Improved neurocognitive performance in FIV infected cats following treatment with the p75 neurotrophin receptor ligand LM11A-31

HIV rapidly infects the central nervous system (CNS) and establishes a persistent viral reservoir within microglia, perivascular macrophages and astrocytes. Inefficient control of CNS viral replication by antiretroviral therapy results in chronic inflammation and progressive cognitive decline in up to 50% of infected individuals with no effective treatment options. Neurotrophin based therapies have excellent potential to stabilize and repair the nervous system. A novel non-peptide ligand, LM11A-31, that targets the p75 neurotrophin receptor (p75NTR) has been identified as a small bioavailable molecule capable of strong neuroprotection with minimal side effects. To evaluate the neuroprotective effects of LM11A-31 in a natural infection model, we treated cats chronically infected with feline immunodeficiency virus (FIV) with 13 mg/kg LM11A-31 twice daily over a period of 10 weeks and assessed effects on cognitive functions, open field behaviors, activity, sensory thresholds, plasma FIV, cerebrospinal fluid (CSF) FIV, peripheral blood mononuclear cell provirus, CD4 and CD8 cell counts and general physiology. Between 12 and 18 months post-inoculation, cats began to show signs of neural dysfunction in T maze testing and novel object recognition, which were prevented by LM11A-31 treatment. Anxiety-like behavior was reduced in the open field and no changes were seen in sensory thresholds. Systemic FIV titers were unaffected but treated cats exhibited a log drop in CSF FIV titers. No significant adverse effects were observed under all conditions. The data indicate that LM11A-31 is likely to be a potent adjunctive treatment for the control of neurodegeneration in HIV infected individuals. Author Summary There are no effective treatments to halt the progression of most neurodegenerative diseases including HIV-associated neurodegeneration. Neurotrophins have the potential to provide strong neuroprotection but it has been difficult to develop usable interventions. A new drug, LM11A-31, that targets the p75 neurotrophin receptor has been developed that provides potent neuroprotection, is orally bioavailable and has the potential to prevent disease progression. The current studies were designed to evaluate the effects of the compound in an animal model of active HIV infection in preparation for a human clinical trial. Treatment of chronically infected animals with LM11A-31 normalized deficits in T maze performance, novel object recognition and open field behavior with no measurable adverse effects. Potential adverse effects associated with natural neurotrophins such as changes in sensory perception and increased systemic viral burden were not observed. A decrease in CSF FIV titers and a slight improvement in the CD4:CD8 ratio suggested that LM11A-31 may have beneficial effects beyond the anticipated neuroprotective effects. These findings are similar to beneficial effects seen in other animal models of neurodegeneration and CNS injury and support the use of LM11A-31 as an adjunctive neuroprotective agent for the treatment of HIV infected individuals.


Introduction
Lentiviruses such as HIV, SIV and FIV rapidly penetrate the central nervous system 3 (CNS) where they establish perivascular macrophage reservoirs and infect microglia 4 and astrocytes. Sustained virus production by microglia and macrophages is difficult to 5 treat with antiretroviral drugs due to poor penetration of the blood brain barrier. The 6 resulting inflammation leads to neural dysfunction and an accelerated aging 7 phenotype(1-8) with varying degrees of cognitive decline in up to 50% of infected 8 individuals (9)(10)(11). Cognitive decline adversely affects quality of life and there are 9 currently few therapeutic options available. Attempts to protect the nervous system are 10 focused, in part, on interventions that reduce damage due to HIV-associated 11 inflammation. We and others have shown that activation of mononuclear phagocytes 12 (microglia and macrophages) results in the release of factors that induce a gradual 13 accumulation of intraneuronal calcium (12)(13)(14)(15)(16)(17)(18)(19). In vitro models of neuroinflammation, 14 show that chronic elevation of calcium