SUMMARY
We studied the spread of excitatory postsynaptic potentials (EPSPs) through electrical synapses in Retzius neurones of the leech Haementeria officinalis. The pair of Retzius neurones in each ganglion is coupled by a non-rectifying electrical synapse. Both neurones displayed synchronous EPSPs of varying amplitudes and rise times. The kinetics of synchronous EPSPs was similar in 79 % of the EPSP pairs. In the remaining 21 %, one EPSP was smaller and slower than the other, suggesting its passive spread from the other neurone. The proportion of these events increased to 75 % in the presence of Mg2+ in the bathing fluid. This spread of EPSPs from one neurone to another was tested by producing artificial EPSPs by current injection into the soma of one Retzius neurone. The artificial EPSPs were smaller and arrived more slowly at the soma of the coupled neurone. The coupling ratios for the EPSPs were proportional to the coupling ratio for long steady-state pulses in different neuronal pairs. Our results showed that EPSPs spread from one Retzius neurone to the other and support the idea that EPSP spread between electrically coupled neurones may contribute to the integration processes of neurones.
Introduction
Electrical activity patterns in neurones arise from the integration of multiple synaptic inputs acting in concert. The original concept of integration predicts that one neurone summates the excitatory and inhibitory influences converging upon it to produce a new output signal (Fatt, 1957; Coombs et al., 1957; Nicholls et al., 1992; Yuste and Tank, 1996). Diverse mechanisms contribute to the integration process in single neurones, including the production of dendritic spikes (Spencer and Kandel, 1961; Andersen et al., 1966; Llinás and Nicholson, 1971; Llinás and Hess, 1976; Llinás and Sujimori, 1980), the activation of voltage-dependent Na+ and Ca2+ currents (Deisz et al., 1991; Hirsch and Gilbert, 1991; Schwindt and Crill, 1995; Stuart and Sakmann, 1995; Haag and Borst, 1996; Lipowsky et al., 1996; Gillessen and Alzheimer, 1997; Seamans et al., 1997; Margulis and Tang, 1998) and non-linear summation of EPSPs (Wessel et al., 1999).
The fact that many central neurones of adult invertebrates and vertebrates are electrically coupled (Bennett et al., 1963; Auerbach and Bennett, 1969; Baker and Llinás, 1971; Nicholls and Purves, 1972; Korn et al., 1973; Llinás et al., 1974; Schmalbruch and Jahnsen, 1981; Marder and Eisen, 1984; Abbott et al., 1991; McMahon, 1994; Hatton and Yang, 1994; Wolszon et al., 1995; Galarreta and Hestrin, 1999; Gibson et al., 1999) makes it possible that the arrival of synaptic potentials from other neurones will affect the integration process. One clear example is the stomatogastric system of crustaceans, where the spread of inhibitory postsynaptic potentials through electrically coupled neurones contributes to the regulation of the timing of the pyloric rhythm (Marder and Eisen, 1984).
In this paper, we studied the spread of synaptic potentials from one neurone to another in pairs of electrically coupled Retzius neurones in the Mexican leech Haementeria officinalis. Each of the 21 segmental ganglia of the leech central nervous system contains a strikingly similar number of neurones distributed in a stereotyped manner (Payton, 1981). Retzius cells are the largest neurones in each ganglion and release most of the serotonin in the animal (Willard, 1981; Henderson, 1983). This modulates behaviour patterns such as swimming (Willard, 1981; Nusbaum and Kristan, 1986; Nusbaum et al., 1987), shortening (Sahley, 1995) and local bending (Kristan, 1982; Lockery and Kristan, 1990). The two Retzius neurones in each segmental ganglion are coupled by a non-rectifying electrical synapse, presumably formed by contacts between the neuropilar branches (Lent, 1973; Smith et al., 1975; Mason and Leake, 1978), and display spontaneous synchronous EPSPs (Hagiwara and Morita, 1962; Eckert, 1963) from which they produce action potentials. In addition, they receive a polysynaptic input from pressure-sensitive neurones (Szczupak and Kristan, 1995). By collecting simultaneous recordings from both Retzius neurones of a pair under different experimental conditions and producing artificial EPSPs by current injection into one of the neurones, we have shown that EPSPs spread from one Retzius neurone to the other. The possible effects of this spread of EPSPs on integration are discussed.
