A novel fluorescent pH probe for expression in plants
© Schulte et al; licensee BioMed Central Ltd. 2006
Received: 19 January 2006
Accepted: 06 April 2006
Published: 06 April 2006
Skip to main content
© Schulte et al; licensee BioMed Central Ltd. 2006
Received: 19 January 2006
Accepted: 06 April 2006
Published: 06 April 2006
The pH is an important parameter controlling many metabolic and signalling pathways in living cells. Recombinant fluorescent pH indicators (pHluorins) have come into vogue for monitoring cellular pH. They are derived from the most popular Aequorea victoria GFP (Av- GFP). Here, we present a novel fluorescent pH reporter protein from the orange seapen Ptilosarcus gurneyi (Pt- GFP) and compare its properties with pHluorins for expression and use in plants.
pHluorins have a higher pH-sensitivity. However, Pt- GFP has a broader pH-responsiveness, an excellent dynamic ratio range and a better acid stability. We demonstrate how Pt-GFP expressing Arabidopsis thaliana report cytosolic pH-clamp and changes of cytosolic pH in the response to anoxia and salt-stress.
Pt- GFP appears to be the better choice when used for in vivo- recording of cellular pH in plants.
Fluorescent proteins have revolutionized the understanding of cellular event cascades, signal transduction, and structure dynamics [1, 2]. The green fluorescent protein from Aequorea victoria (Av-GFP) is the most popular species used by scientists to date. Av-GFP and its corresponding cDNA has been altered many times to give fluorescent proteins of higher quantum efficiency, different spectral characteristics, less temperature sensitivity, improved solubility, and higher expression levels in other organisms [e.g. [3, 4]]. Enhanced variants of Av- GFP are frequently used to decorate cellular structures and proteins in order to observe shape, location and dynamics in vivo [e.g. [5–12]] or to visualize gene expression and/or activity of promoters or enhancers .
One of the advantages of GFPs is their ability to be engineered to indicators for cellular signal transduction studies [e.g. [14–16]]. Engineered GFPs have been used in plants to report cellular concentrations of Ca2+, H+, Cl-, and NO3- [e.g. [17–23]]. No loading of the indicator is necessary with GFP-based probes and they can be precisely targeted to almost any organelle, compartment or tissue in question [e.g. [8, 9, 11, 12, 24]]. This potentially makes GFP-derived probes superior to small molecular weight fluorescent dyes used as ion indicators provided they can replicate the sensitivity and responsiveness of these probes.
In particular two pH-sensitive variants of Av-GFP (so-called pHluorins) have been engineered . These two reporters are called 'ratiometric' and 'ecliptic' pHluorin. They both allow ratiometric in vivo-pH recording. Ratiometric pHluorin is a double excitation indicator whereas ecliptic pHluorin has to be used in the double emission mode . The amendments necessary for sufficient expression of Av-GFPs in plants (i.e. removal of the cryptic intron, changes to A. thaliana codon usage and improvement of solubility) have been combined with the properties of pHluorins. The resulting pH indicators have been successfully expressed and used in Arabidopsis [21, 26, 27].
However, more and more fluorescent proteins (FPs) from other marine organisms are being discovered with other interesting properties [28–35]. Some of these newly discovered FPs unveil advantages when compared with Av-GFP variants. We have expressed the GFP from the orange seapen (Ptilosarcus gurneyi) in bacteria and plants. Here, we compare ratiometric properties of Pt-GFP with those of pHluorins and also with conventional fluorescent dyes often used for ratiometric pH measurements in vivo.
