A rat model of the cognitive impairment from Pfiesteria piscicida exposure.

Pfiesteria piscicida Steidinger & Burkholder, an estuarine dinoflagellate known to kill fish, has also been associated with neurocognitive deficits in humans. We have developed a rat model to determine the cause-and-effect relationship between exposure to Pfiesteria-containing water and cognitive impairment and to determine the neurobehavioral mechanisms underlying the Pfiesteria effect. The rat model of Pfiesteria toxicity can also provide important information concerning the toxin or toxins responsible for neurocognitive deficits resulting from Pfiesteria exposure. With the rat model we have repeatedly documented a Pfiesteria-induced choice accuracy impairment during radial-arm maze learning. The Pfiesteria-induced impairment was relatively specific to the acquisition phase of training. When rats were pretrained, Pfiesteria treatment did not affect performance. However, when these same rats were retrained on another task, the Pfiesteria-induced impairment became evident. Pfiesteria-induced effects were also seen in a locomotor activity test in the figure-8 apparatus and selected components of the functional observational battery. Pfiesteria effects on choice accuracy in the radial-arm maze in rats constitute a critical component of the model of Pfiesteria toxicity, as the hallmark of Pfiesteria toxicity in humans is cognitive dysfunction. Our finding that analysis of the first six sessions of radial-arm maze testing is sufficient for determining the effect means that this test will be useful as a rapid screen for identifying the critical neurotoxin(s) of Pfiesteria in future studies.

Pfiesteria piscicida Steidinger & Burkholder is a toxic estuarine dinoflagellate that causes lethal toxic effects in fish (1,2). Adverse health effects have been found in individuals after accidental Pfiesteria exposure in the laboratory (3). The health effects in these individuals were characterized by cognitive disturbance, fatigue, mood lability, dermal lesions, and respiratory impairment. Similar associations have been seen in a cohort of individuals exposed in field conditions in Maryland (4). In order to rigorously determine the cause-and-effect relationship between Pfiesteria exposure and neurobehavioral impairment, an experimental animal model is necessary. We have developed a rat model of Pfiesteria-induced neurocognitive impairment to demonstrate and characterize the causative relationship and to help identify the critical toxin(s) underlying Pfiesteriainduced impairment.
We conducted a series of studies to characterize Pfiesteria effects on cognitive performance in the radial-arm maze (5)(6)(7)(8)(9). These studies laid the groundwork for the rat model of Pfiesteria-induced cognitive impairment. The route of exposure for humans is not certain. It is likely to be inhalation of the aerosol with some possibility of ingestion or transdermal absorption. The subcutaneous (sc) route was chosen for the current studies to ensure delivery of known amounts of Pfiesteria into the body in a reliable fashion. Because the field is at an early stage in which the chemical identity of the toxin(s) is not known, our intention was to determine the nature of the neurobehavioral toxicity in a reliable manner, then proceed with the more environmentally relevant but more difficult to control routes such as aerosol inhalation when chemical analysis could be used to verify the dose delivered. We documented a consistent Pfiesteriainduced learning impairment when Pfiesteria was administered before training was initiated (6). However, when Pfiesteria was administered after rats were trained on the radial-arm maze, no deficit in retention was observed. Nevertheless, when these Pfiesteria-treated rats were subsequently given a new task to learn, they showed a significant deficit. The learning impairment was evident at least 10 weeks after a single exposure. With this series of studies we developed a reliable model of the neurocognitive deficits caused by Pfiesteria exposure. The detailed results of these studies are presented in this article. They have brought us to the point where we can use the animal model to investigate both the character of the Pfiesteria-induced cognitive deficit and the mechanisms of its action, including the search for the critical toxin or toxins.