Materials and methods
Preparation and dissection
Experiments were carried out in isolated ganglia from adult leeches Haementeria officinalis at room temperature (20–25°C). Intersegmental ganglia were dissected out and pinned in Sylgard-coated dishes containing leech Ringer composed of (mmol l–1): NaCl, 115; KCl, 4; CaCl2, 1.8; glucose, 11; Tris maleate 10, buffered to pH 7.4. Since the ganglia capsules were extremely hard and resisted cell impalements by microelectrodes, they were opened with forceps and the soma of Retzius neurones were exposed to the bathing fluid. Retzius neurones could be unequivocally identified by their characteristic size and position in the ganglion.
Morphology
The morphology of Retzius neurones and the possible locations of their contacts in the ganglion were studied by double intracellular staining, with Lucifer Yellow injected into one neurone and Texas Red into the other. Neither of these dyes crossed from one neurone to the other (see Fig. 1), as has previously been shown for Lucifer Yellow injections into Retzius neurones (Stewart, 1978). The dyes were pressure-injected into the soma of each neurone, and 30 min later the ganglia were fixed in 4 % (w/v) paraformaldehyde in phosphate-buffered saline (PBS) for 2–12 h before being mounted in methyl salicylate and examined by confocal microscopy (Biorad, Hemel Hempstead, UK). Three-dimensional reconstructions of the branching patterns of single neurones or pairs of Retzius neurones were made from serial images taken every 0.5 μm. Reconstructions were carried out using Confocal Assistant 4.02 (Hemel Hempstead, UK) and Metamorph Imaging System 3.6 software (Universal Imaging, West Chester, PA, USA). Final images were produced using Adobe Photoshop 5.0 software (Adobe Systems, Mountain View, CA, USA).
Intracellular recordings
Microelectrodes for intracellular recordings were made with borosilicate glass of 1 mm exterior diameter and 0.75 mm internal diameter pulled in a P97 puller (Sutter Instruments). The microelectrodes had resistances of 18–25 MΩ when filled with 3 mol l–1 KCl. The presence of inhibitory synaptic potentials was explored using microelectrodes filled with 4 mol l–1 potassium acetate. Since there was no evidence of such activity and recordings were noisier than with KCl, the results presented here were obtained with electrodes filled with KCl.
Separate electrodes were used to inject current and to record voltage. This procedure was preferred over single-electrode current-clamp because, in many cases, fast and large biphasic current pulses were applied to produce artificial synaptic potentials and the speed of the switch mode limited the amount of current injected. A constraint of this technique is the decrease in somatic resistance caused by the penetration of the microelectrodes. Nevertheless, after a few minutes of recording, the membrane time constant and the amplitudes and kinetics of synaptic potentials were similar using one or two electrodes, and recordings could be maintained for several hours, suggesting a recovery of the input resistance. Intracellular recordings of up to 4 h showed no decrease in the resting potential, the somatic coupling ratio or the shape of the synaptic potentials.
Recordings were filtered using a costume-designed Bessel filter with a cut-off frequency of 600 Hz, which did not affect the rise time of the synaptic potentials. Data were acquired by an analog-to-digital board Digidata 1200 (Axon Instruments) at a sampling frequency of 20 kHz, using Axoscope software (Axon Instruments), and stored in a PC. Under these conditions, the recording noise was approximately 0.15 mV peak-to-peak, while the amplitudes of the smaller synaptic potentials were approximately 0.4 mV.
Data analysis
The rise times and the amplitudes of EPSPs were measured manually using Axoscope 8.0 and Clampfit 8.0 software (both from Axon Instruments). To express data, a convention was adopted in which V1 was the voltage recording from the neurone into which current was injected or in which a synaptic potential was produced and V2 was the voltage recording from the follower neurone. The coupling ratio either for steady-state voltages or for EPSPs was defined as the ratio of the amplitude of the follower voltage to that of the driving voltage, V2/V1. In the case of steady-state responses, these values were the voltage changes at the end of long current steps. In recordings of spontaneous activity, the nomenclature was used arbitrarily. Data are expressed as mean values ± the standard error of the mean (s.e.m.). Steady-state coupling ratios do not present variability, since only two or three measurements were made in every pair of neurones. The data presented are from the first test of the recording; one or two subsequent tests were made as controls at variable periods.
