Uman-type neurofilament light antibodies are effective reagents for the imaging of neurodegeneration

Abstract

Recent work shows that certain antibody-based assays for the neurofilament light chain detect informative signals in the CSF and blood of human and animals affected by a variety of CNS injury and disease states. Much of this work has been performed using two mouse monoclonal antibodies to neurofilament light, UD1 and UD2, also known as Clones 2.1 and 47.3, respectively. These are the essential components of the Uman Diagnostics Neurofilament-Light™ ELISA kit, the Quanterix Simoa™ bead-based assay and others. We show that both antibodies bind to neighbouring epitopes in a short, conserved and unusual peptide in the centre of the neurofilament light Coil 2 segment of the 'rod' domain. We also describe a surprising and useful feature of Uman and similar reagents. While other well-characterized neurofilament antibodies generally show robust staining of countless cells and processes in CNS sections from healthy rats, both Uman antibodies reveal only a minor subset of profiles, presumably spontaneously degenerating or degenerated neurons and their processes. However, following experimental mid-cervical spinal cord injuries to rats, both Uman antibodies recognize numerous profiles in fibre tracts damaged by the injury administered. These profiles were typically swollen, beaded, discontinuous or sinusoidal as expected for degenerating and degenerated processes. We also found that several antibodies to the C-terminal 'tail' region of the neurofilament light protein bind undamaged axonal profiles but fail to recognize the Uman-positive material. The unmasking of the Uman epitopes and the loss of the neurofilament light tail epitopes can be mimicked by treating sections from healthy animals with proteases suggesting that the immunostaining changes we discovered are due to neurodegeneration-induced proteolysis. We have also generated a novel panel of monoclonal and polyclonal antibodies directed against the Uman epitopes that have degeneration-specific staining properties identical to the Uman reagents. Using these, we show that the region to which the Uman reagents bind contains further hidden epitopes distinct from those recognized by the two Uman reagents. We speculate that the Uman-type epitopes are part of a binding region important for higher order neurofilament assembly. The work provides important insights into the properties of the Uman assay, describes novel and useful properties of Uman-type and neurofilament light tail-binding antibodies and provides a hypothesis relevant to further understanding of neurofilament assembly.

Keywords: ELISA; NF-L; biomarker; neurodegeneration; neurofilament.

Graphical Abstract

Figure 1

Diagrams of NF-L structure, landmarks and epitopes. (A) Domain organization and epitope map of human NF-L protein. The yellow blocks in Coil 1b and Coil 2 indicate the location of 11 amino acid ‘hendecad’ sequences, one of which is referred to here as ‘Stutter 2’. Location of landmarks and relevant antibody epitopes are based on the human NF-L protein sequence. (B) Amino acid sequences including the Uman epitopes and the various peptides used. The core of the UD2 epitope is outlined, as is the extended peptide essential for UD1 binding. The ‘a’ and ‘d’ positions of the hydrophobic heptad repeat are identified below the sequences. The region of the ‘Stutter 2’ where the normal heptad repeat is replaced by a hendecad sequence is indicated in the 333-343 region, with the addition of ‘h’, ‘i’, ‘j’ and ‘k’ amino acids. The charged i, i + 4 type residues are indicated with appropriate + and − signs above the sequences, and each instance is numbered in magenta. The C-terminal cysteine residues added to all peptides are omitted.