The spectrum of a fluorescent ratiometric indicator is in a first approach mainly the sum or overlap of two spectra : First, the spectrum of the free indicator (in case of pH indicators the de-protonated form) and second the spectrum of the bound (protonated) indicator molecules. There are further two different effects on the fluorescence of pH indicators which need to be distinguished when the pH is lowered: First a fluorescence quenching at all wavelengths. Second, the spectral disproportionation (i.e. the attenuation of the spectrum from de-protonated indicator for the benefit of the fluorescence spectrum formed by the protonated molecules). Ratiometric fluorescence measurements have been established to cancel down all side-effects based on variations in indicator concentration, illumination intensity, detector sensitivity etc. [37, 38]. So, only the second effect, namely spectral disproportionation is essential and relevant for ratiometry and, in theory, the ratio solely correlates with the analyte concentration (here [H+]). A quenching at all wavelengths can be considered an apparent decrease in indicator concentration and is thus irrelevant for ratiometry.
Here all spectra are presented normalized by their area for three reasons: First, this is a way to uncover all spectral effects relevant for ratiometric measurements. Second, it is the optimal way to present the potential capabilities of the indicator (i.e. best pair of wavelengths, dynamic ratiometric change etc.) when intended for ratiometry and/or ratio imaging. Third, it does not require an a-priori- knowledge of the isosbestic point.
By calculating the so-called minimax spectrum (i.e. the spectrum that is obtained when the minimum fluorescence is substracted from the maximum fluorescence at each wavelength within a scanned set of spectra) it is possible to derive two important indicator characteristics: First, the real isosbestic point is the wavelength where the fluorescence is independent from spectral disproportionation. Here, the minimax spectrum, in theory, is zero with discontinuous slope, but, in practice, approaches a minimum close to zero. Second, the sensitivity of the indicator which is defined here as the integral of the minimax spectrum. The sensitivity is a number in the range between zero and two. It is two when there is no spectral overlap of the protonated and the de-protonated form of the indicator (ideal ratiometric indicator). It approaches zero when the indicator is less suitable for ratiometry. Consequently, the normalized spectra give the real isosbestic point while in the raw data spectra (not given here) this point is shifted by the superposed quench effect and is then distinguished here as 'apparent' isosbestic point.
The isosbestic points λiso (precisely, the isoexcitation points in case of a double excitation probe or the isoemission points in case of a double emission probe) are the wavelengths where the fluorescence is independent of the indicated analyte (ion) concentration. Here, the real isosbestic point is distinguished from the apparent isosbestic point.
Spectral peaks or shoulders left (λ1) and right (λ2) of the isosbestic point λiso which vary in opposite direction when the analyte concentration is changed.
The maximum fluorescence wavelength (λmax) is the peak in the emission spectrum in case of a double excitation indicator (such as ratiometric pHluorin and Pt- GFP) and the maximum in the excitation spectrum in case of a double emission indicator (such as ecliptic pHluorin).
The maximum and minimum ratios, Rmax and Rmin (i.e. the ratios taken in the absence of the analyte or when the indicator is saturated with analyte).
The apparent pK (i.e. midpoint of the calibration curve where the ratio reaches half maximum between Rmin and Rmax) is dependent on the ratiometric wavelengths chosen (see supplemental data in additional file 1). Ideally, the apparent pK is about the dissociation constant which is defined by the analyte concentration where the indicator is half-saturated.
The useful concentration range (x-axis range) in which the indicator is reasonably applied and where the logarithm of the ratio depends approximately linear on the logarithm of analyte concentration. In the case of pH-indicators this range is given by the pH values where the straight line through the midpoint of the logarithmic calibration curve intersects with log(Rmax) and log(Rmin) (Figure 3).
The sensitivity S of the indicator is defined here (see introduction section) by the area or integral of the minimax spectrum. The sensitivity is excellent when S > 0.5 (Table 1) and it is negligible when S < 0.1 (Table 4).