is quickly followed by cytoskeletal damage, 15 transport deficits and focal swelling (beading) within dendrites and axons(20). These 16 dendritic and axonal swellings accumulate mitochondria, endoplasmic reticulum (ER), 17 p75 neurotrophin receptors (p75 NTR ) and a unique oligomeric form of Tau(21), as well 18 as other proteins, providing an environment poised for mitochondrial dysfunction, ER 19 stress and pathological protein modifications. Interestingly, an almost identical 20 progression is seen in animal models of Alzheimer disease (AD) indicating that this is a  A number of in vitro studies have shown that neurotrophin signaling has the capacity to 3 reduce calcium-induced changes and prevent cytoskeletal disruption (reduced beading, 4 sparing of dendrites) that underlie the initial development of neuronal vulnerability in 5 response to gp120 and other HIV/FIV virion components. For example, in vitro studies 6 have clearly demonstrated that cell culture medium from macrophages or microglia 7 treated with neurotrophins abrogates lentivirus-induced neurodegenerative changes(20, 8 23-28). To capitalize on the protective potential of neurotrophin signaling, the nerve 9 growth factor (NGF) loop 1 mimetic, LM11A-31, was developed to provide a stable, 10 small molecule ligand targeted to p75 NTR (29, 30). This non-peptide ligand crosses the 11 blood brain barrier and promotes neuroprotective signaling through the activation of pro- 12 survival pathways and inhibition of degenerative pathways(31, 32). In vitro, LM11A-31 13 has neurotrophic and neuroprotective effects in neural cultures treated with HIV 14 gp120(20) as well as models of Alzheimer disease(31, 32) and p75 NTR -mediated 15 apoptosis(33). A number of in vivo studies in animal models have clearly demonstrated 16 that LM11A-31 prevents neural damage and cognitive decline in models of aging, 17 Alzheimer disease, Huntington disease, traumatic brain injury and spinal cord injury with 18 no reported adverse effects(32, 34-37). Thus, LM11A-31 has properties that strongly 19 suggest it is useful as a disease modifying intervention to prevent CNS dysfunction and 20 cognitive decline in HIV infected patients. To assess LM11A-31 efficacy in an animal model of HIV, we treated FIV infected cats 23 with LM11A-31 for a period of 10 weeks and assessed cognitive function, virus 6 1 production, T cell numbers, sensory function, and general health relative to 2 pretreatment and placebo controls. Untreated FIV infected cats have neuropathological 3 changes similar to HIV infected humans although the changes are generally less 4 severe(38) resembling the gradual progression in individuals on antiretroviral therapy. 5 A subtle gliosis can be seen as early as 1 week post-inoculation followed later during 6 late asymptomatic or pre-AIDS stages of disease by: diffuse gliosis, microglial nodules, 7 meningitis, perivascular infiltrates, white matter pallor, multinucleated cells (listed in 8 order from more common to more rare findings). Some neuron loss can be detected in 9 asymptomatic FIV-infected cats at ~3 years post infection and matches the pattern of 10 loss seen in humans(39). Other changes include neurological dysfunction(40-44), 11 cortical atrophy revealed by MRI(43), and discrete white matter lesions(41, 43). These 12 changes indicate that the FIV model is a good model to examine effects of compounds 13 on the gradual development of neurological disease in response to lentiviral infection. 14 We therefore treated FIV infected cats with LM11A-31 at a time when cognitive deficits 15 first begin to appear to determine if the compound could suppress disease progression. 16 17

18
LM11A-31 achieves high CSF to plasma ratios and exhibits rapid clearance 19 following intravenous or oral administration 20 To determine the distribution and half-life of the drug and screen for possible adverse 21 effects, the drug was administered orally (n=2), intravenously (n=5) or subcutaneously 22 in pluronic gel (n=5). LM11A-31 reached therapeutic concentrations in CSF for all 23 dosing strategies. Drug delivered subcutaneously was retained longer but achieved 7 1 lower CSF penetration with a CSF/plasma ratio of 0.257. Intravenous and oral delivery 2 showed higher CSF penetration (average CSF:plasma ratio of 7.0 for both).
