INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Myotonic dystrophy (DM) is a multi-system disorder associated with an expanded triplet-repeat in the 3'-untranslated region of the gene for myotonic dystrophy protein kinase (DMPK) (Fu et al. 1992; Harley et al. 1992; Mahadevan et al. 1992). One prominent, disabling symptom of DM is cognitive dysfunction, which includes mental retardation in 60% of patients with neonatal DM (Miller 1992) and a decrease in IQ in those with adult-onset DM (Ashizawa 1998; Censori et al. 1990). Memory loss may be the most vulnerable cognitive function to DM (Rubinsztein et al. 1997). It is not understood why cognitive dysfunction and specifically memory loss occur in DM. One possible explanation would be that DMPK has a role in the cellular basis of cognitive function, perhaps through an effect on the actin cytoskeleton.

Long-term potentiation (LTP) is a use-dependent form of synaptic plasticity that is thought to contribute to the cellular basis of memory storage and other cognitive functions. Changes in the actin cytoskeleton are important for the mechanisms underlying LTP. These include changes in synaptic shape (Desmond and Levy 1988; Kim and Lisman 1999) and the shape of dendritic spines (Engert and Bonhoeffer 1999; Fischer et al. 2000; Toni et al. 1999).

DMPK can modify the actin cytoskeleton. We recently demonstrated that DMPK overexpression in human and rabbit lens epithelial cells induces marked rearrangements of the actin cytoskeleton and plasma membranes (Jin et al. 2000). Also, our biochemical experiments revealed that purified DMPK phosphorylates and inactivates myosin phosphatase (Fig. 6) (Muranyi et al. 2001), thereby increasing the phosphorylation state of myosin and enhancing the assembly and contractility of the actin cytoskeleton. Moreover, both DM and the DMPK null mouse are associated with increases in skeletal muscle basal cytosolic Ca²⁺ concentration (Benders et al. 1996, 1997). Increases in intraneuronal calcium are critical for LTP induction. In addition, at least in muscle, this increased calcium activates calmodulin, which can activate myosin light chain kinase (MLCK; Fig. 6) (Kamm and Stull 2001). MLCK, in turn, can phosphorylate myosin, thereby also increasing its activity. It is not known whether DMPK also has a role in the hippocampal cytoskeletal modulation, though DMPK RNA and protein are present in the hippocampus (Jansen et al. 1996; Whiting et al. 1995), a structure that is critical for memory storage (Scoville and Milner 1957). If DMPK does alter hippocampal cytoskeleton, then decreased DMPK expression could lead to an alteration in synaptic plasticity and possibly impaired cognitive function.

Several other lines of evidence also suggest that DMPK could have a role in synaptic plasticity. DMPK is a member of two networks already implicated in cytoskeletal dynamics and synaptic plasticity. The first network involves cell adhesion molecules such as integrins (Fig. 6: Adhesion Molecules), which are known to be important for short-term memory (Grotewiel et al. 1998). They activate members of the Rho family of small guanine nucleotide-binding proteins (GTPases) (Clark et al. 1998; Schwartz and Shattil 2000). This family includes Rac-1 (Fig. 6), Rho A, and Cdc42. These have been implicated in cell proliferation, differentiation, and extension via effects on the actin cytoskeleton. On the postsynaptic side of neuronal connections, Rac-1 and RhoA specifically regulate dendritic spine and branch numbers, respectively, and their complexity (Li et al. 2000; Nakayama et al. 2000). On the presynaptic side, the Rac GTPases are important for axonal growth, guidance, and branching (Hakeda-Suzuki et al. 2002; Ng et al. 2002). This network must be important for cognitive function since mutations in several members produce mental retardation (Allen et al. 1998; Billuart et al. 1998). We recently showed that Rac-1 GTPase also activates DMPK (Fig. 6) (Shimizu et al. 2000). Since the Rho/Rac-1 network is so important for cytoskeletal dynamics and cognitive function, the finding that they activate DMPK raises the possibility that DMPK is also involved in those functions.