Spectral characteristics of recombinant pH-probes. The real isosbestic point is derived from normalized spectra whereas the apparent isosbestic point is from the raw data. The Stokes-shifts are defined here by the difference between the wavelength of the major (and the minor) absorption peak and the wavelength of the emission peak. Index 'x' of S designates sensitivity calculated from excitation spectra, and index 'm' values are calculated from emission spectra. 'em', 'ex', and 'abs' designate wavelengths in the emission spectrum, the excitation spectrum, and in the absorption spectrum, respectively.
real isosbestic point λiso
426 nm (ex)
495 nm (em)
437 nm (ex)
apparent isosbestic point
428 nm (ex)
489 nm (em)
430 nm (ex)
left peak WL (λ1 < λiso)
395 nm (ex, abs)
464 nm (em)
390 nm (ex)
right peak WL(λ2 > λiso)
475 nm (ex)
511 nm (em)
502 nm (ex)
Stokes-shift (λem - λabs)
113 nm (33 nm)
114 nm (35 nm)
maximum WL (λmax)
508 nm (em)
397 nm (ex, abs)
508 nm (em)
Rmax - Rmin
apparent pK (λ1; λ2)
pH sensitivity (Sx)
pH sensitivity (Sm)
Comparison of camera exposure times needed for ratiometric pHluorin and Pt-GFP when excitation ratios R(F475; F390) are taken in plants under otherwise identical settings.
Comparison of ratiometric pHluorin an Pt-GFP when excitation ratios R(F475; F390) are taken. The optimal pH-range is calculated here from a double log-plot as shown in Figure 3
Rmax - Rmin
apparent pK (390 nm; 475 nm)
4.8 < pH < 7.6
3.8 < pH < 8.2
When doing ratio imaging, a maximum dynamic fluorescence ratio range (i.e. a maximum fold fluorescence ratio increase Rmax/Rmin) is desired in order to gain an optimal signal to noise ratio. However, the experimental conditions – in particular the spectral characteristics of the available filter set and/or dichroic mirror – are often not optimised for the ratiometric probe in use or the Stokes-shift of the indicator is too small for reliably separating fluorescence emission from excitation light. In case of Pt- GFP, for instance, the Stokes-shift is just 6 nm (i.e. right excitation peak λ2 = 502 nm; maximum emission λmax = 508 nm; see Table 1). This is too close to be separated by conventional microscopic filter sets. Hence, wavelengths other than those giving the maximum dynamic ratio range are compulsorily chosen. Thereby, it should be kept in mind that responsiveness and midpoint of the calibration curve (apparent pK) depend on the two wavelengths chosen for ratio measurements. This effect is demonstrated as supplemental data in additional file 1 (Figure S1, Table S2).
For in vivo comparison of ratiometric pHluorin and Pt- GFP we used the F475 nm/F390 nm pair for excitation with an emission range between 510 nm ≤ λem ≤ 560 nm. In Table 2 the optical properties for this particular pair are listed for comparison.
Both, ratiometric and ecliptic pHluorins do not differ significantly in their spectra when other proteins are fused to the N- and/or the C-terminus . This is also true for Pt- GFP (unpublished observations). This property is important since fusions with transit or signal peptides are often used to specifically target the indicator to subcellular locations.
For direct comparison of all three pH-probes the ratios where plotted on a log-log-scale (Figure 3). This allows to determine the area of best indicator responsiveness (dynamic ratio range vs. dynamic pH range). The diagram (Figure 3) clearly shows that Pt-GFP has the best responsiveness (Δlog(R)·pH = 9.8) and the broadest pH-application range. The responsiveness is lower with ratiometric pHluorin (2.9) and ecliptic pHluorin (2.0). Pt- GFP responsiveness also exceeds that of conventional pH-indicators (Table S1 in additional file). However, Aequorea GFPs have a better sensitivity than Pt- GFP (Table 1) but the sensitivities of all three recombinant indicators are of similar magnitude (0.65 < S < 1) when compared with conventional dyes (Table S1).
