An Inducible Expression System to Measure Rhodopsin Transport in Transgenic Xenopus Rod Outer Segments

We developed an inducible transgene expression system in Xenopus rod photoreceptors. Using a transgene containing mCherry fused to the carboxyl terminus of rhodopsin (Rho-mCherry), we characterized the displacement of rhodopsin (Rho) from the base to the tip of rod outer segment (OS) membranes. Quantitative confocal imaging of live rods showed very tight regulation of Rho-mCherry expression, with undetectable expression in the absence of dexamethasone (Dex) and an average of 16.5 µM of Rho-mCherry peak concentration after induction for several days (equivalent to >150-fold increase). Using repetitive inductions, we found the axial rate of disk displacement to be 1.0 µm/day for tadpoles at 20 °C in a 12 h dark /12 h light lighting cycle. The average distance to peak following Dex addition was 3.2 µm, which is equivalent to ~3 days. Rods treated for longer times showed more variable expression patterns, with most showing a reduction in Rho-mCherry concentration after 3 days. Using a simple model, we find that stochastic variation in transgene expression can account for the shape of the induction response.


Introduction
Xenopus photoreceptors have played an important role in understanding the cell biology of membrane assembly [1][2][3][4], retinal disease [5][6][7] and ciliary transport [8]. A major advantage is that Xenopus has much larger photoreceptors than mammals, which make them an exquisite system for highresolution microscopic imaging [5,[9][10][11]. The photoreceptors develop rapidly, forming light-sensitive outer segment (OS) membranes within a week post-fertilization [12,13] and expression of fluorescently tagged proteins are readily expressed [14]. The OS contains high concentrations (~3 mM) of the integral membrane protein rhodopsin [15], which is synthesized in the inner segment and then rapidly delivered to the base of the OS for incorporation into disk membranes [16][17][18][19][20]. Rhodopsin is free to move laterally within a disk membrane [21], however there is no axial movement between disks. OS renewal is accomplished as membrane disks are displaced progressively outwards, with eventual phagocytosis by the retinal pigment epithelium [16][17][18][19]22,23]. Therefore, the distance along the OS axis from the base is linearly related to the time elapsed since incorporation for 4-6 weeks and thus offers an opportune model to investigate membrane protein synthesis and processing.
In order to more precisely control expression of the transgene in Xenopus rods, an inducible expression system is invaluable. There have been three inducible systems developed in Xenopus based upon the Gal4-UAS [24,25], Tet-On [26] and heat shock [27][28][29] strategies. The Gal4-UAS systems (Gal4 DNA binding domain was fused with a ligand binding domain from progesterone [24] or glucocorticoid [25] hormone receptors) respond well to low concentration of inducer during Xenopus development. However, they have significant leakiness in the absence of inducer and pleiotropic variation. The Tet-On inducible system was successfully employed in Xenopus to study thyroid hormone response gene expression [26]. However, rtTa binds to tetO weakly even without doxycycline, which leads to a basal level expression of reporter genes [30,31]. Heat-shock promoters have been used to study Wnt signalling [27,28] and HNF1 related organogenesis [29]. A big advantage of this system is that only a short-term heat shock is needed to turn on the system and no inducer is needed. However, the heat-shock promoter systems are not suitable for long-term treatment. To overcome these limitations, we designed a system that would have tight control of expression and provide reproducible induction responses ( Figures 1A and 2A). This system is based upon a GAL4-VP16 transcriptional activator fused to a fragment of the glucocorticoid receptor, which makes nuclear localization dexamethasone (Dex)-sensitive [32]. An eGFP protein is fused to the C-terminus of GAL4-VP16-GR chimeric protein to allow detection of transcriptional activator translocation. This system was placed under control of the Xenopus rhodopsin promoter (XOP) to direct expression specifically in rods. To follow rhodopsin transport, a Rho-mCherry fusion protein was placed under the control of a UAS-hsp promoter. Thus, Dex addition triggers GAL4-VP16-GR-eGFP (G3) transport into the nucleus and initiates the transcription of the Rho-mCherry reporter. We used high-resolution confocal microscopy of single rods treated with Dex to characterize spatial and temporal characteristics of induction responses and displacement rates of rhodopsin in the OS.

