, 2008) and attention (Buschman and

Miller, 2009). This synchrony-based linking of neurons into ensembles could be an ideal mechanism for cognitive flexibility, allowing ensembles Regorafenib in vivo of task-relevant neurons to be dynamically formed and reformed (Sejnowski and Paulsen, 2006; Womelsdorf et al., 2007). Our results are consistent with recent evidence from humans and monkeys suggesting that beta oscillations play a major role in top-down organization of neural processing (Engel and Fries, 2010; Oswal et al., 2012). There is enhancement of beta oscillations in human sensorimotor cortices when maintaining posture (Gilbertson et al., 2005; Androulidakis et al., 2007) and when competing movements need to be inhibited (Pfurtscheller, 1981; Swann et al., 2009). Beta synchronization between frontal and parietal cortices increases during top-down attention (Gross et al., 2006; Buschman and Miller, 2007, 2009) and with increased working memory load

(Babiloni et al., 2004; Axmacher et al., 2008). Further, beta synchronization Selleck Fulvestrant increases in anticipation of an upcoming stimulus and is stronger when a stimulus is more predictable (Liang et al., 2002; Gross et al., 2006; Zhang et al., 2008). Similarly, we observed that rule-selective beta synchronization in anticipation of the test stimulus was correlated with the animal’s reaction time. Orientation seemed to be the dominant modality. This may be due to its relative saliency, much like word naming in the Stroop test (MacLeod, 1991). We found the orientation ensemble, which was synchronized at beta-band frequencies during the orientation rule, had increased alpha-band synchrony when color was

relevant. Recent studies L-NAME HCl in humans have suggested a role for alpha oscillations in working memory (Jensen et al., 2002; Freunberger et al., 2008; Palva et al., 2011) and visual attention (von Stein et al., 2000; Sauseng et al., 2005; Sadaghiani et al., 2010). In particular, alpha oscillations during attention are suppressed in the task-relevant sensorimotor cortices, enhanced in the task-irrelevant cortices, and can influence discriminability of stimuli (Worden et al., 2000; Gould et al., 2011; Haegens et al., 2011a). Because of this, it has been suggested that enhanced alpha synchronization creates an inhibition of irrelevant processes (Klimesch et al., 2007; Mathewson et al., 2011). Our study is consistent with this model: alpha synchronization may allow the weaker color ensemble to be activated over the stronger (orientation) ensemble when color is relevant. In support, we observed an increase in the activity of color-selective neurons after an increase in alpha in the orientation ensemble. These results suggest a dual model of competition between ensembles of neurons: beta synchrony selects the relevant ensemble, while alpha may deselect the irrelevant, but dominant, ensemble so that a weaker, relevant one can be established.

, 1998a, Dehaene et al., 2003b, Dehaene et al., 2006 and Dehaene and Naccache, 2001). Our proposal is that a subset of cortical pyramidal cells with long-range excitatory axons, particularly dense in prefrontal, cingulate, and parietal regions, together with the relevant thalamocortical loops, form a horizontal “neuronal workspace” interconnecting the multiple specialized, automatic, and nonconscious processors

VE-821 ic50 (Figure 6). A conscious content is assumed to be encoded by the sustained activity of a fraction of GNW neurons, the rest being inhibited. Through their numerous reciprocal connections, GNW neurons amplify and maintain a specific neural representation. The long-distance axons of GNW neurons then broadcast it to many other processors brain-wide. Global broadcasting allows information to be more efficiently processed (because it is no

longer confined to a subset of nonconscious circuits but can be flexibly PD-0332991 in vivo shared by many cortical processors) and to be verbally reported (because these processors include those involved in formulating verbal messages). Nonconscious stimuli can be quickly and efficiently processed along automatized or preinstructed processing routes before quickly decaying within a few seconds. By contrast, conscious stimuli would be distinguished by their lack of “encapsulation” Bumetanide in specialized processes and their flexible circulation to various processes of verbal report, evaluation, memory, planning, and intentional action, many seconds after

