We transfected cultured hippocampal neurons with the glutamate receptor subunits GluR-A(flip) tagged with pHluorin on the extracellular N-terminus (pHluorin-GluR-A; Additional file 1: Figure S1). A similar construct has been employed in a recent study, except the authors used the flop splice variant [17]. In order to highlight the membrane-inserted fraction of these receptors, we combined total internal reflection fluorescence (TIRF) microscopy with pHluorin tagging as has been done previously [17, 18]. The TIRF microscopy selectively visualizes fluorescent molecules located within approximately 0.1-0.2 μm of the cell-bearing glass coverslip (therefore, in or close to the basal PM; Figure 1A-C). Unlike epifluorescence imaging where all fluorescent molecules in a cultured cell are excited by the light passing through the cell (Figure 1D-F), TIRF offers supra-optical vertical resolution superior to that of confocal microscopy [19]. Indeed, individual clusters of pHluorin-GluR-A fluorescence were clearly visible in TIRF (Figure 1C) but not in epifluorescence mode (Figure 1F).
While in some secretory organelles the fluorescence of pHluorin molecules is fully quenched by acidic intralumenal milieu [14, 15], quenching of the ER-residing pHluorin may not occur because pH in the ER is not acidic [16]. To test directly whether the ER-residing pHluorin retains its fluorescence, we overexpressed in cultured hippocampal neurons a pHluorin-encoding construct engineered for expression in the lumen of endoplasmic reticulum (pHluorin-ER). This protein is retained in the ER via retrograde transport based on KDEL receptor (see Methods). We observed bright fluorescence of the pHluorin-containing ER (Additional file 2: Figure S2). Importantly, pHluorin-ER was detectable not only in epifluorescence but also in TIRF mode (Additional file 2: Figure S2), consistent with our previous findings [5, 7]. This observation lead us to conlcude that i) fluorescence of the ER-resident pHluorin molecules is not quenched, and that ii) TIRF imaging mode does not fully exclude the fluorescence of intracellular pHluorin-tagged molecules.
To overcome these limitations, we set out to develop a method for assessing the fractional contribution of the PM-inserted pHluorin-GluR-A receptors versus intracellular ones to the overall fluorescence. We performed repetitive acidification tests, which consisted in perfusing cells for 1-2 minutes with an acidic extracellular solution (pHo = 5.4, titrated with HCl). The overall pHluorin fluorescence was strongly reduced within 30 seconds of the acidification test (compare Figure 2A and 2B). The rate of fluorescence quenching closely corresponded to the time required for near-complete exchange of the bath solution, which was determined in a separate set of experiments (22 ± 3 s, n = 3; see Methods). This observation strongly suggests that the rapid drop in fluorescence was caused by quenching of the PM-inserted pHluorin-GluR-A molecules due to their direct exposure to extracellular acidic solution.
In spite of the rapid initial quenching, a significant fraction of the pHluorin-GluR-A receptors retained their fluorescence during the acidification tests (Figure 2B). On average, in 39 neurons the fluorescence fraction that remained unquenched after 30 seconds of the acidification test constituted 42 ± 3%. The unquenched fluorescence was observed throughout the neuronal soma and neurites (Figure 2B) and often had a reticular appearance resembling that of the ER. To visualize spatial distribution of those pHluorin-GluR-A molecules that were quenched during the acidification test, we digitally subtracted the image obtained during the test (Figure 2B) from the control image (Figure 2A). The resulting image (Figure 2C) showed fluorescence that was distributed throughout neuronal soma and dendritic tree, a pattern consistent with the PM-inserted fraction of pHluorin-GluR-A [18].
When the acidification test was terminated by returning to the control perfusion solution (pHo = 7.4), the fluorescence level was completely restored to its pre-test value within few minutes (time course of a representative experiment is illustrated in Figure 2D). Repetitive 1- or 2-minute acidification tests did not have any irreversible effects on the overall pHluorin fluorescence, because complete fluorescence recovery was reliably observed when we repeated these tests for up to one hour with the 5 minute inter-test intervals (data not shown). It is conceivable that repetitive acid exposure could have caused some subtle changes in those physiological parameters of neurons that were not monitored in our experiments. However, reproducibility of acidification tests argues against accumulating detrimental effects.
