A cytological investigation of the association between Azospirillum brasilense and some C-4 grasses


Material Information

A cytological investigation of the association between Azospirillum brasilense and some C-4 grasses
Physical Description:
ix, 102 leaves : ill. ; 28 cm.
Matthews, Sharon Williams, 1938-
Publication Date:


Subjects / Keywords:
Rhizobium   ( lcsh )
Nitrogen -- Fixation   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis--University of Florida.
Includes bibliographical references (leaves 93-101).
Statement of Responsibility:
by Sharon Williams Matthews.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000014035
notis - AAB7221
oclc - 06081515
System ID:

Full Text








I wish to express my gratitude to the following:

Dr. Stanley C. Schank, the chairman of my committee,

without whose help completion of this manuscript would have

been impossible, for his ideas, advice, and encouragement,

and especially for his patience and enduring friendship

throughout our relationship;

Dr. Rex L. Smith for advice concerning immunology,

for the root material and for the specific antiserum

against Azospirillum and.for serving on my committee;

The Department of Microbiology and Cell Science for

the use of the Biological Ultrastructural Laboratory, and

especially Dr. Henry C. Aldrich for his special efforts to

instruct and assist me in cytochemical techniques, electron

microscopy and dark room techniques, and for serving on my


Dr. Sylvia Coleman, Research Scientist, Veterans

Administration Hospital, Gainesville, for advice concerning

immunological labelling for the TEM and for supplying normal

goat serum;

Drs. Kenneth H. Quesenberry and Victor E. Green, Jr.,

for their ideas, advice and friendship, and for serving on

my committee;

Dr. Max E. Tyler, Department of Microbiology, University

of Florida, for providing experimental bacterial cultures;

My parents, Ruthe and Roy Williams, for their belief

in me, and for always lending their support when it was

needed, for whatever reason;

My husband, Larry, for his patience, love, under-

standing and support, and my children, Kristi, Kevin and

Kyp, for their encouragement, love and extensive help at


To all these people, I dedicate this manuscript.








Nitrogen-Fixation in Grasses .
Diazotrophs and Nitrogenase . .
Root-Bacteria Interactions .. .
Characteristics of Azospirillum .
Evidence of Nitrogen-Fixation in Azospirillum
Occurrence of Azospirillum .. .
Immunocytochemistry . .
Specificity of Antibodies . .
Immunocytochemical Methods for Electron
Microscopy .

Fixation of Roots for TEM . .
Embedding Procedures and Materials ..
Peroxidase-Antiperoxidase Labelling for TEM
Peroxidase-Antiperoxidase Labelling for SEM

Fixation and Embedding of Roots for TEM .
Peroxidase-Antiperoxidase Labelling for TEM
Pleomorphic Forms of Azospirillum brasilense
Invasion of Host Roots by Azospirillum .
Bacterial Longevity and Migration .
Scanning Electron Microscopy .






