X-ray microanalysis of leaf tumors from maize plants experimentally infected with maize rough dwarf virus: Scanning and transmission electron microscopic study

VIROLOGY

103, 357-368 (1980)

X-Ray Microanalysis of Leaf Tumors from Maize Plants Experimentally Infected with Maize Rough Dwarf Virus: Scanning and Transmission Electron Microscopic Study MARIA AUGUSTA FAVALI, **’ NICOLETTA BARBIERI,; ROBERTO BONECCHI,$ AND MAURIZIO *Istifuto

ANNALISA CONTI#

Botaniche, I;wiversit& di Milano; TIstituto di Fisica Tecnicu, Politecnico SCentro “Gino Bozza” CNR, Politecnico di Milano; and Kaboratorio di Fitovirologia Applicata CNR, Torino, Italy

di &‘&me

Accepted Febmary

BIANCH1.f di .k’?lnno;

7, 1SNO

X-Ray microanalysis, performed by both scanning and transmission electron microscopes equipped with energy-dispersive spectrometers (EDS), was used to analyze the element distribution in phloem tumors induced by maize rough dwarf virus (MRDV) in Zeu mays L. The content of both Si and Ca was lower in tumors than in the nontumoral surrounding tissues. The Si content increased in old tumors where this element was mainly concentrated in the trichomes growing at their surface. The peaks of other detectable elements, I’, K, and S, were higher in tumor than in nontumor tissues. The distribution of inorganic cations in the cells was also studied by the potassium pyroantimonate (KPA) precipitation technique. The KPA precipitates have proved to be primarily of calcium salts and they occurred quite frequently in the cells at the tumor periphery. INTRODUCTION

Maize rough dwarf virus (MRDV) is a Fijivirus (Matthews, 1979) which infects maize causing severe dwarfing and typical phloem tumors or enations originating from veins on the abaxial side of the leaves (Lovisolo, 1971; Milne and Lovisolo, 1977). The virus appearance and localization in both plant and vector cells have been described by Gerola et al. (1966), Gerola and Bassi (1966), Lovisolo and Conti (1966), Vidano (1966), and Appian0 and Lovisolo (1979). Further studies have been made on the chemical composition and function of the cytoplasmic inclusions induced by MRDV in plant cells and the sites of synthesis of the virions (Bassi and Favali, 1972; Favali et al., 1974). The cell wall modifications in infected plant cells have been studied by autoradiography (Bassi et al., 1974). The aim of the present work was to analyze the element distribution in MRDVinduced tumors in comparison with nontumoral interveinal leaf tissues. X-Ray ’ To whom reprint

requests should be addressed.

microanalysis was performed by both scanning and transmission electron microscopes equipped with energy dispersive spectrometers and concerned tumors at different stages of growth and surrounding nontumoral tissues. Besides using the current techniques, the specimens were prepared by a method involving only the initial freezing of the samples in liquid nitrogen (Parsons et al., 1974)and by K-pyroantimonate treatment which enables a relatively precise localization of inorganic cations in the cells. (Bulger, 1969; Legato and Langer, 1969; Shiina et al., 1970; Tandler and Kierszenbaum, 1971; Clark and Ackerman, 1971; Hayat, 1975; Favali et al., 1977, 1978; Wooding and Morgan, 1978). As far as we know, this is the first attempt to use these techniques in the study of plant tumors caused by virus infection, and may suggest further application in plant virology. MATERIALS

AND METHODS

Infected plants. Maize plants (Zea rnnys L. cv. Wisconsin 641 AA) were inoculated

368

FAVALI

at the coleoptile stage with MRDV by the planthopper Laodelphax striatellus Fallen. At the time of sampling (1 and 2 months after inoculation), vein tumors were clearly visible on the abaxial surface of maize leaves. Pieces of leaf were cut in such a way as to include one or more vein tumors surrounded by about 1.5 mm of healthy tissue.

