Pronounced cytoplasmic pH gradients are not required for tip growth in plant and fungal cells

Tip growth is the predominant method of cell growth in fungi, and is also exhibited by a number of specialised plant cell types such as pollen tubes, rhizoids and root hairs. Tip growth involves the polarised extension of a walled cell in which the increase in cell length is restricted to a narrow region of a few micrometers at the cell apex (Heath, 1990). Much recent research has focused on how this important mode of growth is regulated.

It is now accepted widely that both intracellular and extracellular ion gradients, notably of H⁺ and Ca²⁺, are associated with polarised growth in plant and fungal cells (Harold and Caldwell, 1990; Jackson and Heath, 1993). Extracellular pH (pH_ext) gradients are commonly associated with the apices of tip-growing cells (Harold and Caldwell, 1990; Gibbon and Kropf, 1991) and cytoplasmic pH (pH_cyt) gradients have also been reported (Turian, 1979; Roncal et al., 1993; Gibbon and Kropf, 1994). While a primary regulatory role for pH_ext gradients has largely been discounted (Harold and Caldwell, 1990; Gibbon and Kropf, 1991), evidence for the existence of pH_cyt gradients and their involvement in regulating tip growth remain controversial (Herrman and Felle, 1995). This is in contrast to the situation with Ca²⁺ where strong evidence supports the ubiquitous presence of a tip-focused gradient of cytosolic free Ca²⁺ in growing pollen tubes (e.g. Pierson et al., 1994; Malhó et al., 1995), and further evidence suggests similar gradients probably exist in other tip-growing cell types (Jackson and Heath, 1993; Berger and Brownlee, 1993; Herrmann and Felle, 1995).

A number of cellular processes central to tip growth, such as cytoskeletal organisation, vesicle fusion and enzyme activity are sensitive to pH and might be regulated at the cell apex by local differences in pH_cyt (Guern et al., 1991; Roos, 1992; Grabski et al., 1994). There is, however, a problem in identifying specific roles for pH_cyt in the regulation of tip growth because of the involvement of pH_cyt in basic cellular processes which have an indirect effect on, or are consequential or incidental to tip growth (Felle, 1996). Localised uptake of nutrients, localised metabolic activity and maintenance of polarised membrane potential are polarised activities in which protons probably do not have a direct regulatory role in tip growth (Harold and Caldwell, 1990). With respect to a possible fundamental regulatory role of pH_cyt in tip growth, two key questions need to be addressed. First, are pH_cyt gradients a general feature of tip-growing cells? Second, how dependent is tip growth on pH_cyt and pH_ext? In the current study we have performed a rigorous quantitative, spatial analysis of pH_cyt in three taxonomically diverse types of tip-growing cells (fungal hyphae, higher plant pollen tubes and fern rhizoids; Fig. 1) which exhibit contrasting lifestyles. This analysis has involved simultaneous dual channel confocal ratio imaging of the pH-sensitive probe cSNARF-1 (5-(and-6)-carboxy-seminapthorhodafluor-1). Our data demonstrate that pronounced, longitudinal pH_cyt gradients (ΔpH >0.1 pH unit) are absent from the apical 50 μm of the three cell types during active tip growth. We have also shown how experimentally induced changes in pH_cyt or pH_ext influence tip growth. Our results suggest that pronounced, longitudinal pH_cyt gradients are not fundamental to the regulation of tip growth but that changes in pH_cyt and pH_ext can significantly influence the growth rate and morphogenesis of tip-growing cells.

Chemicals for cell culture and experimental manipulations were obtained from Sigma Chemical Co. (Poole, Dorset, UK). pH-sensitive fluorescent dyes (free acid, AM-ester and 10 kDa dextran-conjugate of cSNARF-1), fluorescent beads (200 nm nile red FluoSpheres) and Pluronic F-127 were obtained from Molecular Probes Inc. (Eugene, Oregon, USA). For ester-loading, cSNARF-1 AM (20 mM stock in dimethylsulphoxide) was diluted to 200 μM with aqueous Pluronic F-127 (0.04% v/v) detergent solution.

Typical examples of the experimental material used are shown in Fig. 1 and a summary of the methods of dye loading are given in Table 1. Neurospora crassa (wild type [74A], Fungal Genetics Culture Collection, strain 262) was cultured and prepared for loading with dye as described previously (Knight et al., 1993). For cSNARF-1 AM esterloading, hyphae were cultured in half strength liquid Vogel’s medium (Vogel, 1956) and sandwiched between two glass coverslips using the mounting method described by Knight et al. (1993). For pressure microinjection of the cSNARF-1 dextran conjugate, hyphae were excised from the leading edge of a colony cultured on agar medium overlaid with cellophane (gauge 525, uncoated Rayophane from A. A. Packaging, Walmer Bridge, Lancs., UK) and subsequently transferred to slide culture on a thin, even layer of Vogel’s medium (containing 2% w/v agarose) spread over a glass coverslip. Slides were incubated at 25°C in a humid chamber for 5-8 hours. Five to ten minutes prior to microinjection, hyphae were covered with liquid medium.

