Intra- and Extracellular Use and Evaluation of Ammonium-Selective Microelectrodes

One of the latest developments in ion-selective membranes for microelectrodes is an NH₄⁺-selective membrane, which is based on a mixture of the antibiotics nonactin and monactin (Bührer et al. 1988; Ammann, 1986). From calibrations in defined chemical salines it is apparent that potassium interference is considerable, as selectivity coefficients of –0.6 were obtained. This means that about four potassium ions produce the same change in potential as one ammonium ion. So far, this cocktail has not been tested under biological conditions. From the calibration curves it is not clear whether this membrane can be used in invertebrate cells, especially the crayfish stretch receptor neurone (SR) which has a (mean) intracellular K⁺ activity of about 135 mmol I^-1 (Brown et al. 1978; Moser, 1985), corresponding to an intracellular K⁺ concentration [K⁺]_i of about 200 mmol I^-1. However, there are advantages in using this preparation, as there is indirect evidence from intracellular pH (pHi) measurements that entry of NH₄⁺ is large in the SR neurone compared with other cells (Boron and de Weer, 1976; Moody, 1981; Moser, 1985, 1987; Galler and Moser, 1986).

This study is mainly methodological. Its purpose is threefold: (1) to assess the limitations of the NH₄⁺-selective microelectrode (NH₄ISM) during intracellular as well as extracellular use, (2) to demonstrate and to discuss problems specific to the NH₄ISM and (3) to provide some practical hints on its use. We show that NH₄ISMs can be successfully used intracellularly even at a background K⁺ concentration of about 200 mmol l^-1.

Preliminary results were presented as an abstract (Moser et al. 1989) at a satellite symposium of the XXXIst International Physiological Meeting in Helsinki.

A double-barrelled electrode, to measure membrane potential (E_m) and intracellular pH (pHi), and two single-barrelled ammonium electrodes, for extra- and intracellular use, were prepared and connected to the headstages of voltmeters. When all the electrodes had been successfully calibrated (see below), the slowly adapting stretch receptor (SR) of Astacus astacus L. was dissected and mounted in the experimental chamber (Brown et al. 1978; Moser, 1985). A flowthrough system (driven by a peristaltic pump, Pharmacia, model P3) continuously exchanged the saline in the experimental chamber about 12 times per minute.

Chemicals of highest purity were obtained from Merck (Darmstadt, BRD), Sigma (Deisenhofen, BRD) and Fluka (Buchs, Switzerland). The ammonium cocktail was a gift from Professor W. Simon (ETH, Zürich). The pH of all solutions was measured with an Orion 8162 (Ross-type) glass electrode. Two sets of calibrating solutions were used. (1) NAS (normal Astacus saline), containing a low [K⁺] and corresponding to the composition of A. astacus blood, was modified from that devised by van Harreveld (1936). It had the following composition (mmol l^-1): 207 NaCl; 5.4 KC1; 2.4MgCl₂; 13.5 CaCl₂; 10 Hepes adjusted to pH7.4 with NaOH. In NH₄⁺-NAS or K⁺-NAS, NH₄⁺ or K⁺ was substituted for 20mmol l^-1 Na⁺. Small amounts of NH₄C1 or other substances (up to 1 mmol l^-1) were added without compensation. (2) SIS (simulated intracellular saline) consisting of a Hepes-buffered saline (10 mmol l^-1, pH 7.4) with 194 mmol l^-1 KC1 and 21 mmol l^-1 NaCl as background ions. To calibrate the NH₄ISMs, measured amounts of NH₄C1 were added to SIS at pH7.4. The intracellular pH electrode was also calibrated in SIS, using 10 mmol I^-1 Hepes at pH 8.0 and 7.4 and 10 mmol I^-1 Pipes at pH 6.4. The SIS solutions were made alkaline by adding measured amounts of KOH (the amount of K⁺ added being included in the total K⁺ concentration) and then adjusted to the proper pH by adding HC1. The composition of the calibrating solutions is shown in Fig. 1. Additional calibrating solutions are described in Table 1 (see below).

