Inflammation model

Zymosan A, a glucan prepared from yeast cell walls (Saccharomyces cerevisiae), induces a strong inflammation when injected in living tissues (Meller and Gebhart, 1997). The inflammation develops in the first three to four hours after injection.

Meller and Gebhart used concentrations of 0.313 to 6.25 mg Zymosan A per animal.

Because even a very low dosage of 0.15 mg per animal induces measurable thermal hyperalgesia (Hess et al., 2007), a medium dose of 1.25 mg was used for the present study.

In order to induce thermal hyperalgesia, Zymosan A (1.25 mg per animal, SIGMA-ALDRICH) was dissolved and diluted in 0.9% saline and injected subcutaneously into the plantar side of the left hind paw of the rats in each case four hours before starting the fMRI scans (Hess et al., 2007).

5.2.3.4 Behavioral tests

In a first run, each group consisted of 5 animals. In a second run, the measurements were repeated for NhZ and NanZ to reproduce the antinociceptive effects of the corresponding substances. Each individual had one control measurement with 0.9%

saline to get animal-specific baseline values. All behavioral tests were performed in a soundproof room and previously all animals could acclimatize to the environment over seven days. All rats got three days to get acquainted specifically to the particular setup.

At day four, the control experiment with intraperitoneal injection of 0.9% saline was performed. At day five, the venoms were tested with the previously defined dosages (see chapter 5.2.3.2). All behavioral measurements were performed one hour after intraperitoneal injection of the test substances because animals had to calm down after handling and the absorption of the venom via the peritoneum is slow. For animal welfare the behavioral tests were performed without induction of hyperalgesia.

5.2.3.4.1 Hargreaves test

A modified Hargreaves test (Hargreaves et al., 1988) was applied for measuring potential antinociceptive effects of the venom in an acute pain model. A plantar-test

apparatus (7370, UGO BASILE Biological Research) and five single animal boxes (21 cm x 17 cm x 14 cm) with a metal-grid bottom instead of glass were used. Painful heat stimuli (infrared beam) were applied at the plantar sides of the rats` hind paws. Paw- withdrawal latency (PWL) was measured four times for each paw during 30 minutes.

5.2.3.4.2 Tail-flick test

Tail-flick test was performed to measure more peripheral sensitivity to increasing temperature and nociception (D´Amour, 1941). We used the infrared tail-flick apparatus 7360 (230 Volt) from UGO BASILE Biological Research. A special “rat restrainer” made of an empty and black-coated bottle was used to fixate the respective animal with minimal stress and with its tail exposed for positioning over the infrared- heating device. Tail-withdrawal latency (TWL) was measured three times at three different tail locations during 10 minutes.

5.2.3.4.3 Rota-rod test

Rota-rod test was performed to check the motoric integrity of the animals (Kuribara et al., 1977). For this purpose the rota-rod apparatus Thyristor TR 50 was used. The animals had to walk on a rotating rod at a speed of five rotations per minute for a maximum of two minutes. The time until a possible fall was taken with a stop watch.

5.2.3.4.4 Open-field test

Open-field test was performed to estimate locomotor activity of the rats (Prut and Belzung, 2003). Therefore, a slightly shaded box (four-sided area 100 cm x 100 cm, barrier height 40 cm) was used with a camera device on top (CONRAD, 640 x 480 pixel). Each rat had to stay in the box for ten minutes and locomotion and behavior were recorded during this time. The open-field arena was cleaned with an alcoholic solution each time before a rat was placed inside in order to avoid biasing effects from previous animals. The open-source software ImageJ (V. 1.43u, Wayne Rasband, National Institutes of Health, USA) was adapted to the existing data structure for automatic subject tracking. The walked distance during ten minutes was calculated for each animal with a customized plug-in for ImageJ.

5.2.3.5 Functional MRI

Each fMRI group consisted of 8 animals. One group was defined as control group and injected with 0.9% saline instead of venom. Adult male Wistar rats were initially

anesthetized with isoflurane 5% for 5 minutes followed by isoflurane 1% - 2% to maintain a breathing rate of 60 and constant blood-pCO2 levels thus enabling a stable physiological status for functional imaging (Ramos-Cabrer et al., 2005). An intraperitoneal catheter (VASUFLO, 24G) was implanted enabling an injection of the venom during fMRI measurements inside the scanner. The rats were placed on a cradle and the animal body temperature was kept at constant levels (37°C) by warm water circulating in the cradle wall. Extensive physiological monitoring (respiration rate, pulse rate, blood oxygen: STARR MouseOx Life Science Group) was performed.

