Neurofilaments: Neurobiological Foundations for Biomarker Applications

Arie R. Gafson; Nicolas R. Barthélemy; Pascale Bomont; Roxana O. Carare; Heather D. Durham; Jean-Pierre Julien; Jens Kuhle; David Leppert; Ralph A. Nixon; Roy O.Weller; Henrik Zetterberg; Paul M. Matthews


Brain. 2020;143(7):1975-1998. 

In This Article

Abstract and Introduction


Interest in neurofilaments has risen sharply in recent years with recognition of their potential as biomarkers of brain injury or neurodegeneration in CSF and blood. This is in the context of a growing appreciation for the complexity of the neurobiology of neurofilaments, new recognition of specialized roles for neurofilaments in synapses and a developing understanding of mechanisms responsible for their turnover. Here we will review the neurobiology of neurofilament proteins, describing current understanding of their structure and function, including recently discovered evidence for their roles in synapses. We will explore emerging understanding of the mechanisms of neurofilament degradation and clearance and review new methods for future elucidation of the kinetics of their turnover in humans. Primary roles of neurofilaments in the pathogenesis of human diseases will be described. With this background, we then will review critically evidence supporting use of neurofilament concentration measures as biomarkers of neuronal injury or degeneration. Finally, we will reflect on major challenges for studies of the neurobiology of intermediate filaments with specific attention to identifying what needs to be learned for more precise use and confident interpretation of neurofilament measures as biomarkers of neurodegeneration.


Neurofilaments are assembled from a family of five intermediate filaments (Julien and Mushynski, 1983) that are distinguishable based on their relative apparent molecular masses on SDS-polyacrylamide gels. The largest of these is neurofilament heavy chain (NfH), followed (in order of descending molecular weight) by the medium chain (NfM), the light chain (NfL), α-internexin and peripherin (Figure 1). Neurofilaments contribute to growth and stability of axons in both central and peripheral nerves as well as to maintaining mitochondrial stability (Gentil et al., 2015) and microtubule content (Bocquet et al., 2009). Roles for distinct neurofilament isoforms in maintaining the structure and function of dendritic spines and in regulating glutamatergic and dopaminergic neurotransmission synapses have also been discovered (Schwartz et al., 1994, 1995).

Figure 1.

Schematic representation of the structure of neuronal intermediate filament proteins. All intermediate filament proteins have a highly conserved central domain of 310 amino acid residues that is responsible for the formation of coiled-coil structures. Flanking this central rod domain are the amino- and carboxyl-terminal domains. These latter domains confer functional specificity to the different types of intermediate filaments proteins. For example, the NfM and NfH carboxyl-terminal domains contain multiple repeats of phosphorylation sites KSP (Lys–Ser–Pro) that account for the unusual high content of phosphoserine residues for these proteins. The N- and C-terminal regions contain multiple O-linked glycosylation sites. Neurofilament proteins NfL, NfM and NfH are obligate heteropolymers. Although α-internexin or peripherin can form homopolymers in vitro, these intermediate filaments proteins usually co-polymerize with the neurofilament triplet proteins in vivo.

The fundamental importance of neurofilaments to neurons has been highlighted by molecular characterization of diseases of the brain and peripheral nerves associated with abnormal neurofilament structure and function. Mutations in the NEFL gene, which encodes NfL, lead to peripheral neurodegeneration in Charcot-Marie-Tooth (CMT) disease types 2E (Mersiyanova et al., 2000) and type 1F (Jordanova et al., 2003) and G (Zuchner et al., 2004). While polymorphisms in NEFH, encoding NfH, are associated with amyotrophic lateral sclerosis (ALS) (Figlewicz et al., 1994), mutations in this gene also are a cause of CMT type 2 (Rebelo et al., 2016). Neurofilament dysfunction or aggregation also may play roles in the neuropathology of Alzheimer's disease, Parkinson's disease and other neurodegenerative disorders (Khalil et al., 2018).