Figure 2

Western blotting studies. (A) Western blotting of Uman antibodies and MCA-1B11 on recombinant NF-L constructs. Lanes marked S are molecular weight standards of indicated size in magenta. In each case, Lane 1 is full-length human NF-L (1-543), Lane 2 is NF-L rod (80-400), Lane 3 is a proprietary immunogen containing NF-L 311-362 used to generate novel antibodies, and Lane 4 is NF-L Coil 2 (256-400). Original western blot images are in Supplementary Fig. 5 (left two panels) and Supplementary Fig. 6 (right panel) (B). Partial protein cleavage of NF-L 256-400 at tryptophan 279 produces two fragments of expected size 8.5 and 13.9 kDa. The N-terminal 8.5 kDa is identified by a His-tag antibody and the C-terminal 13.9 kDa by anti-IFA (see text). All three NF-L antibodies tested bind to the C-terminal 13.9 kDa band. Original western blot image is shown in Supplementary Fig. 7. (C) Partial cleavage of NF-L 256-400 at cysteine 322 produces an N-terminal 13.2 kDa fragment and a C-terminal 9.3 kDa fragment, identified using antibody to the vector-derived S-tag and anti-IFA. Both UD1 and MCA-1B11 bind the C-terminal 9.3 kDa fragment, while the epitope for UD2 is apparently destroyed. Original blots for UD1 and MCA-1B63 lanes are shown in Supplementary Fig. 8, while UD2, MCA-1B11 and anti-IFA lanes are shown in Supplementary Fig. 9.

Figure 3

Immunofluorescence of control cells and tissues. (A and B) Seven- to 10-day neural cultures from E20 rats stained with UD1 (A) and UD2 (B) in green and both co-stained with RPCA-NF-L-ct in magenta. In both cases, the majority of the Uman-positive material is punctate and negative for RPCA-NF-L-ct, while the RPCA-NF-L-ct antibody stains fibrillar profiles with a typical neuronal morphology. C shows a region of a similar culture stained with UD2 in green and RPCA-NF-L-ct in magenta. The prominent linear array of Uman-positive globular material is suggestive of the remains of a degenerated process, while the RPCA-NF-L-ct–positive profile appears continuous and fibrillar. D shows a section of spinal cord from an uninjured rat stained with UD2 in magenta and RPCA-NF-L-ct in green. A single fibre negative for RPCA-NF-L-ct is revealed with the UD2 antibody. E and F show a region of brain stem from a control rat. One somewhat unhealthy looking neuron and associated processes is positive for UD2 (magenta) but not RPCA-NF-L-ct (green). F shows the same specimen with only the Uman antibody in magenta. The blue signal in D and E shows DNA staining with DAPI. Bar in image F = 50 µM, in A–C = 20 µM, in D = 40 µM and in E = 50µM.

Figure 4

Immunofluorescence of coronal sections from a rat given a spinal cord contusion 3 days previously. A shows a coronal section stained for RPCA-NF-L-ct in magenta and UD2 in green. UD2-positive profiles are particularly obvious in the dorsal columns, corticospinal tracts and rubrospinal tracts, less abundant in the lateral and ventral funiculi and least abundant in the spinal cord grey matter. B shows a ×6 magnified view of the region just above the central canal of the same section shown in A, showing well-defined RPCA-NF-L-ct axonal profiles in magenta in comparison with the more diffuse and generally non-overlapping UD2 profiles in green. C shows a ×6 magnified view of the same region of a similar section stained with RPCA-NF-L-ct in magenta and UD1 in green. Bar in image A = 500 µM and in images B and C = 83 µM.

Figure 5

Immunofluorescence of longitudinal sections from contused rat spinal cord 3 day after injury. Lateral rubrospinal and neighbouring tracts some distance from the lesion. Linear arrays of profiles visualized with UD1 (A) and UD2 (B) shown in green are negative for RPCA-NF-L-ct shown in magenta. The RPCA-NFL-ct profiles are generally well-defined and continuous; the UD1 and UD2 profiles are mostly swollen, globular and discontinuous. Inserts show ×3 magnified sections of each image. C shows a region close to a contusion lesion and shows swollen and apparently degenerating axonal profiles positive for RPCA-NF-L-ct in magenta but also shows UD1 staining in green. Note a few the swollen and unhealthy looking profiles that show staining for both antibodies (arrowed). Bar in image C = 25 µM applicable to images A–C and bar in images A and B inserts = 8 µM.