S 4
S 8
. 11
S 17
S 23
S. 24
S 27

S. 30

S 50

. 86























13t, 125, PAP treatment


Design of field plots . ... .40

Fixation of A. brasilense .. 53

Fixation of A. brasilense ......... 53

Fixation of root sample . 53

PAP treatment on sterile culture .. 55

A pure culture, control, no stain 55

A pure culture, antirabbit control .... 55

A pure culture, PAP control ......... 55

A pure culture, PAP treatment ... 57

A pure culture, Os04 control .... .... .57

Axenic root, 13t, PAP treatment ...... .57

Axenic root, DAB control ... .57

Axenic root, 13t, JM125A2, PAP treatment 59

Axenic root, 13t, PAP treatment ...... .59

Axenic root, control, no stain ... 59

Axenic root, PAP control ... .59

Field root, 13t, PAP treatment ... .61

Field root, PAP control . ... .61

Field root, DAB control . ... .61

20 Field root,

. 61




























Field root, 13t, PAP treatment 63

Field root, Os04 only . ... 63

Field root, antiserum control ... 63

Field root, antirabbit control 63

Axenic root, 13t, JM125A2, PAP treatment 65

Axenic root, 13t, JM125A2, PAP treatment 65

Axenic root, 13t, JM125A2, PAP treatment 65

Field root, 13t, JM125A2, PAP treatment .. 67

Field root, 13t, JM125A2, PAP treatment 67

Axenic root, 13t, JM125A2, PAP treatment 69

Axenic root, JM125A2, PAP treatment ... 69

Axenic root, pleomorphic forms ....... 71

Axenic root, pleomorphic forms 71

Field root, microorganisms, sloughed cells 73

Field root, A. brasilense in cell wall 73

Field root, bacteria in cell and cell wall 75

Field root, A. brasilense in cell .. 75

Field root, young, few microorganisms .. 77

Old field root, microorganisms in cell 77

Young field root, cortex intact ..... .77

Old field root, sloughed cells ....... 79

Young root, cortex intact . 79

Old field root, microorganisms in cell .. .79

Field root, 13t, PAP treatment 81

Field root, invasion of endodermis ..... 81


46 Field root, invasion of stele ........ 83

47 Field root, 13t, JM125A2, PAP treatment .. 83

48 Field root prepared for SEM ....... 85

49 Field root, inset of Figure 48 ..... 85

50 Spiral-shaped bacteria on root surface .85

51 X-ray analysis of osmium scatter ...... 85

Abstract of Dissertation Presented to the Graduate
Council of the University of Florida in Partial
Fulfillment of the Requirements for the
Degree of Doctor of Philosophy



Sharon Williams Matthews

March 1979

Chairman: Dr. Stanley C. Schank
Major Department: Agronomy

The objectives of this study were to experimentally

label bacterial cells using specific antigen-antibody

reactions for observation at the electron microscope level;

to verify that pleomorphic forms of Azospirillum brasilense

exist under axenic conditions; to establish the extent of

specific bacterial invasion, both intercellularly and intra-

cellularly, of the host root; and to monitor the longevity

of the bacteria in the soil and migration of the bacteria

from inoculated to uninoculated plots.

Bermudagrass and pearlmillet root samples which had

been inoculated with autoclaved or live inoculum containing

A. brasilense, strains 13t or JM125A2, were treated using

the immunological peroxidase-antiperoxidase (PAP) method.

Improved techniques were necessary so that the PAP method


could be used in electron microscope studies of field

plants. Grid holders were devised, allowing a greater

number of sections to be treated, and gelatin content was

lowered in order to reduce nonspecific staining.

Bacterial cells of Azospirillum brasilense were success-

fully labelled with PAP and identified using the electron

microscope. Positive identification was possible because

of the heavy outlining of the cells with a dense deposit of

osmium. Cells showing a negative response were not heavily

outlined. In method and specificity controls, when each

step in turn was omitted, the bacterial cells were not

heavily outlined.

Pleomorphic forms of Azospirillum brasilense were

observed in the axenic root culture. Encapsulated forms

were much larger than vibrioid forms, and both types reacted

with antiserum against either strain 13t or strain JM125A2.

Small roots usually had intact cortical regions with

few microorganisms present, while large roots contained

many microorganisms in the cortical region, and cells were

often broken or sloughed. Azospirillum brasilense strains

13t and JM125A2 were immunologically identified, both inter-

cellularly and intracellularly, in roots of pearlmillet.

The continual occurrence of both A. brasilense strains

in uninoculated root sections indicated that the organism

overwinters in north Florida soils, and migrates from

inoculated to uninoculated plots.


For centuries people who tilled the soil have been

aware that certain plants improved the land rather than

depleted it, but they did not know why. In 1886, two German

scientists, Hellriegel and Wilfarth (per Norris, 1956),

showed conclusive evidence that legumes were able to make

use of atmospheric nitrogen which was unavailable to other

plants. Two years later, in Holland, Beijerinck isolated

bacteria from legume nodules and grew them in culture (per

Norris, 1956). He called them Bacillus radicicola, and it

was many years before these nitrogen-fixing bacteria were

considered diversified enough to be placed in a genus of

their own and become known as Rhizobium. It is now known

that Rhizobium are not obligate symbionts, but free-living

saprophytes which can survive in soil for long periods of

time between chance meetings with legumes (Norris, 1956).

Since nitrogen fertilizers are necessary for growing food

and are produced only by using expensive and diminishing

petroleum fuels, microorganisms such as Rhizobium have

increased in importance, since they are capable of con-

verting atmospheric nitrogen into combined nitrogen which

can be assimilated by green plants.

