Characterization of the genetically encoded sensors used

Published values for properties like ΔF_/F₀ and K_d for a particular sensor are useful when choosing which variant to use. However, the actual properties often differ when the sensor is expressed in a system that is different from that which was used to obtain those values (Linse et al., 1991). Since the results from measurements using GESs were going to be used quantitatively and in order to compare properties between cells and preparations, it was deemed necessary to first characterize the sensors in the primary cortical cell culture prepa-rations where they were to be used. For all sensors this meant finding proper expression levels, especially for targeted versions, that were not toxic or produced cells with apparent

changes in morphology due to expression of the sensor. Also, the practical maxΔF_/F₀ or complete dynamic range (Fmax/Fmin) was establishedin cell where possible.

D3cpV and RCaMP1e

For genetically encoded calcium indicators (GECIs), the dynamic range is fairly small and the difference between the resting cytosolic free-Ca²⁺in neurons and the level of free-Ca²⁺after high potassium depolarization or another strong stimuli is large, between∼100nm at rest and several µm after stimuli (Ross, 1989; Marambaud et al., 2009). This means that a GECI with a dynamic range of 10 will still get saturated upon calcium influx in a neuron that is stimulated if the sensor has a high enough affinity for Ca²⁺. Thus, the sensor needs to be chosen based on its K_d, dynamic range and the type of stimuli that it shall be able to measure the response to. For this thesis, the interest lie with physiological stimuli of neurons and the stimuli used was trains of action potentials at10Hz.

For D3cpV and RCaMP1e the dynamic range was established by applying ionomycin which led to very high responses of the respective sensor signals. Ionomycin, a ionophore capable of acting as a mobile ion carrier to transport cations across solvent barriers (C Liu and Hermann, 1978; Erdahl et al., 1994), is commonly used forin cell determination of the dynamic range of both genetically encoded and organic calcium indicators. It facilitates the equilibration of the intracellular calcium concentration with that of the extracellular media (A Takahashi et al., 1999). If a sufficiently high extracellular calcium concentration is applied, it will saturate the sensor while ionomycin is present. To measure the response of a sensor to zero level of free Ca²⁺ within a cell, ionomycin can be applied together with a chelator for calcium ions such ethylene glycol tetraacetic acid (EGTA). Calcium ions that leaks out from the cell, facilitated by ionomycin, will be sequestered by EGTA and eventually all calcium will have left the cell. The dynamic range acquired from ionomycin treatment was compared with the typical response to a physiological stimuli from field stimulation (FS) at10Hz. For both GECIs, the response from FS was within 75% of the sensor range measured with saturating stimuli with ionomycin and both sensors were thus usable for comparison experiments in the primary neuronal cell cultures used. Although not as important for this thesis, the dynamic range (R_max/R_min) measured from D3cpV of^∼4(section 3.1.4) corresponded with previously published values by Palmer et al. (2006). Likewise for RCaMP1e, for which only the maximum ΔF_/F₀was measured (∼3_.5), corresponded with what has recently been reported for RCaMP1e by Akerboom et al. (2013).

When targeted to the mitochondria, the responses from either of D3cpV or RCaMP1e was un-reliable when applying ionomycin. High potassium depolarization produced larger changes of the FRET ratio for 4mtD3cpV and theΔF_/F₀for mtRCaMP1e than did 10mm Ca²⁺ + ion-omycin, suggesting that the Ca²⁺ was not able to properly enter the mitochondria. When neurons were depolarized with high potassium, the response in the mitochondria was nev-ertheless higher than that caused by the FS stimuli used for later experiments. This showed that also in the mitochondria was a depolarizing stimuli using FS not enough to saturate the sensor and proved that it was feasible to use both D3cpV and RCaMP1e for mitochondrial measurements of calcium.

When the dissociation constant Kd, as well as the minimum and maximum response (Rmin

and R_max) values are known, the measured ratio or response from ratiometric and single wavelength GECIs can be converted into approximate free-Ca²⁺ concentrations according to the methods for single wavelength and ratiometric indicators put forth by Grynkiewicz et al. (1985). The K_d values are taken from literature or determined in vitro using purified sensor protein. However, there are no guarantees that the values acquired correspond to actual intracellular calcium levels and there are problems associated with these conversions such as the sensitivity of indicators to photobleaching of the chromophore (Becker and Fay, 1987; Fowler and Tiger, 1997). Also, the binding properties of calmodulin (CaM), the binding protein domain used in both D3cpV and RCaMP1e, can be influenced by ionic strength and pH (Ogawa and Tanokura, 1984; Linse et al., 1991). Because of these issues and since a determination of absolute intracellular Ca²⁺levels was not deemed necessary, the data from experiments with GECIs is presented “raw”, without conversion to calcium concentration.

ATeam1.03

In addition to GECIs, ATeam1.03, a GES sensitive to changes in ATP levels was employed to study the effects of the synucleins on ATP consumption and regeneration from oxidative phosphorylation. Previously, measurements of ATP levels in neurons have established the resting ATP concentration to be between 1 and3mm. These values come from enzymatic measurements on whole brain lysates (Veech et al., 1979) or luciferase luminescence based assays on primary neuronal cell culture lysates (Fukuda et al., 1983) or from intact cells using genetically encoded luciferase (Ainscow et al., 2002; Mironov, 2007).

