Visual discrimination learning in the water maze: a novel test for visual acuity

Establishment and visual analysis of CBA/J-Pde6bY347Y/Y347X and C3H/HeJ-Pde6bY347Y/Y347X mice

In CBA/J and C3H/HeJ mice, retinitis pigmentosa is inherited as an autosomal-recessive trait due to a mutation in Pde6b, which encodes cGMP phosphodiesterase subunit b. In these strains, the Y347X mutation in Pde6b leads to the upregulation of cGMP levels, increased Ca2+ influx induces rod death, and the outer segment and rod cells entirely disappeared by 35 days after birth. In the present study, we utilized the clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated (Cas) 9-mediated gene editing to repair the Y347X mutation in CBA/J and C3H/HeJ mice. Evaluation of the established CBA/J-Pde6bY347Y/Y347X and C3H/HeJ-Pde6bY347Y/Y347X mice, which were confirmed to have normal retinal layers by live fundoscopic imaging and histopathological analysis, revealed improved visual acuity based on the visual cliff and light/dark latency tests. Furthermore, our analyses revealed that the visible platform test was a more effective tool for testing visual behavior in these mice. The results suggest that the established strains can serve as control groups for CBA/J and C3H/HeJ in ophthalmology studies involving retinitis pigmentosa.

Retrosplenial inputs drive diverse visual representations in the medial entorhinal cortex

The ability of rodents to use visual cues for successful navigation and goal-directed behavior has been long appreciated, although the neural mechanisms supporting sensory representations in navigational circuits are largely unknown. Navigation is fundamentally dependent on the hippocampus and closely connected entorhinal cortex, whose neurons exhibit characteristic firing patterns corresponding to the animal's location. The medial entorhinal cortex (MEC) receives direct projections from sensory areas in the neocortex, suggesting the ability to encode sensory information. To examine this possibility, we performed high-density recordings of MEC neurons in awake, head-fixed mice presented with simple visual stimuli and assessed the dynamics of sensory-evoked activity. We found a large fraction of neurons exhibited robust responses to visual input that shaped activity relative to ongoing network dynamics. Visually responsive cells could be separated into subgroups based on functional and molecular properties within deep layers of the dorsal MEC, suggesting diverse populations within the MEC contribute to sensory encoding. We then showed that optogenetic suppression of retrosplenial cortex afferents within the MEC strongly reduced visual responses. Overall, our results demonstrate the the MEC can encode simple visual cues in the environment that can contribute to neural representations of location necessary for accurate navigation.

Automatic measurement of mouse visual acuity based on optomotor response: SKY optomotry

In the field of visual science study using rodents, several assessment methods have been developed for measuring visual function. However, methods such as electroretinograms tests, visual evoked potentials tests and maze tests have limitations in that they measure function of only a specific type of cells, are difficult to quantify or require sufficient training time. The method which uses an optokinetic reflex and optomotor response, a compensatory eye and head movement in response to changes in the visual scene, became the most widely used method. However, this method requires highly trained experimenters and is time consuming. We showed that measured visual acuity values are significantly different between beginner and expert. Here we suggest an automated optometry program, 'SKY optomotry', which automatically tracks rodents' optomotor response to overcome subjectivity and the lengthy scoring procedure of the existing method. To evaluate the performance of SKY optomotry using 8-12-week-old C57BL/6 mice we compared the binomial decision of SKY optomotry with a skilled expert, and the area under the curve of SKY optomotry was 0.845. Comparing the final visual acuity, the intraclass correlation coefficient value between SKY optomotry and an expert was 0.860 (95% confidence interval (CI) 0.709-0.928), whereas that between an expert and a beginner was 0.642 (95% CI 0.292-0.811). SKY optomotry showed an excellent level of performance with good inter-rater agreements based on the visual acuity measured by an expert. With the use of our application, researchers will be able to test an experimental animal's eyesight more accurately while saving time on specialized training.