The purposes of this study were to experimentally

label nitrogen-fixing Azospirillum brasilense bacterial

cells using specific antigen-antibody reactions for observa-

tion at the electron microscope level; to verify that

pleomorphic forms of Azospirillum brasilense exist under

axenic conditions; to establish the extent of specific

bacterial invasion, both intercellularly and intracellularly,

of the host root; and to monitor the longevity of the

bacteria in the soil and migration of the bacteria from

inoculated to uninoculated plots.


Most symbioses between free-living diazotrophs and

other organisms are based on a contribution of carbon com-

pounds by the nondiazotrophic host plant to the diazotroph,

usually via an exudate. In return, the diazotroph con-

tributes nitrogen to its host, usually in the form of N-

containing compounds derived from decomposition of dead

diazotroph cells, according to Burns and Hardy (1975). The

host plant may additionally provide a favorable environment

for N2-fixation by limiting the 02 pressure.

When Barber and Martin (1976) grew wheat and barley
seedlings in sterile soil and introduced 1CO2, they con-

cluded that living plant roots release considerable quanti-

ties of organic materials into the soil, with microorganisms

apparently stimulating the process. In the absence of

microorganisms, 7-13% of the total dry matter was recovered,

while in unsterilized soil, exudates increased to 18-25%.

These exudates contained diffusible, soluble compounds, non-

diffusible, mucilagenous compounds, and sloughed cells.

Barber and Gunn (1974) grew plants in glass beads and in

culture solution in order to compare the influence of solid

or liquid media on the amount of root exudation. In 1 mm

glass beads, the total exudate was equivalent to 9% of the

total dry matter of the plant, while the total exudate from

plants grown in culture solution was only 5%. Martin (1977)

labelled wheat with 4CO and recovered as much as 39% of the
exudates from the soil in the form of 14C-labelled neutral

sugars, amino acids, amino sugars, carboxylic acids and

phosphate esters. Since 1C-labelled materials were

recovered from plants grown in sterile soil, Martin concluded

that cortical tissue degenerates in the absence of micro-

organisms, but a loss of carbon from the roots is accelerated

by soil microflora. This led him to suggest three stages of

root decomposition for wheat in nonsterile soil:

1) Continual release of components with low molecular

weight from degenerate epidermal and cortical

tissue and sloughed root caps. This appears to be

normal but is accelerated by soil microflora in the

absence of invasion;

2) Invasion of epidermal and cortical tissue by soil

microflora, leading to an extensive breakdown of

cell walls;

3) Decomposition of endodermal tissue following the

death of the plant or severance of roots.

Nitrogen-Fixation in Grasses

Since the nitrogen-fixing bacteria on and around the

roots are largely dependent on the roots for their supply of


organic nutrients, nitrogenase activity of the N2-fixers

could depend on the qualitative and quantitative nature of

those root exudates. Some genotypes of plants could be more

favorable for rhizosphere nitrogenase activity than others.

Factors which affect the rate of photosynthesis, such as

light intensity, may also affect stomatal opening and the

diffusion of gasses, especially N2, through the plant. This

will also affect the quality and quantity of the root

exudates (Dommergues et al., 1973 and Balandreau et al.,

1975). Several tropical grasses, including Paspalum notatum

Flugge, Digitaria decumbens Stent., Pennisetum purpureum

Schumach., Brachiaria spp., and Panicum maximum Jacq., which

utilize the highly efficient C-4 photosynthetic pathway, are

able to fix enough dinitrogen to cover a large part of their

needs according to Dibereiner and Day (1976). However, these

grass-bacteria associations seem less specialized than legume

nodules, especially with regard to nitrogenase protection

from 02, and seem more vulnerable to environmental change

(Dobereiner, 1977).

De-Polli et al. (1977) and Dobereiner and Day (1976)

introduced 1N2 into field-grown soil-plant cores and found
low but progressive incorporation of fixed N into the roots,

rhizomes and leaves for up to 17 hours, indicating to them

that fixed N2 is transferred directly to the plant without

requiring bacterial death and decomposition. This was also

shown in a pot experiment, where fixed N2 was transferred

directly to the plant, with none available in the vermiculite

(Dobereiner and Day, 1976). In 1977, Day exposed sugarcane

(Saccharum spp.) and Paspalum notatum, the common type,

batatais to 1N2 and found that 1N was transferred to the

host plant within 24 hours, which Day believed was too rapid

an uptake to be products of bacterial decay.