3 Pharmacokinetic (pK) data following oral or intravenous delivery of 10 mg/kg LM11A-31 4 were similar (Fig. 1). Clearance was rapid in both plasma and CSF with t 1/2 estimates of 5 1.3-2.1 hrs. The CSF concentration was maintained above the minimum target 6 therapeutic concentration of 10 nM for approximately 3.7 hrs. Peak drug concentration 7 and T 1/2 estimates were similar to values obtained in mouse studies with a comparable 8 dose, based on body surface area(32). No significant adverse effects were noted in 9 response to the drug, although a transient decrease in heart rate was noted in some 10 cats. Given these results and to match the proposed route of delivery in humans, we 11 elected to dose the cats orally. 12 13 Effect of LM11A-31 on plasma and CSF viral loads 14 Plasma FIV copies following intracranial inoculation were assessed for each cat (Fig.   15 2A). Samples collected just prior to inoculation were all negative for FIV. Two weeks 16 after intracerebroventricular (icv) delivery of FIV, plasma FIV RNA rose to an average of 17 log 5.86 + 0.21 copies/ml in the group that subsequently received the placebo and log 18 5.18 + 0.32 copies/ml in the group randomly selected to receive the drug. We observed 19 a gradual decline in FIV viral loads over the next 18 months in both groups with no 20 differences between groups, which is consistent with viral kinetics noted in our previous 21 studies(45). Final FIV titers were measured after the completion of testing at 22 approximately 20 months, while the cats were still being administered LMA11A-31. 23 When FIV titers at the end of the 10 week treatment were compared to pretreatment 8 1 levels, we found that plasma FIV rose slightly in the placebo treated cats and remained 2 stable in the LM11A-31 treated cats; however, the changes were not significant.  Body Weights and body conditions cores were unaffected by LM11A-31 treatment 18 Body weight and body condition scores were monitored throughout the study to assess 19 any adverse effects of infection or treatment. The data shown in Figure 3, illustrate that, 20 with the exception of the expected acute drop in body weight immediately following 21 infection (7-8%) and again at approximately 29-33 weeks (3-4%), all cats showed an 22 overall body weight gain at the same rate. Treatment with LM11A-31 had no effect on 23 body weight. Body condition scores paralleled the changes in body weight, again with 9 1 no effect of treatment indicating that the cats were healthy and eating well throughout 2 the experiment. The CD4+:CD8+ ratio improved after treatment 5 As expected, a decline in the CD4 + :CD8 + T cell ratio was seen over time (Fig. 4A). The 6 drop was similar in both FIV infected groups and was significant relative to sham 7 inoculated controls. Treatment with LM11A-31 (20 month data) caused a small, but 8 significant increase in the CD4:CD8 ratio (Fig. 4B) whereas both the placebo and sham 9 groups showed small decreases, suggesting that LM11A-31 may have a favorable 10 effect upon T cell population dynamics. These changes were too small to interpret the 11 physiological relevance, but additional studies should 12 determine if this trend continues with a longer treatment duration. 13 14 No adverse effects were observed in response to drug administration or FIV 15 infection 16 Before initiating the behavior studies reported here, safety studies were performed in a 17 subset of cats. Cats were administered the drug orally (n=4, LM11A -31 10 mg free 18 base/kg) and assessed before and 48 hours after dosing; intravenously (n=2, LM11A 10 19 mg free base/kg) and assessed before and 24 hours after dosing; and subcutaneously 20 (n=2, LM11A 10mg free base/kg) and assessed before and 72 hours after dosing. In 21 these safety studies, a decreased hematocrit was noted in 2 of 4 cats at 48 hours 22 following oral dosing, in 1 of 2 cats at 72 hours following intravenous dosing and in 2 of 23 2 cats following subcutaneous dosing. However, the effect was transient. The cats in 10 1 the current longitudinal study reported here exhibited only sporadic, mild decreases in 2 hematocrit over the entire study period (both control and treated groups). Further, no 3 other significant laboratory changes were noted on the CBC, serum biochemistry profile 4 or urinalysis throughout the entire length of the study. 5 6 Daily activity was unaffected by treatment with LM11A-31 7 A subset of sham inoculated (n=5) and FIV infected (n=6) were monitored for changes 8 in activity using Actical monitors over a period of 24 days. No other testing was 9 administered during this period. On average, FIV infected cats at 18 months post-10 inoculation showed less total daily activity than matched sham controls (Fig. 5).

