The findings of this study indicate that for the specific injury conditions of moderate central FPI, unmyelinated axons undergo responses that are quite distinct from those of myelinated axons. The midline FPI model was selected for this study to match prior experiments showing functional impairments (12
), and to explore nonlethal axonal changes in a white matter environment with minimal fiber degeneration. Although regarded as a moderate intensity injury, the 2.0 atmospheres midline FPI used here elicits less axonal injury and degeneration in the cerebrum than most alternative animal models of TBI, including lateral FPI, controlled cortical impact and impact-acceleration (31
). The microscopic examination in the present study did not demonstrate significant degrees of axonal swelling, bulb formation or Wallerian degeneration, which are reliably elicited in brainstem white matter tracts following impact acceleration injury in adult rats (10
). Indeed, the FPI-induced morphological changes in this study were relatively modest and required a quantitative stereological assessment for their characterization. These results may, therefore, have implications for the clinical condition of mild TBI, which often goes unreported and has been estimated to be approximately 22 times more prevalent than severe TBI (37
). We observed post-TBI time-dependent cellular changes that are not necessarily lethal but may contribute to functional impairment that is concomitant with white matter subjected to mild TBI.
A key objective of the study was to establish the fiber composition of the corpus callosum in sham-injured rats as a valid estimate of the normal condition and to provide a reliable baseline against which to evaluate the effects of injury. This was warranted in view of variability among published reports regarding the percentage of unmyelinated axons in the rodent corpus callosum. Earlier studies of axon types in the rodent splenium estimated that about 45% of fibers were unmyelinated (38
), but more recent studies have consistently estimated the proportion to be approximately 88% (25
); our present estimate of 84% is in line with these reports. The consensus among the more recent estimates is likely due to sampling strategies that systematically cover the dorsal-to-ventral extent of the callosum. Among these prior studies, only Gravel et al (25
) implemented a sampling methodology that was essentially equivalent to the present approach, including estimates at genu, mid-callosum, and splenium. Averaging across all regions, those authors estimated that 81% of corpus callosum axons were unmyelinated, which corresponds closely to our estimate of 78%.
Almost all previous reports of TBI-induced changes in axon caliber have referred to focal axon swellings in myelinated axons that are usually associated with cytoskeletal degradation and resultant impairments of axoplasmic transport. Early events include focal axolemmal perturbations that are probably mechanically induced during the injury and are associated with aberrant permeability (41
). Subsequently, microtubule loss and altered neurofilaments lead to focal swellings in myelinated axons (33
). These changes are observed in white matter that is subjected to greater injury forces than applied in the present study. We hypothesize that changes in the unmyelinated axon population, including the reduction in area observed in the current study, likely coexist with the abnormalities of myelinated axons in models of more severe TBI. Even within myelinated axons, a type of axonal shrinkage that takes the form of focal decreases in internodal axonal diameter has been described (41
). In the current study, caliber measurements were based on sagittal sections that present axons in cross-section; these results do not address whether the observed changes in area occur at focal points along the unmyelinated axons or are expressed uniformly along the fiber length.
The finding of distinctive reactive changes in unmyelinated fibers adds to growing evidence that traumatic axonal injury has more heterogeneous features than previously considered. It is becoming clear that many injured axons do not exhibit the sequence of changes observed in myelinated axons. For example, axons showing cytoskeletal disruption do not invariably progress to swelling (45
), and neurofilament compaction is not invariably associated with impaired axoplasmic transport (9
). One factor underlying such divergence of reactive changes may be axon heterogeneity, i.e. there likely are subpopulations of fibers with distinct molecular or structural properties. It is likely that myelinated and unmyelinated fibers are distinctive in more ways than diameter and/or the presence of myelin. Indeed, freeze fracture electron microscopy study indicated that axolemma in unmyelinated axons differs from that in myelinated axons at the macromolecular level (46
). These authors found that P-face intramembranous particles were more numerous in the internodal myelinated axolemma than in unmyelinated axolemma. Intrinsic differences in membrane constituents may influence fiber vulnerability. Indeed, the pattern and time course of structural changes that we observed in callosal fibers corresponds in several respects to suppression of compound action potentials (CAPs) previously reported using the same FPI model (12
). Post-injury reductions in evoked response amplitude were larger for unmyelinated, than for myelinated, CAPs. CAPs evoked through unmyelinated fibers were significantly suppressed at all time points recorded (3 hours – 7 days), whereas FPI-induced impairments to the myelinated CAPs were more transient and had recovered to control levels by 7 days post-injury. Previous CAPs were acquired in the mid-callosal region (12
); more detailed analyses of CAP impairments at genu, mid-callosal and splenial regions are now in progress in our laboratory.