their disappearance (Baars, 1989 and Dehaene and Naccache, 2001). Dehaene and Naccache (2001) postulate that “this global availability of information (…) is what we subjectively experience as a conscious state. The GNW has been implemented as explicit computer simulations of neural networks (Dehaene and Changeux, 2005, Dehaene et al., 1998a and Dehaene et al., 2003b; see also Zylberberg et al., 2009). These simulations incorporate spiking neurons and synapses with detailed membrane, ion channel, and receptor properties, organized into distinct cortical supragranular, granular, infragranular, and thalamic sectors with reasonable connectivity and temporal delays. Although the full GNW architecture was not simulated, four areas were selected and hierarchically interconnected (Figure 7). Bottom-up feed-forward connections linked each area to the next, while long-distance top-down connections projected to all preceding areas. Moreover, in a simplifying assumption, bottom-up connections impinged on glutamate AMPA receptors while the top-down ones, which are slower, more numerous, and more diffuse, primarily involved glutamate NMDA receptors (the plausibility of this hypothesis is discussed further below).

Common DISC1 variants affecting baseline Wnt signaling may therefore predispose individuals to reduced Wnt signaling as a sort of “first hit,” while the presence of other concomitant risk alleles that also reduce Wnt signaling would serve as a the “second hit,” and would be sufficient for the onset of psychiatric disease. Recently, Kleinman and colleagues demonstrated that individuals carrying the L607F variant had increased DISC1 transcripts lacking exons 3 and/or 7/8, which are very close to or within the GSK3β binding domain

(Nakata et al., 2009). Therefore, it is possible that common DISC1 variants not only reduce Wnt signaling through Selleckchem GW3965 the decreased inhibition of GSK3β, but also through decreased overall levels of DISC1 transcripts. However, although the L607F DISC1 variant demonstrated reduced binding SB431542 ic50 to GSK3β and caused decreased Wnt signaling and progenitor proliferation, this SNP does not lie directly in the mapped GSK3β binding region. One explanation may be that the L607F

phenotype results from a tertiary change in DISC1 conformation that reduces the efficiency of GSK3β binding. Our data suggest that the DISC1 SNPs are divided into two categories: those that affect Wnt signaling (A83V, R264Q, and L607F) and the S704C variant, which does not affect the Wnt pathway. A recent study implicating the R264Q variant in schizophrenia suggests that treatment-resistant schizophrenia, whereby 17-DMAG (Alvespimycin) HCl patients

do not respond to typical antipsychotic medication, are likely to carry the minor allele of the R264Q variant (Mouaffak et al., 2010). This is very interesting, as these findings suggest that this SNP may regulate the efficacy of antipsychotic treatment. One explanation for this finding is that, since antipsychotics partially activate GSK3β and Wnt signaling (Freyberg et al., 2010 and Sutton et al., 2007), in patients with the minor 264Q DISC1 allele, the DISC1-mediated inhibition of GSK3β may be faulty, thus reducing Wnt activation in response to antipsychotic treatment. This concept is supported by mice possessing a mutation in DISC1 exon 2 (L100P) that display reduced neurogenesis and aberrant cortical development, reduced dendrite branching and schizophrenia behavioral phenotypes (Clapcote et al., 2007, Lee et al., 2011 and Lipina et al., 2010a). Interestingly, DISC1 (L100P) in these mice display reduced binding to GSK3β. These findings support the conclusion that mutations in DISC1 that disrupt binding to GSK3β inhibit Wnt signaling, leading to impaired cortical development and psychiatric behavioral phenotypes. However, we cannot exclude the possibility that these DISC1 variants, in particular R264Q, may impact binding to other proteins that may also be required in regulating DISC1 in Wnt signaling.

, 2000 and Parush et al., 2011) of the BG activity. Therefore, these models predict inferior information processing of the BG network upon the emergence of synchronized activity that disrupts these decorrelations. Furthermore, large-scale synchronization of cortical activity could serve as the basis for akinesia (Brown, 2006). Since synchronization and oscillations tend to coincide, manipulations

affecting one can affect the other and therefore 17-AAG the closed-loop stimulation in this study could disrupt synchrony as well. However, previous studies have demonstrated that oscillations and synchrony can exist independently (Heimer et al., 2006). Since theoretical studies have demonstrated the plausibility of closed-loop systems AZD9291 targeted at synchronization of activity (Popovych et al., 2005 and Tass, 2003), further experimental studies are needed. The closed-loop approach suggested in this study may not be limited to PD. Work done on animal models of several neurological