DMPK is also a member of another network that is important for synaptic plasticity. Extracellular ligands (Humoral factors and neurotransmitters; Fig. 6) can activate receptor-tyrosine kinase complexes that activate the small GTPase Ras. Ras can activate Raf-1. This Ras-Raf system is linked downstream to mitogen-activated protein kinase (MAPK; Fig. 6) (Frost et al. 1996), which is required for hippocampal LTP induction (English and Sweatt 1997) and memory (Atkins et al. 1998). We recently demonstrated that Raf-1 also activates DMPK in vitro (Fig. 6) (Shimizu et al. 2000). This again places DMPK in a network that is critical for synaptic plasticity.

DMPK is activated, then, by two systems that are important for actin cytoskeletal dynamics and therefore for synaptic plasticity. Adhesion molecules activate one system while neurotransmitters and humoral factors stimulate the other. DMPK, in turn, can activate cytoskeletal actomyosin. Thus we hypothesized that DMPK could have a role in the cross-talk between the adhesion-dependent and chemically stimulated transduction systems as they impact synaptic function and plasticity (Shimizu et al. 2000). We tested for such a role of DMPK in synaptic function by examining synaptic transmission and plasticity in the DMPK knockout mouse, and we report the loss of a unique phase of synaptic potentiation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

DMPK null mice

Gene knockout models have been very useful for dissecting the mechanisms underlying learning and memory in multiple systems (Steele et al. 1998) including Drosophila (e.g., volado and dunce) and mice [e.g., glutamate receptor subunit 2, the cAMP cascade, Ras guanine nucleotide-releasing factor, and CA1 N-methyl-D-aspartate (NMDA) receptors]. To test the hypothesis that DMPK has a role in synaptic plasticity, we examined hippocampal synaptic plasticity in the DMPK knockout mouse.

Dr. Wieringa's laboratory at the University of Nijmegen (Nijmegen, The Netherlands) developed the DMPK null mice. Jansen et al. (1996) knocked out exons one through eight of the Dmpk gene in a C57BL/6 line of mice. They characterized this mouse line as null for DMPK at both the RNA and the protein levels. DMPK RNA was reduced 50% in heterozygotes and was absent in homozygotes by Northern analysis with Dmpk cDNA. Similarly, Western blots using three different polyclonal antibodies to the DMPK protein demonstrated a reduction of DMPK to undetectable levels in brain.

We crossed the DMPK null mice with normal C57 mice to produce heterozygotes and then mated brother-sister heterozygotes to produce litters with wild-type (+/+), heterozygous (/+), and homozygous null (/) genotypes. Genotypes were determined by preparing DNA from amputated tails, and Southern blotting of BamH I digests. The wild-type and knockout DMPK haplotypes were detected as 4.5 and 6.5 kb BamH I restriction fragments, respectively, using a 500-bp fragment of DMPK cDNA as a probe. There was adherence to the American Physiology Society's policies regarding the use and care of animals.

Hippocampal slice preparation

DMPK protein localizes to the dentate gyrus and the CA1 region of hippocampus (Jansen et al. 1996; Whiting et al. 1995). Hippocampal slices were prepared as follows: 10- to 12-wk-old mice (C57BL/6J, 16-30 g) were anesthetized with a combination of ketamine (42.8 mg/ml), xylazine (8.6 mg/ml), and acepromazine (1.4 mg/ml) at 0.5-0.7 ml/kg ip. When the animal did not respond to tail pinch, a cannula was inserted in their left ventricle, which was perfused with iced saline containing the following (in mM): 125.00 NaCl, 2.50 KCl, 1.25 NaH₂PO₄, 25.00 NaHCO₃, 0.50 CaCl₂, 7.00 MgCl₂, 25.00 dextrose, and 1.00 ascorbate. A cut was made in their right atrium to allow extravasation. When the feet pads and nose were pale, animals were decapitated; their hippocampi were harvested, and 400-µm-thick transverse slices were made using a Vibratome (Technical Products International). Hippocampal slices were cut in iced saline containing the following (in mM): 110 sucrose, 60.00 NaCl, 3.00 KCl, 1.25 NaHPO₄, 28.00 NaHCO₃, 0.50 CaCl₂, 7.00 MgCl₂, 5.00 Dextrose, and 0.60 ascorbate. Added to the cutting saline to block NMDA currents was 100 µM D,L-aminophosphonovaleric acid (APV). A surgical cut between CA3 and CA1 prevented epileptiform activity.

