Module 3: Intermediate Filaments

Unlike actin (two genes) or tubulin (a handful of paralogs), the human genome encodes >70 intermediate-filament genes, grouped by sequence homology of the central\(\sim 310\)-residue α-helical rod domain (Fuchs & Weber 1994,Annu. Rev. Biochem.; Herrmann & Aebi 2004, Annu. Rev. Biochem.). The classification by sequence type is clinically and developmentally informative:

• Type I (acidic keratins): K9–K28 epithelial, K31–K40 hair/nail; obligate heterodimers with a Type II keratin partner; mutations cause epidermolysis bullosa simplex (K5/K14) and pachyonychia congenita(K6/K16/K17).
• Type II (basic keratins): K1–K8 epithelial, K71–K86 hair; partner with a Type I keratin. K1/K10 in cornifying skin, K8/K18 in simple epithelia. Rates of expression track cell differentiation programs.
• Type III (homopolymeric non-epithelial): vimentin (mesenchymal cells, leukocytes, fibroblasts), desmin(muscle Z-disks, intercalated disks), GFAP (astrocytes, Schwann cells, enteric glia), peripherin (peripheral neurons). All assemble as homopolymers and can co-assemble among themselves.
• Type IV (neurofilaments and α-internexin): NF-L (light, 61 kDa), NF-M (medium, 90 kDa), NF-H (heavy, 115 kDa), and α-internexin. Heteropolymeric: NF-L forms the core, NF-M/H projecting tails space neurofilaments radially and determine axon caliber (Hoffman et al. 1985; Eyer & Peterson 1994). Nestin and the lens-specific synemin/paranemin are sometimes grouped here.
• Type V (nuclear lamins): lamin A/C(alternatively spliced from LMNA), lamin B1 (LMNB1), and lamin B2 (LMNB2). The only IFs that polymerize in the nucleus; they form the nuclear lamina, a 10-20 nm thick meshwork lining the inner nuclear membrane (Dechat et al. 2008, Genes Dev.; Gruenbaum & Foisner 2015).
• Type VI (nestin, synemin): expressed in neural/muscular stem cells and in regenerating tissue; exceptionally long tail domains that project into the cytoplasm and recruit signaling components.

Why so many IF genes? Because different tissues see different mechanical loads. Skin keratins have a different stress profile than muscle desmin, which differs from neuronal neurofilaments. IF expression is co-regulated with tissue differentiation and is routinely used in histopathology: a cytokeratin-positive tumor is epithelial, vimentin-positive is mesenchymal or undergoing EMT, GFAP-positive is astrocytic, desmin-positive is myogenic, and neurofilament-positive identifies neurons and their tumors. Every antibody panel in the pathology lab relies on this IF-class logic.

Every IF subunit has a conserved three-part architecture: a variable N-terminal head (intrinsically disordered, glycine/serine-rich, positively charged), a conserved central rod (~310 residues, 45-46 nm long, α-helical with heptad repeats), and a variable C-terminal tail (usually disordered, often acidic). The rod contains four segments (coil 1A, coil 1B, coil 2A, coil 2B) connected by short non-helical linkers (L1, L12, L2); their lengths are identical across all IF classes—a diagnostic signature of the family (Herrmann & Aebi 2004; Parry & Steinert 1999).

Two monomers dimerize by parallel coiled-coilpacking: the rod domains wind around one another in a left-handed superhelix with heptad repeats supplying hydrophobic contact at positions a and d (Crick 1953 knobs-into-holes). The result is a rigid ~46 nm rod with flexible head and tail domains dangling at either end. A single dimer is a polar object—head at the N-terminal end, tail at the C-terminal end—but the assembly pathway very quickly kills that polarity.

The critical feature that distinguishes IF assembly from actin and microtubule assembly is apolar hierarchical lateral association. IF dimers do not elongate head-to-tail into a polar protofilament; instead, two parallel dimers associate antiparallel and half-staggered to form a symmetric, apolar tetramer. The tetramer has two heads at each end and no distinguishable plus/minus chemistry. Eight such tetramers then associate laterally into a unit-length filament (ULF)—a short rod 60 nm long, 16 nm in diameter, containing 32 monomers in cross-section. Finally, ULFs anneal longitudinally and radially compact to give the mature 10 nm filament (Herrmann, Bar, Kreplak, Strelkov & Aebi 2007, Nat. Rev. Mol. Cell Biol.; Herrmann et al. 2009, Annu. Rev. Biochem.).

Because tetramers are apolar, the mature filament is also apolar. This has profound functional consequences: there are no IF-specific motor proteins. Kinesin and dynein need a polar track; IFs lack one. IFs are load-bearing structural elements only, not transport highways. Their mechanical job is to resist deformation.