Methods
Pfiesteria samples for this series of studies were collected from aquaria at the North Carolina State University laboratory of J. Burkholder, in which Pfiesteria were killing fish. The water in the aquaria had ammonia levels less than 200 µg/L and nitrate levels less than 500 µg/L. The aquarium water was injected with no additives into sealed glass test tubes. The tubes were then frozen on dry ice for at least 1 hr. The samples were frozen as a precaution to attenuate possible degradation of the toxin(s). In all the studies, the tubes holding the samples were warmed until no ice crystals remained before sc injection into the rats. In studies 1-4, controls were injected sc with a volume that was the average of the dosed groups. In studies 5-7 all rats in control and dose groups received the same volume of injection-3 mg/kg. Subjects were adult female Sprague-Dawley rats (Zivic-Miller, Allison Park, PA, USA) except for those in the juvenile experiment, which used 24-day-old male and female Sprague-Dawley rats. The rats were housed in groups of 2-4 in plastic cages with wood shavings. All rats in a cage received the same Pfiesteria exposure. They were in an approved vivarium immediately adjacent to the behavioral test facility. They were on a reverse 12:12 light:dark cycle with testing during the behaviorally active, dark phase. All rats had ad libitum access to water, and except for experiment 1, they were on scheduled feeding after testing. These studies were conducted under an approved protocol of the Animal Care and Use Committee of Duke University in an AAALAC-approved facility.

Experiment 1: The Pilot Study
This was an initial evaluation of the acute and persisting behavioral effects of Pfiesteria. Six rats were injected sc with Pfiesteria in doses from 35,600 to 961,200 Pfiesteria cells per kg of rat body weight. Six control rats were not injected. Two days after exposure we began Pfiesteria piscicida Steidinger & Burkholder, an estuarine dinoflagellate known to kill fish, has also been associated with neurocognitive deficits in humans. We have developed a rat model to determine the cause-and-effect relationship between exposure to Pfiesteria-containing water and cognitive impairment and to determine the neurobehavioral mechanisms underlying the Pfiesteria effect. The rat model of Pfiesteria toxicity can also provide important information concerning the toxin or toxins responsible for neurocognitive deficits resulting from Pfiesteria exposure. With the rat model we have repeatedly documented a Pfiesteria-induced choice accuracy impairment during radial-arm maze learning. The Pfiesteria-induced impairment was relatively specific to the acquisition phase of training. When rats were pretrained, Pfiesteria treatment did not affect performance. However, when these same rats were retrained on another task, the Pfiesteria-induced impairment became evident. Pfiesteria-induced effects were also seen in a locomotor activity test in the figure-8 apparatus and selected components of the functional observational battery. Pfiesteria effects on choice accuracy in the radial-arm maze in rats constitute a critical component of the model of Pfiesteria toxicity, as the hallmark of Pfiesteria toxicity in humans is cognitive dysfunction. Our finding that analysis of the first six sessions of radial-arm maze testing is sufficient for determining the effect means that this test will be useful as a rapid screen for identifying the critical neurotoxin(s) of Pfiesteria in future studies. testing rats in the win-shift radial-arm maze task (below) for 18 sessions over 6 weeks. The rats were on an ad libitum feeding schedule to ensure that the Pfiesteria exposure did not adversely affect free-feeding body weight. In later studies, the usual procedure of daily scheduled feeding after testing to maintain body weight at approximately 85% of ad libitum levels was used so that the rats were more motivated to run the maze for food reinforcement.

Experiment 2: The Repeat Study
This study was was a more focused evaluation of the 106,800 cells/kg Pfiesteria dose. The same Pfiesteria sample used in experiment 1 was used in this experiment. It had been stored sealed and frozen at -4°C for 7 weeks. Ten rats were injected sc with Pfiesteria and 10 with saline. Win-shift radial maze training began 2 days after exposure.

Experiment 3: The Fresh Sample Study
In this study we evaluated the effects of a fresh sample of Pfiesteria collected from aquaria at North Carolina State University in Raleigh, North Carolina. The sample was frozen at -4°C overnight. Ten rats were injected with Pfiesteria at the benchmark dose of 106,800 cells/kg and 10 controls were injected with aquarium water collected from tanks that did not contain Pfiesteria. The rats began training in the win-shift radial maze task for 18 sessions over 6 weeks starting 2 days after exposure. After testing, the rats were sacrificed and the brain, blood, lungs, liver, kidneys, and spleen were collected for pathological assessment. Gross and microscopic examination of hematoxylin and eosin (H & E) stained sections were made in search of lesions or signs of pathology. Glial fibrillary acidic protein (GFAP) immunoreactivity was determined to check for more subtle signs of toxicity in Pfiesteria-exposed animals