Next, we focused on kinetic analysis of the pHluorin-GluR-A fluorescence changes during the acidification test. We hypothesized that this analysis may provide a means for repetitive measurement of the PM-inserted fraction of pHluorin-GluR-A receptors. Using a shorter inter-frame interval (6 s), we performed acidification tests and recorded the time course of the fluorescence quenching and its recovery (Figure 3A). The initial decrease in the fluorescence was termed "fast phase" (Figure 3A, green circles) because it developed rapidly with a mono-exponential decay time constant of 13.4 ± 0.8 s (R2 = 0.999; n = 3). Thus, by 30 seconds after the beginning of the test, the fast phase was largely over (i.e., ~90% completed). After the first phase, we observed the second phase of fluorescence quenching that proceeded more gradually (Figure 3A, red circles). We termed this second phase "slow phase" because its time course was slower than that of the fast phase (decay time-constant of 69.2 ± 3.7 s, n = 3; P < 0.0001). While the exponential equation was a satisfactory fit for the slow phase (R2 = 0.989), the linear fit was more suitable (R2 = 0.995) and yielded the linear coefficient of 0.17 ± 0.03%/s (n = 3).
Fractional contributions of the fast and slow phases varied widely between different cells and different subcellular regions. We hypothesized that larger fractional contributions by the slow phase were due to stronger "contamination" of the "signal" (i.e., fluorescence of the PM-inserted pHluorin-GluR-A receptors) by "noise" (i.e., fluorescence of the receptors residing in non-acidic organelles such as the ER). If this is true, then the slow phase of fluorescence quenching (red circles in Figure 3A) should reflect passive acidification of these organelles, which is secondary to the decrease in cytosolic pH caused, in turn, by extracellular acidification. To test this hypothesis, we transfected neurons with a construct encoding for soluble pHluorin expressed in the cytosol (pHluorin-CS). A two-minute perfusion with acidic solution (pHo = 5.4) produced a slow, gradually developing decrease in pHluorin-CS fluorescence (Figure 3B). This decrease was well fitted with a linear regression (R2 = 0.994), with the coefficient of 0.47%/s. This result indicates that, at a later stage of the pH test, the partial collapse of the proton gradient across the PM caused gradual acidification of the cytosol.
The cytosolic acidification should then lead to acidification of those intracellular organelles that lack active proton transport. We transfected cells with the ER-trapped pHluorin (pHluorin-ER), performed acidification tests and compared the time course of fluorescence changes in neurons expressing pHluorin-ER (Figure 3C) to those expressing pHluorin-CS (Figure 3B) or pHluorin-GluR-A (red circles in Figure 3A). Similarly to pHluorin-CS, the pHluorin-ER fluorescence exhibited a slow decrease during the acidification test and was well fitted with a linear regression (R2 = 0.991), yielding the linear coefficient of 0.45%/s. Importantly, the near-linear time course of the fluorescence changes both in the cytosol and ER was very similar to the slow phase of pHluorin-GluR-A fluorescence quenching (red circles in Figure 3A). Indeed, the linear coefficients were close for all three constructs. To compare them, one should take into account the fractional contribution of the pHluorin-GluR-A slow phase that was, on average, 42% (Figure 3A). Thus, the corrected linear coefficient for the slow phase of pHluorin-GluR-A quenching is 0.17%/s/0.42 = 0.40%/s, which corresponds well to the linear coefficients of both pHluorin-CS (0.47%/s) and pHluorin-ER (0.45%/s). Taken together, these results strongly suggest that the slow phase of pHluorin-GluR-A fluorescence quenching was due to intracellular acidification, and that this slow phase developed after the fast phase was largely complete.
The clear temporal segregation of fast and slow quenching phases allowed us to perform accurate separation of the fluorescent signal derived from the PM-inserted pHluorin-GluR-A receptors from that of the intracellular receptors. For our repetitive acidification method, we selected the time point of 30 seconds after the beginning of the acidification test because at this point approximately 90% of the fast quenching phase is completed, whereas the contribution of the slow phase is negligible. We estimated the error due to the slow phase contribution to be below 3%. This estimate is based on two measurements: i) thirty seconds after the beginning of the acidification test, the decrease in pHluorin-ER fluorescence (black arrow in Figure 3C) was 5.8 ± 1.1% (n = 5), and ii) on average, the intracellular fraction of the pHluorin-GluR-A fluorescence was 42%. Thus, the error of our method can be calculated as 5.8% × 0.42 = 2.4%. Importantly, the acidification tests are highly reproducible and thus can be used for monitoring changes in the PM-inserted fraction of pHluorin-tagged channels, such as their insertion or internalization during synaptic plasticity.