Results
Electrical coupling and potential contact sites between Retzius neurones
The characteristics of the steady-state electrical coupling between pairs of Retzius neurones, studied in 16 ganglia by simultaneous voltage recordings following current injection into either neurone, reproduced the earlier observations made in the leech Hirudo medicinalis (Hagiwara and Morita, 1962; Eckert, 1963). The resting potential of Retzius neurones was between –55 and –50 mV, which is close to the threshold for action potentials (–45 mV). Positive or negative currents injected into the soma of either neurone spread equally well in both directions, suggesting no rectification. This was confirmed by the linear relationships between voltage and current amplitudes and between the amplitudes of the voltage changes in both somata (V1 and V2).
The possible sites of electrical contact between Retzius neurones were studied using three-dimensional confocal reconstruction of the arborizations of both neurones after one had been filled with Lucifer Yellow and the other with Texas Red. Presumably, the confocal overlap between the two neurones reflects possible areas of electrical coupling. Individual neurones were also filled with either dye to confirm that this would not spread to the adjacent neurone. Fig. 1A shows a three-dimensional reconstruction of one neurone injected with Lucifer Yellow. Although the soma of the other neurone can be seen, this was the result of intrinsic fluorescence and not of diffusion of dye from one neurone to the other. Similar images were obtained with both dyes, except that the intrinsic fluorescence was more intense in the fluorescein range. The shapes of both neurones were mirror images of each other and, in general, were very similar to the morphology of Retzius neurones in the leech Hirudo medicinalis (Lent, 1973; Smith et al., 1975; Mason and Leake, 1978). A primary process emerged from the cell body and produced four major secondary branches, one to the anterior ganglia, one to the posterior ganglia and the other two travelling to the periphery through the nerve roots. The primary process gave rise to multiple neurites that elongated within the neuropil.
In double-staining experiments, there was extensive overlap in the neuropilar arborizations of pairs of Retzius neurones. Since the somata produced severe optical interference, they were removed mechanically for confocal observations after aldehyde fixation. Fig. 1B shows one example in which the branches in yellow were filled with Lucifer Yellow and the branches in blue were filled with Texas Red. The white dots identify potential contact sites between the two neurones. The number of potential contact sites obtained from 18 different preparations ranged between 90 and 120, all established by fine neurites within the neuropil. Potential contact sites were absent from the somata or large processes of the neurones, suggesting that electrical coupling was restricted to the neuropil.
Synaptic activity of Retzius neurones
Spontaneous pairs of synchronous EPSPs, appearing at an average frequency of 3.36±0.12 s–1 (N=6), were recorded simultaneously from the soma of both Retzius neurones, suggesting a shared common input (Hawigara and Morita, 1962). Fig. 2A shows sequential series of simultaneous recordings from both somata. As can be seen, EPSPs could be abolished in a reversible manner when Ca2+ was substituted by Mg2+ in the external solution (Fig. 2B,C). Synaptic activity was always excitatory, and the characteristics of the EPSPs were similar when the recording electrodes were filled with potassium acetate or with potassium chloride. In addition, synaptic events were not reversed when the resting potential of the neurones was held at voltage levels ranging from –80 to –40 mV by direct current injection into both somata (not shown), excluding the possibility of inhibitory synaptic potentials.
Excitatory postsynaptic potentials varied in their amplitude and kinetics from one event to the next. Fig. 3A shows a typical series of simultaneous recordings from a pair of Retzius neurones. As can be seen, the amplitudes of the synchronous EPSPs varied from one neurone to the other and from one event to the next. The simultaneous recordings at the top of Fig. 3A contain two pairs of synchronous EPSPs with inverse relationships in their amplitudes (marked with the asterisks). While the first EPSP in V2 was approximately twice the amplitude of the corresponding EPSP in V1, the following pair of EPSPs had the opposite relationship. In addition, there were synchronous events in which one EPSP was slower and smaller than that in the other neurone, as indicated by the arrows in Fig. 3A. An interpretation of these observations was that, when asynchronous transmission occurred, the fast EPSPs spread to the other neurone. The fluctuations in the amplitudes and kinetics of the synchronous EPSPs suggested that transmitter release onto each of the neurones was independent and variable, so that when it failed for one of the neurones an asynchronous EPSP was observed in dual recordings. Therefore, upon failures onto either neurone, the asynchronous EPSP could be recorded from both neurones.