Experiment 4: The Pretraining Study
Here we determined if the deficits observed in radial-arm maze performance were due to impairments in learning or memory. Rats were pretrained for 18 sessions on a radialarm maze win-shift task before Pfiesteria administration. Then they were administered Pfiesteria samples at doses of 0, 35,600, or 106,800 cells/kg. As in experiment 3, a fresh sample of Pfiesteria collected from aquaria at North Carolina State University, which had been frozen at -4°C overnight, was used and the vehicle control was aquarium water without Pfiesteria. Two days after exposure, testing on the radial-arm maze win-shift task resumed. The rats were tested for the following 6 weeks for an additional 18 sessions. Then to assess persistent Pfiesteria effects on learning, we changed the rules of the task from win-shift in which there was one reward at the end of each maze arm to repeated acquisition in which there were rewards at the ends of only three of the eight arms (radialarm maze methods are discussed below). Rats were tested for six sessions over 4 weeks, using a repeated acquisition task in the same 8-arm radial maze with the same environmental cues. The rats were also assessed at 1 hr, 1 week, 4 weeks, and 9 weeks postexposure using the Functional Observational Battery (FOB) (below).

Experiment 5: The Cue Structure Study
In this study, we examined the importance of testing environment on the Pfiesteria effects on radial-arm maze choice accuracy. Two similar radial 8-arm mazes were used in this experiment. One maze was located in a standard test room with dimensions of 4.57 × 3.43 m and a ceiling 2.90 m high. The other maze was in a sound-attenuating chamber 3.35 × 3.05 m and a ceiling 1.98 m high. Half the rats in each condition were tested in each room. After 18 sessions of training, the testing room of the rats was switched from one room to the other for six sessions. For a final three sessions the testing room was switched back to the original training room.
The studies were vehicle-controlled experiments using two fresh Pfiesteria samples. The samples were collected directly from an aquarium at the Burkholder laboratory in which P. piscicida cultures were actively killing fish. The Pfiesteria cell concentration was determined by counting identified cells per unit volume under light microscopy (10). Taxonomy was confirmed by scanning electron microscopy. The aquarium water was injected with no additives into sealed glass test tubes and frozen overnight before injection into the rats. Doses of 35,600, 106,800, and 320,400 Pfiesteria cells/kg of rat body weight were injected sc. In addition one group was injected with the equivalent amount of water from the Pfiesteria culture as the 106,800 cells/kg dose from which the cells were removed by filtration (0.02-micron filter size). The 106,800 filtered group was included to test the necessity of administering the Pfiesteria cells for the toxic effects seen. There were two control groups, one injected with saline and one with water from aquaria without Pfiesteria.

Experiment 6: The Sample-Type Study
This provided a screening analysis of the neurobehavioral potency of three Pfiesteria cultures: B-113-3 (Pf-113), B-7-28-B (Pf-728), and B-Vandermere (Pf-Van), which were gathered at three different sites. There were six dose groups. Ninety-six (12/group) adult female Sprague-Dawley rats were injected (sc) with a single dose of Pfiesteria taken from aquarium-cultured Pfiesteria (35,600 or 106,800 Pfiesteria cells/kg of rat body weight). The Pfiesteria-treated rats were compared with groups of rats injected with either saline (saline controls) or aquarium water without Pfiesteria (tank-water controls). The three sample types had the same average concentration (23,700-24,300 cells/mL). The potency of the samples in killing fish per day of exposure varied from 49% lethality for Pf-113 to 93% for Pf-Van and 100% for Pf-728. One control group (n = 12) was injected with saline and one (n = 12) with aquarium water not containing Pfiesteria. We used a neurobehavioral screen consisting of a short six-session sequence of radial-arm maze testing, two 1-hr sessions in the figure-8 activity maze, and two sessions of the FOB.