Mammalian Cell Culture and Luciferase Reporter Assay
HEK293T cells (ATCC, Manassas, VA) were cultured in DMEM with 10% FBS and 1 mM L-Glutamine. Cells were seeded at 75,000 cells/ml one day before transfection. Cell transfections were performed using a total of 1 µg of DNA and 3 µl Fugene 6 (Roche, Branchburg, NJ) in 100 µl DMEM for 2 ml of culture. Cells were transfected with pCMV:G3 and either pUAS:Luc or pUAS:mCherry constructs. Empty pCS2 (D. Turner, University of Michigan) was included in the transfection medium to bring the total DNA to 1 µg. Cells were harvested 48 h post-transfection and luciferase activity was determined with Bright-Glo Luciferase Assay System (Promega, Madison, WI) using a Synergy 2 Multi-Mode Microplate Reader (BioTek, Winooski, VT) according to the manufacturer's instructions.

Transgenesis and induction procedure
Transgenic Xenopus laevis were generated by restriction enzyme mediated integration [35][36][37]. Tadpoles were raised in a 12 h dark/12 h light cycle at 20 °C. During induction, tadpoles had daily water changes and replenished with fresh 10 µM Dex (Sigma-Aldrich). All animal handling and experiments were in agreement with the animal care and use guidelines of the Association for Research in Vision and Ophthalmology (ARVO). This study was done under the approval of the SUNY Upstate Medical University Committee on the Human Use of Animals (CHUA No. 209).

Confocal microscopy image acquisition setting
Cells were imaged with a confocal microscope (LSM510 META; Carl Zeiss, Germany) using LSM acquisition software (Carl Zeiss, Germany). Images were acquired with a Plan-Apochromat 63× oil immersion objective (NA 1.4). The pinhole was adjusted to obtain 1.24 Airy units for the fluorophore of shortest wavelength excitation/emission properties. mCherry fluorescence was detected by using an HeNe1 laser (excitation at 543nm, power 20-60%), a main dichroic beam splitter (MBS) HFT UV/488/543/633-nm followed with a dichroic beam splitter (DBS) NFT490-nm for excitation, and a 650/710-nm band pass filter. GFP fluorescence was detected using an Argon laser with an excitation line at 488 nm (power 0.5-2%), a MBS HFT UV/488/543/633-nm follow with a DBS NFT545-nm for excitation, and a 500/530-nm band pass emission filter. Hoechst 33342 staining was detected using a two-photon Chameleon laser with an excitation at 800 nm (power 4-8%), a MBS HFT KP650-nm follow with a DBS NFT490-nm for excitation, and a 435/485-nm band pass emission filter. For dual-colour acquisition, images were sequentially acquired in line scan mode (average line = 2).

Immunohistochemistry
Dark-adapted tadpoles were fixed with 4% paraformaldehyde in PBS overnight and processed for cryostat section and immunostaining as previously described [5].

Live Rod imaging
Xenopus were dark adapted for at least five hours before being euthanized for the experiment. The retina was isolated and cut into small pieces in oxygenated Ringer's solution, (111 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 0.5 mM MgSO 4 , 0.5 mM NaH 2 PO 4 , 3 mM HEPES, 10 mM glucose, 0.01 mM EDTA, pH 7.6). Portions of retina were loaded into a glass chamber (No. 1 coverslip affixed to the bottom of a Fisher 3 cm petri dish with a 3 mm milled hole) and then sealed with a No.1 coverslip [9]. A dim red light was used for all steps of tissue manipulation.