Vimentin stopped-flow experiments (Kirmse et al. 2007 J. Biol. Chem.; Mücke et al. 2004) reveal that ULF formation is essentially complete within 10–20 seconds at physiological ionic strength, while longitudinal annealing and radial compaction continue for >10 minutes. The final filament is not a covalent polymer—the subunits are held by coiled-coil and lateral electrostatic contacts—so IFs are capable of subunit exchange along their length and at the ends (Ngai, Ring, Johnson, Sun & Lazarides 1990; Vikstrom, Lim, Goldman & Borisy 1992), albeit much more slowly than actin or microtubules.

In the cell, phosphorylation of head domains by protein kinase A, protein kinase C, and Cdk1 drives filament disassembly. At mitosis, nuclear lamins disassemble via Cdk1-mediated phosphorylation of N-terminal head serines, allowing nuclear-envelope breakdown; they reassemble when phosphatases act in telophase (Heald & McKeon 1990).

Single-filament force-extension experiments (atomic-force-microscopy pulling on vimentin IFs in buffer; Kreplak, Bar, Leterrier, Herrmann & Aebi 2005; Qin, Kreplak & Buehler 2009) show that intermediate filaments can be reversibly stretched to 2.5–3.5 times their resting length without rupture—in stark contrast to actin (<2% strain) and microtubules (even less). The stress-strain curve of a vimentin filament exhibits three regimes:

1. Linear elastic up to ∼10% strain (coiled-coil as Hookean spring); effective Young’s modulus\(E \approx 6\) MPa.
2. Plateau from ∼10% to ∼150% strain where the α-helix transitions to β-sheet structure (α \(\to\) β transition), a well-known mechanical switch in keratin hairs (Kreplak & Fudge 2007).
3. Strain-hardening above ∼150% strain as the β-sheets pack; the modulus rises sharply before rupture at ∼250–300% strain.

The persistence length of a single vimentin filament is\(L_p \sim 1\) μm (Nogels et al. 2006; Mücke, Winheim, Merlitz, Buchrieser, Langowski & Herrmann 2004). This is comparable to typical cellular length scales, so IFs are semi-flexible: thermal bending matters, but they are not freely jointed chains.

At the network level, IFs strain-harden strongly at high deformation (Janmey, Euteneuer, Traub & Schliwa 1991)—the more you stretch, the stiffer they become. This is the opposite behavior of most soft materials and explains why IFs can carry large, slow, sustained deformations without rupture: they redistribute load across their network topology. In skin, this is what prevents keratinocytes from lysing under mechanical load; in muscle, desmin keeps the Z-disks registered; in axons, neurofilaments determine caliber and resist axonal shear.

Keratins constitute the largest IF subfamily (54 human genes: 28 type-I acidic, 26 type-II basic). Every epithelial cell expresses a specific type-I/type-II pair, and the pairing is tissue-diagnostic: K8/K18 in simple epithelia (gut, liver), K5/K14 in basal keratinocytes, K1/K10 in differentiated keratinocytes of the epidermis, K6/K16 in wound-healing skin, K6a/K17 in nails.

Keratin filaments are crosslinked to desmosomesand hemidesmosomes by plakin-family proteins (desmoplakin, plectin, bullous pemphigoid antigen 1). These junctions translate IF networks across cell boundaries, creating an intercellular keratin mesh that bears the macroscopic tensile load of skin and gut epithelium.

Mutations in K5 or K14 destabilize basal keratinocyte IFs, producing epidermolysis bullosa simplex (EBS): basal cells lyse under trivial mechanical stress, and the skin blisters. Coulombe, Hutton, Letai, Hebert, Paller & Fuchs (1991 Cell) established this molecular diagnosis and founded the entire field of IF pathology.

Hair is a crystalline array of type-I/type-II hair keratins (K31–K40, K81–K86) embedded in a cysteine-rich protein matrix stabilized by disulfide crosslinks. The α→β transition (Kreplak & Fudge 2007) is why wet hair stretches without breaking and dry hair fractures at lower strain.

Desmin (53 kDa) is the muscle-specific type-III IF. It forms a lattice around Z-disks and interconnects neighboring sarcomeres, transmitting longitudinal contractile force laterally. DES mutations cause desmin-related myopathies and arrhythmogenic right ventricular cardiomyopathy: Z-disks lose registration, sarcomeres misalign, cardiac conduction is disrupted.

Vimentin (57 kDa) is expressed in mesenchymal cells and is the classical marker of epithelial-mesenchymal transition (EMT)— a developmental process co-opted by metastatic carcinoma. Vimentin-null mice are viable but show impaired wound healing, reduced fibroblast motility, and more fragile endothelium (Colucci-Guyon et al. 1994 Cell). Vimentin also functions in stress tolerance: cells with vimentin-null networks are mechanically softer, less able to resist compression.

GFAP (50 kDa, glial fibrillary acidic protein) marks astrocytes; upregulated in reactive astrogliosis after CNS injury. It is the standard diagnostic immunostain for glial tumors. Mutations in GFAP cause Alexander disease, a rare leukodystrophy.