When batatais was transplanted from the field into pots,

the plants grew well and were healthy and green when pro-

vided with a nitrogen-free nutrient solution (Dobereiner and

Day, 1975). Addition of 30 ppm N did not increase growth or

percent N content in the plant. The C2H2 reduction was

concentrated on the roots and was estimated to be responsi-

ble for fixing 84 mg N/pot in two months. Little activity

was found in the soil, and washing the roots did not remove

the activity. Dobereiner et al. (1972) found the bacterium

Azotobacter paspali became permanently established in the

rhizosphere of four tetraploid ecotypes of P. notatum, but

not with diploid P. notatum ecotypes or other Paspalum

species. Benzion (1978) inoculated three diploid and three

tetraploid cultivars of P. notatum with A. paspali, however,

and found a significant increase in dry matter over uninoc-

ulated cultivars in only one tetraploid. He found no

increase in inoculated batatais dry matter.

Extrapolation of nitrogenase activity measured using

the acetylene reduction assay led to an estimated maximum of

93 kg N/ha/yr (DObereiner et al., 1972). However, the

highest rates of ethylene production with maize roots

observed by Burris et al. (1978) were approximately 1/7 of

the rates of von BUlow and Dibereiner (1975). The method of

excising, washing and preincubating roots as reported by

D~bereiner et al. (1972) has been severely criticized by

other researchers. Okon et al. (1977) and Burris et al.

(1978) question whether rates are representative of fixa-

tion in the field, since a 100-fold or greater increase in

the bacterial population has been observed after a 24 hour

incubation. During preincubation, excised roots deplete the

02 supply, fermentative metabolism, shown by rapid CO2
production, is initiated, and organic acids which support

vigorous bacterial growth are probably produced (Okon et al.,

1977; Burris et al., 1978 and van Berkum and Day, 1979).

Testing this theory, van Berkum and Day (1979) found that

roots which had been washed and preincubated produced three

times as much ethylene as those which were not washed or


D'bereiner et al. (1972) found that gentle washing

decreased the activity of A. paspali only slightly, even

though about half of the bacterial cells were removed, and

even vigorous washing under a strong jet of water until no

soil was visible removed only 50% of the activity. Okon et

al. (1977) removed most of the bacterial cells from roots by

shaking the roots in buffer, but many cells remained even

after the roots were vigorously washed free of soil particles,

suggesting a close association between the bacteria and

the roots. Although Okon et al. (1977) found that surface

sterilization greatly reduced the number of Azospirillum on

roots, Tyler et al. (1979) did not observe that surface

sterilization reduced the number of bacteria more than

gentle washing, also demonstrating that the bacteria are

closely associated with the roots.

Most grasslands, varying from moist blue-green algal-

rich meadows to dry semideserts, are characterized by a

high internal rate of nitrogen cycling, with low losses and

low nitrogen-fixing requirements (Paul, 1978). Thus 5 kg

N/ha/yr is probably high for a mature system, although

higher fixation rates are often found in areas reestablishing

a vegetation. However, grasses such as Saccharum ssp.

(sugarcane) and Pennisetum purpureum (napiergrass) may yield

40 metric tons/ha of dry matter during a tropical year

(Dbbereiner, 1977). If half of this requirement were

supplied by biological fixation, at least 10% of the total

dry matter, or four tons of carbon substrate per hectare,

would be needed to support the nitrogen fixation.

Diazotrophs and Nitrogenase

Diazotrophs have little in common except for their

ability to fix nitrogen. The enzyme nitrogenase has been

found to occur only in procaryotic cells, but diazotrophs

may be aerobic, facultative aerobes, or anaerobic, and may

be photosynthetic or not (Burns and Hardy, 1975). They

vary in complexity from the primitive free-living bacterium

Clostridium to the genetically, physiologically and

structurally (Napoli and Hubbell, 1975) intricate Rhizobium

legume-root nodule symbiosis. There are diazotrophs in 26

of 200 bacterial genera, 11 of 47 families, and 3 of 10

orders. They are found both in aquatic and soil species

(Burns and Hardy, 1975).

