We tested whether differential effects would be produced by employing a short inter-dose treatment interval to better model accumulating plasma MA concentrations as predicted by pharmacokinetic models of human exposure than the typical neurotoxic regimen used in rodents that exceeds the plasma half-life of the drug. We compared the two dosing regimens on separate aspects of MA-induced effects: (1) on markers of neurotoxicity 72 h post-treatment, and (2) on behavior at longer post-treatment intervals. The principal new finding was that, regardless of dosing regimen, MA induced path integration learning deficits in the CWM while sparing spatial navigation in the MWM even though a more demanding procedure in the MWM was used than previously employed to assess spatial ability after MA exposure.
Path integration is conserved in organisms ranging from ants (Wittlinger et al., 2006
), to rodents, to humans (Etienne and Jeffery, 2004
). It is a form of egocentric learning that relies upon self-movement cues to locate places in an environment based on direction and rate of movement, i.e., trajectory or vector learning (Etienne and Jeffery, 2004
). Unlike spatial or allocentric (landmark-based) learning, path integration is dependent on movement cues (primarily internal) rather than visual orientation to distal landmarks (Etienne and Jeffery, 2004
). The neural circuits underlying path integration in rats partially overlap with those of spatial navigation inasmuch as some place cells in the hippocampus are activated during path integration, however, path integration depends heavily upon head-direction cells in the presubiculum and grid cells in the entorhinal cortex (Whishaw et al., 1997
;Rondi-Reig et al., 2006
;Witter and Moser, 2006
;Fuhs and Touretzky, 2006
;Sargolini et al., 2006
;McNaughton et al., 2006
). What makes the path integration effects unique following neurotoxic doses of MA is both the magnitude and extent of the deficits: MA-treated animals displayed no evidence of performing as well as controls even after 15 days of testing. By the last day of testing, MA-treated animals had greater than four-fold increases in latency and errors compared to SAL-treated animals. Interestingly, while CWM performance was impaired, no effects on MWM performance, a hippocampally-dependent task (Morris et al., 1982
) were observed, despite the use of an extensive, multiphase testing method that has proven sensitive to other drug effects (Vorhees and Williams, 2006
). Examination of suggests that the animals given MA may perform better in the MWM, although no statistical differences were noted between groups during this phase. Nonetheless, the lack of MWM effects supports the findings of others of no overall effects of MA on acquisition (Schroder et al., 2003
) or reversal learning (Friedman et al., 1998
) in the Morris maze, although Friedman et al. did find an effect on a single test day. Whether this represents a meaningful effect remains to be determined. Given that no effects of MA were seen in the MWM, the data demonstrate that it is possible to functionally separate effects on path integration from spatial mapping, despite neural network overlap (Whishaw et al., 1997
). We have previously demonstrated a similar result following a single day administration of fenfluramine, that is, effects were observed in the CWM but not the MWM (Williams et al., 2002
The significance of the path integration deficits in rats in relation to human MA users is not yet known because no human study has assessed this specific function, but it is noteworthy that MA affects cortical regions in humans (Meredith et al., 2005
;Barr et al., 2006
;Baicy and London, 2007
) and path integration is a cortically-mediated function. Moreover, human virtual path integration tasks have recently been developed and used in fMRI experiments to map the locations of path integration in humans (Wolbers et al., 2007
). Hence, future studies of this function in MA users may now be feasible.
We also replicated the previous finding of NOR deficits (Bisagno et al., 2002
;Schroder et al., 2003
;Belcher et al., 2005
;He et al., 2006
;Belcher et al., 2006
). With the effects on NOR reported here, this effect is now the most widely replicated cognitive effect arising from exposure to a neurotoxic dosing regimen of MA and is further strengthened by the fact that both dosing regimens used here caused the same effect. The similarity across dosing regimens was also seen on path integration, i.e., both dosing regimens induced essentially identical effects.
Other learning and memory effects of MA have also been reported. For example, in a test of route or motor learning (in which animals learned a specific path through corridors without choices), Chapman et al. (2001)
reported impairments in latency to complete the task. In a later study by the same group, latency was unaffected, although they reported that on the last day of testing there was a significant reduction in a measure of ‘directness’ in the MA-treated group (Chapman et al., 2001
;Daberkow et al., 2005
). While it is appreciated that the motor learning task above and the CWM involve learning a sequence to solve the task, and the neostriatum has been implicated in sequence learning (Potegal, 1972
;Cook and Kesner, 1988
), the ability of animals to learn in the CWM may be different. Most notably, the CWM was run under infrared lighting, eliminating distal and local surrounding cues, whereas the motor learning task was run under lighted conditions. We have previously shown that even under low light conditions, animals perform better in the CWM compared to when animals learn under infrared conditions; presumably they use a combination of strategies to solve the maze when light is present [c.f., (Williams et al., 2002
;Able et al., 2006
)]. The CWM does not offer the animals the ability to know when they have reached the end of an alleyway or are in the center of it as in the motor learning task, therefore, not only do the animals have to learn a sequence of turns, but they must also determine the exact location of each turn, otherwise a “correct turn” could lead to entry in a dead-end channel. At the present time, it is unknown whether the CWM involves striatal functions in its solution and future studies will be necessary to investigate this possibility.