11
LM11A-31 treatment over a 17-day period had no effect on average daily activity. 12 Activity patterns were similar across cats with regard to time. The pattern of activity did   Figure 7A. Uninfected sham control cats generally 11 1 showed stable latencies after the first trial with gradual improvement over sessions. By 2 18 months, the average latency times for the control group were consistently low. In 3 contrast, all FIV infected cats increased reversal and repeat reversal latency times with 4 less stability at 18 months relative to 12 months. Taken together, these data clearly 5 indicated the onset of a decline in performance. FIV cats receiving placebo continued 6 to show decreased performance (increased latencies) at 20 months. In contrast, FIV-7 infected cats treated with LM11A-31 showed response patterns that were similar to 8 controls, with low latency times and decreased variability. The mean change in latency 9 illustrates the improvement in the LM11A-31 treated cats (negative differential latencies) 10 relative to the placebo treated cats (increased latencies, Fig. 7B). We utilized an 11 additional crossover session to determine if the placebo FIV cats would improve 12 performance following treatment with LM11A-31. A similar pattern of improvement was 13 observed in the crossover group when compared to their latencies while under placebo, 14 further strengthening the evidence for a positive drug effect (Fig. 7C). A comparison of 15 the number of errors made at 18 months relative to performance at 12 months (Fig. 7D) 16 showed a small improvement for the LM11A-31 cats (decreased errors) whereas the 17 performance of both the sham and FIV cats declined slightly (increased errors). The 18 trend toward improved performance in the treated cats failed to achieve significance 19 (p=0.08) but was consistent with the improved latencies. When the ability of the cats to 20 learn the discrimination and reversal paradigms was evaluated by comparing trials to 21 reach criterion, no significant differences were seen, indicating that all cats were able to 22 learn the task equally well (Fig 7E). A relationship between T maze performance and FIV titers was seen before but 2 not after treatment 3 To evaluate the relationship between viral titers and T maze performance, we correlated 4 CSF, plasma and PBMC FIV titers with mean reversal latencies. A significant positive 5 correlation was seen between pre-treatment plasma FIV (r=0.519, p=0.027) and 6 latencies (Fig. 8A). A trend was also observed between CSF FIV and latencies    18 To determine if latency increases in the FIV infected cats were associated with neural 19 damage, we evaluated expression of microtubule associated protein-2 (MAP-2). Loss (not shown). We then asked if the MAP-2+ hippocampal pyramidal cell density 5 correlated with improvements in performance rather than absolute latencies (i.e. a 6 decrease in average T maze latency) from 12 to 18 months. A significant negative 7 correlation was seen when pyramidal cell density in hippocampal regions CA1 (Fig. 9A) 8 and CA3 (Fig. 9B) was compared to the change in reversal latencies, suggesting that 9 improvements in performance were directly related to hippocampal neuron damage. 10 A similar comparison of Iba-1 immunoreactive microglia to latency change was also 11 performed to examine the relationship between performance changes and inflammatory 12 activity. As shown in Figures 9C and 9D the relationship was the opposite of the 13 predicted outcome. In the hippocampal dendritic field (stratum radiatum and 14 lacunosum), greater numbers of Iba-1 immunoreactive microglia were associated with 15 improved latency scores (negative change scores, r=-0.580, p=0.019). No relationship 16 was seen between the Iba-1 immunoreactive microglia in the pyramidal cell layer 17 (stratum pyramidale) and the latency changes (r=0.354, p=0.215). It is noteworthy that 18 the overall density of Iba-1 immunoreactive microglia in the hippocampus was relatively 19 low as compared to results from more aggressive animal models of disease, indicating 20 that the extent of inflammation was modest. LM11A-31 treated cats sustain their ability to recognize a novel object 14 1 All cats showed strong novel object recognition at all tests. By 12 months 97% of the 2 time spent exploring the two objects in the field was devoted to the novel object with a 3 range of 11.2 -27.7 sec. Exploration of the familiar object was relatively stable with an 4 average range of 0.6-1.6 sec. By 18 months, we recorded a large drop (45%) in time 5 spent exploring the novel object for the placebo FIV infected cats. In addition, 47% of 6 the FIV placebo cats showed a slight increase in the time spent with the familiar object 7 (i.e. less familiar object recognition) which was just short of significance (p=0.086). In   15 placebo. 16 Measures of general activity (total activity, thigmotaxis) and special preference (relative All cats performed equally well on the GTA reaching task 5 All cats performed well on the motor reaching task and efficiently retrieved the reward 6 except at the longest distances. No significant differences between LM11A-31 and 7 placebo groups were observed in the latency and efficiency of reward retrieval. The 8 task appeared to be easy for all cats and any minor differences were most likely due to 9 variability in the natural reaching limits of individual cats. These observations indicated 10 that neither infection nor drug treatment had any effect on basic motor performance for 11 a food reward. 12 13 No significant changes were observed between groups with administration of 14 sensory testing 15 Neurotrophin receptors can play a role in the development of aberrant sensory 16 processing including the facilitation of pain syndromes. Therefore, we asked if LM11A-17 31 administration might affect sensory processing in FIV-infected cats. Cats were 18 tested for sensitivity to both thermal and tactile stimuli. Average thresholds for response 19 to von Frey stimuli and latency for withdrawal in response to a thermal stimulus are 20 summarized in Figure 12. Each group of cats showed similar responses to the von Frey 21 device that were relatively stable over time. Similarly, the average response to the 22 thermal stimulus was the same for all groups with no significant change over time.