and psychiatric disorders indicate that recognizable pathological patterns emerge (Uhlhaas and Singer, 2006). Some bear marked resemblance to the patterns seen in PD; namely, synchrony and oscillatory activity are seen in schizophrenia, a highly prevalent and extremely debilitating psychiatric disorder (Uhlhaas and Singer, 2010). Attempts at using closed-loop approaches for the treatment of other brain disorders will first need to be made in animal models, where the study of the MPTP primate model substantially facilitates the investigation of PD (Langston et al., 1984 and Redmond et al., 1985). We did not carry out a comprehensive investigation to determine the optimal parameters for closing the DBS loop. The aggravation of akinesia during the closed-loop GPtrain|GP stimulus application (with 80 ms delay) may be due to the positive feedback to the ongoing oscillatory activity in the GPi, and further manipulation of the stimulus delay might identify the working regimens for a GPi based feedback paradigm. Using the same location for both reference and stimulation would no doubt reduce the surgical complexity (Rouse over et al., 2011). Moreover, since the neuronal oscillatory activity demonstrated in PD patients includes

higher frequencies (beta band, approximately 15–35 Hz) than those observed in MPTP-treated primates, a delay that will best fit these frequencies should be chosen when attempting closed-loop stimulation in human PD patients (de Solages et al., 2010, Eusebio and Brown, 2009, Hammond et al., 2007, Kühn et al., 2009, Mallet et al., 2008, Weinberger et al., 2009 and Zaidel et al., 2009). Further studies should be performed to ensure the safety and maximal efficacy of different closed-loop parameters in experimental models of PD and human PD patients. These studies should examine the effects of changing the neural location used as the stimulation reference and the stimulated location (e.g., GPi versus STN; Follett et al., 2010 and Moro et al., 2010).

Before the LD cycle shift, PER1 and PER2 levels in the KO mice were not different from those in the WT animals (day 0, Figures 3A and 3C). In contrast,

on day 5 after the 6 hr advancing LD cycle shift, PER1 and PER2 levels were higher in the SCN of the KO mice, suggesting better resynchronization of cellular clocks by day 5 (Figures 3A and 3C). Quantitations of PER levels before and after the light cycle shift are presented in Figures 3B and 3D. One day after the LD cycle shift, PER levels at ZT12 were dramatically decreased in the WT mice. They increased with time and reached preshifted control (day 0) levels 9 days after the light cycle shift (days 1, 3, 5, and 7 versus day 0, p < 0.05;

day 9 Ku 0059436 versus day 0, p > 0.05, ANOVA, Figures 3B and 3D). In the KO mice, PER1/2 at ZT12 decreased to levels similar to those in WT mice following the light cycle shift, indicating a similar degree of desynchronization. Significantly, however, in the SCN of KO mice PER1/2 reached the preshifted levels 5 days after the light cycle shift, ∼40% faster than in the WT mice (days 1 and 3 versus day 0, p < 0.05; days 5, 7, and 9 versus day 0, p > 0.05, ANOVA). Thus, on days 5 and 7 following the light cycle shift, PER levels in the SCN of the KO mice were significantly higher than in the WT mice (days 5 and 7, KO versus WT, p < 0.05, ANOVA, Figures 3B and 3D). Notably, the PER staining data are remarkably find more consistent with the behavioral entrainment data (see Figure 2), showing that the WT mice re-entrained to a shifted light cycle in approximately 9 days, whereas the KO mice reach a new steady phase in approximately 5 days. Taken together, these results support the ADAMTS5 idea

that KO mice re-entrain more quickly because cellular clocks in the SCN of these mice resynchronize faster to the shifted LD cycle. Prolonged exposure to constant light (LL) extends endogenous circadian period and induces arrhythmic behavior in a sizable percentage of animals, depending on the light intensity and animal species (Daan and Pittendrigh, 1976). In the arrhythmic animals, LL disrupts the coupling among individual SCN neurons without affecting intracellular clock function (Ohta et al., 2005). To study the effect of LL on circadian behavior and PER2 expression in the SCN, Eif4ebp1 KO and WT mice were first housed in regular colony cages in LL (200 lx at cage level) for 14 days. Subsequently, the animals were transferred to individual cages equipped with running wheels in LL (55 lx at cage level) and their circadian behavior was recorded for 14 days.