ET AL.

system was used only for thick sections (1 i-4. Other leaf samples, unfixed, were quickly frozen in liquid nitrogen, mounted on a standard goniometer stage, and kept cooled in the microscope during analysis (Parsons et al. 1974; Yeo et al., 1977). The accelerating voltage was 20 kV, the beam current lo- I” to lo-” A, and the distance between Scanning electron WL~CYOSCO~Y (SEMI. specimen and detector about 32 mm. MultiThe samples were fixed in one of two ways: channel analyzer windows were set up to (1) in acetate-buffered 2% potassium pyro- record integral counts for the various eleantimonate (KPA) at room temperature for ments. Counts were generally accumulated 4 hr (Tandler and Kierszenbaum; 1971); or for 200 see for all spectra presented. Ele(2) in 0.1 M phosphate-buffered 3% glutar- ments were identified on each spectrum and aldehyde, pH 6.9, for 2 hr at 4”. energy levels in Kilo-electron volts are After fixation, some samples were dehy- shown on each abscissa. The data were drated in ethanol, critical-point dried with processed with a Hewlett-Packard 9830 CO, using amyl acetate as an intermediate desk-top computer that allows the noise fluid, coated with carbon and gold in a high subtraction from peaks height and the area vacuum evaporator on a rotating tilting integral of peaks. To make easier qualitastage, and observed in a Jeol scanning elec- tive analysis between tumor and nontumor tron microscope, type JSMIUS. tissues, we have reported in histograms the Transmission electron microscopy (TEM). percentage of integrals referring to various Two methods for tissue fixation were com- elements. The observations were repeated pared: (1) in acetate-buffered 2% potassium on several tumors from plants which had pyroantimonate (KPA) at room temperature been infected at various times of the year. TEM X-ray microanalysis. Unstained for 4 hr (Tandler and Kierszenbaum, 1971), and (2) in phosphate-buffered 3% glutaralde- ultrathin sections were placed on a carbon support film on copper grids and further hyde containing 2% potassium pyroantimonate at room temperature for 4 hr, post- coated with carbon before analysis. The fixed in 1% osmium tetroxide, pH 6.9. All grids had a hole diameter of 600 Frn to rethe samples were subsequently dehydrated duce the peak due to copper as much as possible. The cells were analyzed for the in ethanol and embedded in Epon-Araldite. For morphological studies, ultrathin sec- presence of different elements by using a tions were stained with lead citrate and Siemens Elmiskop 1A electron microscope examined in a Siemens Elmiskop 1A electron equipped with an energy-dispersive spectrometer (EDS) system (Laben Montedel). microscope, at 80 kV. SEM X-ray microanalysis. Some speciATlalysis of antimonate precipitates. mens were prepared by the critical-point The nature of the KPA precipitates was drying technique, coated only with carbon, determined in three different ways: (1) thick placed on graphite mounts, and observed in (1 pm) unstained sections of material fixed a scanning electron microscope equipped with KPA and embedded in Epon- Araldite with both a wavelength-dispersive spec- were mounted on carbon collodion-coated trometer (WDS) system (Jeol) and an grids and coated with carbon before analysis energy-dispersive spectrometer (EDS) sys- by the SEM, equipped with the WDS system (Laben Montedel). tem (Favali et al., 1977); (2) ultrathin, Resolution of the EDS system, when unstained sections of the same material using a Si-Li detector, is about 160 eV. were analyzed by the TEM equipped with Since the difference between the Ka line of the EDS system; (3) ultrathin sections, preCa and the LN line of Sb is only 90 eV we viously checked for KPA deposits, were cannot discriminate the two elements. Dis- floated on a 5 mM solution of EDTA at crimination is possible with the WDS sys- 60” (Burglen, 1974). Control sections were tem by using different crystals. The WDS incubated in EDTA-free medium.

MANE;

KOUGH

DWARF

VIRUS

TUMORS

FIGS. l-3. Scanning electron micrographs (SE :M) of the lower epidermis of leaves of Zea rrlaqs infected with MRDV. FIG. 1. Initial tumor cells. FIG. 2. Surface of a vein tumor at advanced st: ige of development. FIG. 3. A %-month-old tumor: note the long tric #homes growing on the surface. FIG. 1. SEM section through a Z-month-old t umor. The trichome (arrow) originates from the epidermis cells. RESULTS

SCU Lrein tumors were observed on infected mai ze leaves from the early stages of developme]nt. The tumor cells appeared abnormally k?-0wn and, in contrast with the surrounding nor ma1 cells, were not regularly aligned

(Figs. 1 and 2). Fully developed tun Lors had stiff trichomes on their surfaces (Fig :. 3) which were absent from the tumor-l %ee parts of veinal and interveinal epidert nis. Transverse sections of tumors showed t,hat the trichomes originated from the epider tmis and that the bulk of the tumor consi: ;ted of misshapen and structurally altered csells

360

FAVALI

ET AL.