Electrophysiologic recordings

Slices were transferred to a Haas-type (Haas et al. 1979) interface-recording chamber (Medical Systems) at 32.5 ± 0.1°C. The recording saline contained the following (in mM): 124 NaCl, 4.4 KCl, 26 NaHCO₃, 1 NaH₂PO₄, 1.3 MgCl₂, 10 dextrose, 2.5 CaCl₂, and 4 µM picrotoxin. All salines were gassed with 95% O₂-5% CO₂. Microelectrodes were pulled from 1.5 mm OD glass tubing using a Flaming/Brown micropipette puller (Sutter Instrument). Extracellular electrodes were filled with artificial cerebrospinal fluid (ACSF; resistances = 1 to 3 M). Stimulation was for 50 µs at 20-s intervals via bipolar, Teflon-coated platinum electrodes. Extracellular recordings were made in area CA1. This technique is particularly well suited to the multi-hour recordings necessary for the results reported. A single stimulus intensity that initially yielded a 1-mV population excitatory postsynaptic potential (pEPSP) was used throughout each experiment. At baseline, synaptic integrity was tested in several ways. This included the input-output curve, which compares the ratios of stimulus strength to pEPSP slope.

Slices received high-frequency stimulation (HFS) at 100 Hz for 1 s, times four, at 20-s intervals. For the HFS, the stimulus intensity was increased one-third of the way from the baseline intensity to the intensity that yielded the maximal pEPSP slope. Data were acquired and analyzed on intel- (Bifrost, Houston, TX) and Motorola-based (NeXT) computers using the NeXTStep operating system, which was running the DAM program written by Costa Colbert. pEPSP slopes were determined by linear regression over the maximum initial slope points. To avoid bias, we analyzed the same time points throughout the experiment. Standard statistical tests were used as noted for each figure (Zar 1984).

Experimentalists were unaware of animal genotypes and no external features suggested the genotype, thereby keeping experimentalists blinded. Each litter was tested neurophysiologically and then was tested for genotype by polymerase chain reaction analysis of the animal's tails. All animals tested were littermates of the same age and genetic background, except for the control group labeled "wild nonlittermates." Only one slice was used from each animal.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Normal baseline synaptic transmission

To assess synaptic transmission in the DMPK null mice, we used three neurophysiologic measures of synaptic integrity: input-output curves, trace morphology, and paired-pulse facilitation (PPF).

Input-output curves quantify the ratio of stimulus intensity-to-pEPSP slope. This formally tests whether a range of stimulus intensities produce a normal range of pEPSPs. Stimulus intensities and pEPSP slopes were averaged to produce the input-output curves of Fig. 1A, and the curves were similar (n = 6 /, 10 +/, and 17 +/+). Normalizing the curves to the maximum stimulus and pEPSP slope values also produced similar plots for each group (Fig. 1B). The similarity of input-output curves across groups suggests normal synaptic physiology in the / group. The morphology of the pEPSPs in / animals was also normal (Fig. 1C). Increasing stimulus intensity produced a family of normal appearing pEPSPs.

View larger version (20K): [in this window] [in a new window]   Fig. 1. Synaptic function is normal in homozygous myotonic dystrophy (DM) protein kinase (DMPK) knockout mice. A: input-output curves are similar for knockout slices (/), heterozygous slices (/+), and wild slices (+/+). B: normalizing input-output curves to the maximum voltage and slope also indicates similar input-output curves, suggesting normal baseline synaptic function. C: raw traces from a representative / and +/+ slice demonstrates normal population excitatory postsynaptic potential (pEPSP) morphology and normal responses to increasing stimulus intensity.

PPF is an increase in a second pEPSP when it follows shortly after a first. Presynaptic mechanisms contribute to PPF expression (Foster and McNaughton 1991; Katz and Miledi 1968). PPF measurement is useful because changes in presynaptic function lead to changes in PPF (Schulz et al. 1994, 1995, 1997). Conversely, a lack of change in PPF suggests normal presynaptic function. Thus we assessed presynaptic function by testing PPF.