Neurofilaments' turnover in healthy neurons is slow. Their expression is regulated by neuronal activity acting through developmentally regulated promoter regions (Yaworsky et al., 1997). Post-transcriptional regulation of neurofilament mRNA stability also may contribute to determining levels of expression of neurofilament protein (Schwartz et al., 1994, 1995). Additional insights into mechanisms for turnover of neurofilament have come through studies of rare diseases arising from gigaxonin E3 ligase mutations causing giant axonal neuropathy (GAN) (Bomont et al., 2000) and mutations in TRIM2 (another E3 ligase) and sacsin (which includes both ubiquitin-like and chaperone domains) that are responsible, respectively for a form of CMT (Ylikallio et al., 2013) and for the cerebellar degeneration occurring in autosomal recessive spastic ataxia of Charlevoix-Saguenay (ARSACS) (Engert et al., 2000). Study of these diseases has elucidated major pathways responsible for the degradation of neurofilament protein, which is a consequence of combined activities of proteasomal and, possibly, autophagocytic mechanisms (Bomont, 2016). Neurofilament or its fragments can be released from neurons secondary to axonal damage or neurodegeneration, although the predominant peptide species released and the mechanisms responsible for the release have not been clearly characterized. Release may occur actively (e.g. by means of exosomes; Faure et al., 2006; Lachenal et al., 2011) or passively with loss of neuronal membrane integrity. Neurofilaments in different supramolecular structures or with different isoforms may show differences in degradation rates (Nixon and Logvinenko, 1986; Millecamps et al., 2007). Studies of pathways for trafficking of other proteins (Szentistvanyi et al., 1984) suggest that degraded neurofilament proteins may enter the peripheral circulation via perivascular drainage along basement membranes of arteries (Carare et al., 2008) to drain into cervical or lumbar lymph nodes and then into the blood.

Reliable, sensitive assays for measuring concentrations of neurofilaments in CSF have been available for many years (Norgren et al., 2002). These provided an early foundation for exploration of the associations of increased CSF neurofilaments with neurological diseases (Rosengren et al., 1996; Lycke et al., 1998). Ultra-sensitive assays of neurofilaments in blood are now available routinely for clinical applications in many centres (Kuhle et al., 2016a). This has enabled several studies assessing the potential utility of neurofilament (particularly NfL) peptide concentrations in the CSF or peripheral blood as clinical biomarkers of neuronal injury or neurodegeneration (Kuhle et al., 2016a; Disanto et al., 2017; Khalil et al., 2018). For example, CSF and peripheral blood neurofilament protein concentrations are increased after stroke or traumatic brain injury (Khalil et al., 2018) and are associated with ageing (Disanto et al., 2017) and primary neurodegenerative diseases including Alzheimer's disease (Mattsson et al., 2016; Weston et al., 2017). Neurofilament concentrations in both CSF and peripheral blood are increased in some individuals with multiple sclerosis (Amor et al., 2014; Kuhle et al., 2016b; Disanto et al., 2017; Novakova et al., 2017; Barro et al., 2018; Hakansson et al., 2018; Piehl et al., 2018) and have potential roles as clinical biomarkers of disease activity, treatment responses or prediction of future disease progression and disability. Both peripheral blood and CSF concentrations are correlated with radiological (Kuhle et al., 2016b; Disanto et al., 2017; Novakova et al., 2017; Siller et al., 2018) and clinical measures of disease activity (Disanto et al., 2017; Novakova et al., 2017; Barro et al., 2018; Hakansson et al., 2018; Piehl et al., 2018; Siller et al., 2018). Evidence for treatment responsiveness further supports a causal link between disease activity and increased neurofilament concentrations in CSF and peripheral blood (Gunnarsson et al., 2011; Disanto et al., 2017; Piehl et al., 2018; Gafson et al., 2019; Kuhle et al., 2019). NfL levels have a potential role in assessing prognosis in multiple sclerosis. For example, the predictive association between increased NfL and longer-term brain and spinal cord atrophy (Barro et al., 2018) likely arises from the sensitivity of NfL to the neuroinflammatory neuronal injury and degeneration that provides a substrate for future disease progression (Matthews, 2019).

While the concentration in CSF is ~20–50-fold greater than in peripheral blood (Bergman et al., 2016), moderate to high correlation between concentrations measured in the two compartments have been reported (Gaiottino et al., 2013; Gisslen et al., 2016). Nevertheless, given that neurofilaments can also be released from peripheral nerves, depending on the pathological context, peripheral blood and CSF measures should not necessarily be correlated (Bergman et al., 2016). Longitudinal measures of CSF and peripheral blood concentration changes after brain injury (Shahim et al., 2016) suggest that turnover times in the blood and CSF compartments are similar. This is consistent with models positing that central and peripheral turnover are linked functionally. However, important questions regarding the neurobiology, mechanisms of turnover and kinetics in the brain and blood compartments remain. These will be highlighted as the current understanding of neurofilament neurobiology in health and disease is reviewed in more detail below (Box 1).