In addition to the small EPSPs already described, approximately 15 % of the total number of events had much larger amplitudes. An example of this is shown in the bottom traces of Fig. 3A. Thus, when the distribution of the amplitudes of the EPSPs was plotted, two components were observed, as shown in Fig. 3B. One component had amplitudes ranging from 0.3 to 2.5 mV while the other had amplitudes ranging from 3.0 to 6.0 mV. While the small EPSPs had monotonic rising phases (Fig. 3C), the large EPSPs had complex rising phases, suggesting the summation of several small synaptic potentials (Fig. 3D). For this reason, only the population of low-amplitude events was used to test whether EPSPs spread from one neurone to another in the rest of this study.
Rise time distribution of EPSPs
The variations in the amplitude of the EPSPs and the possible sources for this variability (different quantal contents, different initiation sites and their spread through the gap junction) made it difficult to base our analysis on amplitude measurements. Instead, we used the rise times of EPSPs as the main source of information about their initiation sites and their spread to the contralateral neurone.
The rise time of EPSPs was variable, suggesting that they were produced at different electrotonic distances. Fig. 4A shows four EPSPs with different rise times and amplitudes (a–d), all recorded from the same neurone. The differences in the rise times could be an indication of the electrotonic distance at which they were produced. The rise times of EPSPs analysed in 22 Retzius neurones ranged from 2 to 14 ms and were distributed in two populations when Gaussian functions were fitted to the data. One population had a major peak at 5.6±0.24 ms (N=11) and the other had a smaller peak at 9.6±0.33 ms containing 10–12 % of the events. An example of the rise time distribution of one neurone recorded in saline solution, is shown in Fig. 4B. The arrows indicate statistically significant peaks, the first at 5.97 ms and the second at 8.75 ms. When recordings were made with an external solution containing 1 mmol l–1 Ca2+ and 1 mmol l–1 Mg2+, instead of the usual 1.8 mmol l–1 Ca2+, the proportion of slow rising events increased to nearly 40 %, as shown in Fig. 4C. In this case, the fast rising peak was at 4.45 ms and the more slowly rising peak at 8.14 ms.
These results suggested that there were two major input domains to Retzius neurones. The fast-rising EPSPs were produced electrotonically closer to the recording site than those with the longer rise times. Considering that long-rise-time EPSPs could have arrived from the contralateral neurone, the clear separation between the two peaks suggested that most inputs were located between the gap junction and the soma of each Retzius neurone. A hypothetical circuit consisting of the pair of electrically coupled Retzius neurones, sharing a common synaptic input, is presented in Fig. 4D.
Rise time relationships between synchronous EPSPs
To test whether the slowly rising EPSPs had their origin in the contralateral neurone, the kinetics of synchronous pairs of EPSPs were compared. The rise times of synchronous EPSPs showed two types of relationships. In 79 % of the synchronous pairs, both EPSPs had similar rise times. The two first pairs of synchronous EPSPs (boxed in Fig. 5A and amplified in Fig. 5B) varied in their amplitudes but had similar rise times. In the remaining 21 % of synchronous EPSP pairs, one EPSP had a fast rise time, which fell within the left side of the rise time distribution shown in Fig. 3B, and the other had a slow rise time. In addition, the amplitude of the slow EPSP was always a fraction of that of the fast one. Fig. 5C shows another segment of the recording shown in Fig. 5A. The amplitudes of the two slow EPSPs illustrated were 43 and 50 % of the amplitudes of their respective large EPSPs. A plot correlating the rise times of 60 pairs of subsequent EPSPs recorded in saline solution is presented in Fig. 5D. When 1 mmol l–1 Ca2+ was substituted for 1 mmol l–1 Mg2+ in the external solution, the percentage of pairs with one fast and one slow EPSP increased to 75.39 %. Under these conditions, there was a good correlation between the fast rise times in one neurone and the slow rise times in the other. One example is shown in Fig. 5E and represents 11 pairs of neurones.