Immunotherapy with Cleavage-Specific 12A12mAb Reduces the Tau Cleavage in Visual Cortex and Improves Visuo-Spatial Recognition Memory in Tg2576 AD Mouse Model

Tau-targeted immunotherapy is a promising approach for treatment of Alzheimer's disease (AD). Beyond cognitive decline, AD features visual deficits consistent with the manifestation of Amyloid β-protein (Aβ) plaques and neurofibrillary tangles (NFT) in the eyes and higher visual centers, both in animal models and affected subjects. We reported that 12A12-a monoclonal cleavage-specific antibody (mAb) which in vivo neutralizes the neurotoxic, N-terminal 20-22 kDa tau fragment(s)-significantly reduces the retinal accumulation in Tg(HuAPP695Swe)2576 mice of both tau and APP/Aβ pathologies correlated with local inflammation and synaptic deterioration. Here, we report the occurrence of N-terminal tau cleavage in the primary visual cortex (V1 area) and the beneficial effect of 12A12mAb treatment on phenotype-associated visuo-spatial deficits in this AD animal model. We found out that non-invasive administration of 12 A12mAb markedly reduced the pathological accumulation of both truncated tau and Aβ in the V1 area, correlated to significant improvement in visual recognition memory performance along with local increase in two direct readouts of cortical synaptic plasticity, including the dendritic spine density and the expression level of activity-regulated cytoskeleton protein Arc/Arg3.1. Translation of these findings to clinical therapeutic interventions could offer an innovative tau-directed opportunity to delay or halt the visual impairments occurring during AD progression.

Significance of visual scene-based learning in the hippocampal systems across mammalian species

The hippocampus and its associated cortical regions in the medial temporal lobe play essential roles when animals form a cognitive map and use it to achieve their goals. As the nature of map-making involves sampling different local views of the environment and putting them together in a spatially cohesive way, visual scenes are essential ingredients in the formative process of cognitive maps. Visual scenes also serve as important cues during information retrieval from the cognitive map. Research in humans has shown that there are regions in the brain that selectively process scenes and that the hippocampus is involved in scene-based memory tasks. The neurophysiological correlates of scene-based information processing in the hippocampus have been reported as "spatial view cells" in nonhuman primates. Like primates, it is widely accepted that rodents also use visual scenes in their background for spatial navigation and other kinds of problems. However, in rodents, it is not until recently that researchers examined the neural correlates of the hippocampus from the perspective of visual scene-based information processing. With the advent of virtual reality (VR) systems, it has been demonstrated that place cells in the hippocampus exhibit remarkably similar firing correlates in the VR environment compared with that of the real-world environment. Despite some limitations, the new trend of studying hippocampal functions in a visually controlled environment has the potential to allow investigation of the input-output relationships of network functions and experimental testing of traditional computational predictions more rigorously by providing well-defined visual stimuli. As scenes are essential for navigation and episodic memory in humans, further investigation of the rodents' hippocampal systems in scene-based tasks will provide a critical functional link across different mammalian species.

Magnesium acetyltaurate prevents retinal damage and visual impairment in rats through suppression of NMDA-induced upregulation of NF-κB, p53 and AP-1 (c-Jun/c-Fos)

Magnesium acetyltaurate (MgAT) has been shown to have a protective effect against N-methyl-D-aspartate (NMDA)-induced retinal cell apoptosis. The current study investigated the involvement of nuclear factor kappa-B (NF-κB), p53 and AP-1 family members (c-Jun/c-Fos) in neuroprotection by MgAT against NMDA-induced retinal damage. In this study, Sprague-Dawley rats were randomized to undergo intravitreal injection of vehicle, NMDA or MgAT as pre-treatment to NMDA. Seven days after injections, retinal ganglion cells survival was detected using retrograde labelling with fluorogold and BRN3A immunostaining. Functional outcome of retinal damage was assessed using electroretinography, and the mechanisms underlying antiapoptotic effect of MgAT were investigated through assessment of retinal gene expression of NF-κB, p53 and AP-1 family members (c-Jun/c-Fos) using reverse transcription-polymerase chain reaction. Retinal phospho-NF-κB, phospho-p53 and AP-1 levels were evaluated using western blot assay. Rat visual functions were evaluated using visual object recognition tests. Both retrograde labelling and BRN3A immunostaining revealed a significant increase in the number of retinal ganglion cells in rats receiving intravitreal injection of MgAT compared with the rats receiving intravitreal injection of NMDA. Electroretinography indicated that pre-treatment with MgAT partially preserved the functional activity of NMDA-exposed retinas. MgAT abolished NMDA-induced increase of retinal phospho-NF-κB, phospho-p53 and AP-1 expression and suppressed NMDA-induced transcriptional activity of NF-κB, p53 and AP-1 family members (c-Jun/c-Fos). Visual object recognition tests showed that MgAT reduced difficulties in recognizing the visual cues (i.e. objects with different shapes) after NMDA exposure, suggesting that visual functions of rats were relatively preserved by pre-treatment with MgAT. In conclusion, pre-treatment with MgAT prevents NMDA induced retinal injury by inhibiting NMDA-induced neuronal apoptosis via downregulation of transcriptional activity of NF-κB, p53 and AP-1-mediated c-Jun/c-Fos. The experiments were approved by the Animal Ethics Committee of Universiti Teknologi MARA (UiTM), Malaysia, UiTM CARE No 118/2015 on December 4, 2015 and UiTM CARE No 220/7/2017 on December 8, 2017 and Ethics Committee of Belgorod State National Research University, Russia, No 02/20 on January 10, 2020.