Pollen tubes of Agapanthus umbellatus were cultured from stored pollen grains as described by Malhó et al. (1994). Briefly, pollen was evenly scattered onto 200 μl agar medium, which had been spread thinly and evenly over glass coverslips, and incubated under humid conditions at 25°C for 1-2 hours to produce 200-2,000 μm long tubes. Growing pollen tubes were loaded with dye, either by the ester method or by ionophoretic injection of cSNARF-1 free acid.

Spores of Dryopteris affinis, morphotype affinis/borreri (Newman) Fraser-Jenkins, collected in the grounds of Edinburgh University, were cultured under conditions modified from those described by Dyer and Cran (1976). After initial rhizoid emergence (typically 5 days after sowing), spores were transferred to modified liquid medium containing 6.1 mM Ca(NO₃)₂, 1.2 mM KNO₃ and 2.0 mM MgSO₄. They were then cultured for a further 24-36 hours under white light (∼100 μEm⁻²s¹) at 19-21°C to allow rhizoids to grow to 150-300 μm in length. Cells were only loaded with cSNARF-1 by the ester method.

Microinjection was carried out on a Nikon Diaphot TMD inverted microscope set up for fluorescence microscopy and fitted with Narishige N-88 micromanipulators. Ionophoretic microinjection involved the use of the equipment described by Knight et al. (1993). Pressure microinjection was performed with the pressure probe system described by Oparka et al. (1991).

Needles for ionophoretic microinjection were pulled on a Narishige PB-7 electrode puller (single stage pull) from GC-150F borosilicate glass (Clark Electromedical Instruments, Pangbourne, Reading, UK); needles for pressure injection were pulled with a Campden 773 micropipette puller (three stage programmed pull) from GC-100F glass.

Manipulation of pH_cyt was performed with modified culture media containing a cell permeant weak acid or base buffer: pH 6.0-7.0 (15, 30 or 50 mM sodium propionate added) and pH 7.0-8.0 (30 mM ammonium chloride or 50 mM trimethylamine added). The pH_ext was shifted by exchanging the standard medium with similar media of different pH adjusted with KOH or HCl. Images were usually collected between 3 and 10 minutes after buffer application.

Confocal laser scanning microscopy (CLSM) was performed with the Bio-Rad MRC-600 system (described by Read et al., 1992) running COMOS, MPL and TCSM software (all version 7.0) supplied by Bio-Rad (Hemel Hempstead, UK).

Intracellular distribution of cSNARF-1 was determined by CLSM imaging in single channel mode with excitation at 514 nm and detection of total dye fluorescence emission (>550 nm). cSNARF-1 ratio imaging was performed with the CLSM in dual channel mode using 514 nm excitation and a custom built SNARF emission filter block (supplied by Bio-Rad). The filter block contained a 610 nm dichroic mirror and 640/40 and 580/30 nm filters. CLSM settings used for confocal ratio imaging of cSNARF-1 are summarised in Table 2. A ×40 (0.95 NA) dry objective and ×60 (1.4 NA) oil immersion objective were used. Whilst the oil immersion objective was superior to the lower NA dry objective in terms of optical resolution and efficiency of light collection, the latter still gave useful resolution and was more amenable to the examination of fast growing cells.

Axial and lateral resolution for the different objectives and confocal aperture settings used were estimated in terms of the FWHM (full width half maximum) value obtained from images of 200 nm nile red FluoSpheres (Cogswell and Larkin, 1995). With a confocal aperture of 30% (Table 2), the ×40 dry objective gave an estimated axial resolution of 1.3 μm and lateral resolution of 0.6 μm; the corresponding values for the ×60 oil immersion objective were 0.8 μm and 0.4 μm, respectively. Confocal apertures were opened to ∼30%, slightly greater than the value reported to give maximum axial and lateral resolution (Lemasters et al., 1993), because of the need to maximise the collection of emitted photons for a given dye loading to improve the signal to noise ratio (R_S/N). In addition, the slow scan mode (integrating over the full pixel dwell time) was used and images were averaged or accumulated over several scans where possible (Table 2).

Ratio images were produced using the TCSM software. From each fluorescence image pair the photomultiplier dark signal was removed by a ‘dark image’ subtraction. Dark images were collected by blocking the laser illumination path. Thresholding, usually at brightness values of 10-15 (grey scale from 0-255), was used to remove background noise from outside the fluorescent specimen. The ratio image was formed by division of corresponding pixels in the image pair Ch-2/Ch-1. Ratio images were scaled such that the pixel range 0-255 corresponded to ratio values between 0 and 4 depending upon the pH range covered. Finally a 3×3 median filter was applied to improve image appearance and images were displayed with a pseudocoloured look up table (LUT).