Calibration of fluorescence protein concentration with confocal microscope
Purified mCherry concentration was determined with a UVvisible spectrophotometer (Beckman, Brea CA) using the extinction coefficient 72 x 10 3 M -1 cm -1 at λ=587 nm [38]. The mCherry protein stock was diluted at various concentrations with Tris-HCl (pH 7.8-8.8) and then loaded into a 15 µl chamber. Fluorescence intensity was measured using the LSM confocal microscope with the same optical settings as described in live rod imaging. The mCherry measurements were obtained 10 µm away from the surface of the cover glass. The mCherry solution was imaged with the HeNe1(543 nm) laser at various power and gain values. The measured fluorescence intensity was plotted versus concentration at different laser power levels. We then used these plots to calibrate mCherry concentration from fluorescence intensity in images of live rods.

Analysis of confocal image data
Images from live rods were analysed with Image J software (NIH). Central axial z-sections of rods were extracted from stacks. Heat-maps of rods were plotted with "Heatmap from stacks" plug-in for Image J (http://www.samuelpean.com/ heatmap-from-stack). Fluorescence intensity of each rod along their axis were measured, constrained normalized with the maximum intensity set to 100% and plotted against the distance from IS/OS junction. The average Rho-mCherry fluorescence intensity distribution in OS from multiple rods was calculated as follows. First, the position corresponding to the 50% maximal intensity in the rising phase of each induction response in each rod was set to 0 µm. The rods were then aligned at this reference position. The fluorescent intensities from different induction responses were then averaged and constrained normalized. All intensity profiles, dot plots and bar graphs were generated with Sigma Plot 11.0 (Systat Software, Inc., Chicago, IL). The statistical analysis was done using Student's t-test in Sigma Plot 11.0 (Systat Software). Gaussian, exponential and sinusoidal curve fitting also used corresponding global fitting function in Sigma Plot 11.0 (Systat Software).

Dexamethasone induced expression: Cell culture
To determine the efficiency of regulation achievable by the G3U inducible system, we first utilized transfected HEK293 cells to measure the magnitude of induction and the leakiness of the UAS promoter. Cells were transfected with pUAS:Luc and either pCMV:eGFP (control) or pCMV:G3 for luciferase assay. Different concentrations of Dex (0-100 μM) were added into the medium and incubated for various periods of time (0-24 h) prior measurement of luciferase activity ( Figure 1B). In the absence of Dex, cells transfected with either pCMV:eGFP (control) or pCMV:G3 had no significant luciferase activity compared to untransfected cells. We observed robust responses (~150-fold induction) in cells transfected with pCMV:G3 when Dex was included in the culture media, even at the lowest concentrations tested (5 μM). We were able to detect expression of luciferase within 6 hours of Dex addition, and the expression steadily increased after that time. Thus, the G3U system exhibits tight control of expression and rapid induction in HEK293 cells. To study G3U translocation using confocal microscopy, we induced cells transfected with pCMV:G3 and pUAS:mCherry with 10 μM Dex and observed the fluorescence distribution pattern over time ( Figure 1C and 1D). The eGFP nuclear/cytoplasm intensity ratio equilibrated within 20 min of Dex addition, with a half time to reach a steady distribution was ~3 min at 37 °C, in close agreement to previously reported values of a GFP-GR protein [39]. Since we intend to utilize this system in transgenic Xenopus, which are housed at lower temperatures, we also measured the rate at 27°C ( Figure 1C and 1D) and found that it is 2.2-fold slower than at 37 °C. We also tested Rho-mCherry, which is suitable for measuring rhodopsin transport in transgenic Xenopus rods [5], for induction response in transfected cells. We first observed Rho-mCherry fluorescence in cells after 4 h of treatment (data not shown) with steady increases after 4 h ( Figure 1E). Thus, these preliminary experiments suggested that G3U system functions as designed in mammalian cells and thus be studied in transgenic Xenopus.