Figure 6

Further immunofluorescence of contused rat spinal cord. A shows a region at the lesion site of an animal given a contusion 3 days previously stained with RPCA-NF-L-ct in magenta and UD1 in green. Fibres positive for RPCA-NF-L-ct alone are visible amongst mostly globular profiles positive for both antibodies or only for UD1. B shows a region of adult mouse spinal cord that was allowed to sit at room temperature for 4 h prior to fixation. Sections were then incubated with RPCA-NF-L-ct in green and the Uman-type antibody MCA-1B11 in magenta. Note prominent beaded, discontinuous, sinusoidal and apparently helical MCA-1B11–positive profiles that are negative for RPCA-NF-L-ct, while RPCA-NF-L-ct staining is attenuated. C shows a longitudinal section from an SCI animal showing beaded, swollen and sinusoidal processes (arrowed) stained with MCA-DA2 directly labelled with Alexa Fluor® 488 and co-stained with UD1 in magenta. Like RPCA-NF-L-ct, MCA-DA2 does not stain most of the UD1-positive material. (D–F) A longitudinal section of spinal cord from an uninjured control rat was searched to find what we propose are spontaneously degenerating or degenerated nerve fibres as in Fig. 3D. D shows typical axonal profiles stained with RPCA-NF-L-ct in magenta. A linear group of globular profiles that were negative for RPCA-NF-L-ct but positive for MCA-1B11 (E, green) directly coupled to Alexa Fluor® 488 were also identically positive for MCA-6H63 coupled to Alexa Fluor® 647 (F, blue). MCA-1B11 and MCA-6H63 bind distinct hidden NF-L epitopes, building confidence that the objects stained are indeed degenerating or degenerated axons. Bar in image D = 10 µM, applicable to images C–F, and in image A = 20 µM, applicable to images A and B.

Figure 7

Protease experiments. Coronal sections of uninjured control rat spinal cords were treated with 0.25% trypsin or with buffer control lacking trypsin under exactly the same conditions. A shows a fibre tract after 10 min in control buffer then stained with RPCA-NF-L-ct in magenta and UD1 in green, while B shows just the green UD1 signal. C and D show a similar region incubated exactly as the section in A and B but with the addition of 0.25% trypsin. C and D were imaged on a confocal microscope, and identical laser and software settings were then used to images A and B. Following enzyme treatment, there is a weaker RPCA-NF-L-ct signal and a significant UD1 signal. E and F show similar control data for UD2 (green) co-stained with RPCA-NF-L-ct (magenta) on a similar section from an uninjured animal, and G and H show the staining of the same two antibodies following 10-min treatment in trypsin. The blue signal in all images is nuclear DNA revealed with DAPI. Bar in image H = 20 µM.

Figure 8

Immunocytochemical studies with novel antibodies. A–D are sections of spinal cord from a rat given a contusion 3 days previously and stained with RPCA-NF-L-Degen in magenta, and respectively in green: (A) the NF-M tail antibody MCA-3H11 that shows little staining of the Uman-positive aggregates. (B) The MAP-τ core domain antibody MCA-5B10 shows that MAP-τ associating with some inclusions (arrowed example) but apparently not others. (C) The phospho-KSP pNF-H antibody MCA-NAP4 strongly recognizes normal axons and the Uman-positive material that therefore appears yellow. (D) The α-internexin antibody MCA-2E3 shows that a part of this protein is also component of the Uman-positive material. E shows a longitudinal section of rat SCI tissue stained with MCA-6H63 in magenta and co-stained with RPCA-NF-L-ct in green. Despite having an epitope distinct from both UD1 and UD2, the MCA-6H63 antibody also clearly stains degenerating material and not healthy processes. F shows MCA-1D44 staining of mixed E20 cortical neural cultures in green co-stained with RPCA-NF-L-ct in magenta. As with the Uman reagents (Fig. 3C), MCA-1D44 fails to stain what are apparently healthy neuronal processes but does stain linear sets of aggregates (arrowed) which we conclude originated from degenerated processes. The blue signal in A–E is nuclear DNA revealed with DAPI. Bar in image H = 20 µM, applicable to all images.