Factors such as temperature, pH and moisture influence

the types of organisms which will dominate an environment,

as well as the extent of growth (Burns and Hardy, 1975);

therefore, although there are few habitats which altogether

exclude diazotrophs, dense populations of N2-fixers

generally propagate or decline as conditions become more or

less favorable. Soils which lack organic matter will support

only low populations of nitrogen-fixing bacteria (Burns and

Hardy, 1975). Although many associations between free-

living diazotrophs and other organisms have been documented,

these are not as easily detectable as the more specific

associations of root or leaf nodule associations, since any

manipulations of the associations are likely to disrupt the

loose association (Burns and Hardy, 1975).

Nitrogenase found in all nitrogen-fixing systems con-

tains two enzyme moieties: a molybdenum-iron protein and

an iron protein. Either one alone is catalytically inert,

but together they catalyze all the reactions attributed to

nitrogenase. The two moieties from different sources form

functional nitrogenases, even across taxonomic lines,

although the more distant the relationship, the less active

the enzyme becomes (Burns and Hardy, 1975).

When ethylene produced by the acetylene reduction assay

is converted into the potential amount of N2 fixed, gains of

up to 1 kg N/ha/day in P. purpureum (Dobereiner et al., 1975)

and 2.4 kg in the best maize lines (von BUlow and Dobereiner,

1975) were reported in Brazil using the washed-root assay.

The major ecological nitrogenase-limiting factors were soil

temperature, ammonium concentration, and 02 partial pressure.

Although von BUlow and DIbereiner (1975) indicated that soil

moisture was not important until the wilting point was

reached, Rao et al. (1978) found Azospirillum populations

greater in flooded conditions than in nonflooded, and Smith

and Schank (personal communication) observed a sharp increase

in ethylene production under flooded conditions.

Assuming 38 moles of adenosine triphosphate (ATP)

derived from one mole (180g) of glucose, and 9 molecules of

ATP to produce 6 e-, then one mole of glucose could fix 1.8

moles of N2, or one g sugar could fix 280 mg N2 according to

Mulder, provided all of the available carbohydrate were

available for N2 fixation. This condition may be possible

in nongrowing Rhizobium bacteroids, but the efficiency

values for free-living nitrogen fixers are estimated to be

one g of glucose consumed for every 5 to 20 mg of N2 fixed

(Mulder, 1975).

Root-Bacteria Interactions

Observations of field-grown roots with the scanning

electron microscope (SEM) reveal microorganisms on the root

surface, while the transmission electron microscope (TEM)

has been used to study the penetration of root tissue by

bacteria, as well as mucigel phenomena (Old and Nicholson,

1975; Rovira and Campbell, 1974; Greaves and Darbyshire,

1972). Old and Nicholson (1975) point out that with the SEM

it is usually difficult to distinguish bacteria from

inanimate particles of similar dimensions, and no immunolog-

ical methods for specific identification had at that time

been developed for the SEM.

However, as early as 1972, LoBuglio et al. reported

using latex spheres noncovalently coated with absorbed

antibody to localize antigens for SEM viewing. Linthicum

and Sell (1975) and Linthicum et al. (1974) have since

modified and simplified the system, and Molday et al. (1974)

and Molday (1976) have covalently coupled antibody to

acrylic latex spheres which can visually mark cell surface

antigen. Methacrylate spheres (Manning et al.,1975) and

Silica spheres (Peters et al., 1976) are also reported to be

effective markers for the SEM.

Rovira and Campbell (1974) report that bacteria can-

not be distinguished from clay particles and root debris at

1500X on the SEM, but at 3000 to 7000X they are easily

distinguished. Old and Nicholson (1975) felt that the SEM

gave a good indication of the amount of microbial activity

on root surfaces, since actinomycete hyphae and sporophores,

as well as spiral-shaped bacteria, could occasionally be

distinguished, but not specifically identified. They found

that electron micrographs of root sections viewed with the

TEM showed that most of the particles of bacterial dimensions

were bacteria, and the numbers corresponded well with the

numbers of microbe-sized particles viewed with the SEM.