Dexamethasone Induced Expression: Transgenic Rods
We generated transgenic Xenopus tadpoles, iXRC, with two separately integrated transgenes: XOP:G3 and pUAS:Rho-mCherry ( Figure 2B). Rod-specific expression is accomplished using the Xenopus rhodopsin promoter (XOP) driving transcription of G3. Dex treatment of iXRC animals induces synthesis of Rho-mCherry that is transported to the rod outer segment (OS) disk membranes. The fluorescent disks are moving outward continually. Thus, the distance of disks from the base of the OS correlates to the time after induction ( Figure  2B). All animals exhibited Dex-dependent expression of Rho-mCherry and one male founder, iXRC1, was chosen for expansion and F1 animals were subjected to detailed imaging analysis. We produced tadpoles (Stage 54-58) from the iXRC1 founder and treated them with various concentrations of dexamethasone (0 to 500 µM) for seven days, during which time we did not detect any significant adverse effects on tadpole health (data not shown). Dexamethasone treatment had no obvious effect on organization of the retina or cell number, and the OS appeared normal (data not shown).
Transgenic tadpoles were examined for Dex induction response in whole retina. iXRC1 tadpoles were treated with 10 µM Dex for three days and placed in Dex-free water. After three more days, tadpoles were fixed, retina were cryosectioned and examined using DIC and fluorescence confocal microscopy ( Figure 3A). Dex induced the expression of Rho-mCherry in rods and the Rho-mCherry fluorescence was found primarily near the base of the OS ( Figure 3A, right panel). In contrast, there was no detectable Rho-mCherry in iXRC1 tadpoles only treated with DMSO for the same period of time ( Figure 3A, left panel). Epifluorescence micrographs show G3 translocated into nucleus after induction (data not shown). The majority of rods (at least 70%) in transgenic tadpoles responded to Dex induction (data not shown).
To quantify the expression level of the Rho-mCherry reporter, we used high-resolution confocal imaging on live rods [5,9,21,40]. Using constitutive promoters, several features of the Rho-mCherry distribution have been described in Xenopus rods that will influence the response profile to Dex ( Figure 2C) [5,40]. First, the rhodopsin transgene is synthesized and transported to the OS in a diurnal cycle producing an axial banding pattern with a ~1.5 μm periodicity in animals housed in 12 h dark /12 h light cycles [5]. Second, there is a long-term variation in transgene expression along OS axis with an approximate period of 7-10 days in animals housed in 12 h dark /12 h light cycles [5]. The transgene variation is unsynchronized between cells in the same retina and we call this stochastic variation, since it reflects temporal variation in transgene expression [5]. Thus, we expect a combination of diurnal banding ( Figure 2C, middle) and long-term variation ( Figure 2C, bottom). For example, in rods, which constitutively express Rho-mCherry under the XOP promoter, periodic axial banding and long-term variation are observed ( Figure 3B and  Figure 3B and 3D) and in isolated cells ( Figure  3C and 3E), but the latter gives the highest resolution images and were used for all the analysis to follow.
To examine background expression and induction response magnitude, we continuously treated tadpoles with 10 µM Dex for seven days ( Figure 2D) and sacrificed the tadpoles for live rod imaging immediately after the treatment. Non-transgenic cells had undetectable fluorescence intensity ( Figure 3F and 3G) while cells from transgenic that constitutively expressing Rho-mCherry showed fluorescence throughout the OS ( Figure  3B and 3C). Dex treatment did not alter the fluorescence intensity distribution in these two groups. In cells from iXRC1 animals, the fluorescence intensity in the distal (pre-induction) region was undetectable and equivalent to a concentration of Rho-mCherry of <0.2 µM (SD = 0.26, n = 68 rods) which is not significantly different from non-transgenic rods (p = 0.396, ttest). In rods from iXRC1 tadpoles that received 7-day Dex induction, Rho-mCherry fluorescence was only detected at the base of OS where newly synthesized OS membranes are located ( Figure 3D, 3E and 3H). Rho-mCherry extended 7-8 µm away from IS/OS junction, close to the distance expected from metabolic labelling studies for OS disk displacement over this period [16,17] (more details see below). It is possible to detect both types of axial variation in these examples (Figure 3  H).
There was a range in the maximal concentration achieved in these rods ( Figure 3I and 3J). The average peak Rho-mCherry concentration was 64.6 µM in rods received 7-day induction (SD=54.4, n=28 rods, Figure 3I and 3J), representing a >300fold increase of Rho-mCherry concentration after induction. This value represents a significant fold increase over the preinduction levels. However, since the uninduced Rho-mCherry fluorescence level is below our detection limit, we cannot confidently establish the magnitude of the fold increase; in most cells, it was >100. It is important to note that even though there is a large increase in Rho-mCherry concentration following Dex induction, the levels of Rho-mCherry are significantly lower than the 3 mM endogenous rhodopsin [41]. Rods that were treated for a shorter length of time had lower peak Rho-mCherry concentrations but had less variability ( Figure 3I).