α-synuclein overexpression in the retina leads to vision impairment and degeneration of dopaminergic amacrine cells

The presence of α-synuclein aggregates in the retina of Parkinson's disease patients has been associated with vision impairment. In this study we sought to determine the effects of α-synuclein overexpression on the survival and function of dopaminergic amacrine cells (DACs) in the retina. Adult mice were intravitreally injected with an adeno-associated viral (AAV) vector to overexpress human wild-type α-synuclein in the inner retina. Before and after systemic injections of levodopa (L-DOPA), retinal responses and visual acuity-driven behavior were measured by electroretinography (ERG) and a water maze task, respectively. Amacrine cells and ganglion cells were counted at different time points after the injection. α-synuclein overexpression led to an early loss of DACs associated with a decrease of light-adapted ERG responses and visual acuity that could be rescued by systemic injections of L-DOPA. The data show that α-synuclein overexpression affects dopamine neurons in the retina. The approach provides a novel accessible method to model the underlying mechanisms implicated in the pathogenesis of synucleinopathies and for testing novel treatments.

An Automated Water Task to Test Visual Discrimination Performance, Adaptive Strategies and Stereotyped Choices in Freely Moving Mice

We describe an automated training/testing system for adult mice that allows reliable quantification of visual discrimination capacities, adaptive swimming strategies, and stereotyped choices with minimal human intervention. The experimental apparatus consists of a hexagonal swimming pool with an internal decision zone leading to three interior arms with two software-controlled platforms inside of each arm. Each experimental trial consists in projecting a "positive" conditioned discriminative stimulus (SD) in one randomly chosen arm, whereas the other two arms project non-reinforced stimuli (the delta stimuli, SΔ). By employing a classical behavioral training schedule, the mice learn to swim toward the arm that displays the SD, because it predicts the presence of two elevated platforms located symmetrically to the left and right side of the projecting monitor. Separate behavioral components for discriminative and stereotyped choice behavior can be identified through this geometric arrangement. In addition, the projection in real-time of either static or dynamic visual stimuli allows the usage of training programs contingent on current behavioral performance. We validated the system by characterizing the visual acuity and contrast sensitivities in a group of trained mice. By employing pharmacological manipulations, we found that the mice required an intact functioning of the primary visual cortex (V1) to solve the hexagonal pool. Overall, the automated training system constitutes a reliable, rapid, and inexpensive method to quantify visual capacities of mice. It can be used to characterize visual and non-visual factors of choice behavior. It can also be combined with manipulations of visual experience and pharmacological micro-infusions to investigate integrated brain function and learning processes in adult mice over consecutive days.

Longitudinal Performance of Senescence Accelerated Mouse Prone-Strain 8 (SAMP8) Mice in an Olfactory-Visual Water Maze Challenge