Glass for both single-(GC150) and double-barrelled (TGC 150) electrodes was obtained from Clark (Reading, UK) and pulled vertically (model 700C, Kopf Instruments, USA). Single-barrelled NH₄ISMs were made after the method of Bührer et al. (1988). The pulled pipettes were kept overnight in an oven at 160–180 °C and then a small quantity of silane was added to the glass beaker covering them. The pipettes were baked in the silane vapour for an additional 30 min and allowed to cool. The cold pipettes were back-filled under pressure with filling solution, consisting of 10 mmol I^-1 NH₄C1, before the NH₄⁺ cocktail was sucked in by applying negative pressure, to give cocktail columns of between 100 and 300 μm. After the electrodes had been conditioned (see Results) in a 20 mmol I^-1 NH₄C1 solution to stabilize them, they were ready for calibration. Double-barrelled electrodes were made and silane-coated following the instructions given by Grafe et al. (1985). For pHi measurements we preferred the tri-n-dodecylamine cocktail (Fluka), which proved to be much less susceptible to interference from a variety of substances, such as various ion-transport blockers. The filling solution of the E_m electrode, which also served as an internal reference, was a mixture of 0.6moll^-1 K₂SO₄ (85%) and 1.5moll^-1 KC1 (15%). The resistance of the E_m barrel was 12–30MΩ. The calibration of this electrode is described in the Results section.

The signals from the NH₄ISM were recorded either with an electrometer (WPI, model FD 223) or with a varactor amplifier (Analog Devices 311 J) according to the design of Thomas (1978). The signals from the E_m electrode and from an additional bath electrode were measured with an electrometer (WPI 750) against a grounded calomel electrode (Amagruss, Helmstadt-Bargen, Germany) connected to the bath through a 3 mol I^-1 KCl/agar bridge. All signals were fed into an oscilloscope (Tektronix, model 5113), a personal computer for further calculations and pen-recorders (BBC-Goerz: Servogor 220; SE130), from which the records were photocopied for illustration.

Conditioning time should be kept to a minimum or problems will arise when fine electrodes (0.5–0.8μm tip diameter) penetrate the membrane. The conditioning saline appears to affect the mechanical stability of the electrode tip. We found that conditioning times below 2 h sufficed and we could further improve the number of successful penetrations by using normal-walled GC150 glass (instead of thinwalled GC150T). However, if ‘blunt’ electrode tips were used, the prepared NH₄ISMs could be stored in the conditioning solution for at least 1 week.

Calibration of NH₄ISMs is rather time-consuming, taking at least 40 min (without conditioning). We combined the calibrating salines in such a way as to calibrate both NH₄⁺-sensitive and pH electrodes simultaneously. Fig. 1 shows the second part of such a calibration curve, the first part (not shown) can be thought of as a mirror image on the left side of the picture. In the part shown, the electrodes are usually less ‘jumpy’ and drift is less pronounced and, therefore, we took these values as precalibration curves. Some other features should also be noted. In the case of the pH electrodes, we regularly found (1) that in the presence of NH₄C1 the response shifted in the acid direction and (2) that the calibrations in SIS gave values that were shifted towards the alkaline direction by about 0.05 pH units compared with similar calibrations in NAS. In the case of NH₄ISMs, a change from SIS to NAS was accompanied by a potential change of about 72 mV, mainly due to the change in [K⁺] (from 194 to 5.4 mmol l^-1). Potassium was the dominant interfering ion, the interference from other ions being small by comparison.

Most NH₄ISMS showed a short, transient sensitivity to acidic pH. As two electrodes were calibrated simultaneously we could choose the one that was less sensitive to pH for intracellular measurements. In Fig. 1 this is the NH₄ISM shown in the upper trace. The temporal resolution of NH₄ISMs is about the same as that of the pH electrode. It seems to be limited by the mixing time in the experimental chamber.

Both pre- and postcalibration curves for NH₄ISMs are shown in Fig. 2. Despite a background concentration of 194 mmol I^-1 K⁺, NH₄⁺ could be measured at concentrations as low as 5 mmol I^-1. When electrodes were exposed to SIS and NH₄⁺ concentrations were increased the potentials of the NH₄ISMs also increased. Precalibrations of 130 NH₄ISMs showed that a change from SIS to SIS containing 5 mmol I^-1 NH₄C1 resulted in a potential change of approximately 3–4 mV, whereas the potential change resulting from a transfer from SIS to SIS containing 50 mmol I^-1 NH₄C1 was between 20 and 25 mV. A comparison between pre- and postcalibration curves obtained before and after intracellular measurements lasting 4–11 h revealed two important points. Pre- and postcalibration curves were never identical; the electrode response decreased by about 10–15 % after intracellular measurements. The potential difference when SIS was replaced with NAS could decrease by 50 % in postcalibration measurements with electrodes that had been used intracellularly; this did not occur in electrodes that remained outside the cell. It seems possible that this phenomenon explains most of the drift seen when the electrode was withdrawn from the cell. This point will be discussed below as it is important in evaluating the experiments.