Spread of artificial EPSPs
Indirect evidence that the slower and smaller EPSPs had arrived from the contralateral neurone came from the spread of artificial EPSPs produced by injection of brief biphasic current pulses into the soma of one neurone. Fig. 6A shows the protocol for current injection used to produce the artificial EPSPs. Note the scale values compared with the duration of the EPSPs. As shown in Fig. 6B, the shape of artificial EPSPs was similar to that of natural EPSPs recorded at the soma. In seven pairs of Retzius neurones, artificial synaptic potentials arrived at the contralateral soma attenuated and delayed. In the example shown in Fig. 6C, the coupling ratio of the artificial EPSP was 0.31±0.04 (N=10), smaller than the 0.42±0.01 (N=32) coupling ratio of pairs of naturally occurring EPSPs (Fig. 6D), and both were smaller than the 0.59 steady-state coupling ratio in these neurones, as expected for a passive spread of EPSPs through dendrites.
Steady-state and EPSP coupling ratios
To provide further information about the transynaptic flow of EPSPs, we tried to uncouple the electrical synapse pharmacologically. Unfortunately, heptanol, octanol and acetic acid, which uncouple gap junctions in other preparations, had no effect on this synapse at concentrations ranging from 0.1 to 20 mmol l–1. Moreover, serotonin and octopamine, which modulate the electrical coupling in Retzius cells of the leech Hirudo medicinalis (Colombaioni and Brunelli, 1988), also failed in our hands. As an alternative, we compared the coupling ratios of EPSP in pairs of neurones with different steady-state coupling ratios. In the 12 pairs tested, the rise time distributions and the range of amplitudes of EPSPs were very similar. We expect, therefore, that the variations in the steady-state coupling ratios were due to the coupling resistance and not to the dendritic membrane properties.
The results showed that, as expected, in these 12 pairs of Retzius neurones, the coupling ratios of EPSPs were always smaller than the steady-state coupling ratios. For example, in a pair of neurones that had a steady-state coupling ratio of 0.62 in the first 5 min of recording and at the end of the recording, the average coupling ratio of 34 EPSPs at the peak voltage values was 0.45±0.04. In the example shown in Fig. 7A, the steady-state coupling ratio was 0.31, while the coupling ratios of EPSPs were between 0.20 and 0.26 (Fig. 7B). Note that, despite the low coupling ratio, the amplitudes and shapes of EPSPs corresponded to those described above. Mean coupling ratios of 35 EPSPs in 12 different pairs of neurones with coupling ratios from 0.31 to 0.72 are plotted in Fig. 7C. A line of slope 1 is drawn to highlight the fact that the coupling ratios of EPSPs were, in all cases, smaller than those of steady-state pulses.
Discussion
We have shown that electrical coupling between Retzius neurones serves as a bridge for EPSP spread from one neurone to another. The presence of synchronous EPSPs in pairs of Retzius neurones suggests a common and as yet unidentified synaptic input (or inputs) with spontaneous activity. Similar results were obtained in Retzius neurones (Hagiwara and Morita, 1962) and touch-sensitive neurones of the leech Hirudo medicinalis (Baylor and Nicholls, 1969). The amplitudes of synchronous EPSPs varied independently from one neurone to the other and from one event to another, suggesting probabilistic release of transmitter from presynaptic endings. These variations also suggest that asynchronous transmission may result from transmission failures onto either of the neurones. This is supported by the increase in the proportion of asynchronies in the presence of Mg2+ in the recording solution.
The rise time distribution of EPSPs recorded from individual neurones suggests two major input domains. Excitatory postsynaptic potentials with short rise times could, therefore, be produced by inputs between the gap junction and the recording electrode. The equivalent input to the other neurone would produce the long-rise-time EPSPs (see Fig. 4). The overlapping zones of the rise time distribution may reflect inputs closer to the gap junctions. Since the kinetics of these EPSPs would be very similar in both neurones, without pharmacological uncoupling it is extremely difficult to determine on which side of the gap junction they were produced.
Electrical properties and coupling resistance of Retzius neurones
Fundamental factors for the integrative processes are the electrical properties of the neuronal membranes. Contact sites between Retzius neurones in other species are supposed to be established by proximal dendrites rising from the primary segmental axon of the neurones (Lent, 1973; Smith et al., 1975; Mason and Leake, 1978). Our morphological analysis suggests that EPSPs have to spread for distances of between 100 and 200 μm to arrive at the soma. Since action potentials are initiated in the primary process of these neurones (F. F. De Miguel, unpublished observation), the spread of EPSPs to the soma predicts a large space constant. However, the fact that the coupling ratio of natural or artificial EPSPs was smaller than the steady-state coupling ratio indicates that membrane filtering contributes to the EPSP coupling ratio.