, 2008, Brun et al., 2008 and Mizuseki et al., 2009), which makes it impossible under our conditions to certify if a cell is a grid cell or not. The largest fraction of grid cells were recorded from superficial layers of medial entorhinal cortex (Hafting et al., 2005, Sargolini et al., 2006 and Boccara et al., 2010). In agreement with these studies, we also observed a large fraction of cells with multipeaked firing behavior in layer 2 and 3, a firing pattern that is reminiscent of grid cells tested in linear environments

(Hafting et al., SB203580 2008, Brun et al., 2008 and Mizuseki et al., 2009). The spatial firing across laps was highly stable in some cells (Figure S5A) but less reproducible in other neurons (Figures S5B and S5C).

The complete novelty of the environment might contribute to the instability of spatial firing, as previously suggested (Hafting et al., 2005, Langston et al., 2010 and Wills et al., 2010). Finally, we cannot rule out that external, potentially uncontrolled nonspatial stimuli contributed to the observed spatial modulation because we did not perform spatial manipulations (such as cue-card rotation experiments) that address such possibilities. We describe a system of large patches at the dorsal border of medial entorhinal cortex, which covers the entire mediolateral extent of medial entorhinal cortex and overlaps more medially with the selleck kinase inhibitor parasubiculum (Caballero-Bleda and Witter, 1993). Because of their similarity and continuity with the parasubiculum, we suggest that the large dorsal patches should best be viewed as an extension of this structure. We note, however, that while previous authors described the parasubiculum as a multilayered structure (Witter and Amaral, 2004 and Boccara et al., 2010), we found no evidence that the large dorsal patches were associated in any way with deep cortical layers. To our knowledge, the large dorsal patches have been largely unrecognized previously. Classically, the dorsal border of medial entorhinal cortex had been defined by the sudden increase in layer 2 width, which extends into

layer 1 and forms a “club-like thickening” (Amaral and Witter, 1989 and Insausti et al., 1997). Here, we provide several lines of evidence suggesting that this for dorsal-most structure is organized in patches and contains a distinct neuronal subpopulation from the rest of medial entorhinal cortex. Cells in this structure have unique morphology and connectivity, are strongly theta modulated, show different theta-phase preferences, and are more head-direction selective than superficial layer neurons. In our study we assessed head-direction selectivity in an “O”-shaped linear arena, and we cannot exclude that our directionality measures were influenced by the special geometry of this environment, in particular by the constraints imposed on the rat’s heading direction.

After a delay, the participant was asked whether, for one of the 16 locations, a red dot was presented. From these data, we calculated a K value, reflecting the amount of information that the participant can store in working memory. For details of the task and analysis, see McNab and Klingberg (2008). Participants received TBS over the right dlPFC, left dlPFC, and vertex on three separate occasions, with site order counterbalanced across 24 participants, and the 25th participant received a randomly

selected session order. We identified stimulation sites as follows: the MNI coordinates for the right dlPFC (x = 37, y = 36, z = 34) were taken from a previous study that used a combination of individual anatomy and fMRI results to pinpoint

the dlPFC (Feredoes et al., 2011). For the left dlPFC (x = −37, y = 36, z = 34), we took the negative Imatinib chemical structure of the right dlPFC x-coordinate. These MNI coordinates were transformed to coordinates in native space by taking the inverse normalization parameters from unified segmentation of a previously acquired T1w structural image as implemented in SPM8 (Wellcome Trust Centre for Neuroimaging, UCL, UK). We visually confirmed that the coordinates in native space corresponded to middle frontal gyrus (as in Feredoes et al., 2011). These coordinates were then entered as targets into Visor2 (ANT B.V.), which uses a 3D camera to guide the stimulation coil (Magstim) IPI-145 chemical structure to the target coordinate. The vertex

was set to the Cz of the 10-20 system. To mimic the stimulation experience for the participant, we entered the vertex coordinates into Visor2 and used 3D navigation to target the stimulation coil. We administered stimulation in 5 Hz bursts of three pulses set 20 ms apart, for 40 s, amounting to a total of 600 pulses. Stimulation intensity was set for each individual participant as 90% of active motor threshold (AMT). AMT was defined as the lowest stimulation intensity, expressed as a percentage of max output of the Magstim equipment that reliably (3/5 times) yielded a visible muscle twitch in the hand when stimulating the hand area of the contralateral motor cortex with a single pulse. During this procedure, participants held (lightly) an item in the hand contralateral to the stimulation site. For technical Ribonucleotide reductase and safety reasons, the maximum stimulation intensity was set to 51% of maximum output; as such, any participant with an AMT > 56% received TBS at 51% of maximum output. Note that such reduced stimulation will make it less likely to find significant effects of TBS. The average stimulation intensity was 49% (range: 40%–51%) of maximum output. We analyzed stay-switch behavior on the first choice of each trial to dissociate model-based and model-free control. A model-free reinforcement learning strategy predicts a main effect of reward on stay probability.