To check the influence of several variables on the response of the NH₄ISMs additional calibrations were performed. As may vary between 160 and 90 mmol I^-1 from one SR neurone to another (Brown et al. 1978; Moser, 1985) we calibrated NH₄ISMs in SISI or SIS2, simulating a high-K⁺ or a low-K⁺ cell. The results are shown in Fig. 3A, where two families of curves can be distinguished. The total voltage response was smaller in SISI (filled symbols) than in SIS2 (open symbols), since larger amounts of the main interfering ion (K⁺) were present in SISI. As SISI was always used for calibrations when working with the SR neurone, the results overestimate the ammonium concentration. Overestimation could be about 30% at 5 mmoll^-1 NH₄⁺ and about 15% at 50mmoll^-1 NH₄⁺. Table 1 shows that ionic strength was changed from 181 to 275 mmol I^-1, resulting in a change in the activity coefficient for monovalent ions of about 0.03, calculated according to Ammann (1986, page 76). As the effects are rather small even when amplified under the condition of low [K⁺] in SIS2, we need not go into detailed descriptions of all the variables investigated. In summary, we conclude that NH₄ISMS are sensitive to a number of different variables, especially at low NH₄⁺ concentrations.

Compared with control calibrations we find that osmotic pressure is of some importance for measurements with NH₄ISMs at low [NH₄⁺] (about 7 % at the 5 mmol I^-1 NH₄⁺ level), but the most important variable is background [K⁺], Cell volume changes causing alterations of intracellular [K⁺] are critical as they constitute one of the main sources of erroneous measurements for [NH₄⁺], while alterations in intracellular [Na⁺] (by up to 44 mmol I^-1) can be ignored.

There are two reasons for puncturing a cell with the E_m electrode before inserting an NH₄ISM. (1) The intracellular K⁺ interference produced a potential shift of about +70 mV. The magnitude of this potential was about the same as the membrane potential (E_m) and, since the potential of the intracellular NH₄ISM is the sum of and E_m, this value will be close to 0. It was, therefore, difficult to detect the entry of the NH₄ISM, since a potential change of only a few millivolts could be expected. (2) When the NH₄ISM touched the cell surface, a gradually increasing positive potential shift of up to 30 mV developed as the tip pressed against the membrane. This can be seen in Fig. 4. Comparable effects have never been found with other ISMs measuring intracellular pH and Na⁺ and Ca²⁺ concentrations.

Fig. 4 shows voltage recordings from an E_m electrode and an NH₄ISM before and after cell penetration and from an extracellular NH₄ISM. At point A the E_m electrode crossed the membrane, reaching a value of about —65 mV. As [NH₄⁺]_i is measured differentially, penetration of the E_m electrode caused its reading to move in the ‘opposite’ direction. To show this more clearly a small portion of the recording of the [NH₄⁺]_i electrode has been replaced by a dashed line. At B the NH₄ISM touched the cell and, as it was pressed against the membrane, a positive potential shift of +20 to +30mV occurred. Several gentle attempts to push the NH₄ISM into the cell failed. A stable potential value seems to be reached at point C. Exposure to K⁺-NAS (25.5 mmol l^-1) at point D caused the cell to depolarize by about 25 mV, a value similar to that recorded by the extracellular NH₄ISM. The small change in potential indicates that the supposed ‘intracellular’ NH₄ISM was not yet within the cell. At E the NH₄ISM was pushed inside the cell, as shown by the second depolarization induced by K⁺-NAS at F. Isopotentiality resulted, which strongly suggests that both electrodes were in the same cell. During the experiments we included several depolarizations induced by K⁺-NAS to ensure that the NH₄ISM remained within the cell.