Our evidence also supports a contribution of the coupling resistance to the somatic coupling ratio. First, since EPSPs are produced relatively close to the gap junction, their decay to both somata should be similar and, therefore, a large proportion of their coupling ratio must be due to the coupling resistance. Second, the coupling ratios of EPSPs and steady-state pulses of different pairs of neurones with similar electrophysiological properties suggest that these differences arise from the coupling resistance value. This is in agreement with previous evidence showing the combined contribution of cable properties and coupling resistance to the somatic coupling of Retzius neurones (Yang and Chapman, 1983).
Expected effects of electrical coupling to integration
An expected effect of the coupling resistance value is that, when it is high, the dendritic impedance will increase, tending to the formation of a ‘sealed end’ (Rall, 1958), improving current spread towards the soma of the same neurone. In these conditions, each Retzius neurone would integrate almost independently from the other. In contrast, when the coupling resistance value is low, the two dendrites would tend to be in electrical continuity, thus resembling a semi-infinite cable and leading to reduced dendritic impedance but improved trans-junctional current spread. The EPSPs arriving from the other neurone (alien EPSPs) would, therefore, be more influential for integration. A benefit of a low coupling resistance would be that the firing frequency of the neurone would increase as a result of summation of local with alien EPSPs.
An expected consequence of the synchrony of most EPSPs in Retzius neurones is that, if the inputs were distant from each other, part of the current of one EPSP would flow towards the contralateral soma, adding mostly to the decay phase of the other EPSP. This would reduce the membrane filtering effect of the EPSPs, improving their conduction. In contrast, if synchronous EPSPs were produced on either side of the gap junction, the summation would affect the amplitudes and not the wave shape. Therefore, with short distances between the initiation sites of EPSPs, summation of EPSPs would affect the gain, whereas an increase in the distance between the inputs would reduce the effect on the amplitude and would affect the decay phase. Experiments combining artificial synaptic potentials with computational modelling should help to solve this problem.
The different electrotonic distances at which EPSPs are produced and their flow through gap junctions may explain why the amplitude distribution of the population of EPSPs does not reflect a quantal phenomenon (Fig. 3B). The increasing evidence of electrical coupling between mammalian neurones makes it interesting to explore whether, in some of the quantal analysis mismatches in central vertebrate neurones, electrical coupling has obscured the interpretation of results (see Faber et al., 1998).
Possible general significance
An increasing amount of evidence from electrophysiological experiments and from the expression patterns of gap junction proteins indicates that a large proportion of central neurones in embryonic and adult nervous systems of vertebrates, including mammals, display electrical coupling (Bruzzone et al., 1996; Nadarajah et al., 1996; Dermietzel and Spray, 1993). Among the functions of electrical synapses, the mediation of fast behavioural responses in crustaceans (Furshpan and Potter, 1959) and fishes (Lin and Faber, 1988) and also the synchronization of groups of neurones in vertebrates, including mammals (Christie et al., 1989; Valiante et al., 1995; Ishimatsu and Williams, 1996; Mann-Metzer and Yarom, 1999; Galarreta and Hestrin, 1999; Gibson et al., 1999), have been demonstrated. Other possible functions, including transmitter coupling (Vaney et al., 1998) and second-messenger coupling (Dermietzel and Spray, 1993), have also been proposed. Even though electrical coupling in invertebrates is mediated by a different set of proteins (Bacon et al., 1998), the fact that some functions of electrical coupling are conserved between invertebrates and vertebrates makes it very likely that EPSP spread may also be conserved in higher animals.
Acknowledgements
We are greatly indebted to Mr Bruno Mendez for his invaluable technical assistance in our experiments. We also wish to acknowledge the Computing and Microscopy Units at our Institute for their continuous support during this project. We thank Dr Lidia Szczupak for her critical discussion on the data. E.G.P. was supported by CONACYT and DGEP fellowships. M.V.C. and E.G.P. received complementary fellowships from a Human Frontiers Science Program grant (RG-162/98) to F.F.M. Human Frontiers Science Program (RG-162/98), CONACYT (1285-N9204) and PAPIIT (IN-207593) grants to F.F.M. have supported this project.