Peripherin (57 kDa) is expressed in peripheral and some CNS neurons; its upregulation in motor neurons is a marker of amyotrophic lateral sclerosis (ALS), and peripherin aggregates are found in Bunina bodies.

Axons transmit information at speeds that scale with diameter: conduction velocity\(v \propto \sqrt{d}\) for unmyelinated and\(v \propto d\) for myelinated fibers. Axon diameter is set by neurofilament density and composition. The neurofilament triplet (NF-L + NF-M + NF-H) forms a core with projecting phosphorylated side arms from NF-M and NF-H that push neighboring filaments apart, giving the axon an actively-maintained caliber. Phosphorylation of the KSP repeats in NF-H and NF-M tails by CDK5 and MAPK is what extends the projections and increases inter-filament spacing (Julien & Mushynski 1998, Prog. Nucleic Acid Res. Mol. Biol.).

Neurofilament abnormalities are central to several neurodegenerative diseases. NF-H/NF-M aggregates appear in amyotrophic lateral sclerosis (ALS), and giant axonal neuropathy (GAN) is caused by mutations in the E3 ubiquitin ligase gigaxonin that regulates IF turnover. Quantifying neurofilament light chain (NfL) in plasma or CSF is now a standard biomarker of axonal injury in multiple sclerosis, traumatic brain injury, and neurodegenerative disease.

The nuclear lamina is a 10–20 nm thick meshwork of lamin A/C (encoded by LMNA) and lamin B (B1 LMNB1, B2 LMNB2) filaments lining the inner nuclear membrane. It is the only place in the cell where IFs assemble inside a membrane-bound organelle. Unlike cytoplasmic IFs, lamins carry a nuclear localization signal (NLS)in their tail and terminate in a CaaX motif(farnesylated and proteolytically processed by ZMPSTE24).

Lamins dimerize as coiled coils, form tetramers, and assemble into filaments that weave a square-grid-like mesh at the nuclear surface (Turgay, Eibauer, Goldman, Vignjevic & Medalia 2017, Nature: cryo-electron tomography of theXenopus oocyte lamina). The mesh has a characteristic spacing of ~50-80 nm.

Laminopathies are a family of >10 distinct disease phenotypes caused by mutations in a single gene, LMNA, depending on mutation location and mechanism:

• Emery-Dreifuss muscular dystrophy (LMNA missense).
• Dilated cardiomyopathy with conduction-system disease.
• Dunnigan familial partial lipodystrophy.
• Charcot-Marie-Tooth type 2B1 (peripheral neuropathy).
• Hutchinson-Gilford progeria syndrome (HGPS): G608G silent substitution activates a cryptic splice site, producing progerin, a truncated lamin A lacking the ZMPSTE24 cleavage site. Permanently farnesylated progerin integrates into the lamina and stiffens it, disrupts heterochromatin, and produces a premature-aging phenotype with death from cardiovascular disease around age 13 (Dechat et al. 2008; Eriksson et al. 2003 Nature).
• Mandibuloacral dysplasia.

The working model (Dahl et al. 2008; Swift et al. 2013 Science) is that lamin A/C tunes nuclear stiffness to match the tissue mechanical environment. Cells in stiff tissue (muscle, bone, heart) upregulate lamin A/C; cells in soft tissue (brain, marrow) have low lamin A/C. Loss of this mechanical buffer leads to nuclear-envelope rupture, DNA damage, and premature senescence—this is the core mechanism linking nuclear mechanics, mechanotransduction, and laminopathy disease.

The nuclear lamina is not an isolated compartment. It is physically coupled to the cytoskeleton by the LINC complex (LInker of Nucleoskeleton and Cytoskeleton): SUN1/SUN2 proteins span the inner nuclear membrane and bind lamins intranuclearly; their lumenal SUN domains interact across the perinuclear space with KASH-domain proteins (nesprin-1, -2, -3, -4) that span the outer nuclear membrane and connect to cytoplasmic actin (nesprin-1/-2), microtubule motors (nesprin-4), or plectin (nesprin-3) and thus to IFs.

Through the LINC complex, force applied to the cell surface is transmitted to the nuclear lamina, where it can deform chromatin and alter gene expression (Crisp et al. 2006, J. Cell Biol.; Lombardi et al. 2011). This is the physical substrate of nuclear mechanotransduction—how cells on a stiff substrate differentiate into osteoblasts while cells on a soft substrate become neurons or adipocytes (Engler et al. 2006 Cell).

Because IF gene expression is cell-type-specific and stably maintained through differentiation, IF immunohistochemistry is the backbone of diagnostic pathology. A few representative fingerprints:

Because a single gene mutation can cause an entire disease syndrome (laminopathies being the clearest case), understanding the IF classification is not merely academic— it is the entry point to molecular diagnosis in several dozen heritable disorders.

It is worth pausing to compare intermediate filaments with the two polymers from Modules 1 and 2. The table below summarizes structural, kinetic, and mechanical contrasts.