Ratio values were calculated by dividing average pixel intensity values extracted from defined areas at corresponding positions in the 580 and 640 nm fluorescence images, using COMOS and MPL commands (Length, Histogram and Stats). Numerical data were displayed graphically through Fig P (version 2.7) from Biosoft (Cambridge, UK).

In vitro calibration of cSNARF-1 ratio values was performed by imaging the fluorescence of the appropriate dye-buffer mixtures over the pH range 6.0 to 8.0. Image collection and processing were as described above. The buffer media compared were a simple 20 mM MES/Hepes buffer and a ‘pseudocytosol’ medium designed to mimic the intracellular environment (Fricker et al., 1994). The dye concentration used in each of the different buffers was adjusted to give fluorescence image intensities of about 100 (on a scale of 0-255) in both channels at pH 7.0.

Average pH_cyt of cell populations was determined from the average pH_cyt measured over large sample areas (>5,000 pixels) within individual cells. Pixel variation within individual images was not considered and variation in pH_cyt within the population was assumed to be normally distributed and reported as ± s.e.m values.

Whilst autofluorescence was negligible and photomultiplier dark signal was subtracted during ratioing, the degree of intrinsic noise in the number of detected photons (Pawley, 1995) could not be controlled. Intrinsic noise, leading to random variation in individual pixel values, arises from several sources during the process of digital imaging (Sheppard et al., 1995). One of the principal sources of such random variation between pixel values is the Poisson distributed variation in photons detected from the specimen, known as ‘shot’ noise (Pawley, 1995), which imposes a fundamental limit upon the R_S/N of images with respect to the number of photons detected at each pixel. The variation between individual pixel values within an image becomes increasingly important when extracting quantitative data from small image areas (of the order of hundreds of pixels), and when making comparisons between the values from different regions within the same image. Therefore, a means of quantifying the magnitude of the variation between pixel values is required to provide: (a) estimates of the precision of quantitative data for particular sample sizes (and thus the spatial resolution) and (b) confidence limits to allow meaningful comparison between different regions of the same image. Hence the need for rigorous statistical analysis.

The extent to which the two mean ratio values differ is itself dependent upon the degree of noise in the fluorescence images; the more noisy the greater the difference.

Mean ratio values must, therefore, be calculated by dividing the mean fluorescence values from corresponding sample areas in the λ₂ and λ₁ images. The confidence limits for the mean ratio value must be calculated by taking into account the variation in the pixel values of both fluorescence images. Unfortunately this variation could not be analysed effectively by conventional statistical methods and thus a Bayesian approach had to be adopted (see Appendix).

Experimental analysis of the spatial resolution and precision of pH measurement within individual images was initially done in vitro by ratio imaging cSNARF-1 free acid buffered (MES/Hepes) at pH values (accurate to ±0.005 units) over the range pH 6.5-7.5 (Fig. 2). Imaging parameters used were as defined in Table 2.

In in vitro test samples, the average ratio determined from all pixels in a fluorescence image pair was considered as the ‘true mean’ ratio value. However, variation in ratio values at the level of individual fluorescence pixel pairs influenced how well a small sample of pixels estimated the true mean. This uncertainty, which represented the precision with which pH could be estimated for a particular sample area size, was assessed according to the Bayesian theory of inference (Kappenman et al., 1970). An algorithm for determining the posterior density function of sample mean ratio values (see Appendix) was written in S-plus from MathSoft (StatSci Europe, Oxford, UK). The 95% Bayesian intervals for variation in ratio values of sample areas of different sizes were determined and from this the optimum precision of pH measurement and spatial resolution in vitro were determined (Fig. 2 and Table 3).

Average signal strength in fluorescence images and the use of Kalman and Accumulative filters all influenced the precision of pH measurement. Increasing average fluorescence brightness values from 25 to 150 (on a 0-255 scale) gave an improvement in precision of >150%. Accordingly, images with high average fluorescence brightness (typically >100) were used where possible. Kalman filtering (n=3) gave an improvement of ∼40% over direct image collection. Accumulative filtering (n=3) to increase signal strength also improved precision but was inferior to the direct capture of images at a signal strength equivalent to the accumulated image. Standardised image capture (Table 2) made data more amenable to the application of systematic statistical analysis.

Living cells were examined with a Reichert-Jung Polyvar light microscope using differential interference contrast (DIC) optics and a ×40 (1.0 NA) or ×100 plan apo (1.32 NA) oil immersion objective. Micrographs were recorded on Kodak TMAX 400 35 mm black and white film.

The health of cells during pH_cyt imaging was assessed on the basis of their apical morphology (Figs 1, 3, 4 and 5) and growth rates (Table 1). Stressed cells exhibited increased apical vacuolation, altered apical morphology, and reduced growth rates. The phototoxic effects of laser-irradiating intracellular dye was found to be the most significant threat to cell health. However, the dye-loading and low dosages of laser irradiation employed (Tables 1 and 2) did not generally cause stress responses in cells over the usual period of 20-30 minutes during which images were obtained. Adverse affects on tip growth (e.g. from dye toxicity and pH buffering) were also considered negligible as similar growth rates were obtained with loaded and unloaded cells. Based on a comparison with fluorescence images of dye/buffer mixtures in vitro the cytoplasmic concentration of loaded dye was estimated to be <100 μM.