As shown in Fig. 2A, the morphologies of paired-pulses were similar in +/+, +/, and / slices. Formal assessment of PPF, defined as the ratio of the initial slope of the second pEPSP to the initial slope of the first pEPSP, demonstrates that PPF was the same across groups (Fig. 2B). It was 46.9 ± 2.1% [mean ± standard error (SE)] in +/+ nonlittermates (, n = 18), 43.0 ± 4.3% in +/+ littermates (, n = 5), 38.3 ± 2.9% in +/ mice (, n = 10), and 38.3 ± 4.3% in / mice (, n = 7). As shown in Table 1, PPF did not differ significantly between the / mice and either the +/ mice (t = 0.0026 with 15 df, P = 0.94), the +/+ littermates (t = 0.68 with 10 df, P = 0.51), or the +/+ nonlittermates (t = 1.94, 23 df, no significant difference; Zar 1984).

View larger version (16K): [in this window] [in a new window]   Fig. 2. Normal presynaptic function in / mice. A: paired-pulse facilitation (PPF) was tested using a 55-ms interstimulus interval. The morphology of paired pulses is similar across groups of slices. B: quantification of PPF demonstrates that it is the same in +/+ nonlittermates (), +/+ littermates (), /+ mice (), and / mice ().

Baseline synaptic transmission, then, was normal in the / mice as assessed by the three electro-physiologic measures of input-output curves, pEPSP trace morphology, and PPF. The effects of DMPK knockout on synaptic plasticity were then tested.

Comparing LTP across groups

In these example slices, LTP was elicited by first obtaining stable baseline pEPSPs and then giving HFS (Fig. 3A, at the arrows). In the literature, LTP is defined as an increase in the pEPSP at various times after HFS, including 2 min (Isaac et al. 1995; Liao et al. 1995), 30 min (Oliet et al. 1997), and 40-45 min (Misner et al. 2001; Xu et al. 2000). We initially used a later time point, i.e., 60 min after HFS. In this example, HFS to a +/ slice produced >100% LTP at 1 h after HFS. In contrast, HFS to a / slice produced only 35% LTP. To the right, we compare baseline traces to those obtained 20 (top) and 60 (bottom) min after HFS for +/ (left) and / (right) slices (scale bars = 10 ms and 1 mV). While we assessed pEPSP slope in these experiments, it is also obvious that there is a smaller increase in amplitude in the / versus the +/ slice at time "c."

View larger version (24K): [in this window] [in a new window]   Fig. 3. An example of the effect of DMPK knockout on decremental and sustained potentiation. A: after obtaining a stable baseline, high-frequency stimulation (HFS) was given (arrows, 4 times at 100 Hz for 1 s at 20-s intervals). A representative experiment with single +/ and / slices are shown with 1 h of follow-up. To the right are averaged traces at 20 and 60 min after HFS. This experiment suggests that there is less LTP in the / mice. B: follow-up of the same slices for an additional 2 h, however, indicates that sustained potentiation is similar between slices: only the decremental phase of potentiation is reduced in the / slice. Traces to the right indicate that potentiation at 150 min after HFS is equivalent. C: plot of the fiber volley area for the slices of B demonstrates that the fiber volleys are stable during the recordings of B. Traces to the right are the average of all traces before and after HFS from B for the heterozygous slice (top traces) and the homozygous null slice (bottom traces).

Decremental phases of potentiation may accompany the expression of LTP. One form, with a duration <6-15 min, is referred to as short-term potentiation (STP) (Huang and Kandel 1994; Schulz and Fitzgibbons 1997). This potentiation is usually absent after multiple HFS such as those used here (Schulz and Fitzgibbons 1997). Longer duration, decremental phases of LTP are referred to as E-LTP, I-LTP, and protein synthesis-independent potentiation (Abel et al. 1997; Frey et al. 1993; Winder et al. 1998). They have durations of 40 min to several hours. To be certain that we were studying LTP and not a decremental phase of potentiation, we followed slices until the pEPSP slopes were constant for >30 min.