To demonstrate applicability of the repetitive acidification method, we used it to monitor translocation of pHluorin-GluR-A receptors to the PM. Membrane insertion of GluR-A-containing AMPA receptors associated with LTP is a well-established phenomenon (for review see [20]). In a recent study, Yudowski and colleagues used pHluorin tagging to visualize membrane insertion of GluR-A(flop) receptors [17]. Here, we focused on the "flip" splice variant of this receptor. On hippocampal neurons transfected with pHluorin-GluR-A, we performed repetitive acidification tests with a 10 minute interval. During each test, we measured the amplitude of the fast quenching phase 30 seconds after the onset of acidification.
Under control conditions, the amplitude of the fast phase remained stable for at least 40 minutes (Figure 4A). We stimulated NMDA receptors by removing extracellular Mg2+ and adding glycine (200 μM) for 5 minutes. After this stimulation commonly used to trigger the so-called "chemical LTP" [17, 21], we observed a long-lasting increase in the fast phase amplitude (Figure 4C). Thirty minutes after the glycine/0 Mg2+ application, fast quenching phase was significantly larger compared to control conditions (after glycine: 112 ± 2%, n = 5; control: 99 ± 3%, n = 4; P < 0.01). Significant difference was maintained at 40 minutes (P < 0.05). At some subcellular locations we observed much larger increases in the fast phase amplitude (up to 180%; data not shown). However, the detailed analysis of region specificity of GluR-A membrane insertion was beyond the scope of this study. Taken together, these results demonstrate persistent insertion of pHluorin-GluR-A(flip) receptors to the PM in response to brief activation of NMDA receptors. They also show that the repetitive acidification method can be used as a tool to monitor kinetics of activity-dependent AMPA receptor membrane trafficking.
Two observations presented below illustrate advantages of our method over the previously used approach where changes in the overall pHluorin-GluR-A fluorescence were used as the index of membrane insertion [17, 18]. First, we found that repetitive acidification method has an improved sensitivity due to the increase in the signal-to-noise ratio. This is apparent in the example shown in Figure 5A. Thirty minutes after gly/0 Mg2+ application, insertion of additional pHluorin-GluR-A receptors to the PM resulted in i) increased overall fluorescence as well as ii) increased amplitude of fast quenching phase. The increase in the overall fluorescence constituted 14%: fluorescence level grew from 100 to 114 on the normalized fluorescence scale, (114-100)/100 = 14%. The increase in the fast phase amplitude was considerably larger and constituted 24%, calculated as: first test 100-38 = 62, second test 114-37 = 77, change (77-62)/62 = 24%. The reduced sensitivity of the conventional method is most likely caused by the "contamination" of the overall fluorescence by intracellular pHluorin-GluR-A receptors. In contrast, the fast phase amplitude only reflects membrane-inserted pHluorin-GluR-A molecules, thus making the sensitivity of the repetitive acidification method superior. An alternative way to remove the background fluorescence that is contributed by unquenched intracellular pHluorin consists in pre-bleaching and has been previously employed by von Zastrow's group [17]. In this approach, prolonged light exposure is used to destroy fluorophore of the unquenched pHluorin-GluR-A molecules (PM-inserted and intracellular alike) while sparing the fluorophore of the quenched pHluorin-GluR-A molecules and preserving their ability to become fluorescent upon PM insertion [17].
The second notable advantage of the repetitive acidification method is that it allows distinguishing between PM-insertion-related changes in pHluorin fluorescence from those caused by e.g. changes in the pH of the intracellular organelles containing pHluorin-tagged molecules. In the example of Figure 5B, the neuron was monitored under control conditions without NMDA receptor activation. During the 30 minutes between the first and the second acidification tests (left and right traces, respectively) the overall pHluorin-GluR-A fluorescence increased by (112-100)/100 = 12%. In contrast, the amplitude of the fast quenching phase remained unchanged: first test 100-62 = 38, second test 112-74 = 38, change (38-38)/38 = 0%. This observation suggests that interpreting the increase in pHluorin fluorescence as an indication of membrane insertion may lead to erroneous conclusions and should, therefore, always be verified.