Intracellular dye distribution was found to depend upon the dye-loading method, cell type and time after the start of loading (Figs 3, 4 and 5). cSNARF-1 was successfully introduced into all cell types as its AM-ester. However, dye sequestered from the cytosol into organelles. In Neurospora hyphae and Agapanthus pollen tubes, continuous loading was necessary to maintain suitable fluorescence levels, and even then after 20-40 minutes the fluorescence intensity began to decrease (Figs 4A,B and 6). Dryopteris rhizoids could not be continuously loaded due to extensive vacuolar accumulation of dye so loading was halted once suitable cytoplasmic fluorescence levels were reached (Fig. 5A and B).

Ester-loaded hyphae of Neurospora exhibited bright dye fluorescence throughout the cytosol (Fig. 3A and B). Dark areas in subapical hyphal compartments corresponded to exclusion of dye from large vacuoles. Bright fluorescent spots, ∼0.5 μm in diameter, were evident throughout ester-loaded hyphae. These spots represented dye sequestered within unidentified, roughly spherical organelles. Within the apical 50 μm of hyphae, dye fluorescence was uneven and elongated tubular structures could be discerned. These structures correlated with the morphology and distribution of mitochondria (Fig. 1).

Ester-loading of Agapanthus pollen tubes initially resulted in a very uniform distribution of dye fluorescence except within the 5-10 μm apical region where the fluorescence was reduced (Fig. 4A). The latter corresponded to the apical ‘clear zone’ (Derksen et al., 1995; Fig. 1) in which the high concentration of vesicles probably excluded dye. Sequestration of dye was more obvious after 30 minutes (Fig. 4B′ and B′′) when bright tubules, possibly part of the vacuolar system, became apparent. Initially, Dryopteris rhizoids rapidly accumulated ester-loaded dye in the cytoplasm with the vacuolar system appearing non-fluorescent (Fig. 5A and B). Brighter patches of fluorescence within the cytoplasmic regions were visible very early during loading (Fig. 5B) and were probably mitochondria (Parton, 1996). Gradual redistribution of dye from the cytoplasm to the vacuolar system occurred from the start of loading but was only significant after >30 minutes (Fig. 5C and D).

The fluorescence distribution in Neurospora hyphae pressure-injected with the 10 kDa dextran conjugate of cSNARF-1 differed in several important ways from hyphae ester-loaded with dye (compare Fig. 3C with 3A and 3D with 3B). First, there was no indication of dextran-dye associated with putative mitochondria. Second, the dextran-dye was excluded from both the large subapical vacuoles and the small unidentified organelles which sequestered dye after ester-loading. Third, the fluorescence of dextran-dye within hyphae diminished much more slowly with time. Fortuitously, the dextran-dye tended to move forward from the subapical injection site towards the tip during growth.

Ionophoretic microinjection of Agapanthus pollen tubes with cSNARF-1 free acid resulted in a pattern of dye loading and sequestration within organelles similar to that obtained after ester-loading (cf. Fig. 4C and A).

The pH-dependent responses of cSNARF-1 free acid and the 10 kDa dextran conjugate of cSNARF-1 were assessed in vitro for our specific imaging set up using: (1) simple MES/Hepes buffer solution, and (2) ‘pseudocytosol’ solution designed to mimic the intracellular environment (Fig. 7).

In both calibration solutions, cSNARF-1 fluorescence ratios for the two dyes gave an approximately linear response over the range pH 6.9-7.3. Between pH 6.0 and 8.0 the response was sigmoidal; below pH 5.5 dye precipitation occurred. Although the four calibration curves were broadly similar, two significant differences were noted. First, the ratio values obtained for the dextran-dye were lower than those obtained for the dye free acid. Calibration of data from dextran injected cells was, therefore, performed with the dextran in vitro. Second, with both the free acid and dextran dyes, ‘pseudocytosol’ solution caused a shift in ratios to slightly lower values relative to MES/Hepes solution. The effects of the ‘pseudocytosol’ appeared to be greater at alkaline pH resulting in a slight change in the slope of the calibration curve relative to MES/Hepes. In addition, the absolute fluorescence intensities recorded for the same concentration of dye at each pH were roughly three times higher in the pseudocytosol than in the simple buffer solutions.