One typically observes the three longer lasting decremental potentiations after HFS in the presence of inhibitors of transcription and translation (Frey and Morris 1997) or after single HFS (Huang et al. 2000), and thus we did not expect to observe long-lasting decremental potentiation in the absence of such inhibitors or after giving multiple HFS. It was surprising to find, then, that decremental potentiation with a duration of 2-5 h always accompanied LTP expression. This produced the unexpected result shown in Fig. 3B. This is a plot of the same slices shown in Fig. 3A with an additional 2 h of recording. Instead of 100 and 35% LTP, it now appears that there was similar LTP in each slice (Fig. 3B). Traces from the +/ and / slice also indicate that LTP magnitudes were similar when measured late after HFS (right). This figure demonstrates that even defining LTP as the potentiation present 60 min after HFS does not rule out the presence of decremental potentiation. It also demonstrates that the majority of potentiation present shortly after HFS may be decremental as opposed to sustained.

The decremental component of LTP could result from dropout of axonal fibers during the experiments. To rule this out, we monitored the fiber volley during each experiment. An example is shown in Fig. 3C. Here, the fiber volleys from the experiments shown in Fig. 3B are plotted over time. The fiber volley was defined as the area under the curve marked by vertical lines in the traces of Fig. 3C (* marks the stimulus blanking). The plots in Fig. 3C indicate stability of the fiber volley for the 3 h of the recordings. The traces to the right are the average of all traces before and after high-frequency stimulation: the upper traces are the heterozygous slice and the lower traces are the DMPK null slice. The overlap of fiber volleys in these traces also indicates that fiber volley did not change during these experiments. Thus changes in fiber volley do not appear to account for the decrease in decremental LTP in the DMPK null slice of Fig. 3B.

Knockout mice have normal LTP, but decreased decremental potentiation

All slices showed long duration, decremental potentiation. When this decremental potentiation fully decayed, some slices showed LTP, but others had returned to baseline. LTP was defined as the potentiation present at the end of each experiment when pEPSP slopes were constant for 30 min. This usually occurred 2.5 h after HFS. We compared the magnitudes of both decremental and sustained potentiation across groups of slices.

Figure 4A compares the time courses of synaptic potentiation for the slices that showed LTP at the late time point after HFS. At 2.5 h after HFS, sustained potentiation was equivalent between homozygous [, 141.8 ± 13.6% (100% = baseline), n = 5] and wild slices (, 153.0 ± 14.7%, n = 7, P > 0.05). This potentiation is smaller than is sometimes reported at earlier times after HFS since we waited until the decremental phases had decayed. The probability of developing LTP was also similar for all three groups (Table 1; 7/16 for +/+, 3/7 for +/, and 5/7 for /).

View larger version (40K): [in this window] [in a new window]   Fig. 4. Normal sustained potentiation, but decreased decremental long-term synaptic potentiation (LTP) in DMPK null slices. A: sustained potentiation was compared between wild () and knockout slices (). Only those slices that showed LTP were included in this plot of means ± SE. At 2.5 h after HFS, sustained potentiation was equivalent between homozygous and wild slices. B: we plotted the decremental potentiation from A on an expanded y-axis in B after subtracting sustained potentiation at 150 min after HFS (see text for details). There was a 64.9% reduction in decremental potentiation measured 5 to 20 min after HFS. C: averaging all +/+ and / slices, including those that showed sustained potentiation after HFS and those that returned to baseline, demonstrates that sustained potentiation was identical in / and +/+ slices. D: expanding the y-axis from C and subtracting sustained potentiation demonstrates a 68.7% reduction in decremental potentiation. E: averaging all +/ and / slices, including those that showed LTP after HFS and those that showed only decremental potentiation, shows that sustained potentiation was identical in the 2 groups. F: there was a significant decrease in decremental potentiation between the 2 groups. The +/+ and +/ slices did not differ in decremental potentiation.

Despite similar LTP amplitudes, decremental potentiation differed significantly between groups. Subtracting sustained potentiation from each slice in Fig. 4A and plotting the resulting decremental potentiation on an expanded y-axis produced Fig. 4B. We subtracted 141.8% from each homozygous point in Fig. 4A and 153.0% from each wild point. There was a reduction in decremental potentiation at 5 to 20 min after HFS from 51.9 ± 18.4% in +/+ slices (, n = 7) to 18.2 ± 7.0% in / slices (, n = 5). This 64.9% reduction was significant (Table 1, P < 0.05).