FIGS. 5, 7, AND 9. Transmission electron micrographs of tumor cells after KPA fixation according to method 2. Ultrathin sections are unstained. FIG. 5. Low magnification oftumor tissue. Note the peripheral cells (PC) containing electron-opaque material on which KPA deposits are visible.

MAIZE

ROUGH

DWARF

VIRUS

TIJMORS

361

different techniques. Figures lO- 13 show some sample spectra obtained by analyzing the surfaces of control and tumor areas. Figures 14-17 show spectra obtained from Trnrlsmissiorl Electron Microscopy cross sections of tumor-free and tumorTissues prepared for TEM were better affected veins. These spectra reveal the elepreserved when fixed by method 2 than by ment present, the energy levels in kilomethod 1; photographs included here are all electron volts, and the principal emission from preparations obtained with method 2. lines: silicon 1.740 KU; phosphorus 2.013 Method 1, however, not including the use of Ka; sulfur 2.307 Ka; potassium 3.312 Ka; osmium tetroxide which can disturb the de- calcium 3.690 Ka; the chromium and iron tection of elements, gave the best results peaks, when present (Figs. 12 and 13), were for TEM X-ray microanalysis. due to X-rays originating from the specimen On the abaxial side of infected leaves, the holder. To show the differences between phloem showed an uncontrolled growth of spectra from tumor and nontumor tissues, undifferentiated or partially differentiated the results are presented as histograms cells. Xylem and other tissues were appar- (Figs. 18-21). In nontumoral areas, P, S, ently not involved. The cell walls in the and K showed lower peaks than in tumors; neoplastic tissues were distorted and no Si and Ca, on the contrary, appeared more intercellular spaces were visible. At the abundant. In fully developed tumors, the tumor periphery, typical cells containing Si peak increased considerably, becoming amorphous electron-opaque material were higher than that of the surrounding nonobserved (Fig. 5, PC). tumor area. X-Ray microanalysis showed The KPA precipitates in tumor tissues that Si was mainly concentrated in tumoral were scarcer and smaller than those ob- trichomes (Figs. 4 and 25). served in nontumor tissues, but no significant differences were noted in their distri- X-Ray TEM Microarzalysis bution in the two types of tissue. Small The spectra shown in Figs. 6, 8, and 27 deposits were visible in the vacuoles, nuclei, cytoplasm, intercellular spaces, plasma- were obtained on material prepared accordlemma and fibrils of the hydrolyzing pri- ing to method 2 and refer to: peripheral mary cell wall of xylem elements. Larger tumor cells (Fig. 5, PC), viroplasm (Fig. 7, precipitates were associated with the cell V), and cell walls (Fig. 26, CW). The elewall of xylem elements and the cuticle. In ments detected were: Si, P, and Ca in petumor tissue, KPA precipitates were rela- ripheral tumor cells (Fig. 6), and Si and P, tively abundant only over the electron- but not Ca, in the viroplasms (Fig. 8). X-Ray analysis of KPA precipitates on opaque material of peripheral cells; they xylem walls revealed the presence of Si, P, were never observed on virus particles and large amounts of Ca (Fig. 27). (Figs. 7 and 9) or viroplasm (Fig. 7).

(Fig. 4), while the tracheids normal.

X-Ray SEM Microanalysis

appeared

Analysis

qf Antinzonate Precipitates KPA

The analysis performed with the E;DS No consistent differences were noticed by analyzing tissues prepared according to system on l-pm-thick sections (Fig. 22) FIG. 6. X-Ray spectrum obtained by analysis of deposits observed in the peripheral cells (PC) of Fig. 5, showing the presence of Si, P, and Ca. FIG. 7. Tumor cells containing electron-dense material representing MRDV viroplasm (V) and virus crystals (0). No KPA deposits are present on the virus particles. FIG. X. X-Ray spectrum obtained analysing the viroplasm of Fig. 7, showing the presence of Si and P. No Ca was detected. FIG. 9. High magnification of cross-sectioned cytoplasmic tubules, showing a circular profile and surrounded by MRDV virions. One virus particle is visible inside each tubule. No KPA deposits are risible on virions or tubules.

FAV.4121 ET AI,.