The fluorescence quench at all wavelengths by low pH has already been mentioned above. This is due to reversible protonation of Av- GFPs in the range 7 > pH > 5 and by irreversible conformational changes leading to protein instability in the range pH < 5 [39, 40]. The latter effect is undesirable when GFPs are used as pH-probes in plants. The apoplast of plant cells is usually acidic (pH < 6.5) [21, 41, 42], and also some vacuoles have low pH. Thus, cytoplasmic pH changes can be drastic in plants (see e.g. in vivo experiments below). Therefore it is good to have a pH indicator with high acid stability and ratiometric responsiveness in the lower pH range.
The predicted protein masses of Pt-GFP and pHluorins are approximately 27 kDa. We confirmed this by denaturating SDS PAGE (data not shown). However, when native proteins were run on FPLC different masses were detected. Pt-GFP exhibited a mass of approx. 105 kDa, whereas pHluorins were detected at around 55 kDa. This indicates the formation of dimers in case of pHluorins and of tetramers in case of Pt-GFP.
Cross-sensitivities of recombinant pH-probes. Index 'x' designates sensitivity calculated from excitation spectra, and 'm' calculated from emission spectra
However, quantum efficiency or brightness of GFPs cannot be directly compared in vivo since a lower quantum efficiency could be compensated by a higher expression level. But, the excitation energy or dose is crucial for practical work because it may lead to photodamage or -bleaching when too high. Therefore, we compared exposure times necessary to reach a reasonable signal at the two wavelengths used for in vivo ratiometry (Table 4). CCD camera-based in vivo ratio imaging systems allow wavelength-independent adjustments of exposure times.
In Figure 2 absorption is normalized by the protein concentration (i.e. by A280). This allows direct comparison of absorption in the visible and demonstrates that Pt- GFP has a higher peak absorption here than pHluorins. This promises the need of lower excitation energy. But its very low absorption at 390 nm requires the F390 nm signal having threefold the exposure time of the F475 nm signal to be in the optimal range (Table 4). Fluorescein derivatives like BCECF and FITC have similar asymmetric spectra like Pt- GFP and also require appropriate adjustments. For AtpHluorins in contrast such asymmetric adjustment of exposure times is not needed because of the greater symmetry in their spectra. Hence, the higher quantum efficiency of Pt- GFP does not necessarily carry forward in a lower excitation energy needed for balanced signals. A way to circumvent this may be to choose instead of 390 nm a wavelength at or closer to the isosbestic point.
Microenvironmental parameters such as viscosity, hydrophobicity, protein mobility, and binding interactions  as well as spectral imbalance can be attributed to shifts in the response of an indicator when going from in vitro (spectrometer) to in situ or to in vivo (microscope) recording. Noticeable here is a shift of the apparent pK when comparing Figures 1B and 6C. However, this is not disastrous as long as recordings of calibration data and in vivo data are identical.
Pt-GFP is readily expressed by Arabidopsis without any cDNA modifications (Figure 5). Although the cDNA is derived from a distinct and totally unrelated organism, the codon-usage is accepted by Arabidopsis and the plant constitutively expresses the protein with high yield under the control of a single 35S-promotor. The use of Av-GFP in higher plants, in contrast, was initially limited. Alterations of the codon usage and the removal of a cryptic intron were found necessary to express Av-GFP in Arabidopsis .
Pt-GFP is readily soluble and distributes well when expressed in the cytoplasm of plants (Figure 5) whereas for Av- GFPs modifications where found beneficial to increase protein solubility in the plant cytoplasm .
The fluorescence excitation ratio of Pt- GFP has a maximum dynamic range (Rmax - Rmin) 15 times and a maximum ratio increase (Rmax/Rmin) three times higher than that of ratiometric pHluorin when using excitations at 390 nm and 475 nm (Table 2). It further has a broader area of responsiveness (Figure 3) and thereby also exceeds conventional fluorescein derivatives used for ratiometric in vivo pH-measurements (supplemental data in additional file Table S1).