A dedicated animal-MR scanner (BRUKER BioSpec 47/40; free bore of 40 cm, 200 mT/m), a whole-body birdcage resonator for homogenous excitation and an actively radiofrequency-decoupled anatomically shaped four-channel-array surface coil system was used for imaging. The surface coil was placed directly above the head of the rat to maximize the signal-to-noise ratio. For thermal stimulation both hind paws of the respective rat were fixed with their dorsal sides attached to two Peltier elements.

Contact-heat stimuli (40°C, 45°C, 50°C, 55°C, plateau for 5 s after 15 s ramp, 2 min recovery) were generated by a customized computer-controlled Peltier heating device not interfering with the MR scanner. Four stimuli were applied to both hind paws in an alternating fashion.

Abb. 31: Fig. 1: Time schedule for fMRI experiments

Four hours before the fMRI measurements, Zymosan A was injected subcutaneously into the left hind paw of the rat in a dosage of 1.25 mg per animal. Then, the injection of the venom/saline followed. Ten minutes after injection, the challenge measurement was started (1500 sets of functional volumes within a 100 min period). By this time schedule the expected effects of the venoms take place in the middle of the functional experiment. Therefore the measurement was divided in phase I (Ph I) and phase II (Ph II).

During the fMRI experiments a defined stimulation protocol applied four increasing temperatures (T:

40°C, 45°C, 50°C, 55°C) alternately to the right (R; light gray) and to the left hind paw (L; dark gray)

Before the challenge measurement (Fig. 1), MRI-specific global setting steps were performed. First the three spatial directions (coronar, axial, sagittal) were recorded with a Gradient-Echo-based TriPilot sequence (TE = 6 ms, TR = 100 ms, matrix 128 x 128, inplane resolution 390 µm x 390 µm, FOV 50 mm x 50 mm, 2 mm slice thickness) and the orientation of the head inside the MR tomograph was optimized by slightly moving and rotating the cradle inside the scanner between two TriPilot measurements. Next, a fast coronar T2-weighted RARE (rapid acquisition relaxation enhanced) sequence (TEeff = 56 ms, TR = 2000 ms, matrix 256 x 128, inplane-resolution 137 µm x 273 µm, FOV 35 mm x 35 mm, 1 mm slice thickness, 16 coronal slices) was measured (Hennig et al., 1986). These coronar images were used to position the 10^th of the 22 axial slices of the functional measurement between the posterior tip of the striatum and most anterior part of the hippocampus serving as a reliable spatial reference. Next, to check for possible motion artifacts due to suboptimal mounting, an EPI (echo-planar imaging) sequence (TEeff = 25.3 ms, TR = 200 ms, matrix 64 x 64, inplane-resolution 390 µm x 390 µm, FOV 25 mm x 25 mm, 1 mm slice thickness, 1 axial slice) was measured with only one slice 300 times at 200 ms intervals. This dataset was visually inspected for motion as a movie and, if needed, the rat was remounted.

Then the venom (NhZ or NanZ group) or saline (control group) injection followed and ten minutes after injection the challenge measurement was started (Fig. 1): 1500 sets of functional volumes during 100 minutes with six stimulation cycles (one cycle: four temperatures applied alternately to each paw) acquired using a Gradient-Echo-based EPI Technique (TEeff = 24.4 ms, TR = 4000 ms, matrix 64 x 64, inplane-resolution 390 µm x 390 µm, FOV 25 mm x 25 mm, 1mm slice thickness, 22 axial slices). By this time schedule the expected effect of the venom took place in the middle of the challenge measurement, thereby dividing it in phase I (before expected maximal effect of NhZ/NanZ, Ph I) and phase II (with expected maximal effect of NhZ/NanZ, Ph II).

Finally, as an anatomical reference, 22 axial T2 images were acquired at the identical position of the EPI slices with a RARE sequence (TEeff = 47.11 ms, TR = 3000 ms, matrix 256 x 256, inplane-resolution 98 µm x 98 µm, FOV 25 mm x 25 mm, 1 mm slice thickness, 22 axial slices).

5.2.3.6 Statistical analysis 5.2.3.6.1 Behavioral tests

The data collected for each test and animal were averaged per paw/tail and evaluated in MS Excel for NhZ/NanZ and control groups. PWL values, TWL values, duration on the rota-rod, and calculated distances covered in the open-field arena were averaged over 10 animals for NhZ and NanZ, over 5 animals for PaZ and WaeZ. Group-mean values were compared by the Student`s t-test. P < 0.05 was considered to indicate a statistically significant difference.