Consistent with a role for MD-PFC connectivity in task learning, LFP coherence between these two structures increased with task acquisition in both CNO- and saline injected MDhM4D mice (Figures 6A and S6C). Strikingly, coherence in beta-frequency range increased hand in hand with performance (Figure 6B). In control mice,

increases in coherence and choice accuracy occurred simultaneously at the end of the second session (trials 15–20). In contrast, both coherence and choice accuracy started to increase later in CNO-treated MDhM4D mice (session 5, trials 41–45) (repeated ANOVA followed by Fischer correction, difference versus trials 1-5: °p < 0.1, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). In addition, we Autophagy Compound Library nmr did an analysis of correlation between coherence and performance for each session and each individual animal and found that coherence correlated significantly with behavioral performance for both saline- and CNO-treated MDhM4D mice (Figure 6C). Finally, the parallel increase of coherence and performance was also seen in the theta-frequency range (Figures S6D–S6E). In contrast, while gamma-frequency coherence also increased

between early and late training phases in both saline and CNO-treated MDhM4D mice, these increases did not correlate with increases in choice accuracy (Figures S6B–S6E). These data, together with the phase-locking findings, demonstrate that inhibition of Pfizer Licensed Compound Library MD disrupts both MD-PFC functional connectivity and working memory behavior in parallel, during task acquisition as well as task performance. Imaging studies have repeatedly reported only deficits in MD and PFC activation in schizophrenia patients

performing cognitive tasks. However, whether a decrease in MD activity can cause cognitive deficits is unclear. Moreover, the potential role for MD-PFC dysconnectivity in the generation of cognitive symptoms still remains unexplored. Here, by inducing a subtle decrease in the firing of MD neurons, we triggered selective deficits in two prefrontal-dependent tasks that address reversal learning and working memory, cognitive processes impaired in patients with schizophrenia. We further found that beta range synchrony between the MD and the mPFC is modulated by working memory, and that this modulation is disrupted by the decrease in MD activity. Finally, during DNMS task acquisition, the magnitude of MD-PFC coherence increased in tight correlation with choice accuracy and both increases were delayed by inhibition of MD activity. Together these data demonstrate that decreased MD activity disrupts functional communication within the MD-PFC circuit and causes deficits in prefrontal-dependent cognition. Our findings are consistent with a role for MD-PFC synchrony in working memory tasks and further support the possibility that thalamic deficits can causally contribute to cognitive dysfunction.

Channelrhodopsin-2 (ChR2) was expressed in vM1 neurons, either by injecting AAV-encoding ChR2 focally into vM1 or by driving Cre-dependent ChR2 expression from the EMX1 locus (EMX-Cre:ChR2). To tonically stimulate vM1 neurons, we delivered prolonged (1–5 s) ramps of light at the

vM1 dural surface while recording network activity in S1 ( Figure 2). In waking mice (n = 8 mice total, n = 6 AAV-mediated ChR2 expression, n = 2 EMX-Cre:ChR2 mice, data combined), vM1 stimulation activated S1, causing a significant decrease in S1 delta power and increase in MUA (1–4 Hz power: 58% ± 6% decrease, p < 0.05; MUA: 21% ± 8% increase, p < 0.05; 30–50 Hz power: 31% ± 29% increase, p = 0.4). In a subset of mice (n = 6), electromyogram (EMG) recordings from the contralateral whisker pad enabled us to monitor whisker movements (Figures 2A and 2B). We found that whisker activity was enhanced with vM1 stimulation, Inhibitor Library supplier compared to matched spontaneous