ATeam1.03 is a newly developed sensor created and characterized by Imamura et al. (2009).

It relies on FRET between CFP (mseCFP) and YFP (cp173-mVenus) facilitated by binding

of ATP to the ε subunit of bacterial F₀F₁. When the work for this thesis started, the only publication mentioning the usage of this sensor was the original description. Therefor, a more in depth characterization of ATeam1.03 in the primary cortical cell cultures used here was deemed necessary. In section 3.1.6 an attempt was made to calibrate ATeam1.03 in cells of primary cortical cell culture according to a method described for other sensors (Dittmer et al., 2009). Cells were permeabilized with β-escin and superfused with media containing different extracellular ATP concentrations. The parameters from a fit of this calibration gave values for K_d(3_.13mm) and Hill coefficient (2.16) that corresponded with the published values fromin vitrocharacterization (3_.3mm and 2.1 respectively) (Imamura et al., 2009). However, the FRET ratio initially reported by the sensor, before permeabilization, was never recovered even with 10 or 100mm ATP, which should saturate the sensor according to Imamura et al. (2009). Some possible reasons for this behavior was already mentioned in section 3.1.6.

Another reason might be what Willemse et al. (2007) reports on regarding ATP and FRET.

While creating a FRET based GES for ATP, they found that a sensor construct incorporating CFP and YFP with an protease cleavage linker, which should not interact with ATP, responded to changes in ATP concentrationsin vitro. They show that this is likely to be related directly to FRET since it was not present in individual CFP or YFP units and it did not change the ratio measured from constructs were the two FPs where located far enough from each other to not allow FRET to occur. They also found that the effect most likely stems from interaction with Mg²⁺ since variations in Mg²⁺/ATP concentrations had dose dependent effects on the FRET ratios. This conclusion is not unreasonable considering the fluorescent properties and their modification by Mg²⁺of ATP (Amat et al., 2005). It has also been reported that ATP can quench the fluorescence of other fluorophores (C Li et al., 2005). That this does not occur with single FPs can be explained by considering the protected nature of the chromophore in FPs, due to the surrounding β-barrel. This also makes this unlikely to be an issue for single FP based sensors. This has consequences for all FRET based sensors however, and arguably, the issues for ATeam1.03 are of least concern since it is a sensor which has been characterized in regards to its behavior in the presence of ATP, something GESs usually have not. As for the other FRET based sensor used in this thesis, D3cpV, this is unlikely to affect the results.

The results presented in this thesis shows that ATP levels in the cytosol of active neurons, i.e. stimulated with either kainate or FS, is hardly perturbed if the mitochondria is active and able to compensate with increased ATP production. This means that for measurements using D3cpV in the cytosol, no changes in ATP levels that might interfere with the FRET efficiency of the sensor was present. However, it is known that stimuli that causes uptake of Ca²⁺to the cytosol also increases oxidative phosphorylation, leading to transient increased ATP levels

in the mitochondria (Nakano et al., 2011). From the results in section 3.1.6 we can infer such an increase is happening as well since the production rate was higher after cells had been stimulated with kainate. Figure 3.12d shows this especially well. Nakano et al. (2011) targeted a modified version of ATeam1.03 to the mitochondria of HeLa cells and measured increases in ATP levels in response to elevated cytosolic Ca²⁺levels. They did not calibrate their sensor in the mitochondria but comparing their measured FRET ratio change within the mitochondria to theirin vitro data for the sensor, the response in the mitochondria that they measured corresponded to an increase of approximately 20% or∼3_.5mm. The data by Willemse et al.

(2007) shows that for a FRET construct incorporating a short Xa protease-sensitive cleavage linker, a change from 1 to5mm ATP produces a 24% decrease in FRET ratio. However, for a construct with a inosine monophosphate dehydrogenase type II (IMPDH2) domain linking the two FPs, the change was only∼4%. This estimation tells us that in the worst case scenario, a FRET sensor such as D3cpV will, when targeted to the mitochondria and responding to changes induced by a large increase of cytosolic Ca²⁺, report with∼24% smaller ratio change due to interference from elevated ATP levels. For the results in this thesis this can be ignored since:

1. the response of D3cpV expressed in the mitochondria was within ^∼50% of the sensor range when using FS. Assuming that this is 24% less due to ATP interference, the actual sensor response would be^∼74% which is still significantly below saturation.

2. the production rates of ATP from oxidative phosphorylation was the same between all synuclein treatments and control. This suggests that the ATP increase in the mi-tochondria is also the same and thus the interference from ATP was the same for all treatments and is not hiding differences in Ca²⁺accumulation in the mitochondria.

Interestingly,in cell calibration of ATeam1.03 has since been tried by another group (Ando et al., 2012) who did not report any inconsistencies. They used digitonin to permeabilize Huh-7 cells expressing ATeam1.03 in a fusion construct that targeted the sensor to hepatitis C virus (HCV) ribonucleic acid (RNA) replication sites. A ATP concentration of∼5mm was reported at these sites with a cytosolic concentration of1mm.

4.1.3 Genetically encoded sensors for use in the study of