Morris water maze (MWM) is widely used to assess cognitive deficits in pre-clinical rodent models. Latency time to reach escape platform is frequently reported, but may be confounded by deficits in visual acuity, or differences in locomotor activity. This study compared performance of Senescence Accelerated Mouse Prone-Strain 8 (SAMP8) and control Senescence Accelerated Mouse Resistant-Strain 1 (SAMR1) mice in classical MWM, relative to performance in a newly developed olfactory-visual maze testing protocol. Performance indicated as the escape time to rescue platform for classical MWM testing showed that SAMP8 mice as young as 6 weeks of age did poorly relative to age-matched SAMR1 mice. The olfactory-visual maze challenge described better discriminated SAMP8 vs. SAMR1 mice than classical MWM testing, based on latency time measures. Consideration of the distance traveled rather than latency time in the classical MWM found no treatment effects between SAMP8 and SAMR1 at 40 weeks of age and the olfactory-visual measures of performance confirmed the classical MWM findings. Longitudinal (repeat) assessment of SAMP8 and SAMR1 performance at 6, 20, 30, and 40 weeks of age in the olfactory-visual testing protocol showed no age-associated deficits in SAMP8 mice to the last age end-point indicated. Collectively, the results from this study suggest the olfactory-visual testing protocol may be advantageous compared to classical MWM as it avoids potential confounders of visual impairment in some strains of mice and indeed, may offer insight into cognitive and behavioral deficits that develop with advanced age in the widely used SAMP8 murine model.

The C57BL/6J mice offspring originated from a parental generation exposed to tannery effluents shows object recognition deficits

The main aim of the present paper is to assess whether the parental generation exposure to such discharges could cause object recognition deficits in their offspring. Male and female C57Bl/6J mice were put to mate after they were exposed to 7.5% and 15% tannery effluents or water (control group), for 60 days. The male mice were withdrawn from the boxes after 15 days and the female mice remained exposed to the treatment during the gestation and lactation periods. The offspring were subjected to the object recognition test after weaning in order to assess possible cognition losses. The results of the analysis of the novel object recognition index found in the testing session (performed 1 h after the training session) applied to offspring from different experimental groups appeared to be statistically different. The novel object recognition index of the offspring from female mice exposed to tannery effluents (7.5% and 15% groups) was lower than that of the control group, and it demonstrated object recognition deficit in the studied offspring. The present study is the first to report evidences that parental exposure to effluent of tannery (father and mother) can cause object recognition deficit in the offspring, which is related to problems in the central nervous system.

15th Annual Meeting of the Safety Pharmacology Society: Focus on traditional sensory systems

INTRODUCTION

This report summarizes and comments key talks on the five traditional senses (ear, vestibular system, vision, taste, olfaction, and touch) which were delivered during the 2015 Annual Meeting of the Safety Pharmacology (SP) Society.

AREAS COVERED

The functional observational battery (FOB) can detect major candidate drug liabilities only on ear, touch and vision. Anatomy, physiology, pharmacology, and pathology notions on each sensory system introduce speaker talks. Techniques for evaluating drug effects on hearing functions are reviewed. Nonclinical approaches to assess vestibular toxicity leading to balance deficits are presented. Retinal explants studied with multielectrode arrays allow the identification of drug liability sites on the retina. Routinely performed Safety Pharmacology assays are not powered to address candidate drug-induced disturbances on taste and smell. This weakness needs correction since unintended pharmacological impairment of these sensorial functions may have serious health consequences. Neuropathy produced by chemotherapeutic agents may cause multiple sensorial perception distortions.

CONCLUSIONS

Safety Pharmacology studies should ensure the safety of any candidate drug on the five sensorial systems.

Behavioral phenotypic properties of a natural occurring rat model of congenital stationary night blindness with Cacna1f mutation

Cacna1f gene mutation could lead to incomplete congenital stationary night blindness (iCSNB) disease. The CSNB-like phenotype rat is a spontaneous rat model caused by Cacna1f gene mutation. The present study explored the phenotypic properties of behavior performance in CSNB rats further. The vision-related behaviors of CSNB rats were assessed with a Morris water maze (MWM), passive avoidance tests, and open-field test. Motor ability was evaluated with a rotarod test and a wire hang test, and mechanical pain and thermalgia were used to evaluate sensory system function. Electroretinograms (ERGs) were recorded to evaluate the function of the retina. The vision-related results showed that longer latencies of escape and reduced probe trial in MWM for CSNB rats. There were more errors in avoidance test; CSNB rats were more active in the open field and presented a different pattern of exploration. The locomotor-related behaviors showed shorter falling latencies in the rotarod test and shorter gripping time in CSNB rats. And mechanical thresholds of pain increased in CSNB rats. The ERGs indicated that both the amplitude and latency of rod and cone systems were impaired in the CSNB rats. In summary, Cacna1f gene mutation changed the performance of various behaviors in the CSNB rat aside from vision-related phenotype. Cacna1f gene might play a role in a wide range of responses in the organism. These results confirm the importance of a comprehensive profile for understanding the behavior phenotype of Cacna1f gene mutation in CSNB rat.

