In the stationary beam test, the animal either stays put or locomotes along the beam, usually closed at either end to prevent escapes. Distance traveled and number of slips are used as auxiliary measures to distinguish immobile from moving subjects. The beam should be placed at least at a height of 30 cm from a padded surface, to prevent temptations of jumping off deliberately and any type of injury. For optimal performances, the beam should be covered with masking tape or some other material to increase traction. The beam is usually circular, since square-shaped ones facilitate clinging to the side.

The stationary beam test is sensitive to natural mutations with ataxia, as seen in Grid2 ^Lc (78, 79), Grid2 ^ho-Nancy (14, 79), Grid2 ^ho-4J (14), Rora ^sg (79–81), nervous (82), Reln ^rl-Orl (83), pogo (84), and Dst ^dt-J (85, 86), as well as non-ataxic Cacna1α ^rol heterozygotes (87). When a thick square-shaped beam was substituted for a round one to facilitate clinging against the side, Rora ^sg mutants were still impaired (86), but not Grid2 ^Lc (78) or Agtpbp1 ^pcd-1J (87). Thus, a partial loss of Purkinje cells in Rora ^sg caused a worse outcome than their nearly total disappearance in the other two mutants, presumably because the presence of the remaining cells causes worse dysfunction than their absence.

The stationary beam test is also sensitive to cerebellar anomalies in genetically modified mutants, such as ATXN3/Q84 ataxic transgenics (88), as well as Spnb3 (66) and Calb1 (75, 89) KOs, as well as non-ataxic Cacna1α heterozygotes (90), in addition to ATXN1/Q82 transgenics prior to ataxia (57). The bar deficit in the ATXN3/Q84 transgenic was alleviated with dantrolene, a calcium stabilizer (88), indicating its sensitivity to pharmacologic intervention.

In the wire suspension test, the animal either stays put or locomotes upside down while clinging to a thin metal or plastic wire closed at either end. Because the animal must carry its own body weight, this test is more dependent on muscle strength than the previous one. To obtain measurements of movement time and climbing abilities, the animal may reach and then climb at either end on a diagonal wire of a coat-hanger. As with beam tests, the horizontal wire should be placed at least at a height of 30 cm from a padded surface. Despite cerebellar damage, the animal may land on four feet, or on its side, as the air-righting reflex is still active. Some cerebellar mutants are susceptible to convulsions when placed upside down. In such cases, the animal’s fall should not count. Instead, the animal should rest awhile and the trial repeated after recovery.

Latencies before falling from the coat-hanger were lower than normal in natural mutations with ataxia, such as Grid2 ^Lc (78, 79), Grid2 ^ho-Nancy (79), Rora ^sg (79–81), Agtpbp1 ^pcd-1J (87), Reln ^rl-Orl (83), and Dst ^dt-J (85, 86). Latencies before falling were also lower in Grid2 ^Lc mutants placed on the suspended wire (91). On the contrary, ataxic nervous mutants did not fall sooner than controls, probably helped by relatively preserved Purkinje cells at the level of the vermis, crucial in the control of posture and equilibrium, though their movement times were longer (82). However, non-ataxic Cacna1α ^rol (90) and Cacna1α ^ln (92) heterozygotes were deficient in the wire suspension test, though ataxic Spnb3 (66) and Hfz (76) KOs were not. Thus, on one hand, ataxia does not necessarily lead to deficits in hanging from a wire. On the other, some non-ataxic mutants are deficient, indicating no obligatory relation between the two signs.

In the inclined or vertical grid test, the animal is placed on a metal surface oriented from 0 to 180°. As with the previous test, the grid should be placed at least at a height of 30 cm from a padded surface. In comparison to non-ataxic controls, the number of falls from a vertical grid increased in Grid2 ^Lc (78), Rora ^sg (81), and Dst   ^dt-J (85, 86) mutants, but not in Agtpbp1 ^pcd-1J (87). As noted above, the superior performance of Agtpbp1 ^pcd-1J relative to Rora ^sg on the same grid may be explained by nearly total as opposed to partial Purkinje cell depletion, the former being more favorable to behavioral performances. Their superior performance relative to Grid2 ^Lc may be ascribed to later onset of Purkinje cell depletion or to relative preservation of cerebellar granule cells. Their superior performance relative to Dst ^dt-J is attributed to the latter’s spinocerebellar and lemniscal atrophy leading to crawling in addition to ataxia.