Differences in size and the presence or absence of myelin are likely to be primary factors in the contrasting injury responses of the axon types. Axons are the subcellular compartment with the highest membrane-to-cytoplasm ratio, which likely places them at risk to membrane-targeting pathomechanisms of TBI, including lipid peroxidation (47
), rapid proteolysis of voltage-gated sodium channels (49
), and more protracted proteolytic events attacking submembrane ankyrin (51
) and spectrin (52
). We hypothesize that, because of their higher average axolemma-to-axoplasm ratio, unmyelinated axons are at a greater risk to these processes than myelinated axons. This relationship is illustrated in , which models axons as uniform cylinders of arbitrary length and plots the present measured values of mean axon diameter on the curve relating surface-to-volume ratio to diameter. This demonstrates that the mean diameter of unmyelinated axons (0.22 μm) is approximately 60% smaller than the mean myelinated diameter (0.57 μm); this corresponds to a 160% increase in the surface-to-volume ratio. Intracellular calcium loading, a key TBI pathomechanism, may especially challenge small axons with less cytoplasmic volume and calcium buffering/sequestration capacity, which is known to be critical in white matter injury (55
). These factors might contribute to a more significant injury response in unmyelinated axons as a class, as compared to the relatively mild response in myelinated fibers, which are typically much larger. Whereas it remains possible that unmyelinated axons of all sizes underwent some degree of post-injury shrinkage, the present results (, , ) suggest that within the population of unmyelinated fibers, comparatively large diameter may comprise an additional risk factor.
Figure 9 Morphological properties of unmyelinated axons may place them at elevated risk to traumatic brain injury (TBI). (A) Modeling axons as uniform cylinders of arbitrary length, a decrease in diameter corresponds to an increase in surface-to-volume ratio (upper (more ...)
Irrespective of axonal size, the presence of myelin itself, which provides physical support and influences fiber subtype vulnerability may account for some degree of axonal protection. In contrast, unmyelinated axolemma is extensively exposed to a post-traumatic extracellular environment with aberrant ionic composition (56
), reactive matrix metalloproteinases (57
), and infiltrating peripheral factors after blood-brain barrier disruption (51
) (). The intimate exposure of unmyelinated axolemma to the extracellular compartment could also underlie differential responses to drug treatments, such as a greater functional protection for unmyelinated, over myelinated, axons conferred by a calpain inhibitor (13
Assessments of fiber density in this study also confirmed rostrocaudal heterogeneity of axonal composition, as previously described (25
) and cross-sectional area measurements provided, for the first time, a stereological quantification of regional variation in mean diameter for both fiber types. The effects of FPI were regionally specific, with reduced unmyelinated size appearing first in the splenium and then progressing to more anterior zones. This injury sequence may reflect the intrinsic rostrocaudal gradient in unmyelinated axon density. This result may also in part reflect the proximity of the FPI location overlying the mid-callosum and splenium (), which showed axonal shrinkage prior to the genu. However, it would be difficult to attribute the different latencies to maximum shrinkage (1 day for splenium and 3 days for mid-callosum) to different distance from the FPI location. In the cuprizone demyelination, pathological changes were expressed in a rostrocaudal sequence (59
), consistent with the concept that properties intrinsic to specific callosal regions may determine the timing and magnitude of pathological alterations.
In summary, the present results provide evidence that moderate FPI applied to the midline of adult rats induced selective morphological changes in unmyelinated axons of the corpus callosum. A time-dependent reduction in mean axonal caliber, and its subsequent recovery, was highly significant. Reductions in unmyelinated axon density were also detected but were variable. Our findings have implications for current concepts of axonal injury that have tended to focus on pathology in larger myelinated axons. The diminutive size of unmyelinated fibers presents challenges to any monitoring of these structures in clinical, and even experimental, applications but a comprehensive model of axonal injury must account for distinctive alterations that affect the numerical majority of cerebral white matter axons. Injury-induced reductions in average axonal size would likely reduce the mean conduction velocities in these fibers and possibly contribute to the cognitive and memory impairments that are frequently observed after TBI. Although quantitative magnetic resonance imaging has confirmed TBI-induced shrinkage of white matter area (61
), the basis of these changes has not been identified at the cellular level. Fiber atrophy, perhaps related to the axonal changes described here may contribute to post-traumatic the white matter shrinkage observed in patients.