The mouse model of Down syndrome Ts65Dn presents visual deficits as assessed by pattern visual evoked potentials

PURPOSE

The Ts65Dn mouse is the most complete widely available animal model of Down syndrome (DS). Quantitative information was generated about visual function in the Ts65Dn mouse by investigating their visual capabilities by means of electroretinography (ERG) and patterned visual evoked potentials (pVEPs).

METHODS

pVEPs were recorded directly from specific regions of the binocular visual cortex of anesthetized mice in response to horizontal sinusoidal gratings of different spatial frequency, contrast, and luminance generated by a specialized video card and presented on a 21-in. computer display suitably linearized by gamma correction.

RESULTS

ERG assessments indicated no significant deficit in retinal physiology in Ts65Dn mice compared with euploid control mice. The Ts65Dn mice were found to exhibit deficits in luminance threshold, spatial resolution, and contrast threshold, compared with the euploid control mice. The behavioral counterparts of these parameters are luminance sensitivity, visual acuity, and the inverse of contrast sensitivity, respectively.

CONCLUSIONS

DS includes various phenotypes associated with the visual system, including deficits in visual acuity, accommodation, and contrast sensitivity. The present study provides electrophysiological evidence of visual deficits in Ts65Dn mice that are similar to those reported in persons with DS. These findings strengthen the role of the Ts65Dn mouse as a model for DS. Also, given the historical assumption of integrity of the visual system in most behavioral assessments of Ts65Dn mice, such as the hidden-platform component of the Morris water maze, the visual deficits described herein may represent a significant confounding factor in the interpretation of results from such experiments.

Classical MHCI molecules regulate retinogeniculate refinement and limit ocular dominance plasticity

Major histocompatibility complex class I (MHCI) genes were discovered unexpectedly in healthy CNS neurons in a screen for genes regulated by neural activity. In mice lacking just 2 of the 50+ MHCI genes H2-K(b) and H2-D(b), ocular dominance (OD) plasticity is enhanced. Mice lacking PirB, an MHCI receptor, have a similar phenotype. H2-K(b) and H2-D(b) are expressed not only in visual cortex, but also in lateral geniculate nucleus (LGN), where protein localization correlates strongly with synaptic markers and complement protein C1q. In K(b)D(b-/-) mice, developmental refinement of retinogeniculate projections is impaired, similar to C1q(-/-) mice. These phenotypes in K(b)D(b-/-) mice are strikingly similar to those in beta2 m(-/-)TAP1(-/-) mice, which lack cell surface expression of all MHCIs, implying that H2-K(b) and H2-D(b) can account for observed changes in synapse plasticity. H2-K(b) and H2-D(b) ligands, signaling via neuronal MHCI receptors, may enable activity-dependent remodeling of brain circuits during developmental critical periods.

The influence of visual ability on learning and memory performance in 13 strains of mice

We calculated visual ability in 13 strains of mice (129SI/Sv1mJ, A/J, AKR/J, BALB/cByJ, C3H/HeJ, C57BL/6J, CAST/EiJ, DBA/2J, FVB/NJ, MOLF/EiJ, SJL/J, SM/J, and SPRET/EiJ) on visual detection, pattern discrimination, and visual acuity and tested these and other mice of the same strains in a behavioral test battery that evaluated visuo-spatial learning and memory, conditioned odor preference, and motor learning. Strain differences in visual acuity accounted for a significant proportion of the variance between strains in measures of learning and memory in the Morris water maze. Strain differences in motor learning performance were not influenced by visual ability. Conditioned odor preference was enhanced in mice with visual defects. These results indicate that visual ability must be accounted for when testing for strain differences in learning and memory in mice because differences in performance in many tasks may be due to visual deficits rather than differences in higher order cognitive functions. These results have significant implications for the search for the neural and genetic basis of learning and memory in mice.

