littermates (Fig. 8a). A similar partial rescue at 9 months was also observed for αMN survival (Fig. 8b). These results suggest that a single injection of hEN1 slows the degenerative processes that normally occur in the En1-Het mice.
To examine whether late administration of hEN1 would have positive effects on the En1-Het phenotype, 1μg of hEN1 or vehicle was injected in 9-month-old En1-Het mice. Before injection, En1-Het mice manifested significant reductions in forepaw grip strength, time of holding onto the inverted grid and extensor reflex score (Fig. 7d-f). Compared to WT siblings, vehicle and hEN1-injected En1-Het mice continued to be significantly impaired in the three tests, although, one may notice a very small improvement of the hEN1-treated En1-Het mice, two months post injection, for the forepaw grip strength and extensor reflex tests. At the end of the experiment the now 15.5-month-old mice were dissected to analyze NMJ occupancy and count the surviving αMNs. As shown in Figure 8c-d, injection after 9 months had no effect on either morphological criteria evaluated at the lumbar enlargement or for hindlimb muscles.
DISCUSSION
This study establishes the importance of EN1 expression in αMN survival and motor strength. It also shows that one intrathecal injection of hEN1 can rescue the En1-Het phenotype with a long-lasting activity when the injection is made before degeneration has reached a no-return point, as in 9-month-old En1-Het mice. In this study, we used the human recombinant protein, but the human and mouse proteins are 91.1% identical and were used with the same efficiency in mDA neuron survival experiments 19–21,24.
A day after injection hEN1 is present around vascular structures probably in the perivascular space from which it penetrates into the parenchyma and reaches the cell interior. Indeed, it has been shown before that EN1 and EN2, similarly to many HPs 5,9 translocate across the plasma membrane to reach the cell cytoplasm through a mechanism distinct of endocytosis 8. The examination of the ventral horns suggests a preferential accumulation in neurons, not precluding the possibility that other cells types, in particular astrocytes, also capture the protein. Among the neurons, the large neurons with a morphology evoking αMN are systematically more strongly labeled than the medium and small size ones. Although this requires a more detailed study, it might reveal that, physiologically, EN1 expressed by the V1 neurons is secreted and internalized by αMNs expressing EN1-binding sites.
The latter hypothesis is in line with what was demonstrated in the case of OTX2, a HP which is specifically recognized in the cerebral cortex by a class of interneurons expressing parvalbumin 10. OTX2 internalization is due to its recognition by glycosaminoglycans (GAG) present at the surface of these interneurons thanks to the GAG binding domain (RKQRRERTTFTRAQL) overlapping with the first helix of the homeodomain 40. Interestingly, similar GAG-binding domains are present upstream of the homeodomain in many HPs 3,5, including EN1 and EN2 for which this putative GAG-binding sequence is KEDKRPRTAFTAEQL. The potential role of this domain in the specific targeting of EN1 and the presence of EN1 binding sites at the αMN surface will be investigated in further studies.
A striking observation is that, in the En1-Het mouse, the extensor reflex score is reduced at 2 months and muscle strength between 2 and 3 months. This corresponds to the decrease in fully occupied endplates first observed at 3 months and this precedes the loss of αMNs observed at 4.5 months. This slight temporal discrepancy probably reflects that the αMNs are experiencing retrograde degeneration with the terminals, thus strength, affected before cell body loss. This replicates what occurs in En1-expressing mDA neurons in the same En1-Het mouse line: their terminals show degeneration signs at 4 weeks, whereas the neurons only start dying at 6weeks of age 18,41. In this context, it is important to note that hEN1 injection at 3 months not only prevents αMN death for weeks, but also restores normal endplate morphology and muscle strength. This suggests that the neurodegenerative process is slowed or halted at least for a limited period of time since no recovery is possible after 9 months. This time limit is interesting as the time course of αMN degeneration is very slow with 40% of the cells still alive at 15.5 months and seems to reach a plateau. It is thus possible that αMNs are heterogeneous in term of their dependency to EN1. This is reminiscent of the SNpc and Ventral Tegmental Area (VTA) mDA neurons, the two populations expressing En1, but the latter being more resistant to En1 hypomorphism 18. It is also interesting to note that forelimb muscle strength was restored suggesting that even the low levels of hEN1 that attain cervical spinal cord have beneficial effects on the remaining αMNs.