Experiment 7: The Juvenile Study
We examined the doses of the Pfiesteria sample type Pf-728, which was most effective in experiment 7 in juvenile rats. Seventy-two male and female Sprague-Dawley rats 24 days of age were injected (sc) with a single dose of Pfiesteria sample type Pf-728 from the Burkholder laboratory. Each of the following conditions was tested with 12 male and 12 female rats: tank-water controls, 35,600 cells/kg, and 106,800 cells/kg. The behavioral assessment began 2 days after injection.
Radial-arm maze training. A radial 8-arm maze, constructed of wood and painted black, was used in this study. The maze consisted of a central arena 50 cm in diameter with eight 10-cm × 60-cm arms extending radially. Food cups were located 2 cm from the end of each arm. The maze was 30 cm above the floor and was located in a testing room that contained many extra-maze visual cues. The cues were kept in the same position throughout training. The rats in each Pfiesteria treatment condition were tested on the win-shift radial-arm maze procedure 3 days per week. Before each session, all the arms of the maze were baited with 1/3-1/2 piece of sugarcoated cereal (Kellogg's Froot Loops; Kellogg Company, Battlecreek, MI, USA). A radialarm maze test session was started when the rat was placed in a circular plastic ring in the central platform. After 10 sec the ring was lifted and the rat was allowed to freely explore the maze. Arm choices were recorded when the rat had placed all of its paws beyond the threshold of the arm. The reinforcers (Froot Loops) were not replaced during the session. The radial-arm maze test session continued until the rat had entered all eight arms or 5 min had elapsed. Because it has been determined to be a sensitive and reliable index through more than 15 years of study, the number of entries until an error was made (entries to repeat) was used as the choice accuracy measure. Entries to repeat is a better measure than the number of errors to finish the maze in that it indexes when the first error occurs. Because the radial-arm maze becomes more difficult as the session progresses and fewer reinforcements remain, it is important to determine at what stage of the session the first error occurs. During the early stages of training in the radial-arm maze rats often do not finish the maze during the time allotted. The entries to repeat score is a way to index choice accuracy even when the maze is not completed. Random chance performance for the entries to repeat measure on an 8-arm maze as determined by computer simulation is 3.25 (11). If the rat did not repeat an entry or enter at least five arms within 5 min, no choice accuracy score was taken for the session and the average value for the other sessions was entered in the session block. The response latency measure was the total session duration divided by the total number of entries (seconds per entry). In the repeated acquisition procedure, three of eight of the arms were baited. On a particular day the same arms were baited for five consecutive trials. Each day the arms baited were changed in a pseudorandom order. The trial was continued until either the rat entered all three baited arms or 3 min elapsed. The dependent measure for repeated acquisition was errors per trial. The testers were blind to the treatment conditions of the rats.
The Functional Observational Battery. The FOB is a series of observations and tests used to evaluate the overall neurological integrity of the rat. Testing was conducted at 7 and 13 days after Pfiesteria exposure. Detailed descriptions of the procedures and scoring criteria have been published elsewhere (12,13). Home-cage observations included any abnormal motor movements as well as activity level. Lacrimation, salivation, piloerection, ease of removal, and handling reactivity were ranked, according to the defined criteria, as the rat was removed from the cage and held in the observer's hand. The rat was then placed on the top of a laboratory cart (60 × 90 cm) and allowed to freely explore for 3 min. During that time, the observer ranked and/or described any gait abnormalities, arousal, activity level, abnormal motor movements, and excretion level (urination, defecation). The number of rearing responses was also counted. Next, the rat's reactions to the sound of a metal clicker, a pinch near the end of the tail, approach of a pen, and touch on the rump were rated. The aerial righting reflex and pupillary response to light were also tested. Finally, forelimb and hindlimb grip strength, landing foot splay, rectal temperature, and body weight were measured. The same observer conducted all portions of the study and was blind to the treatment condition of each rat. All rats from each cohort were tested in one day.
Locomotor activity in the figure-8 apparatus. The rats were tested for locomotor activity in the figure-8 apparatus in a quiet test room. The figure-8 locomotor activity test has been widely used in behavioral toxicology studies (14). Each figure-8 locomotor apparatus consisted of a continuous enclosed alley 10 cm × 10 cm in the shape of an 8, which was 70 cm long and 42 cm wide. There was a central arena 21 cm × 16 cm with a ceiling 20 cm high and two blind alleys extending 20 cm from either side. Eight photobeams crossed the alleys to index locomotor activity; one was located on each of the two blind alleys and three on each of the two loops of the figure-8. The number of photobeam breaks in each 5-min block in a 1-hr session were tallied by a microcomputer.

Experiment 1: Pilot Study
The Pfiesteria-treated rats had significantly lower average entries to repeat scores (p < 0.005) than controls averaged over 18 sessions of testing (6). The controls averaged 5.5 ± 0.2 entries to repeat, whereas the Pfiesteriatreated rats averaged 4.8 ± 0.1. Latency was not significantly affected by Pfiesteria exposure. This effect was replicated in later studies (below) in which we tested the effects of Pfiesteria exposure on radial-arm maze acquisition over 18 sessions ( Figure 1A,B).