Visual detection, pattern discrimination and visual acuity in 14 strains of mice

Based on the procedure of Prusky et al. (2000, Vision Research, 40, 2201-2209), we used a computer-based, two-alternative swim task to evaluate visual detection, pattern discrimination and visual acuity in 14 strains of mice from priority groups A and B of the JAX phenome project (129S1/SvImJ, A/J, AKR/J, BALB/cByJ, BALB/cJ, C3H/HeJ, C57BL/6J, CAST/Ei, DBA/2J, FVB/NJ, MOLF/Ei, SJL/J, SM/J and SPRET/Ei). Each mouse was tested for eight trials/day for 8 days on each of the three tests. There was a significant strain difference in visual ability in all three tests. Mice with reported normal vision (129S1/SvImJ, C57BL/6J and DBA/2J) and one albino strain (AKR/J) performed very well in these tasks. The other albino strains (A/J, BALB/cByJ and BALB/cJ) took longer to learn the tasks than mice with normal vision and did not reach the criterion of 70% correct. Mice with retinal degeneration (C3H/HeJ, FVB/NJ, MOLF/Ei and SJL/J) performed only at chance levels as did the three strains with unknown visual abilities (CAST/Ei, SM/J and SPRET/Ei). Because many behavioral tasks for rodents rely on visual cues, we suggest that the visual abilities of mice should be evaluated before they are tested in commonly used visuo-spatial learning and memory tasks.

Long-term retention and overshadowing of proximal and distal cues following habituation in an object exploration task

The object displacement task is a popular tool used to investigate spatial learning and memory. However, little attention has previously been given to long-term retention of spatial information following habituation to this task. Furthermore, the role of both proximal and distal cues in this type of passive retention of spatial information is controversial. In Study 1, we examined habituation in the object displacement task across 4 days and examined reactivity to spatial change 7 days post-acquisition. We found that rats habituated rapidly to the environment and retained this environment for the 7 days. Furthermore, this experiment shows that both proximal and distal spatial cues are important in the encoding of the environment during object displacement learning task. In Study 2, we examined the effect of overshadowing and demonstrate that proximal visual spatial cues can overshadow distal spatial cues if a conflict arises between both set.

Definition of a new maze paradigm for the study of spatial behavior in rats

The present work defines a simple behavioral paradigm to evaluate spatial representation in rats. In two experimental conditions differing in the richness of distal visual cues, rats learned to locate a food goal in a cross maze from various starting points. We conducted different challenges consisting of (i) reaching the same goal from a modified arrangement of the maze arms (geometric challenge), (ii) reaching the goal within a 90 degrees rotated maze, herein checking the use of a place strategy, and (iii) investigating the effect of central cholinergic blockade over the retention of the task. Results showed that rats needed 12-30 trials to learn a place response, depending upon the richness of the visual environment. The maze rotation did not produce any impairment whereas the geometric challenge affected the performance specifically under the visually richer environment. Scopolamine injection (i.p.) produced a significant impairment in place recognition. Our present work shows that this maze procedure constitutes a useful paradigm to assess learning and processing of a place representation by rats. Similarly to what has been shown in other popular maze paradigms, our results show that rats mostly rely on distal extra-maze cues to solve the task, but also compute intra-maze information.

Testing the spatial- versus object-learning distinction: water-maze performance of male rats exposed to ethanol during the brain growth spurt

This study investigated the effects of exposure to ethanol during the brain growth spurt on a visual-discrimination (VD) and a place-learning task (PL) using intra-maze cues in the water maze. Artificially reared male Long-Evans rats were exposed to ethanol (ET) in a binge pattern from postnatal days 6-9 (6.5 g kg(-1) x day(-1); BAC approximately 330 mg/dl) or an isocaloric maltose-dextrin solution (gastrostomy control). A third suckled control group was reared by lactating dams. In experiment 1, rats were trained to discriminate horizontal- (H) versus vertical-striped (V) cues, with the positive cue providing escape from water. Groups did not differ with V+, but ET rats made more errors with H+. In experiment 2, the ET group was impaired in learning the spatial location of a submerged platform relative to intra-maze cues. In both tasks, acquisition deficits among ET rats were characterized by impairment emerging at trial 2, with intact reference memory on trial 1, and the ET group reached a comparable level of performance to controls by the end of training. In summary, because impairment was related to task characteristics, a clear distinction between impaired spatial- versus cue-based learning was not supported. However, these findings do support an effect of exposure to ethanol during the brain growth spurt on recent event, but not reference, memory.