In NH₄⁺-NAS an equilibrium between NH₄⁺, NH₃ and H⁺ will be established, the dissociation constant (pK) being 9.25. Hence, in NH₄⁺-NAS (at a pH of 7.4) about 98 % of the ammonium will be present as NH₄⁺ and about 2 % as NH₃. The lipophilic, electroneutral NH₃ and the hydrophilic, charged NH₄⁺ exert specific effects on the membrane, and they may pass through different pathways at different velocities. Depending on which enters the cell earlier, pHi will change in a particular direction. NH₃ entry causes a decrease in intracellular proton concentration, whereas NH₄⁺ entry elicits an increase in intracellular proton concentration. Boron and de Weer (1976) attribute the initial alkalization to the rapid entry of NH₃, and the subsequent slow acidification (degradation of alkalization, Moser et al. 1988) to the entry of NH₄⁺. Upon return to NAS, a period of fast acidification starts, due to a rapid loss of NH₃ converted from intracellular NH₄⁺ that had previously entered the cell. An example of simultaneous measurements of intracellular pH, [NH₄⁺]_i, E_m and [NH₄⁺]_e is given in Fig. 5. The cell was punctured with a double-barrelled and a single-barrelled microelectrode and kept in NAS for about 15 min. When all variables had stabilized, the cell was ready to be acid-loaded by exposure to NH₄⁺-NAS.

Measurements in more than 30 cells are qualitatively and quantitatively similar to the example given in Fig. 5. NH₃ entered the cell rapidly by diffusion, as shown by the fast alkalization. To account for this alkalosis one must assume an initial increase in [NH₄⁺]_i of several millimoles per litre. But such an initial, rapid change, which should be well above the detection limit of the NH₄ISM, was not found. Instead, [NH₄⁺]_i increased slowly in the SR neurone, and this seems to reflect an entry of NH₄⁺ rather than a conversion of intracellular NH₃ (see above and Discussion). NH₄⁺ may be transported either through channels or by carriers (e.g. a Na⁺/K⁺ pump; Moser, 1987). During a 3min exposure to NH₄⁺-NAS, intracellular [NH₄⁺] increased to 19.6±4.94mmol I^-1 (mean±s.D.; N=142). At the same time, the depolarization increased, but pHi, intracellular [NH₄⁺] and E_m did not reach steady-state values within 12min of exposure. During washout in NAS these three variables changed rapidly: pHi exhibited fast acidification and the process of acid extrusion, E_m repolarized and/or hyperpolarized with additional after-effects, and [NH₄⁺]_i fell exponentially and reached a minimum value close to the initial one. At the point of maximum acidification, nearly all the intracellular NH₄⁺ had left the cell as NH₃, which followed its outwardly directed chemical gradient. (E_m and the electrical gradient tend to keep cations inside the cell; from the experiments shown in this study, we cannot exclude any participation of processes transporting NH₄⁺ out of the cell.) In a few measurements we even found that the potential of the NH₄ISM transiently undershot its original value, a phenomenon that will be discussed below.

20 mmol I^-1 NH₄C1 caused the external NH₄ISM to change its potential by about 72 mV (Fig. 5). This indicates that the detection limit for NH₄⁺ is lower in an extracellular solution such as NAS. We carried out several calibrations to determine the detection limit and the effect of adding of 0.1 or 1 mmol I^-1 KC1 to NAS. The results are shown in Fig. 6. NH₄⁺ concentrations as low as 0.02 mmol I^-1 were obtained by diluting NH₄⁺-NAS with NAS. A change from NAS to NAS containing 0.02 mmol I^-1 NH₄⁺ caused a potential shift of about 2 mV. The potential difference between 0.02 and 0.2 mmol I^-1 NH₄⁺ is about 3.5 mV, between 0.2 and 2 mmol I^-1 NH₄⁺ it is about 21 mV and between 2 and 20 mmol I^-1 NH₄⁺ it is about 46 mV. Compared with SISI, the detection limit for NH₄⁺ was at least 10 times (about 100μmoll^-1) better in NAS (at a background¹ K⁺ concentration of 5.4mmoH^-1). At specified NH₄⁺ levels known concentrations of KC1 were added (Fig. 6). At 0.2mmoll^-1 NH₄⁺, addition of O.lmmoll^-1 KC1 produced a negligible potential change. At the same NH₄⁺ concentration addition of 1 mmol I^-1 KC1 caused a positive potential change which could give rise to an error of about 50 % for [NH₄⁺]. In contrast to this, addition of 1 mmol I^-1 KC1 when [NH₄⁺] was 2 mmol I^-1 produced an error of only 15 %. This result again indicates the importance of the main interfering ion (K⁺) for the detection limit of the NH₄ISM. For exact measurements with the NH₄ISM, a knowledge of background [K⁺] and its possible alterations during experimentation seems essential.