Pt-GFP is much more robust at low pH (Figure 4). This makes it also suitable for monitoring pH in acidic subcellular compartments or under conditions when the cellular pH is shifted towards the acidic.
Taken together, Pt-GFP is an excellent pH indicator for excitation fluorescence ratio imaging and in some respects superior to pHluorins when used in plants.
DNA coding for Pt- GFP (Acc.No. AY015995) has been subcloned from a pUC18 vector (Nanolight Technologies, Pinetop, AZ, USA) into the bacterial expression vector pRSETb (Invitrogen GmbH; Karlsruhe, FRG). Protein production was induced with 1 mM IPTG when OD600 = 0.6 and expressed at 20°C/300 rpm over night (i.e. 15 h). For protein isolation bacteria were cracked by sonification (HD2200&MS73, Bandelin, Berlin, Germany) in 200 mM phosphate buffer (pH 7.5). Bacterial lysate was pre-cleared at 4,000 × g for 2 h at 4°C. Remaining debris was removed from the supernatant by filtering through a 0.45 μm nylon filter. The 6xHis-tagged fluorescent protein was purified and concentrated through a Ni2+/NTA-agarose column (Qiagen, Hilden, Germany). Gel filtration through a NAP-25 column (Pharmacia Biotech, Freiburg, Germany) was performed to remove imidazol from the eluted protein. The purified indicator proteins were assessed spectroscopically. Fluorescence spectra (Figure 1) of the proteins were taken with a fluorescence spectrometer (F-2500, Hitachi) in 150 mM KCl and 50 mM appropriate organic buffers (Mes, Pipes, Hepes, Taps) adjusted to the desired pH with NaOH. Absorption spectra (Figure 2) were taken in phosphate buffer (pH = 7.5) with an absorption spectrometer (2100, Hitachi).
For FPLC protein was bound to a 0.5 ml column of Toyopearl resin (AF-Chelate-650 M; Tosoh Bioscience) washed with Tris/HCl-buffer (50 mM Tris/HCl pH = 8) and treated with Enterokinase (EKMax, Invitrogen) for 24 h at RT. Protein with His-Tag cleaved was washed from the column with 2 ml 1 × PBS and used for fast protein liquid chromatography (FPLC). Briefly, proteins were subjected to FPLC at 4°C with an HiLoad-16/60 Superdex200 column (Amersham Biosciences), equilibrated with 1 × PBS (Medicago AB; Uppsala, Sweden), pre-fitted with a column guard, and driven by a HPLC pump (Äkta-Explorer; Amersham Biosciences) at a flow rate of 1 ml/min. The column was calibrated using a mixture of four proteins of known molecular mass, i.e. catalase (232 kDa), aldolase (158 kDa), chymotrypsinogen A (25 kDa), and ribonuclease A (13.7 kDa).
pHluorins for expression in plants were constructed as described in Gao et al. . Pt-GFP cDNA (Nanolight Technologies, Pinetop, AZ, USA) was expressed in ecotype Columbia-0 of Arabidopsis thaliana under the control of the CaMV 35S promoter using the pART7/pART27 cloning/expression system . Full functionality of Pt-GFP in pART7 was assessed by biolistic bombardment and transient expression of young Arabidopsis plants (ecotype Col-0) before subcloning the cDNA cassette from pART7 into the binary vector pART27. For agrobacterium-mediated transformation of Arabidopsis thaliana (Col-0) the floral dip method  was applied.
Transient and stable expressions of GFPs (Figures 6) were assessed by CLSM as described  using a Leica TCS SP confocal laser scanning system. For Pt- GFP excitation the 476 nm beamline of the Argon laser was chosen; emission at 500–540 nm (green channel) for GFP fluorescence and 600–660 nm (red channel) for chlorophyll autofluorescence; HC PL APD objective (40× oil).
For assessment of acid stability (Figure 4) and for in situ-calibration of the pH-indicators (Figure 6) protein was bound to Ni2+-agarose beads (Qiagen, Hilden, FRG). Fluorescent beads were sandwiched between two layers of cellulose (Cellophane) and dialysed on the microscope against the buffer solutions indicated in figures.