Bacterial cover on Lolium and Plantago spp. roots was

4 to 10% by direct microscopy counting, a 10-fold increase

over corresponding plate counts. Rovira et al. (1974) and

Bowen and Rovira (1976) suggest that there is an increase

because not all the bacteria will grow on the laboratory

medium, all cells on the root may not be viable, and a

single colony may arise from a group of cells which does not

disperse, but is counted by the direct method as many


According to Diem et al. (1978a), exudation sites or

sloughing sites are not evenly distributed, and therefore

bacterial growth varies widely in space due to soil and

environmental factors. This leads to a dense colonization

of some root areas in contrast to sparse colonies in other

areas. In addition, the growth rate of bacteria in natural

environments is slower than the growth rate of the root,

due to bacteriostasis; thus colonies of bacteria should

occur behind root tips (Diem et al., 1978).

Greaves and Darbyshire (1972) examined roots of peas

(Pisum sativum L.) and mustard (Sinapis alba L.) grown

axenically, nonaxenically, and inoculated. In root hair

zones and basal root zones, especially when colonized by

microorganisms, the outer cortex and epidermis showed tissue

damage similar to that found when tissue is ineffectively

fixed, but when they compared it with undamaged apical

meristems and elongation zones, they concluded that the

damage took place before fixation and occurred naturally

during growth and aging. In both SEM and TEM studies, Old

and Nicholson (1975) and Bowen and Rovira (1976) observed

mucigel on roots with a complete epidermal layer, varying

from an almost complete cover of 1 to 10 pm, to scattered

threads of material adhering to cell surfaces. Bacteria

were usually embedded in the mucigel, either as individual

cells surrounded by a sheath, or many bacteria contained in

a complex zone. Rovira and Campbell (1974) found the

mucilaginous coating thin and compact, adhering the bacteria

firmly to the root, or more open, and surrounding the

bacteria. They suggested that the filamentous coating was

of bacterial origin, and the "sheets" of mucigel were of

plant origin. Since the bacteria were often embedded in

this mucilage, there was little bacterial mobility and

therefore large areas of root surface were devoid of micro-

bial cover.

Rovira (1956) and Bowen and Rovira (1976) found

epidermal cell junctions were an apparent preferential site

of growth and migration along the root, and suggested that

the preference corresponded with these being greater exuda-

tion sites, and therefore containing more moisture and

mucilage. Where mucigel was lacking, bacteria and actino-

mycetes colonized cells and root hairs extensively,

adhering firmly to the root surface (Old and Nicholson,

1975). The majority of root surfaces showed sloughing of

epidermal cells and sometimes cortical cells. When only the

outer surface of the epidermal cell had been lost, what

remained of the cell lumen was colonized by bacteria.

Foster and Rovira (1976) found that most of the outer

cortex had collapsed and decayed in field-grown wheat roots

by flowering time, even if cells of the stele were normal

and functional. Fahn (1974), however, reports that in many

monocotyledonous roots the cortex is not shed as long as the

root remains viable. Instead, schlerenchyma cells develop

in addition to the parenchyma cells, although intercellular

spaces caused by lysigeny and schizogeny are common in

Graminae. Furthermore, Holden (1975) studied the death of

cortical cells of barley and wheat roots growing under

natural conditions. Sloughing of cortical cells did not

occur, but nucleate cells were reduced from 100% in one-

week-old roots to 25-44% in four-week-old roots, with

nuclei disappearing from the outer layers of cortical cells


Using the SEM, Rovira and Campbell (1974) observed

sparse colonization of microorganisms on five-day-old wheat

plants growing in soil. On 14-day-old plants, bacteria

were on older parts of the root, especially on and around

root hairs. After 28 days in sand, dense colonies of

bacteria covered with mucilage were present. Root tips of

wheat and oats were almost devoid of microorganisms (Rovira,

1956 and 1973), but root hairs supported large numbers of

bacteria, and numbers increased as the age of the roots

increased. In the older basal portions of the root, the

outer cells were devoid of cytoplasm, distorted, and lysed

by bacteria. Bacterial penetration into the living host was

confined to the outermost layers of the cortical cells of

older roots, but these contained a variety of microorganisms,

including single cells and small colonies, short rods,

longer curved rods, Bacillus-like and yeast-like cells.

Ranging in size from 0.3 pm to 1.5 jm, the organisms had

smooth or crenulate cell wall margins, were sometimes

stellate, and were with or without capsules (Foster and

Rovira, 1976). Based on these observations, Foster and

Rovira (1976) outlined typical field-grown wheat root

morphology as follows:

a) Inner cortical cells: minimal amount of

cytoplasm containing few organelles, plasmalemma

and tonoplast closely associated.

b) Outer cortical cells: devoid of cytoplasm,

containing only a few osmiophilic granules.

c) Cells (especially outermost ones) are collapsed

and flattened with lysed walls, sometimes

associated with lysis by microorganisms.