NH₄ISMS show a remarkable drift within 4–11 h, mainly in the positive direction (22.1±15.9mV; mean±s.D.; N=26) when the baseline (in NAS) is considered before and after intracellular measurement. Because of this drift, the NAS value cannot be used as a reference for calibration of intracellular [NH₄⁺]. The source of this drift could be either changes in the electrode, caused by its penetration through the plasma membrane, or ionic alterations in the preparation during the experiment.

As mentioned above and partially shown in Figs 1 and 2, postcalibration curves differ from precalibration curves in the magnitude of their potential changes (about 10%) and in their response to changes from NAS to SIS (about 50%). Each factor by itself, and both together, can cause electrode drift. Intracellular ionic alterations can be seen if we follow the intracellular baseline (IB) of the NH₄ISM. Values of IB can be obtained from the potential of the differential signal of the intracellular NH₄ISM before an NH₄⁺ exposure or after recovery of the cell from an NH₄⁺ exposure. An hour-by-hour comparison of the mean value of the IB of 31 cells shows that within 7 h of experimentation the mean values differed by a maximum of 1 mV, whereas the standard deviation increased from ±3.4 mV (after 2h) to ±5.6mV (after 7h). From the relatively stable IB we conclude (1) that minor changes occur in [NH₄⁺]_i and [K⁺]_i when the cells are allowed to recover from an exposure; (2) that the amount of NH₄⁺ produced or taken up by various metabolic pathways (Kvamme, 1987) is either precisely regulated, shown by its tendency towards a constant value during recovery, or it is below the detection limit of the NH₄ISM; (3) that the main changes in the electrode response occur during impalement or when the electrode is withdrawn. We propose to take the IB as a reference for the evaluation of the [NH₄⁺] measurements. In doing so, we cannot give absolute values of [NH₄⁺]_i (which would be difficult in any case owing to interference from mostly unknown amounts of intracellular K⁺), but only apparent values. The apparent value must be corrected by the basic background level of NH₄⁺ to obtain absolute values. The presentation of apparent values instead of absolute values is, in principle, similar to measurements with chloride-sensitive microelectrodes (e.g. Deisz and Lux, 1982).

We have not been able to measure background [NH₄⁺]_i for several reasons: (1) the amount of intracellular K⁺ is unknown, (2) there is a positive potential shift when attempts are made to penetrate the cell, (3) there is a severe drift of the response of the NH₄ISM with respect to the reference level in NAS, (4) NH₄⁺ level could be below the limit of detection of the NH₄ISM, and (5) it is questionable whether all metabolic pathways in which NH₃/NH₄⁺ is involved can be blocked without killing the preparation. Thus, an NH₃/NH₄⁺-free cell cannot be obtained (in contrast to a Cl^--free cell). Points 2–4 also mean that intracellular [K⁺] can be neither estimated nor calculated from the IB.

For evaluation, the postcalibration curve was used and matched to the actual value of the IB. Therefore, the apparent values given must be considered to be maximum values (Fig. 2). In contrast, during extracellular application of the NH₄ISM the measurement of absolute values of [NH₄⁺] seems to be possible, especially when [K⁺] is known during calibration and when it is ascertained that [K⁺] does not change.

The detection limit of the NH₄⁺-sensitive electrode is affected by the concentration of K⁺ present in the solutions (Figs 2 and 6). From calibration curves in SISI salines with high (194 mmol I^-1) background K⁺ levels (Fig. 3), we deduce that NH₄⁺ concentrations below 5 mmol I^-1 are estimates rather than measurements. Between 5 and 20 mmol I^-1 NH₄⁺ a potential change of l mV corresponds to a concentration change of approximately 20 %. In other words, the differential potential measurements must be accurate to 1 mV to give estimates for [NH₄⁺]_i within a 20 % error limit. Between 20 and 50 mmol 1^-1 NH₄⁺ a 1 mV potential shift represents a 10% change in NH₄⁺ concentration.