Unlike the above-named tests, animals placed on the rotorod move in synchrony with a moving surface. An animal with postural deficits but with preserved muscle strength may cling to the rotating beam without moving, though cerebellar lesions usually cause hypotonia. To distinguish moving from immobile animals, the latter may be removed from the beam after turning twice around, as if it fell, since passively rotated animals tend to repeat this strategy until the end of testing. A more time-consuming approach is to videotape and measure time spent in passive rotation. In most models, the rod is placed lower than 30 cm above an unpadded surface, since animals are less inclined to jump away from a moving beam. However, it is judicious to start the test at the slow pace of 4 revolutions per minute (rpm), accelerating up to 40 rpm, because at faster initial speeds animals tend to jump off deliberately. If the beam is not rotated at the start, animals sometimes fall off immediately when surprised by the first jerk of beam movement.

The rotorod test is sensitive to natural mutations with ataxia, as seen in Grid2 ^Lc (91, 93–95), Grid2 ^ho-Nancy (14, 95), Grid2 ^ho-4J (14), Rora ^sg (80, 95), Agtpbp1 ^pcd-1J (87), nervous (82), and Reln ^rl-Orl (83), as well as non-ataxic Cacna1α ^la (92) and Cacna1α ^rol (90) heterozygotes.

The rotorod test is also sensitive to cerebellar anomalies in genetically modified mice. ATXN1/Q82 transgenic mice were impaired on the rotorod as early as 5 weeks of age, prior to signs of ataxia, when the main histopathology is the accumulation of cytoplasmic vacuoles inside Purkinje cells (57). Moreover, non-ataxic Atxn1/Q78 KI mice were impaired in the same test (59). Rotorod performance improved in the conditional ATXN1/Q82 model with delayed transgene expression relative to the original transgenic (58). After stopping transgene expression in the conditional model, rotorod performance was also ameliorated (96). Non-conditional ATXN1/Q82 transgenics crossbred with mice overexpressing the inducible form of rat HSP70, encoding a molecular chaperone involved in protein folding, had Purkinje cells with thicker and more arborized dendritic branches than the single transgenic and better rotorod performance (97). Rotorod performance also improved after RNA interference (98), intranasal insulin-growth factor (99) or injections of taurine-conjugated ursodeoxycholic acid (100). As with the SCA1 model, FRDA null mutants were impaired on the rotorod prior to ataxia (55).

Other ataxic mutants showing impairments on the rotorod include ATXN3/Q71 (64), ATXN7/Q92 (68), and ATXN7/Q52 (70) transgenics, as well as Spnb3 (66), Atxn8os (71), Hfz (76), and Plcb4 (101) KOs. Moreover, non-ataxic ATXN2/Q75 (60), ATXN3/Q148 (65), and ATXN7/Q90 (69) transgenic mice, as well as Unc13c (73) KOs mice were deficient on this test.

Rotorod impairments are also found after drug injections directly into the cerebellum. Indeed, intracerebellar injections of CB₁ receptor agonists reduced rotorod performances in rats (102, 103).

Motor activity and spontaneous alternation tests (Table 2) are relevant to cerebellar pathology. Motor activity in the open-field and T- or Y-mazes provides an estimate of motor capabilities as well as general arousal levels. Spontaneous alternation testing provides an estimate of the motivation to explore unfamiliar environments.