In vitro calibrations using MES/Hepes buffers were simple, quick, easy to perform, and very reproducible. These calibrations provided good estimates of ΔpH as well as a means of comparison between different sets of data by taking into account day-to-day and cell-to-cell variations, and differences between dyes. The similarity between the pseudocytosol and simple buffer in vitro calibrations provided further support that a reasonable estimate of ΔpH_cyt could be obtained. With this calibration the pH_cyt determined was similar to previously reported values of pH (Guern et al., 1991). Small pH_cyt changes, which had been imposed on cells by permeant weak acid treatments (ΔpH_cyt ∼0.2 unit), were also reported convincingly by the in vitro calibration. Despite the corroborative evidence gathered to support the value of our calibrations, there is no guarantee that the estimated pH values quoted provide a truly accurate measure of absolute pH_cyt. The values given are, therefore, the best estimate currently available without the use of an independent method of pH determination such as pHsensitive microelectrodes.

An in situ calibration using ionophores to equilibrate pH_cyt and pH_ext (Davies et al., 1990) could not be achieved reliably here because of two major limitations. First, it was difficult to stably and reproducibly clamp pH_cyt at a particular level. Second, the ionophore treatment disrupted cell structure, caused dye sequestration within organelles and increased dye loss from cells.

No pH_cyt gradient with a ΔpH >0.1 pH unit was detected along a midline through the central region of the apical 50 μm nor between different regions around the periphery of the apical 15 μm (covering the region of tip extension) of any of the three growing cell types examined (Fig. 8). In Fig. 8D apparent acidic regions at the extreme (<1 μm) cell periphery were attributed to dye binding to the cell wall.

Ratio images of Dryopteris rhizoids proved difficult to analyse because of the thin layer of peripheral cytoplasm within cells and dye sequestration into acidic vacuoles (Fig. 5). With the ×40 dry objective no significant differences in pH_cyt were found between cytoplasm in the extreme apical 3 μm and at the cell periphery up to 50 μm behind the tip (n=7). Imaging with the ×60 oil objective and Kalman filtering were used to provide increased spatial resolution to better distinguish the sparse cytoplasm from contaminating vacuolar signal and gave similar results (n=6). Fig. 8D′, which shows a clear gradient in apparent pH_cyt, is included to demonstrate how easily ratio images can be misinterpreted without proper reference to cytological features (Figs 5 and 8D′).

Estimated values of average pH_cyt over the first 30 μm of each of the three cell types are given in Table 4. The average pH_cyt values obtained for ester-loaded cells were very similar (pH 7.15-7.25; also see Fig. 8A,C and D). No significant difference in average pH_cyt measurement was obtained by ratio imaging ester-loaded Neurospora hyphae with either the ×40 dry or ×60 oil immersion objective. However, the pH_cyt value obtained for Neurospora hyphae loaded with the dextran-dye was significantly higher (pH 7.57; also see Fig. 8B). The lower pH_cyt estimated for ester-loaded cells could not be attributed to oxygen deprivation because similar pH_cyt values were obtained from Neurospora hyphae sandwiched between two coverslips (liquid medium) and from hyphae (on agarose) lacking an overlying coverslip (Table 4). No correlation was found between the growth rates and average pH_cyt values of individual Neurospora hyphae which varied from pH 7.02 to 7.29 in ester-loaded cells and from pH 7.51 to 7.59 in dextran-dye loaded cells (data not shown).

Ratio imaging revealed that the application of a cell permeant weak acid (50 mM sodium propionate) at pH 7.0 resulted in the cytoplasm of Neurospora hyphae being acidified evenly (Fig. 9A-H). Immediately after this treatment, hyphal tips became swollen causing a slight increase in hyphal length (Fig. 9A-E). Five to ten minutes after treatment a new tip was generated, narrower than that of the parent hypha, which gradually increased its growth rate (Fig. 9A and F-H). Eventually this new tip with acidified pH_cyt acquired a stable growth rate which was slower (2-3 μm minute⁻¹) than its parent hypha (14-15 μm minute⁻¹, Fig. 9A, n=5). Similar treatments with propionate at pH 6.0 or 6.5 resulted in even greater acidification of pH_cyt, an initial hyphal swelling and complete inhibition of subsequent tip regeneration (n=5). Addition of a cell permeant weak base (50 mM trimethylamine) at pH 8.0 caused increased alkalinisation of hyphal cytoplasm and a similar pattern of hyphal swelling and tip regeneration (n=5) as observed after weak acid treatment at pH 7.0. The shifts in internal pH due to weak acid (pH 7.0) or base (pH 8.0) treatment were only stable for approximately 10-15 minutes. Thereafter, the pH_cyt slowly returned towards its original value. Agapanthus pollen tubes proved much more sensitive than Neurospora hyphae to propionate and were very prone to bursting in the presence of 30 mM sodium propionate. Treatment with 15 mM propionate at pH 6.0 or 7.0 evenly acidified the cytoplasm and resulted in tip swelling but a new tip was not regenerated (n=8). Addition of a cell permeant weak base (30 mM ammonium chloride) caused even alkalinisation of pollen tube cytoplasm, swelling of the tip but again tip growth did not resume (n=8).