5.2.3.6.2 Functional MRI

The fMRI analysis was performed following a standardized protocol. For more details see Knabl and coworkers (Knabl et al., 2008).

First, standard preprocessing steps (motion correction, slice scan time correction, 2D- Gaussian smoothing (2 pixel kernel), high-pass filtering at 9 cycles and low-pass filter of 12 seconds) were applied in BrainVoyager QX (BRAIN INNOVATION, Netherlands, version 2.0.1480). Then, a contrast-specific general linear modeling (GLM) analysis using individual predictors for every stimulus (algesic or hyperalgesic stimulation; four different temperatures; phase I or phase II of the challenge measurement) was calculated. GLM analysis is mathematically identical to a multiple regression analysis but is more flexible to incorporate multiple qualitative and quantitative independent variables (Friston et al., 1995). The GLM analysis, performed independently on the time course of each individual voxel, led to contrast-specific statistical values (t-values) per voxel which were called statistical parametric maps (SPMs). The SPMs were corrected for multiple comparisons using false-discovery rate (q = 0.05).

For each animal the brain was manually segmented from surrounding tissues creating individual brain masks using Amira 5.3.2 (VISAGE IMAGING Inc.). In order to perform a fixed-effects group analysis of the SPMs over multiple animals, they had to be registered. For this purpose, the first volumes of the functional experiments were used and registered across the different animals by an affine transformation scheme with six degrees of freedom with the application-specific Software MagnAn 2.3 (BioComGbR, Uttenreuth in IDL 6.4, EXELIS). The transformation matrix obtained

was applied to the SPMs and to the anatomical data as a spatial reference. Averaging the registered SPMs resulted in group- and predictor-specific SPMs (Fig. 6) which were used to display group-average results. As a second, quantitative analysis strategy the SPMs per animal and predictor were overlaid with an electronic version of the Paxinos rat brain atlas (Paxinos, 2007) thus labeling each significantly activated voxel as belonging to one of the 202 different brain structures. These voxels were counted as the respective activated volume per brain structure. Next, for each animal, each single brain structure, and contrast, the stimulus-induced relative change of BOLD amplitude, activated volume, peak time (time to BOLD amplitude maximum) and probability (probability of voxel activation) was calculated with respect to baseline.

All quantitative parameters together generated a high-dimensional data space.

In an initial analysis approach, in order to evaluate, if, in principal, NhZ/NanZ datasets can be separated from the control data, a principal component analysis (PCA; Pearson, 1901) of this high-dimensional data space (built from brain-structure specific response parameters: BOLD amplitude, activated volume, peak time and probability separate for phase I and phase II for NhZ/NanZ and saline), was performed in MagnAn 2.3 separately for algesic and hyperalgesic responses for the two painful stimuli (50°C, 55°C) of the challenge measurement. The PCA is a data-reduction method in multivariate statistics creating a new coordinate space in which the first principal component has the largest possible variance and the succeeding components are sorted according to their explained variance of the original data space under the constraint that the components are orthogonal to each other. The factor load values of the PCA components highlight parameters which are most relevant for group separation and should therefore be examined in more detail. Each single brain structure is again represented as one data point in the new PCA-coordinate system. For graphical representation, the center of gravity of all brain structures per experimental group was displayed (Fig. 3). To evaluate which brain structures are most relevant for this separation, Euclidian distances were calculated between NhZ/NanZ and saline values per brain structure (Fig. 4). Brain structures with Euclidian distances larger or less than mean value plus standard deviation were selected for closer examination because the PCA rendered them to be relevant for the group separation.

In order to compare NhZ/NanZ and saline group in general, BOLD amplitude and activated volume values were averaged over all brain structures for each temperature (Fig. 5). The next step was to average the values for different functional systems in the brain (e.g. sensory input, thalamus, sensory cortex or amygdala; Table 1). Mean values were compared by the Student`s t-test. P < 0.05 was considered to indicate a statistically significant difference.

5.2.4 Results

5.2.4.1 Dose determination

At the highest injected dose (half of LD50) no harmful side effects were observed in the experimental animals. Because of this, each raw venom could be injected at this dosage: I) 90 µg/kg (Naja haje/NhZ), II) 200 µg/kg (Notechis ater niger/NanZ), III) 60 µg/kg (Walterinnesia aegyptia/WaeZ), and IV) 90 µg/kg (Pseudonaja affinis/PaZ).

In the Hargreaves test only NhZ and NanZ showed statistically significant increase of paw-withdrawal latency at this dosage. PaZ and WaeZ instead showed a decrease of the paw-withdrawal latency at the highest dose.