Experiment 2: Repeat Study
Significant learning took place, but there was no significant effect of Pfiesteria exposure (6). Over 18 sessions of training, controls averaged 6.1 ± 0.3 entries to repeat; Pfiesteriaexposed rats averaged a slightly lower 5.7 ± 0.3. We hypothesized that the 7-week storage time of the sample attenuated its potency. J. Burkholder and her group have also found that the toxic effects of Pfiesteria on fish decline rapidly over 48 hr after removal of the cells from the aquarium (15), which is consistent with this interpretation. In subsequent experiments only samples stored frozen overnight were used.

Experiment 3: Fresh Sample Study
There was a significant effect of Pfiesteria exposure on choice accuracy in the radialarm maze (Figure 2) (6). The main effect of Pfiesteria exposure was significant (p < 0.025), with the controls averaging 6.2 ± 0.2 entries to repeat and the Pfiesteria-exposed rats averaging 5.4 ± 0.2 entries to repeat over the 24 sessions of testing. There was a significant effect of session block (p < 0.0001) and a significant session block × Pfiesteria interaction (p < 0.025). Analyses of the simple main effects of Pfiesteria at each of the session blocks showed significant Pfiesteriainduced deficits during sessions 10-12 (p < 0.05), 13-15 (p < 0.005) and 16-18 (p < 0.005). After we had trained the rats for our standard 18 sessions for acquisition, we Environmental Health Perspectives • VOLUME 109 | SUPPLEMENT 5 | October 2001 Session block (no.) • added an additional phase of testing, for a total of 24 sessions because the Pfiesteriatreated rats had not shown any learning during the standard 18-session training period; the shorter training period was sufficient for control rats. The Pfiesteria-treated rats improved during the additional phase of training such that they overcame the significant deficits seen earlier. This improvement shows that Pfiesteria-treated rats were not incapable of learning the maze, but just retarded in their acquisition. The delayed acquisition is not a demonstration of recovery since additional training was required. With true recovery, training on a new task would proceed at the same rate as that for controls. This was tested in experiment 4 (below). No significant effects of Pfiesteria exposure were seen in terms of response latency. No significant effects of Pfiesteria exposure were seen in the complete blood count assessment and white blood cell differential counts. Gross and microscopic examination of H & Estained sections did not reveal any obvious lesions or signs of pathology. GFAP immunoactivity was not increased in the brains of Pfiesteria-exposed animals. However, these pathological tests were performed approximately 8 weeks after dosing.

Experiment 4: Pretraining Study
With the postacquisition win-shift testing, there were no significant Pfiesteria-induced deficits (8). Averaged over the 18 sessions of testing after dosing, the controls averaged 6.1 ± 0.2 entries to repeat, whereas the low-dose group averaged 6.6 ± 0.2 and the high-dose group averaged 6.6 ± 0.2 entries to repeat. This showed that Pfiesteria treatment at a dose that significantly impaired win-shift radial-arm maze performance during the acquisition phase did not impair performance during the postacquisition phase. Since the same sensorimotor, motivational and working memory functions are required, this result supported the existence of a relatively selective Pfiesteria effect on learning. Pfiesteria was not without effect during this phase of testing. There was a significant Pfiesteria effect on response latency (p < 0.05). The 106,800-cells/kg dose caused a significant decrease in latency relative to either the controls (p < 0.25) or the 35,600 cells/kg dose group. Controls averaged 25.9 ± 3.3 sec per entry, the 35,600-cells/kg group averaged 24.2 ± 2.9 sec per entry, whereas the 106,800-cells/kg dose group averaged 16.4 sec per entry.
To further assess the persistent effects of Pfiesteria on learning, the rats were switched to the repeated acquisition procedure in the radial-arm maze. The three groups performed equally poorly during the first phase of training just after the switch (Figure 3). However, there was a significant Pfiesteria-induced learning deficit caused by the higher dose during the second training block (p < 0.05). There were significant Pfiesteria decreases in latency in both session block 13 (p < 0.025) and session block 46 (p < 0.01). Before switching to repeated acquisition training, the rats had a total of 36 sessions of training on the same maze with the same cues. Thus, the Pfiesteria-induced deficit did not seem to be due to problems associated with familiarization with handling, the maze, or environmental cues. The Pfiesteria-induced learning deficit seen in the repeated acquisition task was 10 weeks after the time of Pfiesteria exposure, providing evidence for persistent effects. Future studies will carry out the testing for more extensive periods to determine if the Pfiesteria-induced repeated acquisition deficit is related to the shift from prior win-shift radial-arm maze testing.
The FOB was also used to assess the effects of Pfiesteria exposure in experiment 4. The rats were tested on the FOB 1 hr, 1 week, 4 weeks, and 9 weeks after Pfiesteria exposure. The 1-week and 4-week time points were at the same time as win-shift radial-arm maze retesting, and the 9-week time point was at the same time as repeated acquisition radial-arm maze testing. There were no significant differences on the measures of sensorimotor function, no abnormal motor behaviors, and no changes in physiological parameters (e.g., body temperature). The only significant Pfiesteria-induced changes were differences in habituation across repeated testing sessions. There was significant habituation in the controls and low Pfiesteria groups (p < 0.005) on measures of arousal and rearing (Figure 4). Rats receiving the high dose of Pfiesteria, however, showed significantly less habituation (p < 0.05). The lack of habituation in Pfiesteriatreated rats in the FOB could be a representation of the cognitive deficits also seen in these rats, i.e., a learning impairment. However, there are possible noncognitive explanations as well. Future studies will help to determine the relationship of learning deficits to other neurobehavioral changes caused by Pfiesteria exposure.