P

12

13

FIGS. 10 AND 11. Spectra obtained by analyzingthe srwbces ofthe nontumoral (Fig. 10) and tumoral areas (Fig. II). Tnetissues were fixed inglutaraldehyde and critical-point dried according to method 1. FIGS. 12 MC 13. Spectra of the surface of nontumoral (Fig. 12) 2nd uf :I ti~rnoral :trea (Fig. 13). The tissues, unfixed, were frozen in liquid nitrogen. The Cr and Fe peaks are due to the support in the rlcctmn microscope.

revealed the presence of Si, P, S, Cl, and Ca (Fig. 24). The Cl peak is due to the organic chloride contained in the embedding medium (van Steveninck et nl., 1916). The Ca peak was consistently higher than those of the other elements. The map related to Ca, performed by the WDS system, reveals a prevalent concentration of the ion in the KPA deposits (Fig. 23, scintillation dots).

IJnstained ultrathin sections were floated on EDTA for variable periods of time and examined in the electron microscope. Similar sections, floated in EDTA-free medium, were examined as controls. The KPA precipitates visible in untreated tissues (Fig. 26) disappeared after a Xi-min treatment with EDTA (Fig. 28) but remained unaltered in the controls. The spectrum of

MAIZE ROUGH DWARF VIRUS TTlMORS

a

16

17

FIGS. 14 AND 15. Spectra of cross sections through a control vein (Fig. 14) and tumor (Fig. 15). The tissues were fixed in glutaraldeh~de and critical-point dried according to method 2. FICX. 16 AND 17. Spectra of cross sections through a control vein (Fig. 16) and tumor (Fig. 17). The tissues were fixed with KP.4 according to method 1.

KPA precipitates revealed a high Ca peak (Fig. 27). DISCUSSION

X-Ray microanalysis appeared to be a suitable technique to detect some differences in elemental distribution between

tumors and tumor-free areas in MRDVinfected maize leaves. Comparing the heights of peaks from spectra obtained in uniform experimental conditions, it can be observed that nontumor areas contain more Ca and Si, and less P, S, and K than young tumors. The Si content, however, increased in

364

FAVALI

ET AL.

too-, x Csntro’lllllli q TmlDut=Z.Z= I “i ‘”

19

I

60

501 4010 I

‘;I Can’rP~ /llll/l 90-

Turnour = gg

60-

70.

SO-

50

40

30

FIG. FIG. FIG. FIG.

18. 19. 20. 21.

Histogram Histogram Histogram Histogram

related related related related

to to to to

the the the the

spectra spectra spectra spectra

10 (nontumoral) 12 (~~ont.umora~) 14 (nontumoral~ 16 (nontumorai)

and and and and

11 (tumoral). 13 (tumoral). 1,5 (tumoral). 17 (tumoral).

21

MAIZE

,

I

0

2

3

ROUGH

I

/

4

5

DWARF

I

’ KeV’

VIRUS

r0

,

TUTORS

I

2

3

I

t

4

5

’ KeV

25’

FIG. 2% SEM rni~rog~~ph of thick (1 pm) unstained section of leaf tissue fixed with KPA and embedded in Epon- Araldite. Note the KPA precipitates (arrows). E‘IG. 23. Map for the Ca pe~ormed by the WDS system on the Same thick se&ion of Fig. 22. Note the concentration of the scintillation dots on the KPA deposits. FIG. 24. Spectrum obtained by EDS analysis of KPA deposits of Fig. 22. Note the high peak of Ca. FIG. 26. Spectrum obtained by analysing a single trichome (see Fig. 4, arrow). Note the high peak of Si.

FIGS. 26-28. Analysis of KPA precipitates FIG. 26. Deposits associated with the cell wall (CW) of a xylem section. FIG. 27. X-Ray spectrum obtained by analysis of the deposits Si, P, and a large amount of Ca. FIG. 28. Consecutive ultrathin section of Fig. 26 after EDTA in Fig. 26 are absent in this section floated for 15 min on a 5 mkf agent. 366

element (xy) in an unstained ultrathin visible in Fig. 26. Note the presence of treatment. The KPA deposits present solution of EDTA, a calcium-chelating