23
Since there was considerable individual variation in these measures, we asked if a 16 1 subset of cats might show changes in response to FIV infection and/or drug treatment 2 and found no evidence for changes that might suggest any unique individual sensitivity 3 to the drug. Thus, neither FIV infection nor drug treatment produced any significant 4 changes in sensory perception even when assessed for each cat individually. There are currently few reliable interventions proven to prevent progressive HIV-8 associated neurodegeneration in response to HIV in the nervous system. Studies 9 report that neurotrophin supplementation is likely to offer significant neuroprotection, but changes were just beginning to appear. We found that LM11A-31 had beneficial effects 19 on the behavior of FIV infected cats, in the absence of any significant adverse effects. In 20 addition, it also had modest but beneficial effects on virus production and T cell counts.  The pharmacokinetic profile of the compound in cats, with high CSF penetration and an 3 estimated half-life of 1.3-2.1 hours, was similar to studies in mice that showed rapid and 4 efficient penetration of the brain with a half-life of 3-4 hours in brain tissue(32). In vitro 5 studies have shown that the compound has nanomolar efficacy, which supports neurite  One concern with modulating neurotrophin signaling was the potential to have adverse  The studies were designed to initiate treatment during the development of mild cognitive 16 impairment, in an effort to approximate the current status of HIV infected individuals.

17
Because cats have less robust neuropathology relative to humans, the FIV was directly  T maze deficits were restricted to increased latencies and more variable performance 16 particularly during reversal testing, consistent with impaired information processing.

17
The increase in latencies in the FIV infected cats was less than anticipated, partly due 18 to testing at an early stage of CNS disease and the ability of the cats to run the maze 19 very rapidly, even with obstacles designed to slow progress (high hoops in the  Collectively, these studies provide strong support for broad neurocognitive stabilizing 16 effects of LM11A-31.   The effect was largely due to the stabilization of novel object recognition versus a 12 decline in this measure in the placebo cats. Thus, these results are consistent with an 13 effect that prevents disease progression.

15
The open field was designed to assess general motor activity, exploration and anxiety 16 through the measurement of total movement, activity around the edges of the maze 17 (thigmotaxis), time in the center of the maze and activity concentrated in the vicinity of 18 the door (desire to escape). It has been widely used in rodents to test the effects of 19 anxiolytic compounds(86) and has been successful for efficacy evaluation of some 20 drugs (e.g. benzodiazepines) but less successful for others (e.g. SSRIs). Uninfected     or damage to brain tissue. The cannula was then slowly raised out of the brain and any 2 signs of blood or reflux of CSF noted. Bleeding from brain tissue was not observed in 3 any cat, and reflux of CSF was negligible. The opening in the cranium was sealed with 4 bone wax and the incision was closed with buried absorbable sutures to minimize 5 postsurgical complications from the wound (e.g. scratching). The cats were given a 6 dose of buprenorphine (0.02 mg/kg, sublingually) and closely observed during recovery. 7 The cats were alert and ambulatory within an hour post surgery and minor 8 complications such as mild incisional discharge, were rare. Postmortem analysis of the 9 brain at the end of the study showed no signs of damage in the brain and skull for any 10 of the cats, indicating full recovery from the surgery and no CNS trauma. Uninfected 11 control cats underwent the same surgical procedure without virus inoculation, to control 12 for any effects of the surgical procedure and to insure that investigators were blinded 13 regarding which cats were infected. To verify that the cats appeared healthy and 14 normal, the testers and veterinary technicians, who were blinded to treatment 15 conditions, were surveyed at the end of the study to indicate which cats they thought 16 were FIV infected. The average responses were at the chance level (51% correct for 17 infected vs not infected) and randomly mixed indicating that personnel could not 18 discriminate between the infected and uninfected cats. to match the strategy for human clinical trials. Based upon the mechanism of action, it 6 is likely that the modulation of neurotrophin signaling has effects that greatly outlast the 7 circulating drug; however, the long term stability of these effects is currently unknown. 