Homeoproteins can act cell and non-cell autonomously and this raises two possible mechanisms leading to the phenotype observed in the En1-Het mice and protection by exogenous hEN1. A direct cell autonomous effect would mean that spinal V1 interneurons are dysfunctional in the En1-Het mouse. Since these cells are an essential component of the reciprocal inhibitory αMN circuit, loss of their inhibitory input places αMNs at risk for degeneration 42. Deficits in inhibitory interneurons in the spinal cord of motor neuron disease models have been reported 43 and in particular Renshaw cell pathology has also been observed following αMN loss 44. Alternatively, but not mutually exclusive, EN1 might act in a non-cell autonomous fashion. Such non-cell autonomous activities can be of several sorts. Firstly, EN1 may regulate within spinal V1 interneurons the expression of classical trophic factors that support, following their secretion, αMN survival. Secondly, EN1 might be secreted and internalized by intermediate cell types, such as astrocytes or microglial cells, enhancing their αMN-directed trophic activity 45,46. Finally, V1 interneurons may provide EN1 to αMNs where the protein would exert its direct protective activity as observed for OTX2 in RGCs 11. It is possible that EN1 protects the αMNs that internalize it through one or several of the EN1-controled mechanisms that protect mDA neurons of the SNpc from oxidative stress 20,21. Among these mechanisms are the regulation, through local protein translation, of Complex I mitochondrial activity 19, DNA break repair, heterochromatin maintenance and the repression of long interspersed nuclear elements (LINE-1) expression 20,24. The evaluation of these possibilities will be the object of future studies.
The En1-Het mouse presents muscle weaknesses, abnormal spinal reflex, NMJ denervation and αMN loss, all phenotypes highly reminiscent of changes in Amyotrophic Lateral Sclerosis (ALS) patients and in many ALS mouse models. This raises the question as to whether the results reported here are relevant for motor neuron diseases. It is important to note that En1-Het mice do not present a number of important symptoms such as weight loss, respiratory distress, and complete paralysis or endpoint death. The absence of weight loss in the En1-Het mice may be an effect of the Swiss background mouse line. In contrast to virtually all mouse lines whose body weight asymptotes in mid-adulthood (Jackson Laboratory; https://www.jax.org/#) both the En1-Het mice and their WT littermates continue to gain weight reaching about 80 grams at 2 years of age. Thus, it is possible that metabolic defects specific to this genetic background may mask weight losses due to αMN degeneration.
It is also possible that the extensive αMN loss up to about 60% at 15.5 months of age has little or no impact on feeding motor behaviors, although this is difficult to reconcile with the severe phenotypes observed in mouse models of ALS. ALS patients and mice in some ALS models have respiratory or cardiac failure 47–49. We did not observe any obvious breathing difficulty in the En1-Het mice and the long-life span of the mice (at least up to two years thus far) indicates no cardiac failure. The phrenic αMNs that drive respiration are concentrated at the C3-C4 level in mice 45. Thus, interneuron inhibition on phrenic αMNs or αMNs themselves may be insensitive to reduced EN1. At the thoracic level that provides cardiac modulation 47,50,51, inhibitory Ia interneurons are absent 52. We did not assess the number or phrenic αMN in the thoracic spinal cord. It would be interesting to determine if the absence of phrenic and thoracic αMN loss correlates with the absence of breathing difficulties or cardiac function. With regards to ALS patients, since all αMNs derive from the same lineage and transcription factor program, it is possible that phrenic and thoracic αMNs might respond to exogenous hEN1.
Acknowledgements
We thank the College de France Imaging facilities and animal service for assistance in some of the analyses. We are also grateful to Drs. Anne Bousseau and Yoko Arai for very helpful discussion during the study and Angélique Bibimbou for excellent technical assistance.
Footnotes
Conflict of Interest Statement: A.P and K.L.M. are listed on several patents for the use of HPs for the treatment of neurodegenerative disease. They are co-founders and hold shares in a company developing HPs for therapeutic use.
Funding sources: MemoLife Labex PhD fellowship to SEVA, Association Nationale de la Recherche et de la Technologie (Cifre/ANRT n° 2017/0488) to ML, BrainEver, HomeoSign ERC -2013-AdG n°339379