Figure 4C shows the result of averaging all slices from the wild and knockout groups, including those that eventually showed LTP and those that showed only decremental potentiation (i.e., those that returned to baseline). The average sustained potentiation was identical between these two groups of slices (125.9 ± 14.3% in / slices vs. 123.6 ± 9.9% in +/+ slices, P > 0.05). Subtracting sustained potentiation from the points in Fig. 4C and expanding the y-axis (as was done for Fig. 4B) produced Fig. 4D. This again demonstrates a significant decrease in decremental potentiation from 77.7 ± 11.7% (n = 15) in +/+ slices to 24.3 ± 6.5% in / slices (n = 7, 68.7% reduction, Table 1, P = 0.0003).

Figure 4, E and F, shows the result of averaging all slices from the heterozygous and homozygous null mice, including those that eventually showed sustained potentiation and those that showed only decremental potentiation, i.e., that returned to baseline. The two groups showed identical sustained potentiation (Fig. 4E, 125.9 ± 14.3% in / slices, n = 7, vs. 112.6 ± 9.1% in +/ slices, n = 7, ). There was also identical potentiation among slices selected for showing sustained potentiation (i.e., >20% potentiation at 2.5 h after HFS), which was 141.8 ± 13.6% in / slices () and 134.8 ± 7.3% in +/ slices (, P > 0.05, data not shown). There was a significant reduction, however, in decremental potentiation from 62.6 ± 12.2% in +/ slices (Fig. 4F, n = 7) to 24.3 ± 6.5% in / slices (Table 1, n = 7, 62.1% reduction, P = 0.007).

Finally, we compared the amplitude of decremental potentiation between wild ( in Fig. 4C) and heterozygous slices ( in Fig. 4E). There was 62.6 ± 12.2% in the +/ slices (n = 7) versus 77.7 ± 11.7% in the +/+ slices (n = 15), which is not significantly different (Table 1, P > 0.05).

The data indicate that if one defines LTP as the potentiation present between 2 and 60 min after HFS, then there is more LTP in the wild and heterozygous slices than the knockout slices. If, however, one defines LTP as sustained potentiation, which is probably more relevant to sustained memory, then LTP across groups is equivalent and decremental potentiation is decreased approximately 70% in the / group.

Lack of a relationship between the magnitudes of decremental and sustained potentiation

The data thus far indicate that baseline synaptic transmission and sustained potentiation are similar between the three groups of slices, but decremental potentiation differs. The relationship between the magnitudes of the sustained and decremental phases of LTP across the three groups was explored further. We hypothesized that if decremental and sustained potentiation were expressed through the same mechanisms, then the increased expression of one would decrease the expression of the other. Thus a plot of one versus the other would produce a significant inverse relationship.

Figure 5 shows the results of plotting the magnitude of sustained (x-axis) versus decremental (y-axis) potentiation for each individual slice from each group. Contrary to expectations, there was no significant inverse relationship between sustained and decremental potentiation for any subset of slices. This included the +/+ littermates and nonlittermates (Fig. 5A,  and , n = 16, r² = 0.03, P > 0.05), heterozygotes (Fig. 5B, , n = 7, r² = 0.12, P > 0.05), homozygotes (Fig. 5C, , n = 7, r² = 0.57, P > 0.05), or a combination of all groups (Fig. 5D, n = 30, r² = 0.01, P > 0.05).

View larger version (21K): [in this window] [in a new window]   Fig. 5. Absence of a significant relationship between decremental and sustained LTP magnitudes. There was no relationship between decremental and sustained LTP in +/+ slices (A), +/ slices (B), / slices (C), or a combination of all groups (D). It is also obvious that there was less decremental potentiation (plotted on the y-axis) in the / group (C) vs. the other 2 groups (A, B, and D). This lack of a relationship between the magnitude of decremental and sustained potentiation suggests that differing mechanisms underlie the expression of these 2 phases of LTP.

Inspection of these plots also confirms that there is less decremental potentiation in the / slices: the decremental potentiation plotted on the y-axis in Fig. 6, C and D, (/ slices, ) is much less than the potentiation plotted on the y-axis in 6A (wild), B (+/ slices), or D (all slices).