Reproducibility of induction responses
To study the reproducibility of Dex responses in iXRC1 rods, we performed 3-day repetitive induction on transgenic tadpoles ( Figure 2E). In this experiment, animals were treated with 10 µM Dex for three days and then without Dex for five days, and then repeated twice more. Finally, the tadpoles were placed in water without Dex for seven days and then sacrificed for imaging. In most rods, there were three bands of Rho-mCherry fluorescence appear in the OS ( Figure 4B). These Rho-mCherry fluorescent bands reflect the inductions. They were equally separated by non-fluorescent areas. Because OS disk displacement is unidirectional outward [42,43], the outermost fluorescent band represents the first induction. The fluorescence of Rho-mCherry along the axis of OS was measured and profiled with the distance to OS base (defined as OS/IS junction) to examine kinetics and compare maximal intensity of the various induction responses ( Figure 4C). We also found rods with fewer Rho-mCherry bands (see below) in cells from the same retina. To compare the reproducibility of the induction responses for repetitive treatments, we chose rods that had two to three detectable responses, which comprised approximated 70% of all responding rods. We normalized the Rho-mCherry induction responses to its peak fluorescence intensity and then averaged them ( Figure 4D and Figure S1). There were close overlap in all responses during most of the response period ( Figure 4D). However, most responses do not reach pre-induction fluorescence levels during the five-day resting period between inductions and it took seven days for the Rho-mCherry levels to return to preinduction intensities after the last induction. The Rho-mCherry concentration in the troughs between responses ( Figure 4C) was 2.1 μM (SD = 2.1, n = 117) but still above F 0 . These experiments show that iXRC1 rods are able to reproducibly respond to Dex treatment over several weeks.
We examined the reproducibility of the peak magnitude of the induction between the first, second and third inductions. We measured the fluorescence intensity in the distal part of OS (F 0 ) (arrow in Figure 4C) of all rods, which indicates the background expression level of Rho-mCherry before induction, which is <0.22 µM (SD = 0.03, n = 68) ( Figure 4E) and is at the same level as in non-transgenic rods. In the first induction, the peak response was 13.9 µM (SD = 9.4, n = 61), representing a 130fold increase over F 0 ( Figure 4E). The second and third induction responses had Rho-mCherry peak concentrations of 19.2 (SD = 13.6, n = 66) and 16.1 µM (SD = 12.0, n = 45), respectively ( Figure  4E). The peak Rho-mCherry concentrations in the three induction responses were significantly above F 0 (P < 0.001) but not statistically different from each other (one way ANOVA, alpha=0.05). Across all responses, the mean peak response was 23.8 µM (SD = 13.1, n = 68), a 222-fold increase over F 0 (Figure 4 E). Altogether, the activation and inactivation response profiles of G3U are reproducible in most iXRC1 rods. However, in some rods, there were fewer than three responses ( Figure 5A and Figure 5B). The shapes of the responses were not affected, as the remaining responses overlay well ( Figure 5C). The most likely reason for the lack of responses to some but not all Dex treatments is the long-term variation in transgene expression, which is stochastic and not correlated between cells in the same retina ( Figure 2E). Thus, while the response kinetics and peak magnitude were similar in all responses, iXRC1 rods exhibits transgene variation that in turn can have a significant influence on individual responses. The quantitative impact of the long-term variation in transcriptional activity is considered further in the modelling section.