Experiment 5: Cue Structure Study
During the initial phase of training there was a Pfiesteria-induced impairment in radial maze acquisition in the standard test room. The improvement in choice accuracy (entries to repeat) from the first session block (session 13) to the second session block (session 46) was analyzed. Analysis of entries to repeat during the early phase of acquisition showed there was a significant treatment × session block interaction (p < 0.05) ( Figure 5A). Follow-up analysis of the improvement showed there was a significant main effect of  Pfiesteria treatment (p < 0.05). The planned comparisons of the control groups versus the Pfiesteria-treated groups showed that the 106,800-cells/kg, the 320,400-cells/kg, and 106,800-cells/kg filtered groups each had significantly (p < 0.05) less improvement than either the saline and tank-water-treated control groups. In fact, each of these groups had, on average, a decline in choice accuracy from the first to the second training session block ( Figure 5B). This result is similar to that seen in experiment 3 in this series ( Figure 2). The lowest dose group (35,600 cells/kg) was not significantly different from either control group. This shows the same threshold for effect as was seen in experiment 4 ( Figure 3).
In contrast, the rats tested on the same task in the sound-attenuating chamber did not show a deficit. The controls showed similar rates of acquisition in the two rooms (p = 0.38). The differential effect was seen only with the Pfiesteria-exposed rats (p < 0.005). With the maze in the sound-attenuating chamber the controls showed an improvement of 0.72 ± 0.55 entries to repeat, whereas the Pfiesteriatreated rats showed a similar improvement of 0.93 ± 0.29 entries to repeat (p = 0.68). In contrast, with the maze in the standard open test environment, the controls showed an improvement of 1.26 ± 0.23 entries to repeat, whereas the Pfiesteria-treated rats actually showed a worsening of performance with -0.54 ± 0.32 (p < 0.005). During the 18-session acquisition period, the rats trained in the standard test room showed a magnitude of Pfiesteria-induced deficit similar to that seen in previous studies ( Figure 6). When the rats were shifted from testing in one room to the other, the Pfiesteria-treated rats shifted from the sound-attenuating chamber to the standard test room had a significantly greater decrement in performance than the controls (p < 0.05). As shown in Figure 7, the controls averaged a net loss of -0.87 ± 0.38 entries to repeat from sessions 16-18 to 19-21, whereas the Pfiesteria-treated rats averaged a net loss of -1.71 ± 0.27 entries to repeat over the same period. In contrast, there was no difference between the Pfiesteriatreated and control rats switched from testing in the standard test room to the soundattenuating chamber. The change from the standard test room to the chamber did cause a significant overall decrease in accuracy (p < 0.05) simply because the deficit with this shift was substantially less than the deficit when animals were shifted from the chamber to the standard test room.
The hypothesis that the test environment was important in the expression of the Pfiesteria-induced effect was tested over the course of sessions [16][17][18][19][20][21][22][23][24][25][26][27]. For the first block of sessions (16)(17)(18) during this period, the rats continued to be tested in the test room in which they had previously been trained. During the next two session blocks (sessions 19-21 and sessions 22-24) they were switched to the test room opposite the one in which they had initially been trained. Then, during the final session block (sessions 25-27), the rats were switched back to their original test room. Thus, all the rats were tested in both test rooms A and B on either an ABBA or a BAAB schedule over the final four blocks of testing. The results from the initial training phase showed that the Pfiesteria-induced deficit was present in the standard test room (A) but not in the soundattenuating chamber (B), so the difference in choice accuracy performance in the two rooms was analyzed. This analysis showed that the groups given 35,600 cells/kg (p < 0.05) and 106,800 cells/kg (p < 0.025) were significantly worse than the tank-water controls when they were tested in the standard test room versus the sound-attenuating room. Restriction of possible distracting cues in the sound-attenuating chamber may have reduced expression of the Pfiesteria effect on learning. The attentional explanation is a hypothesis that will be tested explicitly in the operant attentional testing in future studies.
There were no overt signs of generalized debilitation in the animals during the period of maze testing. No significant Pfiesteria treatment effects were seen in terms of average response latency. The FOB measurements taken during this period did not detect neurological deficits or overt toxicity.
Locomotor activity measurements were taken in the figure-8 maze 11 weeks after exposure. Although no substantial Pfiesteriainduced alterations were observed in average locomotor activity, a significant Pfiesteriainduced change was observed in habituation. The middle-dose and high-dose Pfiesteria groups had significantly greater rates of habituation than the tank-water controls (Figure 8). No difference was seen between the tank-water-treated and the saline-treated controls or the low-dose Pfiesteria group. There was a nearly significant greater habituation in the filtered middle-dose Pfiesteria group. The greater linear trend in the Pfiesteria groups comprised slightly greater activity counts during the early time blocks and slightly lower during the later time blocks.