Exposure of Dryopteris rhizoids to 30 mM sodium propionate at pH 6.0 altered pH_cyt to an even acidic level throughout cells within 3 minutes (n=10). The pH change was associated with a rapid redistribution of apical cytoplasm and cessation of growth. Cells only remained viable if the duration of propionate treatment was <15 minutes. Rhizoid growth was not resumed until 12-18 hours after returning to normal medium (pH 5.6) and generally involved the generation of a new tip at a position slightly subapical to the original. Similar effects on growth were observed after treating rhizoids with 30 mM sodium propionate at pH 7.0 or 30 mM ammonium chloride at pH 7.0 or 8.0. Propionate treatment at pH 7.0 was unstable and gave apparent pH values of <7.0 (n=5). With ammonium chloride at pH 8.0, the pH_cyt elevation stabilised after 3 minutes. The pH_cyt change after treatment at pH 7.0, however, was unstable and after an initial rapid alkalinisation the pH_cyt value gradually increased to a similar level as for the pH 8.0 treatment (n=20). Recovery after exchange to normal medium was similar to that observed after propionate treatment.

Extracellular pH treatments were also found to change the pH_cyt and growth rate of the three cell types.

In Neurospora hyphae ester-loaded with dye, an increase in the pH_ext from 5.9 (normal medium) to 7.0 resulted in a gradual, transient elevation in pH_cyt by 0.2-0.3 pH units within 10-15 minutes and a reduced growth rate which stabilised after 20-30 minutes (Fig. 10, n=5). Increasing the pH_ext to 8.0 immediately led to an increase in pH_cyt by 0.5-0.6 pH units which declined towards normal values over ∼30 minutes (data not shown). At pH_ext 8.0, hyphal growth rate was reduced to <4 μm minute⁻¹ (n=5), and tips tended to exhibit an undulating growth pattern and underwent increased apical branching (Fig. 11A).

Agapanthus pollen tubes responded differently to pH_ext manipulation. Pollen tubes were normally grown in medium at pH 6.0. Treatment with unbuffered medium at pH 8.0 or 4.5 caused a transient shift in pH_cyt of <1.0 pH units (+ or −,respectively) which was associated with a disruption of tip morphology and growth rate (n=20). After changing the medium to pH 4.5, the pH_cyt recovered to normal levels within one minute of treatment, growth continued but the tip diameter became transiently narrower, recovering to its full width within two minutes (Fig. 11B). After changing the medium to pH 8.0, tip growth temporarily ceased and then within 1-5 minutes the tip underwent swelling, the pH_cyt recovered, and a new tip emerged from the swelling within 3-7 minutes.

Dryopteris rhizoids required strong buffering of pH_ext (5-25 mM MES/Hepes) before effects on growth were observed. Changing the normal, unbuffered pH 5.6 medium to buffered medium at pH 6.0 had little effect on growth. However, buffered medium at pH 7.0 or 8.0 caused cytoplasmic rearrangement and withdrawal from the tip, apical swelling and temporary cessation of polarised growth in all cases (Fig. 11D′). Tip growth resumed 6-12 hours later in cell populations (Fig. 11C and D′). Technical difficulties prevented examination of pH_cyt continuously over the period from the alteration of pH_ext to the time at which cells had undergone changes in apical morphology. At the two time points examined, immediately upon pH modification and at the first signs of changes in apical morphology, no shift in pH_cyt could be detected (n=6).

This study has shown that pronounced longitudinal pH_cyt gradients (ΔpH >0.1 pH unit) are not present within the apical 50 μm of actively growing fungal hyphae, higher plant pollen tubes and fern rhizoids indicating that pH_cyt gradients of this type are not a general feature of tip-growing cells.

The existence of pronounced pH_cyt gradients in tip-growing cells has been a controversial topic. The best published data showing a pH_cyt gradient associated with tip growth has come from the analysis of rhizoids in the brown alga Pelvetia using confocal ratio imaging and ion-selective microelectrodes (Gibbon and Kropf, 1994). Evidence for pH_cyt gradients in other tip-growing cells (e.g. Turian, 1979; Roncal et al., 1993) is poor because the methodologies used in these studies have since been found to be inadequate as a result of the development of improved methods of pH_cyt analysis (such as employed in the present study) which provide less equivocal results. In support of our conclusion that pH_cyt gradients are not a general feature of tip-growing cells, Herrmann and Felle (1995) failed to detect a gradient (>0.1 pH unit) in growing root hairs by using ion-selective microelectrodes. Although the absence of pronounced pH_cyt gradients would appear to be the norm, it cannot be discounted that they may exist in certain cell types such as algal rhizoids.