View larger version (24K): [in this window] [in a new window]   Fig. 6. Proposed model for the involvement of DMPK in decremental synaptic potentiation. Rho guanine nucleotide binding protein family members, including Rac-1, can be activated by mechanically sensitive adhesion molecules (Li et al. 2000; Nakayama et al. 2000). We recently demonstrated that Rac-1 can activate DMPK (Shimizu et al. 2000) and that purified DMPK can phosphorylate and inactivate myosin phosphatase (Muranyi et al. 2001). This reduces its activity, which has the effect of increasing the phosphorylation state of myosin light chains (Myosin) and enhancing the assembly and contractility of the actin cytoskeleton. The myosin phosphorylation state may also be increased by myosin light chain kinase (MLCK), which is activated by Ca2+ and calmodulin. Raf-1 can be activated by Rac-1, humoral factors (Brambilla et al. 1997), and neurotransmitters. It activates a pathway required for LTP which involves mitogen-activated protein kinase (MEK) and mitogen-activated protein kinase (MAPK). (English and Sweatt 1997; Watabe et al. 2000). We recently demonstrated that Raf-1 can also activate DMPK in vitro (Shimizu et al. 2000). DMPK, then, is activated by 2 major stimulus systems, one of which is required for sustained LTP. DMPK null slices, however, showed normal LTP. The selective reduction in decremental LTP in those slices suggests that the putative effectors of DMPK are independent of those involved in sustained potentiation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Baseline synaptic transmission was normal in the DMPK null mouse as assessed by several neurophysiologic measures: input-output curves, voltage trace morphology, and paired-pulse facilitation (Table 1). LTP was also normal 3 h after induction; however, there was a significant decrease in the decremental phase of LTP in the / group (Table 1). The +/ group, however, was identical to +/+ mice. It was also shown that decremental LTP has a duration of several hours and typically accompanies the expression of sustained LTP.

Mechanisms for altered decremental LTP

We initially hypothesized that DMPK would be important for synaptic plasticity because it can alter the actin cytoskeleton and is a member of at least two networks that are important for modulating the actin cytoskeleton and for synaptic plasticity. The specific hypothesis tested was that the DMPK null mouse would show a decrease in sustained potentiation. Contrary to expectations, however, we found that decremental LTP decreased significantly.

How did knockout of the DMPK gene reduce decremental LTP? Cytoskeletal changes contribute to the expression of sustained potentiation (Kim and Lisman 1999) and could conceivably contribute to the expression of decremental potentiation as well. A recent finding of ours may be relevant to the underlying mechanism. DMPK phosphorylates the myosin-binding subunit of myosin phosphatase (Muranyi et al. 2001), which inhibits its dephosphorylation of myosin. The phosphorylation state of myosin affects the assembly and function of the actin cytoskeleton (Schoenwaelder and Burridge 1999). The loss of DMPK in our experiments, then, may have led to a reduction of actin-dependent changes in the cytoskeleton and hence a decrease in decremental LTP. Figure 6 illustrates this model for the involvement of DMPK in decremental synaptic potentiation. Several connections in this cascade have yet to be demonstrated experimentally in hippocampus. Nonetheless, our working hypothesis is that reduced DMPK expression in DM could decrease the phosphorylation state of myosin and hence interfere with any myosin-mediated cytoskeletal contributions to decremental plasticity.

DMPK null mice exhibit several features of DM

Several disease mechanisms are proposed to account for the symptoms of DM, including decreased DMPK production (haploinsufficiency), decreased expression of neighboring genes (Gennarelli et al. 1999; Klesert et al. 2000; Korade-Mirnics et al. 1999), and a toxic gain-of-function due to altered RNA processing (Amack et al. 1999; Mankodi et al. 2000; Miller et al. 2000). The DMPK null mouse is an important test of whether decreased DMPK production may contribute to the symptoms of DM.

Several experimental results suggest that haploinsufficiency can account for some symptoms of DM. First, there is evidence for decreased DMPK expression in DM (Carango et al. 1993; Fu et al. 1993; Hofmann-Radvanyi et al. 1993; Ueda et al. 1999). Second, there is an inverse relationship between DMPK dosage and the cardiac conduction defect frequency in DMPK null mice (Berul et al. 1999). Those arrhythmias also increase with age, mimicking another feature of DM. Third, DMPK null mice have at least some myopathic features of DM (Jansen et al. 1996; Reddy et al. 1996). Fourth, muscle fibers and/or cultured skeletal muscle cells of DM patients exhibit a decreased resting membrane potential and increased basal cytosolic Na⁺ and Ca²⁺ concentration (Benders et al. 1996). Muscle from DMPK null mice also shows higher resting intracellular calcium than wild-type muscle (Benders et al. 1997). Finally, this study demonstrates changes in synaptic plasticity that may be relevant to the cognitive dysfunction associated with DM.