Correlation of spatial and temporal induction response profile
We measured the axial position of the Rho-mCherry fluorescent bands in OS of rods following repetitive 3-day inductions ( Figure 6A). The average location of the first induction peak is 24 µm (SD = 5.1, n = 44) from the OS base, the second response is 16 µm (SD = 3.5, n = 47) and the third is 8 µm (SD = 3.1, n = 42) ( Figure 6B). The first response had a slightly larger variance, which is probably due to OS stretching or swelling during ex vivo imaging. Because all rods received the same Dex treatment, we can attribute each Rho-mCherry fluorescent band to a specific induction period and thus estimate the time for the Rho-mCherry band to migrate from OS base to these positions. The fluorescent bands showed a strong linear correlation of distance along the OS and time of induction (R 2 = 0.99), with a slope of 1.0 µm/day (SE = 0.3) ( Figure 6B) and an intercept on the time axis at 1.9 days. There is some uncertainty in this measurement because of the uncertainty in determining accurately the position of response initiation. To better estimate disk movement rates, we also analysed the peak-to-peak distances for sequential responses. The distribution of peak-to-peak distance passed the normal distribution test ( Figure 6C) and had an average distance between peaks of 8.0 µm (SD = 2.4, n = 72), which encompassed a period of 8-day (three day treatment with dexamethasone and a five day recovery interval). These results suggest that the daily disk displacement towards the OS tip is 1.0 µm, which matches the estimation of daily OS displacement rates in Stage 54 Xenopus raised at 20°C in 12 hours dark/light cycle [13,16,17,22,[44][45][46]. The disk displacement rate among rods is very close, with a standard deviation 0.3 µm/day.

Activation phase of the induction response
We investigated the kinetics of activation by analysing the rising phase of the induction response and response peak position in iXRC1 tadpoles that are treated with Dex for seven days prior to imaging. The induced rods were imaged and analysed for profiling the information of fluorescent intensity and distance (Figure 7 A and B). The fluorescence intensities of different rods were normalized and profiles were aligned. We found two groups of rods, group one with an early termination profile ( Figure 7A) and group two with a prolonged response ( Figure 7B). Both groups had a similar activation phase ( Figure  7C) which was also similar to that observed in 3-day repetitive induction experiments ( Figure 7D and 7E). In fact, even rods treated for either one or two days showed the stereotypical activation shape ( Figure S3).
To estimate the time for the induction response to reach its peak, we measured the distance between the initiation of the response and the peak of Rho-mCherry fluorescence intensity ( Figure 7C). We defined the initiation site as the point where average Rho-mCherry fluorescence intensity was >2 standard deviations above the background fluorescence ( Figure 7C and Figure S2A). In 7-day inductions, we found an average distance of 3.0 µm (SD = 1.0, n = 28) ( Figure S2B) while for the first response in rods treated repetitively for 3-days, the average distance to reach peak fluorescence was 3.2 µm (SD = 1.0, n = 33) ( Figure S2E). Together, the average distance to reach peak fluorescence in rods was 3.2 µm (SD = 0.96, n = 61). Thus, using the estimate of disk displacement rate 1.0 µm/ day, it appears that Rho-mCherry expression takes approximately 3 days to reach the peak concentration following Dex treatment. Thus, there is significant delay in achieving maximal rates of Rho-mCherry incorporation into the OS. This delay may result from a combination of pharmacokinetics of Dex in rods and kinetics of transgene expression (See Discussion). Based on the above rate of disk displacement, the first detectable fluorescence in 7-day treated rods should be ~7 µm from the IS/OS junction. The measured distance was less at 6.5 µm from IS/OS junction (SD = 1.2, n = 28) ( Figure 7F). The 0.5 µm difference translates to ~0.5 day suggesting this as an estimate for the delay between initiation of Dex treatment and the appearance of detectable Rho-mCherry assembled in OS.