MAIZE ROUGH DWARF VIRUS TUMORS

tumors since these, when fully developed, showed a higher Si peak than nontumor areas. The larger amounts of P and S in tumors probably depend on the fact that these contain large quantities of virus particles and associated viroplasms (Bassi and Favali, 1972; Redolfi and Boccardo, 1974; Appiano and Lovisolo, 1979). It is worth noting that the distribution of K and Ca is similar to that observed in some animal tumors where viral infection is believed to alter cell membrane permeability (Grigolato et al., 1977). The direct KPA-fixation technique (Tandler and Kierszenbaum, 1971; Hayat, 1975; Favali et al., 1977, 1978) was used to precipitate the cations, and the composition of precipitation products was analyzed by both X-ray microanalysis and treatment with the Ca-chelating EDTA. After this analysis, the KPA precipitates consisted primarily of Ca salts (Diculescu et al., 1971; Ravazzola et (LI., 1976; Favali et al., 1977, 1978). In tumors, the KPA precipitates were less abl~~ldantthan in nontumor tissues with the only exception being the peripheral cells containing electron-opaque material. Similar large Ca deposits have been observed in necrotic cells of animal tumors (Grigolato et (2. ? 1977) and in necrotic local lesions induced by t.obacco mosaic virus in tobacco leaves (Favali et al., 1978). An interaction between Ca and growth reguIators has been demonstrated by several ph~~siolo~icalinvestigatioIls. Recently studied by electron microanalysis has been the effect of giberellic acid on the ion rations in a dwarfed maize mutant that exhibited normal growth after treatment with minimal al~o~lnts of giberellic acid (~eunlan~l and Janossy, 1977). The ion concentration in the dwarfed mutant was lower than in normal plants and the mutant was found unable to produce giberellic acid in detectable amounts. This deficiency might lead to changes in membrane properties and then in ion concentration, thus preventing the plant from growing normally. It may be of interest to investigate whether dwarfing in maize affected by MRDV could depend on an altered metabolism of growth regulators induced by the virus and resulting in an abnormal ion distribution.

367

X-Ray analysis combined with the other techniques described in this paper proved to be a suitable method of detecting the relative differences in ion distribution between tumor and nontumor tissues and may be further applied in the future to studies on plants showing similar peculiar alterations caused by virus infection. ACKNOWLEDGMENTS We wish to thank Dr. G. Piazzesi and G. Sala (Istituto di Fisica Tecnica, Politecnico di Milan0 and Centro “Gino Bozza” CNR I’olitecnico di Milano) for helpful assistance in computation of the data obtained by X-ray microanalysis. REFERENCES APPIANO, A., and LOWSOLO,0. (19791.Ultrastructurr of maize roots infected with maize rou& dwarf virus and presence of virus particies in vacuoles with lysosomal activity. Mierobiologieu t, 37-50. BASSI, M., and FAVALI, M. A. (1972). Electron microscopy of MRDV assembly sites in maize. Cytochemical and autoradiopaphic observations. J. f&r{. Viral. 16, 153-160. BASS, M., FAVALI, 3%.A.. and APPIANO,A. (19741.An autorddiogrdphic study of the cell wdll modifications induced by maize roug-hdwarf virus in the host cells. Riv. Pat. Veg. X, 19-25. BULGER, R. E. (1969). Use of potassium pyroantimonate in the ~o~~~i~~tion of sodium ions in rat kidney tissue. d. Cell Biot. 40, 79-94, RURGLEN.M. J. (1974). Quelques prCcisions tecniques concernant l’emploi de la coloration rQressivtr a I’EDTA. d. .~ficmscopie 21, 193-196. CLARK, M. A., and ACKER~~AN,G. A. (1971). Altcration of nuclear and nucleolar pyruantimonateosmium reactivity by glutaraldehyde fixation. d. Nistochew. C”!ytochertl.19, 3X8-390. DICCTLESCI:,I., POPESCU,L. M.. IONESCU, N., and BUTUCESCU,N. (1971). ~~lt~~structural study of cdlcium djstributio~l in cardiac muscle cells. %. %~ilfh-sch. .Ifikrosk. hat. 121, I’ll-i%. FA\-ALI, M. A., BARBIERI. N., and BONECCHI, R. (1977). Electron microscopy and X-ray microanalysis of antimonate precipitates in healthy and virus infected leaves in tobacco and maize. C~~*~o~ogi(~ JO(4). FAVALI, 11. A., BARBIERI, N., and BONECCXI, R. (1978). Electron microscopy and X-ray microanalysis of antimonate precipitates in healthy and virus infected leaves of tobacco. Cnqoiogirc 31(3), 331-342. FAVALX, I’& A., BASS. M.. and APPIANO, A. (1974). Synthesis and migration of MRDV in the host ceil: an autoradiographic study. J. Gex Viro/. 24, %3565. GEROLA, F. M., and BASSI, N. (1966). An elwtron microscopy study of leaf vein tumours from ma&t>