An alternative biophysical approach, using a combination of in vitro experimentation and theoretical analysis, has been used to investigate the possible existence of pH_cyt gradients in cells (Al-Baldawi and Abercrombie, 1992). This work with animal cytoplasm showed that estimates of the rate of proton diffusion, the buffer capacity of the cytoplasm and rates of proton flux across membranes would tend to favour a pronounced ΔpH (>0.1 unit) only over distances >25 μm. However, their study did not preclude the possibility of more localised pH domains, which could arise from localised fixed buffering components (e.g. cytoskeletal proteins). Clearly, polarised plant and fungal cells are capable of generating large proton currents involving spatially separated proton transport activities (Weisenseel et al., 1979; Harold and Caldwell, 1990; Feijo et al., 1995) but in light of the absence of a pH_cyt gradient in root hairs (Herrmann and Felle, 1995), and in the tip growing cells studied here, such proton flux must be dealt with by a combination of the metabolic activity (Sanders and Slayman, 1982), proton pumping (Guern et al., 1991) and cytosolic buffering of cells (Al-Baldawi and Abercrombie, 1992) which together make up the cellular buffering capacity.

Simultaneous, dual channel confocal ratio imaging is the most suitable method for visualising pronounced pH_cyt gradients in tip growing cells. However, this procedure is potentially fraught with artefacts and thus requires rigorous controls. In order to validate our conclusion that pronounced pH_cyt gradients were absent in the three types of tip-growing cells studied, we carefully assessed the specific problems of cell perturbation during imaging, dye partitioning within organelles (Seksek et al., 1991; Opitz et al., 1994), R_S/N and ‘edge’ artefacts in ratio images (Bright et al., 1987; Pawley, 1995), and calibration of cSNARF-1 (Fricker et al., 1994). A pronounced longitudinal pH_cyt gradient was defined in this study as having a ΔpH of >0.1 pH unit over a distance >2 μm for two reasons: first, these were the limits of the precision of pH measurement and spatial resolution which could be routinely achieved by confocal ratio imaging of cSNARF-1 within healthy cells. Second, ΔpH of >0.1 pH unit are within the range considered to be of physiological significance (Busa and Nuccitelli, 1984; Guern et al., 1991). Our results do not discount the possible presence of highly localised pH gradients which might be associated with the plasma membrane (Roos, 1992). Previously reported pH_cyt values for plant and fungal cells usually lie between pH 6.8 and 7.9 (e.g. see Sanders and Slayman, 1982; Guern et al., 1991). In our study, the estimated values of mean pH_cyt with ester-loaded dye were between pH 7.15 and 7.25 for the three cell types. With Neurospora hyphae loaded with the dextran dye by pressure injection, however, the estimated mean pH_cyt was pH 7.57. It is generally believed that the results obtained with dextran dyes provide a better estimate of pH_cyt because of their superior localisation and retention within the cytosol (Haugland, 1992; Miller et al., 1992). It is significant that, although the pH_cyt values estimated with esterloaded free dye and pressure-injected dextran dye in Neurospora differed, no significant pH_cyt gradient was detected with either.

With each dye type, the estimated pH_cyt was very reproducible between individual cells and variation in average pH_cyt between individual cells was within the limits of precision of measurement for the imaging technique. This is in agreement with the general finding that pH_cyt is tightly regulated within growing cells (Guern et al., 1991).

For the work reported here, cSNARF-1 proved to be the most appropriate dye. Another ratio dye widely used to measure pH_cyt is BCECF (Haugland, 1992), but this has a number of disadvantages compared with cSNARF-1. First, in Neurospora it exhibits more pronounced sequestration within vacuoles (unpublished results; Slayman et al., 1994). Second, BCECF is a dual excitation dye requiring sequential excitation at two wavelengths. Improved temporal resolution can be achieved with cSNARF-1 which is excited at one wavelength and detected simultaneously at two wavelengths. However, using confocal ratio imaging of BCECF it has recently been demonstrated that pronounced pH_cyt gradients are not associated with growing Lilium pollen tubes (M. D. Fricker, personal communication) which provides further confirmation of our findings.

The lack of pronounced pH_cyt gradients in the cell types examined here during active tip growth and additionally, the work of Herrmann and Felle (1995) in root hairs, argues strongly against a fundamental role for gradients of this type regulating the polar organisation and active processes of tip growth. This conclusion is further supported by our finding that Neurospora hyphae continued to grow in the presence of relatively high concentrations of cell permeant propionic acid (pKa 4.87) which would be expected to collapse any pH_cyt gradient. If the results obtained by Gibbon and Kropf (1994) are correct then a tip-focused pH_cyt gradient may be a feature peculiar to growing rhizoids of fucoid algae such as Pelvetia. We attempted to confirm their results by examining the growing rhizoids of the related alga Fucus serratus. However, our attempts proved unsuccessful due to problems in attaining a suitably high level of dye loading relative to chloroplast autofluorescence in these cells (R. M. Parton and J. Love, unpublished results).