8 Oral dosing of the cats was accomplished by adding a concentrated solution of the drug 9 or placebo (78-128 µl) to a small "meatball" of canned feline cat food that the cats would 10 readily ingest. Cats were treated in the morning (7:00-8:00 AM) prior to testing and at 11 the end of the day after all testing was completed (5:00-6:00 PM).   Olmsted), we can detect between 50 and 100 copies of FIV provirus using primers FIV-7 7 and FIV-8 and as few as 5 proviral copies using FIV-1 and FIV-2 as nested primers. 8 All oligonucleotide primers were synthesized in-house using an Applied Biosystems 9 DNA Synthesizer, Model 391. 10 For PCR amplification of provirus, PBMCs were isolated, washed 2 times with ice cold 11 PBS, and re-suspended at 1 x 10 5 cells/100 uL in PCR buffer with 600ug/mL proteinase 12 K. The cells were digested at 56 o C for 2 hours, boiled for 10 minutes, and stored at -     Behavioral and physiological assessments 16 T maze design. A feline-adapted T-maze designed by CanCog Technologies 17 constructed of plywood sealed with polyurethane to conform to laboratory standards, 18 was used to provide a simple test of cognitive and motor ability(50). Cats were adapted 19 to the maze as well as transport and handling by the testers prior to the start of 20 testing(51). The T-maze was housed in a behavior testing suite within the same 21 building where the cats were housed. None of the testers participated in invasive 22 procedures such as anesthesia and blood collection, to facilitate positive interactions 23 with the cats. Cats were placed in a start box with a manual sliding door opening to a 31 1 runway containing a series of hoops, which the cats had to jump through. At the end of 2 the runway, the cats had to decide to turn into the left or right reward arm, which then 3 turned again leading back to the start area. Doors with magnetic latches positioned in 4 each reward arm, were remotely closed after the arm choice was made to prevent path 5 reversal. The reward area was hidden from view until the cat was in the reward arm 6 and contained a disposable well with a food reward. At the end of each trial, subjects 7 were able to pass directly back into the start box from either reward area, when a 8 connecting door was raised. Fitted acrylic sheets covered the top of each section of the 9 maze to prevent escape but allowed the animals' behavior to be continuously observed. 10 The tester stood behind the start box, positioned at the middle point, and could visualize 11 the cat but did not provide cues or interact with the subject. A specific computer  were manually controlled by the experimenter. The order of the cats testing was 16 randomized daily and all testers were blinded to both infection status and treatment 17 condition. Inter-tester reliability by the three trained testers was evaluated during the 18 study using video recordings of tester performance. An assessment at the end of testing 19 verified that the testers could not distinguish which cats were infected (a chance level of 20 51% correct was measured in their assessment of infected vs. non-infected). To provide 21 motivation, cats were trained and tested prior to daily feeding from 0800-1100 hours and Reversal 2 (described below). Cats were tested 6 days per week by technicians 5 who were both familiar to and with the cats during the behavioral conditioning period. 6 The maze was cleaned between sessions and left to air-dry. 7 8 Adaptation and reward approach. During adaptation, cats were allowed to explore the 9 runway and both arms of the maze with small food rewards placed throughout the 10 maze. After the initial sessions, the maze doors, including those to the start box, were 11 opened and closed manually by the experimenter to acclimate the cats to the sound and 12 associated air movement. When the subject moved reliably throughout the maze, from 13 the start box to both reward arms over a ten-trial session, treats were restricted to the  Preference testing. Following successful completion of the reward-approach stage, 2 each cat had one day of preference testing to determine its preferred side. This was 3 established empirically as the side that a cat went to ≥6 times during one session of 10 4 trials, when both arms contained rewards. Each cat demonstrated a preferred side, 5 which was used as the first rewarded side in discrimination training. Using each cats 6 preferred side improved performance consistency between cats and allowed us to  the tester started the software timer when the door out of the start box was opened. 12 The timer was stopped after all four feet of the cat crossed a pre-determined point in 13 either reward arm. The recorded values represented the latency to complete the trial.