368

FAVALI

plants experimentally infected with MRDV. Caryologia 19, 13-40. GEROLA,F. M., BASSI, M., LOVISOLO,O., and VIDANO, C. (1966). Virus like particles in both maize plants infected with maize rough dwarf virus and vector Laodelphax striatellm Fallen. Phytopathol. Z. 56, 97-99.

GRIGOLATO, P., SALA, N., BONECCHI, R., and VALAGUZZA.E., (1977). Valutazione ultrastrutturale della membrana delle cellule neoplastiche in via di disseminazione. I1problema delle metastasi. III “Atti XIV”Congr. Naz. Sot. Ital. Patologia, Catania, Sot. Ed. Universe Roma,” pp. 95-107. HAYAT, M. A. (1975). “Positive Staining for Electron Microscopy,” pp. 84-93. Van Nostrand Reinhold, New York. LEGATO, M. J., and LANGER, G. A. (1969). The subcellular localization of calcium ion in mammalian miocardium. J. Cell Biol. 41, 401-423. LOVISOLO, 0. (1971). Maize rough dwarf virus. CMIlilAB

ld LIescrip. Planf

Viruses

72.

LOVISOLO,O., and CONTI, M. (1966). Individuazione al microscopio elettronico de1 virus de1 nanismo ruvido de1 mais (MRDV) in piante di Zea mays L. sperimentalmente infettate. Atti Accad. Sci. Torino 100, 63.

MATTHEWS, R. E. F. (1979). The classification and nomenclature of viruses. I?ltervirology 11, 133-135. MILNE, R. G., and LOVISOLO,0. (1977). Maize rough dwarf and related viruses. Aduan. VITALS Res. 21, 267-341.

NEUMANN, D., and JANOSSY,A. G. S. (1977). Effect of giberellic acid on the ion rations in a dwarf maize mutant (Zea mays L. d,). Planta 134, 151-153.

ET AL. PERSONS,E., BOLE, B., HALL, D. G., and THOMAS, W. D. E. (1974). A comparative survey of techniques for preparing plant surfaces for scanning electron microscope. J. Microsc. 101, 59-75. RAVAZZOLA, M., MALAISSE-LAGAE, F., AMBERDT, M., PERRELET, A., MALAISSE, W. J., and ORCI, L. (1976). Patterns of calcium localization in pancreatic endocrine cells. d. Cell. Sci. 27, 107-117. REDOLFI, P. L., and BOCCARDO,G. (1974). Fractionation of the double-stranded RNA of maize rough dwarf virus subviral particles. Virology 59,319-322. SHIINA, S., MIZUHIR.4, V., AMAKAWA, T., and FUTAESAKU, Y. (1970). An analysis of histochemical procedure for sodium ion detection. d. Histochern. C’ytochern. 18, 644-649.

TANDLER, C. J., and KIERSZENBAUM, A. L. (1971). Inorganic cations in rat kidney. Localization with potassium pyroantimonate-perfusion fixation. J. Cell Biol. 50, 8304339. VAN STEVENINCK,M. E., VAN STEVENINCK,R. F. M., PETERS, D. P., and HALL, T. A. (1976). X-Ray microanalysis of antimonate precipitates in barley roots. Protoplasma 90, 47-63. VIDANO, C. (1966). I1 Maize Rough Dwarf Virus in ghiandole salivari e in micetoma di Laode/phax striatellus Fall&n. Atti Accad. Sci. Torino 100, 731. WOODING,F. B. P., and MORGAN,G. (1978). Calcium localization in lactating rabbit mammary secretory cells. d. CT/trastr7&. Res. 63, 323-333. YEO, A. R., L~UCHLI, A., and KRAMER, D. (1977). Ion measurements by X-ray microanalysis in unfixed, frozen, hydrated plant cells of species differing in salt tolerance. Planta 134, 3-38.