A toxic gain-of-function also may account for some symptoms of DM. Transgenic mice expressing an expanded CUG repeat in muscle, for example, show classical myotonia and the morphology of DM. They do not, however, appear to have weakness of skeletal muscle (Mankodi et al. 2000). Expanded repeats may also not account for mental retardation in DM since the DM-related human disorder DM2 has a repeat expansion in a transcribed but noncoding sequence (Liquori et al. 2001), but it is not associated with mental retardation. Knockout of the neighboring Six5 (DMAHP) gene produces cataracts, a characteristic finding in DM, but no other features of DM are evident (Sarkar et al. 2000; Klesert et al. 2000). Thus each of the three proposed mechanisms for DM may account for some but not all symptoms of DM, thereby supporting the hypothesis that a combination of mechanisms may be necessary to account for all symptoms.

What form of decremental potentiation did DMPK knockout reduce?

The decremental potentiation reduced by DMPK knockout has a time course of 30 min to several hours (Figs. 4 and 5). This time course is similar to that of E-LTP, which has a duration of 40 min (Winder et al. 1998) to 3 h (Frey et al. 1993). There appear to be several differences, however, between E-LTP and the decremental potentiation under study here. E-LTP is typically induced by "a single high-frequency train of stimuli" (Huang et al. 2000) or is observed "in the presence of inhibitors of transcription and translation" (Frey and Morris 1997). The potentiation observed here, in contrast, was induced by four high-frequency trains in the absence of inhibitors.

Longer HFS are also suggested to convert E-LTP to L-LTP (Frey et al. 1993). This hypothesis predicts that there would be an inverse relationship between the magnitudes of E-LTP and L-LTP. This was not the case here, however, where there was no relationship between the magnitudes of decremental and the sustained potentiation (Fig. 5). There appears, then, to be several differences between E-LTP and the decremental potentiation under study here.

Regardless of the type of decremental potentiation under study here, this appears to be the first example of a gene knockout reducing decremental potentiation while leaving sustained potentiation intact. The fact that decremental potentiation could be reduced independently of sustained potentiation suggests that the two potentiations involve independent mechanisms.

Insights into the mechanisms of mental retardation

Our understanding of the physiologic basis for mental retardation is limited. Including the DMPK knockout mouse reported here, there appear to be six mouse models of a disorder associated with mental retardation. Two animal models have normal LTP: our DMPK knockout and fetal alcohol syndrome (Bellinger et al. 1999). Two models show decreased LTP: Down syndrome (Siarey et al. 1997) and Angelman syndrome (Jiang et al. 1998). The Fragile X syndrome mouse model shows an increase in LTP with impaired conditioned fear (Gu et al. 2002), an increase in metabotropic glutamate receptor mediated LTD (Huber et al. 2002), and a transient postnatal increase in spine numbers and length (Nimchinsky et al. 2001; O'Donnell and Warren 2002) [those changes, however, may be more long-lasting (Irwin et al. 2000)]. The DMPK knockout presented here appears to be the first model of cognitive dysfunction in which synaptic physiology and LTP are normal, while another phase of synaptic potentiation is abnormal, i.e., decremental potentiation. It has been difficult to understand how some models of mental retardation could have normal classical measures of synaptic plasticity, such as LTP and LTD. Here we raise the possibility that more than one form of synaptic plasticity exists, and some alterations in cognitive function may be due to impairment of decremental potentiation.

There is limited understanding of the normal functions of DMPK and of the cellular basis of cognitive dysfunction in DM. The demonstration of a change in synaptic plasticity in this animal model is an important step toward elucidating the functions of DMPK and the mechanisms of cognitive dysfunction in DM. Understanding the cellular basis of cognitive dysfunction is particularly important in DM because such symptoms significantly limit the quality of life and life span of persons with DM (de Die-Smulders et al. 1998; Mathieu et al. 1999).