14 When stopped, the software began a 30-second inter-trial interval, which allowed the To assure consistency in performance while leaving some flexibility for daily variation, 3 cats were tested for a minimum of four days and a criterion of 27/30 correct responses 4 on three consecutive days was used in order to advance to the next phase. This was 5 because pilot studies indicated that an individual cat's performance may vary from 6 session to session. For example, 10/10 correct on one day may be followed by 8/10 7 correct on a subsequent day. After the discrimination training was completed, Reversal  that was a Panasonic DMR-EH75V DVD recorder with audio sourced from a Canon 5 HDV 1080i camcorder. The video was displayed on a small RCA television, which 6 allowed the tester to watch the cat remotely without disturbance. 7 8 Each cat was filmed in the open field for 10 minutes a day for 5 consecutive days. 9 Testing was repeated every 6 months. The setup minimized any sounds from the tester 10 or external sources. Vocalizations, rears, escapes and jumps were recorded during 11 testing. Once the testing was completed, the DVD was transferred to a PC and In session 4, an additional object C scented with a synthetic feline facial pheromone 4 (Feliway, Ceva Animal Health) was used to assess exploration of a novel smell. During 5 testing, the tester waited outside the room to prevent the cat from seeking human 6 interaction. The time sniffing each object over a period of 3 min was recorded with a 7 video camera set up on a tripod. An inter-trial interval of 3 min within the cat carrier was 8 imposed between each session. The video was scored later by a single technician who 9 recorded the number of sniffs per object and time spent sniffing. 10 11 General Test Apparatus (GTA) 12 Initially cats were screened for their ability to perform discrimination learning tasks in a 13 modified general test apparatus (GTA). Tests were conducted using previously 14 established protocols and a technical staff experienced in conducting these tests. The 15 tests included simple motor tasks, object discrimination learning, object reversal and 16 delayed non-matching to position. Although the cats were able to perform the 17 discrimination learning tasks, the reliability of the behavioral performance was too low to 18 be of value to discriminate changes with drug intervention. A simple reaching task to 19 pull in a reward tray initially showed the greatest potential. The cat was required to 20 reach with its paw and pull a sliding coaster towards them to obtain the food reward.

21
Testing was over a period of 3 days. On day 1 the baited coaster was initially set at 22 position "1" and the distance from the cat was subsequently increased each trial by one 23 position (2.5 cm each) until at position "12". The computer was used to time latency to 37 1 reward retrieval. A tester observed the cat to record the paw used to retrieve the 2 reward, and paw changes (using one paw and then the other to retrieve the reward).

3
This test had 12 trials with 30-second inter-trial intervals. On day 2, the distance from 4 the cat was varied randomly between positions 5, 7, 9 and 11 for 12 trials followed by a 5 distance of 12 for 3 trials, which was beyond the reach of most cats. Latency to 6 retrieval, maximum distance achieved, and efficiency of retrieval at each distance was 7 compared across groups. 8 9 Sensory testing 10 Quantitative sensory testing was performed in a dedicated space adjacent to the 11 housing area (5x5 meter isolated room with one door and no windows) at the NCSU 12 College of Veterinary Medicine. Quantitative sensory thresholds were collected at 13 baseline (prior to any procedures) and then every 6 months for the study duration using 14 the electric von Frey and Thermal probe. cell connected to modified 0.5 mm diameter pipette tip, as described previously(52, 53). 19 The amount of force applied to a test site was measured and displayed with a resolution 20 of 0.1g and maximum of 1000g (although the max force to be applied was set a priori at 21 500g). combined water reservoir. The temperature range of the probe tip could be altered to 5 have a temperature between 0 to 50 o C, maintained to within 0.1 o C. This device was 6 used to deliver a hot thermal stimulus (46.5 o C). 