Gibbon and Kropf (1994) found that 10 mM propionic acid (at pH 7.0) dissipated the pH_cyt gradients and drastically reduced growth of Pelvetia rhizoids. They suggested that these observations provided evidence for a pH_cyt gradient regulating tip growth in these cells. The validity of this interpretation is open to question. Herrmann and Felle (1995) performed similar weak acid treatments on root hairs and recorded similar inhibitory effects on growth. However, in the absence of a detectable pH_cyt gradient within growing root hairs, they concluded that the inhibitory effects of the weak acids on growth were due to a disruption of the normally maintained

pH_cyt. Treatment of cells with membrane permeant weak acids may have a number of non-specific effects relevant to tip growth, including: (1) both organellar pH and cytosol pH will be affected and changes in organellar pH may be important for certain processes involved in tip growth (e.g. the exocytotic pathway). (2) Acidifying cells can increase the cytosolic calcium concentration (Felle, 1988; Guern et al., 1991). Artificially induced increases in cytosolic free calcium have been shown to inhibit tip growth in pollen tubes (Franklin-Tong et al., 1993; Malhó et al., 1994). (3) The membrane potential will be changed by cytoplasmic acidification which will affect ion transport across membranes. Such non-specific effects may explain the inhibition of tip growth in both Agapanthus and Dryopteris cells, although the possible existence of highly localised unseen pH_cyt gradients cannot be ignored.

It is clear from our results that both pH_cyt and pH_ext have a significant influence on the growth of the three cell types analysed. Tip growth was sensitive to any alteration in the normally maintained pH_cyt. Changing pH_ext influenced the rate of tip growth, the width of cell tips, and the formation of new tips (branch formation). Altering pH_ext was commonly accompanied by a transient change in pH_cyt. However, it is unlikely that the effects of altering pH_ext were due to this transient influence on pH_cyt alone, consider increased branching in Neurospora which continues throughout the period of culture at increased pH_ext. Changes in membrane potential and transport activities may also be important.

pH_cyt regulation is undoubtedly an important aspect of cellular physiology. It has been proposed as a mechanism by which cells co-ordinate the regulation of various processes that lack any other common factors and also may provide a regulatory link between metabolic state and physiological responses, such as those involved in tip growth (Busa and Nuccitelli, 1984; Felle, 1996). However, the pervasive nature of H⁺ throughout all cellular processes makes it difficult to determine strict cause and effect relationships and to assign specific second messenger functions.

Statistical analysis by Bayesian inference was used to obtain 95% error bounds for estimates of the mean ratio value of images obtained by sampling small (14×14 pixel) areas. The Bayesian approach permits probability statements to be made regarding unknown population parameters such as mean ratios, variances and correlations, and provides a popular paradigm for statistical inference (O’Hagan, 1994; Leonard et al., 1989; Leonard and Hsu, 1992).

The statistical analysis here assumed that the distribution of fluorescence pixel intensity values in both the Ch-1 and Ch-2 images could each be approximated by a Normal distribution. Hence our model of the data contained two ‘unknown’ population means and variances, which needed to be estimated by analysis of the small sample area data. We also permitted the possibility that a correlation exists between the fluorescence intensity values of corresponding pairs of pixels from Ch-1 and Ch-2 images (i.e. an unknown population linear correlation between pairs of observations). The assumptions were completed by assuming that any linear combination of pairs of pixel values in a fluorescence image pair were also normally distributed.

Error bounds were required for our estimates of the ratio R of the mean intensities of the populations of pixels in a fluorescence image pair obtained from sampling small areas of the image pair. These could, in general, be calculated from a sample of n (in this case n=14×14) independent pairs of pixel values. As an extension of Kappenman et al. (1970), we employed a Bayesian approach, under particular uniform prior assumptions carefully selected to guarantee an almost exact frequency coverage for our 95% intervals if n≥100.

A mathematical solution was derived and incorporated into a computer program which provided plots of the ‘posterior probability density function’ (posterior p.d.f.) for R which summarise the information about R, a posteriori (i.e. after incorporating the data) when there is uniform information, a priori (i.e. prior to viewing the data). An example is given in Fig. 12 for sampling an in vitro calibration image (pH 7.0) with a 14×14 pixel box size. The area under the curve between R values of 0.961 and 1.022 is the 95% Bayesian interval (i.e. there is a 95% posterior probability that R is in that interval). Convenient error bounds were provided by equal-tailed Bayesian 95% and 99% intervals (Fig. 12). A convenient point estimate of R is the posterior median (R*) = 0.991. Note that this probability distribution is not a Normal distribution and is unusual in that standard deviation and mean values are not appropriate (because they do not exist). In extreme cases such distributions can be quite obviously skewed, bumpy or bimodal.

We thank Emeritus Professor David Finney and Dr Bruce Worton of the University of Edinburgh Statistical Laboratory for practical statistical advice. We are also grateful to Dr Lionel Jaffe (Marine Biological Laboratory, Woods Hole, Mass., USA) for his critical reading of the manuscript, Dr Mark Fricker (Dept of Plant Science, University of Oxford) for helpful discussions, and to Dr Tony Collins (University of Edinburgh) for helpful advice on computing matters. The research was supported by an AFRC Research Studentship (to R.M.P.), a TMR Fellowship (ERBFMBICT 950455) from the EC (to S.F.) and a BBSRC Studentship (to T.C.J.).