Experiment 6: Sample-Type Study
There was a significant effect of all three Pfiesteria samples (p < 0.05), impairing choice accuracy over the first six sessions of radial-arm  maze testing (Figure 9) (9). No differences were seen between the different doses of Pfiesteria or between the two control groups. One of the three Pfiesteria samples caused a significant latency increase in the radial-arm maze, but the interpretation of this effect was clouded by the finding of significant differences in the latency of the saline-and tankwater-treated control groups. At the time of the radial-arm maze choice accuracy impairment, no overt Pfiesteria-related effects were seen using the FOB, indicating that the Pfiesteria-induced choice accuracy deficit was not due to generalized debilitation. In the figure-8 maze, the same Pfiesteria treatment, which increased latency in the radial-arm maze, caused a significant mean decrease (p < 0.05) in activity over the 1-hr locomotor test ( Figure 10). The rate of habituation (linear trend of decreased locomotion) was not significantly affected. Pfiesteria effects on choice accuracy in the radial-arm maze and activity in the figure-8 maze and in early radial-arm maze training may be useful in a rapid screen for identifying the critical toxin(s) of Pfiesteria in future studies.

Experiment 7: The Juvenile Study
Rats given the higher dose of Pfiesteria showed a significant (p < 0.05) impairment in radial-arm maze choice accuracy relative to that of the control group averaged over the six sessions of testing ( Figure 11) (9). No sex-related differences in response to Pfiesteria were seen. Unlike the adults given the same Pfiesteria cultures, the juveniles did not show significant Pfiesteria-induced increased response latency in the radial-arm maze or decreased activity in the figure-8 activity apparatus.