periods (30% ± 9% increase in EMG signals, p < 0.05). To determine the relationships between vM1 stimulation, S1 activity, and whisking, we parsed S1 responses into whisking and nonwhisking trials based on whisker pad EMG signals (Figure 2C). vM1 stimulation caused similar decreases in delta power for whisking and nonwhisking trials (nonwhisking: 54% ± 7% decrease, p < 0.05 compared to spontaneous; whisking: 57% ± 11% decrease, p < 0.05; p = 0.5 comparing whisking and nonwhisking) (Figure 2D), suggesting that vM1 modulation of S1 activity can be dissociated from whisking. However, MUA was significantly larger in whisking than MLN2238 supplier nonwhisking trials (33% ± 9% larger, p < 0.05), suggesting the recruitment of additional S1 inputs during whisking. To whatever eliminate the contribution of behavioral changes to network state, we conducted stimulation experiments in anesthetized mice. These experiments utilized

only focal AAV-mediated ChR2 expression to ensure selective stimulation of vM1 neurons as opposed to fibers of passage, and we confirmed that this approach did not produce retrograde expression of ChR2 in somata of S1 neurons (n = 5 injected mice; Figures S3A–S3C). vM1 stimulation in anesthetized mice dramatically altered S1 network dynamics, abolishing the slow oscillation and activating S1 (n = 43 mice) (Figures 3A and 3B). Varying the intensity of vM1 stimulation caused graded decreases in delta power of the S1 LFP and graded increases in both gamma band power and multiunit spiking (Figures 3C–3E) (comparing control to largest vM1 stimulation: 1–4 Hz power, 60% ± 6% reduction, p < 0.0001; 30–50 Hz power, 94% ± 22% increase, p < 0.001; MUA, 235% ± 52% increase, p < 0.001; n = 9). These measurements of network activity had different sensitivities to vM1 stimulation, with delta power being most sensitive, followed by MUA and then gamma power (Figures S2C and S2D).

The redistribution of complex lipids for membrane repair and other metabolic roles undoubtedly relies on apoE through a process termed secretion-capture (Ji et al., 1994; Mahley and Ji, 1999; Mahley et al., 2009), in which secreted apoE scavenges lipids from the local environment click here and targets them to cells requiring lipids for normal metabolism or membrane repair. The secretion-capture role for apoE was first demonstrated in peripheral nerve injury and regeneration

(Boyles et al., 1989; Ignatius et al., 1987; Mahley, 1988) and later in the CNS following hippocampal injury (Poirier et al., 1991). When the sciatic nerve was injured, macrophages responding to the injury rapidly began secreting very large quantities of apoE (200-fold over the level seen in the uninjured nerve) and “capturing” the lipids in the local environment of the injured nerve. ApoE–lipid complexes were shown to be delivered to the growth cones of the regenerating nerves and to Schwann cells for myelin formation through lipoprotein receptor uptake. The secretion-capture process has been further Compound C datasheet established in the liver, where apoE captures lipoproteins

and targets them for receptor-mediated uptake. In fact, apoE secreted by hepatocytes and macrophages has been shown to bind to cell-surface heparan sulfate proteoglycans where it is available to capture lipids and lipoproteins; the heparan sulfate proteoglycans themselves acting as a receptor or part of a receptor complex (Ji et al., 1994; Mahley and Ji, 1999). Thus, apoE secreted by injured neurons may be serving this critical role in lipid redistribution in the repair process. Alternatively, or in addition, apoE may have a role in cell signaling, 3-mercaptopyruvate sulfurtransferase as has also been

suggested (Hayashi et al., 2007; Herz and Bock, 2002). Although apoE may play an important role in repairing damaged neuronal membranes, it is also associated with neurodegeneration. This assertion is supported by a vast array of structural, molecular, cellular, and behavioral data showing that the three isoforms of apoE display key variations in their protein structure and stability that, in turn, differentially impact neuropathology. The single amino acid interchanges that distinguish the apoE isoforms result in differences in protein stability as well as the propensity to display a unique structural property called domain interaction (Dong et al., 1994; Huang, 2010; Mahley et al., 2006; Zhong and Weisgraber, 2009). ApoE2 has a cysteine residue at position 158 whereas apoE3 and apoE4 each have arginine. While this substitution in apoE2 results in defective lipoprotein-receptor binding and the development of the lipid disorder type III hyperlipoproteinemia (Mahley, 1988; Mahley et al.