Inactivation phase of the induction response
We investigated the inactivation kinetics which is the falling phase of the induction response from the peak intensity. As mentioned above, rods that received 7-day induction exhibited highly variable inactivation falling phases. Group I had responses that had returned substantially to baseline levels even though Dex was still present ( Figure 7A and C). This group had very similar inactivation phases to 3-day induction rods ( Figure 7D). By contrast, Group II had more sustained responses ( Figure 7B and C). Nonetheless most responses in this group had reductions in fluorescence intensity after three days ( Figure 7E). To estimate the time for the induction response to return to baseline, we measured the distance between the peak of Rho-mCherry in OS to the basal level (less than 10% of peak intensity) ( Figure S2A). The 3-day induction rods have an average inactivation phase of 5.5 μm (SD = 1.9, n = 26) ( Figure S2F). This result indicates additional 5-7 days after the peak response is required for the rods return to pre-induction status. All average inactivation phases from different induction paradigms could be fit by an exponential decay function (R 2 > 0.97) (Figures S3 and 7G). The inactivation phase for the 3-day induction has an estimated decay rate of 0.40 μm -1 . This decay rate suggests that these rods took more than 5.6 days to return to 10% peak fluorescence intensity, which is consistent with above estimations. The decay constant calculated for each induction paradigm ( Figure 7H) shows a positive correlation between length of treatment with Dex and the half time for decay. This suggests that there may be some process regulating the recovery of the system that is affected by long-term treatment with Dex.