Discussion
We have documented a persistent learning deficit in rats after exposure to water containing Pfiesteria piscicida. The effect is reproducible and robust (6,8,9). The radial-arm maze learning deficit is now very well established, with a total of 36 control and 48 Pfiesteria-treated rats being run and a significant Pfiesteria-induced deficit of p < 0.005. The impairment did not appear to be secondary to generalized health impairment of the animals. Clinical measures of health, blood cell counts and initial histopathological screens did not detect any effects. The FOB did not detect abnormal unconditioned behavior or reflexes, which would explain the deficits in choice accuracy. The roles of memory and nonassociative factors in the Pfiesteria-induced deficit were evaluated. When the rats were administered Pfiesteria after pretraining, they remembered the previous learning and performed just as well as controls. The acute Pfiesteria exposure did not appear to produce persistent sensory, motor, motivational, or memory deficits sufficient to impair radial-arm maze choice accuracy once the task had been learned. Only when the rats were later trained on a different task (repeated acquisition) in the radial maze did they show a significant Pfiesteria-induced deficit. Chronic exposure may well have more pervasive effects on both learning and memory, as seen in chronically exposed humans (3).
In experiment 6, the findings were extended to show that the deficit occurs in both adult and juvenile rats and in both male and female juveniles. The effect in the current study of a significant Pfiesteria-induced choice accuracy deficit during the average of the entire first six sessions was more pervasive than in one of our previous studies (8). There was a significant interaction of Pfiesteria treatment × session block. Follow-up tests of Pfiesteria effects on the rate of acquisition, showed a significant Pfiesteria-induced deficit in the difference between the first and second blocks of three sessions. However, an analysis including all of the data for the previous four studies (6,8) shows a significant (p < 0.025) Pfiesteria-induced deficit for the average choice accuracy over the first six sessions of training in the radial-arm maze.
Although all Pfiesteria samples tested in the neurocognitive studies caused a significant impairment in choice accuracy, only the Pf-728 sample induced a significant increase in response latency in the radial-arm maze. The robustness of this effect was evidenced by the finding that the same Pfiesteria sample was the only one that caused significant hypoactivity in the figure-8 apparatus. The Pf-728 sample may have had a different array of toxins, which caused the additional effect of hypoactivity. Juvenile rats appear to be resistant to the hypoactivity caused by the Pf-728 Pfiesteria in the adults. In contrast to adult rats, no effect of Pfiesteria was seen on locomotor activity in either the radial-arm maze or the figure-8 apparatus in the juveniles.
These data again provide evidence for the specificity of the Pfiesteria-induced impairment on learning. A prominent symptom of Pfiesteria intoxication in humans is cognitive disturbance (3,16). The current results provide additional support for this observation. However, more neurobehavioral studies are needed to determine the critical mechanisms of action of Pfiesteria.
for rapid screening of different Pfiesteria extracts. Radial-arm maze choice accuracy was the most sensitive of the behavioral assessments used. This measure detected significant Pfiesteria-induced deficits with all three of the Pfiesteria samples tested in adult female rats and with male and female juvenile rats tested with the Pf-728 sample. In contrast, no effects were seen with the FOB and only Pf-728 affected radial-arm maze response latency or figure-8 apparatus locomotor activity in adult animals.
There are a variety of negative effects of Pfiesteria intoxication in humans, but the hallmark is cognitive impairment (3,14). It is therefore essential that models of Pfiesteria intoxication include cognitive impairment as a component. The Pfiesteria-induced radialarm maze choice accuracy deficit seen in the current and previous studies appears to fulfill this requirement. We have found that the first six sessions of radial-arm maze testing are sufficient for detecting the Pfiesteria-induced cognitive impairment. This test can be of great use in the determination of the specific toxin or toxins responsible for Pfiesteriainduced cognitive impairment.
In vivo neurobehavioral tests are needed to determine the functional toxic effects of Pfiesteria exposure. Other assays are important but cannot demonstrate the cause-andeffect relationship between Pfiesteria exposure and functional impairment. In vitro cell assays are useful for examining intracellular mechanisms of action but cannot define the functional consequences of Pfiesteria exposure. The potency of Pfiesteria in fish lethality is important for ecotoxicological assessment but does not predict the extent of neurobehavioral effect in mammals. Human studies are important for monitoring possible health effects but often the exposure information is scanty or missing and the causal link cannot be determined. The radial-arm maze choice accuracy measure gathered in six test sessions over a period of two weeks after Pfiesteria exposure has been shown in the current studies as well as previous studies (6,8) to be a sensitive indicator of Pfiesteria-induced cognitive impairment. This test can serve as a sensitive and relatively efficient indicator of neurotoxicity in the search for the critical neurotoxin(s) of Pfiesteria.  Figure 11. Pfiesteria-induced deficit of initial learning in the radial-arm maze (mean ± SEM) in juvenile rats. n = 23-24 rats/treatment group (9). *p < 0.05 vs control group.