For in vivo recording of fluorescence ratios (Figures 7, 8, 9) transgenic Arabidopsis were grown in 9 cm Petri dishes on vertical agar as described  and used when 6 to 14 days old. Cytoplasmic pH was measured in the hairy root segments near the hypocotyl. Experimental conditions, perifusion technique, and fixation of plant material were described previously . Roots were placed in a volume of 1.6 ml and perifusion flow was adjusted to 2.4 ml/min. The perifused buffer contained KCl, MgCl2, and CaCl2, 0.1 mM each and 5 mM MES/NaOH adjusted to pH = 5.4. For pH-clamp this buffer was supplemented with different concentrations of sodium butyrate as indicated by the top bar of Figure 7.
Fluorescence imaging was performed essentially as described [21, 23]. Briefly, fluorescence images at excitation wavelengths of 475 nm and 390 nm were taken every 12 s with a ratio imaging system from TILL-Photonics http://www.TILL-photonics.de fitted to an inverted microscope (Diaphot, Nikon) using light from a monochromator (Polychrome IV, TILL). For the emission path a filter block with beamsplitter 500 dcxr and emission filter HQ535/50 (AHF-Analysentechnik, Tübingen, Germany) was used. TILL software (TILLVision 3.3) was used for processing raw data. The fluorescence ratio F475/F390 was taken with Pt-GFP and the ratio (F390/F475) was taken with ratiometric pHluorin as a measure for pH.
Each spectrum is normalized by its integral (i.e. the sum of fluorescence values over all wavelengths λ). The minimax spectrum Fminimax(λ) of a set of spectra is determined by substracting the minimal fluorescence from the maximal fluorescence at each wavelength within the obtained set of spectra (i.e. within the scanned analyte concentration range). The isosbestic point (λiso) is determined by looking for the minimum in the minimax spectrum. Ideally, the minimax spectrum is zero at the isosbestic point (i.e. Fminimax(λiso) = 0). The sensitivity S is defined here by the integral of the minimax spectrum. The sensitivity S of each reporter protein is calculated for the set of its excitation spectra (Sx) as well as for its corresponding set of emission spectra (Sm). The Boltzmann fit has been chosen here for fitting sigmoidal curves to calibration data since the Boltzmann equation can directly be derived from the Grynkiewicz equation  describing the relation of analyte concentration on fluorescence and fluorescence ratios. The fit parameter of the Boltzmann include Rmin, Rmax, and the apparent pK of the calibrated indicator. Fitting has been performed using Origin 7.0 (OriginLab Corp., Northhampton, MA, USA).
Seeds from Pt- GFP expressing Arabidopsis are freely available from the European Arabidopsis Stock Centre (Nottingham, UK; http://arabidopsis.info/). All other novel material described in this studies can be obtained for non-commercial purposes from the corresponding author on request.
At = Arabidopsis thaliana, CLSM = confocal laser scanning microscopy, DTT = di-Thiotreitol
GFP = green fluorescent protein, Hepes = N-2-hydroxy-ethyl-piperazine-N'-2-ethane-sulfonic acid, Mes = 2-[N-morpholino]-ethane-sulfonic acid, PBS = phosphate buffered saline
We are grateful to Bruce Bryan of Prolume Ltd. (Eagar, AZ, USA) for the generous gifts of Pt- GFP cDNA and Toyopearl resin. We thank Gero Miesenböck and James Rothman (Memorial Sloan-Kettering Cancer Centre, New York) for the generous gift of pHluorin cDNA and Jim Haselhoff (Cambridge University) for the binary vector pBINm-gfp5-ER. Also many thanks to Dana Schöneberg and Steffi Schnell for technical assistance and to Hartmut Kaiser for critically reading the manuscript.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.