The mechanism(s) of MA-induced deficits in path integration are not yet known. We demonstrated increased GFAP in the neostriatum and decreased monoamines at 72 h after MA administration in both the neostriatum and hippocampus as have others (Bowyer et al., 1994
;Cappon et al., 1997
;O’Callaghan and Miller, 2002
;O’Dell and Marshall, 2002
). Given that brain monoamines were reduced 3 days post-treatment and were still reduced (albeit to a somewhat lesser degree) at the end of behavioral testing 2 months later, these data taken together indicate that monoamines were reduced throughout the course of the behavioral experiment; raising the question of whether these reductions mediate the behavioral changes. While it seems likely that DA reductions were involved in the hypolocomotion initially observed in the animals after MA treatment, their role in path integration and NOR remains unknown.
The present experiment does not rule out the possibility that test order or time since treatment may have contributed to the lack of effect in the MWM, but this seems unlikely. No effects on spatial learning following MA with a shorter treatment-to-test interval were seen in another study that used a neurotoxic MA regimen nearly identical to the 2 h regimen in the present experiment (Schroder et al., 2003
). Moreover, the deficit in CWM performance observed herein persisted throughout the 15 days of testing. Given that only 2 days elapsed between the end of CWM and the beginning of MWM testing, it is unlikely that 4-fold learning deficits lasting 29 days after treatment would disappear 48 h later. It is clear, however, that an experiment to specifically rule out the possibility of test order or treatment-to-test interval questions will be needed. Furthermore, the present experiment does not address the question of how long the path integration deficits remain.
In addition to the cognitive deficits in the CWM, we also demonstrated that MA-treated animals had reduced locomotor behavior, as shown previously (Wallace et al., 1999
). This reduction in locomotion may be a result of the decreased DA observed after MA treatment as suggested above, but this change had no effect on swimming speed measured in either straight channel swimming or in the MWM. We also demonstrated a heightened stimulatory response followed by hypoactivity in MA-treated animals following MA challenge. This suggests increased sensitivity to the DA-releasing effects of MA, but precisely why the response was biphasic is unclear. It does not appear similar to receptor supersensitivity (Iwazaki et al., 2007
). An examination of repetitive beam breaks during testing revealed no evidence that MA-induced hyperactivity was replaced by increased focused movements. While repetitive beam breaks are not synonymous with stereotypy, they do capture aspects of focused movements that are part of the spectrum of stereotypic behaviors.
In the present study we demonstrated altered adrenal function as evidenced by increased corticosterone and hypotrophic spleens and thymuses 3 days after MA treatment. Given that the time since treatment exceeded the half-life of MA by more than 60 times, this indicates that no detectable drug would be present to explain the extended corticosterone increase. If increased corticosterone is the result of an altered circadian response or a direct effect on adrenal sensitivity is unknown, but previous studies have demonstrated that MA treatment can alter circadian rhythms (Honma and Honma, 1986
), although it seems unlikely that this alone could explain the magnitude and persistence of the CWM impairment.
In order to model human use of MA, several factors have to be considered. These include amount of drug taken, frequency of use, chronicity and ADME (absorption, disposition, metabolism, and elimination). Species differences in elimination rates have been suggested to have significant impact when repetitive dosing is considered as with chronic drug use; that is, when one considers total exposure based on internal dose and area under the curve (AUC) calculations (Cho et al., 2001
). Such considerations led us to test the concept that the neurotoxic and behavioral sequelae of the typical neurotoxic dose regimen might be different if a dosing model were used that was designed to produce an internal dose that accumulates to steady-state rather than fluctuating dramatically from one dose to another. We did not take blood samples to measure plasma MA, but rather relied on the modeling data reported previously (Cho et al., 2001
) comparing 4 doses given every 2 h to 24 doses given every 15 min matched for total dose, 40 mg/kg. We found no differences on any measure of neurotoxicity (monoamine or GFAP), or learning, or any other behavioral measure except for one difference in stereotypy 1–2 h after the last dose.
The present data demonstrate an unrecognized effect of MA on path integration learning and verified effects on brain GFAP, DA, and 5-HT, and on peripheral corticosterone release, but the results do not establish which of these may be important in the learning effects. We already summarized the evidence that DA is unlikely to be involved. However there is no evidence ruling out roles for 5-HT or corticosterone. The CWM may prove to be useful in future investigations of the above-mentioned or other possible mechanisms underlying the cognitive deficits reported in current and abstinent MA users (Meredith et al., 2005
; Barr et al., 2006