Discussion
We have developed an inducible expression system in Xenopus and implemented this system for rod-specific expression, characterizing the magnitude and kinetics of induction and recovery in a cell that uniquely records reporter expression levels for weeks. We showed that the G3U system exhibits very tight control, having no detectable expression in the absence of Dex yet exhibiting peak concentrations of membrane protein reporter of >10 µM after induction. Besides the iXRC1 line characterized here, we also found similar induction responses in four other iXRC transgenic lines (data not shown). Furthermore, the G3U system allows repetitive inductions with quantitatively similar responses in a majority of cells.
This inducible system also provides a new approach to study rod OS disk displacement. Here, we induced Rho-mCherry expression, whose history was recorded along the OS axis. We found a displacement rate 1.0 µm for tadpoles at stage 54-56 (raised in 20°C with 12 h light/12 h dark cycle). Traditional methods studied disk displacement using pulse-chase methods with radioactive amino acids to label newly synthesized protein followed by autoradiography [42,43]. Our method allows us to use much higher resolution of microscopy via live imaging. Disk displacement is very sensitive to temperature and varies from 0.65 µm/day at 18 °C to 2.4 µm/day at 28 °C [47]. Our results agree well with those values. However, other membrane proteins cannot be easily studied using radiolabeling methods because of their low abundance and the lack of specificity of the radiolabeling approach. Thus, the G3U system opens the way to extend work testing other genetically altered membrane proteins via induced expression.
We used a mathematical model to simulate induction response and to study potential reasons for the variation in the inactivation phase (Figure 8 A). The system produces detectable Rho-mCherry in hours after treatment of animals with dexamethasone. However, it is slower at reaching steady state and recovery than expected, typically taking several days to reach peak synthesis rate and longer to return to baseline after removal of inducer. The reasons for the relative slow rate to reach peak synthesis are unclear. Dex enters the eye equilibrates in the retina within hours in mammals [48] and it has a relatively long half-life (36-54 h) [49] so a steady level of Dex should be reached and maintained relatively rapidly. Although we were not able to use live imaging to determine the rate of transfer of G3 into the rod nucleus, we estimate that the half time for G3 transport is less than 1 hour at 20°C base on the cell culture results ( Figure 1B). G3 binds cooperatively to UAS and activates transcription at a concentration as low as 5 μM ( Figure 1B) and the rate of Gal4 binding is, ~10 min [50,51]. Rho-mCherry is transported from the Golgi to OS within 1-2 h [20]. Thus, it appears that the concentration of G3 is an important determinant of the induction rate and suggests that the long-term variation in G3 levels could limit the rate of Rho-mCherry production. The long-term variation of transgenic expression in most rods constitutively expressing Rho-mCherry can be fit with a sinusoidal function ( Figure S4). Thus, we explored the implications of this type of variation on the induction response using a simple linear model.
In this model ( Figure 8B, Appendix S1), transcription and translation were combined to a single synthesis step to yield a model with only five parameters: variable sinusoidal synthesis of G3, concentration of Dex, rate of synthesis k act and two degradation steps γ cyto and γ nuc for degradation (Details in Appendix S1). A random phase in the sinusoidal function was selected to represent Rho-mCherry expression unsynchronized among rods in the same retina [5,52]. Since induction may occur at any point in the varying G3 expression cycle, we expect a variety of Rho-mCherry induction response shapes ( Figure 8C). The shapes of induction responses were sensitive to the G3 half-life used ( Figure S5). Since we were not able to measure the G3 degradation rate, we used a half-life of four hours based on pulse-chase measurement of Gal4-VP16 degradation rates [53]. This value also generated relatively stable responses in our simulation. Using these parameters, we found that application of Dex during the phase of G3 cycle when concentrations are increasing generates a single peak. Conversely, when induction starts at later points of the cycle where G3 concentrations are falling, several weaker peaks are generated ( Figure S5). When we averaged many induction responses that were generated at random phases of the G3 cycle and aligned them at the 50% peak intensity, we found an induction response with fast activation and slow inactivation phase which was similar to the actual induction responses in Xenopus rods ( Figure 8C represent 7-day induction). The rate of inactivation was relatively insensitive to the length of Dex treatment, as was found in the experimental results. Thus, the stochastic long-term variation in G3 expression could explain the basis for the shape of the induction response.
Many regulatory proteins expressed in the vertebrate eye have functions early during development, cellular differentiation and later maintaining normal function in the adult. Disturbance in the structure or function of the protein may alter both differentiation and cell maintenance. In order to overcome this experimental difficulty for studying transcription factor function in adult rods, a tightly controlled inducible system is needed. The system described here has potential utility for these future studies. Average fluorescence distribution in rods from iXRC1 tadpoles that received 1-day, 2-day, 3-day and 7-day induction. The average line of 3-day induction was drawn in other three plots for comparison (red). Error bars are 95% confidence levels.
Long-term and diurnal variation in rods constitutively expressing Rho-mCherry . A rod constitutively expressed Rho-mCherry shows considerable axial variation in the fluorescence intensity distribution (Upper panel). The axial fluorescence intensity profile of Rho-mCherry along the axis of OS is shown below (black line). The smoothed fluorescence intensity profiles (red line) had two components that could be isolated. First the long term variation (green line) can be fit with sinusoidal function (black dashed line) and the diurnal variation (pale green line) which is less well fit by a sinusoidal function with a shorter period. This rod was from an animal housed at 22 °C and has a faster disk displacement rate than those in the Dex experiments. (TIF) Figure S5.
Simulation of induction responses with different G3 half-life. Simulation of induction responses with 10-day induction and a G3 has a 10-day sinusoidal expression pattern. The G3 expression levels (red line) are shown for four unsynchronized hypothetical rods as a function of time. Note that Dex induction (blue line) occurs at different phases in the G3 expression cycle. The calculated Rho-mCherry expression level is shown (black line) as a function of the G3 degradation rate. The gray indicates a duration of 7 days to aid in comparisons with experimental results. (TIF)