7 8 Testing was performed at the dorsal metatarsal region on both hindlimbs. The von Frey 9 device was applied perpendicular to the dorsal surface of the metatarsus, between 10 metatarsal bones III and IV as previously described in dogs(52). The Physitemp probe 11 was applied to the dorsal mid-metatarsal area following removal of the hair over a 2 x 2 12 cm area. 13 The end point for both thermal and mechanical stimuli was defined as a behavioral  Protocol for sensory testing. The ambient testing room temperature was maintained 21 between 20.0-23.0°C. All cats were acclimated to the testing room for 30 minutes each 22 day for 1 week prior to testing each month. On testing days, the cats were allowed to 23 explore the room for 10 minutes prior to testing. Testing was always performed between 39 1 12pm and 2pm by the same individual, and gentle restraint of the cats was provided by 2 an assistant (same assistant at all times). Gentle restraint of the cats consisted of the 3 assistant standing beside the exam table, with one arm under the cat's abdomen while 4 the assistant's other hand was used to hold the cat in position. The von Frey data were 5 collected on one day, and the thermal latency data collected 1-3 days later. Both limbs 6 were tested, and the order was randomized at the start of the study 7 (www.randomizer.org). Three readings (replicates) were collected from each limb with a 8 1-minute interval between replicates. Mechanical stimuli (EVF) were applied as ramped stimuli, with steadily increasing force 11 applied until the behavioral response was elicited, or the maximum value was reached. 12 A maximum value of 500g was set a priori prior to the study. Thresholds were measured 13 in grams. 14 The hot stimuli were applied as fixed intensity stimuli. Hair over the site was clipped 15 creating a 2 x 2 mm area prior to any testing. Hot stimuli represented probe settings of 16 46.5° C (cut-off time 20 s). The latency to respond was measured in seconds for 17 thermal stimuli using a digital stopwatch (Seiko W073, Seiko, Japan) with the precision 18 of one hundredth of a second.  months to 20 months. The primary comparison across all procedures was drug vs 10 placebo by t-test. Sham infected cats were used to provide an index of performance in 11 uninfected cats as a secondary comparison to evaluate the extent of protection. If the 12 data were not normally distributed, nonparametric analysis was applied. In three cases, 13 cats were not included in the analysis as they failed to perform a task reliably, showed a 14 behavior that was greater than 3 standard deviations from the mean of the other cats in 15 the same group, or were not tested due to a medical condition that interfered with 16 testing (such as gastrointestinal discomfort associated with hairballs and a paw injury).

17
Specific comparisons are provided for each analysis.   The compound rapidly reached peak concentrations of 140-977 nM with higher 7 concentrations in CSF than plasma. A similar rate of decline was seen in both  12 A. FIV RNA in plasma peaked at 5.2-5.8 log copies/ml and decreased to 2.98-3.77 log 13 copies/ml by 18 months with no significant differences between groups. B. FIV RNA in 14 CSF peaked at 4.8-5.8 log copies/ml and dropped to 2.6-2.8 log copies/ml by 18 15 months. A significant decrease to 2.1 log copies/ml in the treated group (FIV+LM11A-16 31) was seen relative to an increase to 3.4 log copies/ml in the vehicle group (FIV) 17 (unpaired T-test of pre to post differential, treated vs. untreated, p=0.032). C. FIV DNA 18 in PBMCs peaked at 3.7-4.5 log copies/ml and thereafter remained relatively stable at 19 3.4-3.9 log copies/10 6 PBMCs. No changes were seen after treatment with LM11A-31.  A. A small decrease in body weight was seen following infection and again for the 23 infected cats at 29-33 weeks. By 38 weeks, rate of gain was restored and by 50 weeks 43 1 body weights were the same for all groups. Treatment with LM11A-31 (bar) had no 2 effect on body weight. B. Body condition scores paralleled the body weight data, again 3 with no effect of treatment.  6 A. As expected, we observed a large drop in the CD4:CD8 ratio in the FIV infected cats 7 relative to sham inoculated controls. B. After treatment with LM11A-31, we noted a 8 small increase in the CD4:CD8 ratio relative to placebo treated cats (*unpaired t-test of 9 changes in placebo vs. treated cats, p=0.016).   19 No significant changes in activity patterns were detected. During treatment with LM11A-31 (yellow circles, lower), latencies decreased and 2 became much more stable, resembling sham inoculated control cats (blue circles, top).

3
High first trial latencies for many sessions of the sham control cats were due to one cat 4 that often showed a high latency on the first trial only. B. Analysis of the change in 5 mean latencies after LM11A-31 treatment versus before treatment illustrating the 6 increase in reversal latencies seen in the FIV infected cats in contrast to the decrease in 7 reversal latencies in the cats treated with LM11A-31. C. Crossover of placebo cats to 8 treatment with LM11A-31 also showed an improvement in latencies (matched t-test, 9 pre-treatment vs. post-treatment, p=0.007). D. A slight reduction in the average 10 number of maze errors per session was seen in the LM11A-31 treated cats relative to 11 placebo controls but just failed to reach significance (p=0.08). E. No changes were 12 seen in the number of trials required to reach criterion.  16 A. A significant positive correlation was seen between average latencies and 17 pretreatment plasma FIV (r=0.519, p=0.027). B. A similar trend was noted between 18 pretreatment CSF FIV and latencies but did not reach significance (r=0.421, p=0.082). 19 C. Pretreatment PBMC FIV DNA also correlated with latencies (r=0.640, p=0.004).