In addition to these limitations and the relatively poor resolution when NH₄ISMS are applied intracellularly, there are additional sources of error which may mimic the presence of, or changes in, [NH₄⁺]_i. These include changes in [K⁺], ionic strength, osmolarity, volume and pH. Their influence is discussed in the text and is apparent from Fig. 3. It is almost impossible to exclude changes in all these factors during an experiment. To improve the detection limit of the NH₄ISM, simultaneous measurement of intracellular [K⁺] is essential and the measured [K⁺] should be used for calibrations.

Extracellular measurements with NH₄ISMs are not only easier to perform but also much more accurate, because in NH₄⁺-NAS the slope of the electrode response is steeper and the detection limit is at least 25 times lower than that of intracellular measurements (Figs 5 and 6). Between 0.2 and 0.5 mmol I’¹ NH₄⁺-NAS a potential change of 1 mV corresponds to a change in NH₄⁺ concentration of 20%. For comparison, this is the same value as that found for intracellular calibration in NH₄⁺-SIS between 5 and 20 mmol I^-1 NH₄⁺. Again, the detection limit can be improved when the background [K⁺] is less than 5.4 mmol 1 ¹ or when [K⁺] and ionic strength do not change during the experiment.

In a number of experiments we found that, after removal of NH₄C1, the voltage of the NH₄⁺-sensitive electrode did not return to its original value of the internal baseline but showed a transient undershoot lasting for a few minutes. Drift of the NH₄ISM can be excluded because of the transient nature of the undershoot. The effects of intracellular acidification (Fig. 1) can be excluded since the time courses are too different. The increase in intracellular [Na⁺] due to pHi regulation is too small (Fig. 3) to have an effect and, more importantly, it is in the opposite direction to the undershoot. For potassium the situation is as follows: a rather wide range of intracellular K⁺ activities (90–165 mmol l^-1) has been found in this preparation (Brown et al. 1978; Moser, 1985). An assumed change of 10 mmol I^-1 K⁺ in a background of 90 mmol l^-1 will be sensed by the NH₄⁺-sensitive electrode as larger than the same change in a background of 160 mmol l^-1, as verified by calibrations. Changes in [K⁺]_i can be caused to some extent by substitution of NH₄⁺ for K⁺ (Moser, 1987) or by volume changes (e.g. Ballanyi and Grafe, 1988). Before drawing any conclusions about this factor it is essential to measure [K⁺]_i.

The undershoot phenomenon could also mean that the intracellular baseline includes a substantial amount of NH₄⁺. To exclude or to verify this factor, a systematic analysis of the undershoot phenomenon should be carried out. However, this was not the subject of the present work.

Touching the plasma membrane with an NH₄ISM (Fig. 4) causes a positive potential to develop. This is not a general phenomenon for all ISMs but seems to be specific for NH₄ISMs. The potential could arise either from NH₄⁺ or from K⁺. We assume that the potential is caused by K⁺ accumulated close to the external surface of the membrane. This effect may be linked to the destruction of small glial cells which surround the sensory neurone.

When comparing directly measured intracellular [NH₄⁺] with that (indirectly) predicted from pHi measurements by the Boron and de Weer (1976) model, we find clear quantitative differences. Although the quantitative aspect will be examined more closely in a separate paper, Fig. 5 shows only a slow increase in [NH₄⁺]_i during the period of initial alkalization. According to the Boron and de Weer model, a rapid increase in [NH₄⁺]_i of 10–20 mmoll^-1 was to be expected (depending on pHi). In agreement with the model, we see a slow increase in [NH₄⁺]_i during the period of slow acidification and a rapid decrease in [NH₄⁺]_i during the period of fast acidification.

It is too early to decide what kind of additional assumptions have to be included in the model. For example, one can argue that the differences between our results and those of the model might be due to some kind of intracellular NH₄⁺ buffering and/or involvement of NH₄⁺ in glutamine and glutamate metabolism, but this cannot yet be confirmed. This study provides evidence that the quantitative difference in determination of [NH₄⁺]_i cannot be attributed to the temporal and/or quantitative limitations of NH₄ISMs.

We wish to thank Dr Harald Völkl for reading the manuscript, Gabriele Buemberger for linguistic improvements, the FWF (project 6177) and the University of Innsbruck for financial support.