The lurching gait so characteristic of Grid2 ^Lc mutants on dry ground is no longer evident in the water, as these show good swimming abilities (119). Nevertheless, Grid2 ^Lc mutants were deficient in spatial orientation toward a hidden platform as well as visuomotor guidance toward a visible one, distance swum and escape latencies being increased (91, 112, 120–122) (Table 3). These results confirm a role for the cerebellum in the visual guidance of movement (123). The same pattern was evident in Reln ^rl-Orl mutants with ectopias in cerebellum and hippocampus (83) as well as Cacna1α ^la heterozygotes (92). In contrast, Rora ^sg (113), Agtpbp1 ^pcd-1J (124), and nervous (82) mutants were deficient only in the hidden platform subtest, concordant with a role for cerebello-limbic or cerebello-neocortical connections in spatial orientation, further supported by findings of the same dissociation in rats with immunotoxin-induced degeneration of Purkinje cells (125) as well as electrolytic lesions in lateral cerebellar cortex or dentate nucleus (126–128).

Due to their susceptibility to drowning, Dst ^dt-J mutants cannot be tested in the hidden platform version (129). In the visible platform subtest, escape latencies were higher than controls but path lengths were lower. Unlike the cerebellar mutants described above, the Dst ^dt-J line is characterized by slower swim speed and intact directional guidance, presumably due to spinocerebellar as opposed to cerebellar atrophy.

The effects of cerebellar damage on brain metabolism have been examined by high-resolution cytochrome oxidase (COX) histochemistry. COX is the fourth enzyme of the mitochondrial electron transfer chain of oxidative phosphorylation, immediately prior to ATPase. COX labeling is a specific marker of neuronal activity, since the contribution of glial cells on oxidative metabolism is minimal (130). Because COX histochemistry is expressed as a function of tissue weight, a loss in neuron numbers does not necessarily lead to reduced enzymatic activity. Studies in brain metabolism offer insights into the functional impact of cerebellar dysfunction on remaining cells of this structure as well as cerebellar-related pathways.

In cerebellar cortex of 15- and 25-day-old Grid2 ^Lc mutants, Purkinje cells could be observed prior to their degeneration. COX activity in these cells was higher than that of wild-type, but unchanged in the molecular layer (131). Hypermetabolism in Purkinje cells probably reflects chronic depolarization of their membrane potential (5). In adult mutants where these cells have disappeared, COX activity of shrunken molecular and granule layers did not differ from that of wild-type, indicating normal metabolic activity relative to tissue weight but higher metabolic activity relative to absolute levels (132). COX activity relative to tissue weight was higher than controls in deep cerebellar and lateral vestibular nuclei (Table 4), attributable to lost gamma-amino butyric acid (GABA)-ergic inhibitory input from Purkinje cells, since excitatory synapses have higher energy demands than inhibitory ones (130). COX activity of Grid2 ^Lc mutants was also elevated in several regions directly connected with the cerebellum, probably as a result of the hypermetabolic deep nuclei, the only pathway out of the cerebellum. In an opposite fashion, COX activity decreased in inferior olive, precursor to retrograde degeneration as a consequence of Purkinje cell atrophy.

Regional brain metabolism of Grid2 ^ho-Nancy mutants differed substantially from Grid2 ^Lc in view of their milder cerebellar atrophy (133). COX activity in the molecular layer of Grid2 ^ho-Nancy mutants was higher than that of non-ataxic controls, probably due to increased neural activity of cerebellar afferents in response to defective dendritic organization of Purkinje cells. Unlike Grid2 ^Lc , COX activity in deep cerebellar nuclei and inferior olive was unchanged in Grid2 ^ho-Nancy probably because of the preservation of Purkinje cells.

Rora ^sg mutants share many neuropathological characteristics with Grid2 ^Lc , in particular massive degeneration in cerebellar cortex. Like Grid2 ^Lc , COX activity of Rora ^sg mutants was higher than that of non-ataxic controls in deep cerebellar nuclei and cerebellar target structures such as red and interpeduncular nuclei, but unchanged in granular and molecular cerebellar cortex, though elevated in Purkinje cells (80). Once again, hypermetabolism in deep nuclei is probably due to decreased GABAergic input from depleted Purkinje cells. Unlike Grid2 ^Lc , COX staining of Rora ^sg mutants was normal in inferior olive, presumably because of a stabilization in retrograde degeneration. In contrast to the previous two mutants, COX activity decreased in thalamic nuclei, probably as a direct consequence of Rora deletion, since this gene is highly expressed there in normal mouse brain (134, 135).