According to Old and Nicholson (1975) modes of entry


1) A variety of bacterial rods, spheres and helices

channelled and perforated epidermal cell walls;

2) Some penetrated along the middle lamella between

adjacent epidermal cells;

3) When sloughing was advanced and cortical cells

were exposed or lost down to the endodermis,

bacterial colonization was extensive, and pits in

the cortex and endodermis cells functioned as

modes of entry;

4) Penetration of the root cells by endomycorrhiza

also allowed entry of other microorganisms.

Diem et al. (1978a) observed that colonization of

diazotrophs in rice roots was greatest in the first 5 cm

from the base of the plant. Surface sterilization did not

kill all of the diazotrophs in this basal zone, suggesting

that they were somehow protected. The SEM and TEM pictures

showed bacteria which appeared to be embedded in the mucigel

of the apical zone of rice roots grown in agar, but not

those grown in soil. The SEM pictures suggested that the

bacteria were protected from sterilization techniques not

only by the mucigel, remnants of dead root cells, and

crevices, but also by colonization inside the root between

cortex cells. The epidermis of rice roots growing in soil

may have been damaged, especially in the basal zone, and

thus internal colonization in this zone by diazotrophs

could be possible when decomposition of the root epidermis

and cortex occurred. Electron micrographs of soil-grown

roots showed internal colonization of the cortex in old

parts of roots, where cortical cells devoid of cytoplasm

appeared to enclose various types of soil debris and

bacteria (Diem et al., 1978a).

Characteristics of Azospirillum

The diazotrophic Azospirillum (known as Spirillum

lipoferum in the literature before Tarrand et al., 1978),

according to Dbbereiner, seems to be the major organism

responsible for nitrogenase activity occurring in the roots

of Digitaria spp., Zea mays, and several other C4 grasses

(Dobereiner, 1978). An inter- or intracellular root-

bacteria association is suggested, since nitrogen is fixed

even after surface sterilization of roots (von BUlow and

D'bereiner, 1975). Nitrogenase activity was not observed

by Dbbereiner (1978) in soils without roots, even though the

bacteria were abundant, suggesting that substantial nitrogen

fixation by Azospirillum is restricted to the root environ-

ment and dependent on available carbon substrates. However,

Tyler et al. (1979) did not find Azospirillum to be more

abundant in the rhizosphere than in soil.

Azospirillum is defined by D'bereiner and Day (1976) as

a highly motile, gram negative, curved rod with one-half to

one spiral turn. It is mainly vibrioid, but under certain

conditions helical cells may occur (Tarrand et al., 1978).

It characteristically has monotrichous flagellation in

liquid media, but on agar there is a single polar flagellum

with numerous lateral flagella of shorter wave length

(Tarrand et al., 1978), and the flagella or flagellar

apparatus is lost in older, nonmotile cultures of some

strains (Ruscoe et al., 1978). The cells usually contain

several light, electron-transparent bodies which have been

identified as poly-B-hydroxybutyrate (PHB), a common

storage product of nitrogen-fixing organisms (Burris et

al., 1978; Dobereiner and Day, 1976; Okon et al., 1976b

and Umali-Garcia, 1978), and there are often dark, electron-

dense bodies, usually near the center of the organism,

identified as polyphosphate bodies (Ph)(Umali-Garcia, 1978).

Azospirillum growth is best supported by the organic

salts malate, lactate, and succinate (Day and Dbbereiner,

1976). Umali-Garcia et al. (1978) found that the bacteria

did not grow on media containing cellulose as its main

carbon source, but they did grow in various pectin media,

suggesting a possible ability of the bacteria to hydrolyze

the middle lamella of root tissue.

Umali-Garcia (1978) observed Azospirillum as small

rods in the early exponential phase of growth, with an

increase in size in the mid-to-stationary phase. After 72

hours, the cells were smaller and more rounded. After 48

hours in tripticase soy broth, Fahreus solution or pectin

broth, Azospirillum often formed rosettes or became aligned

end-to-end. Ruscoe et al. (1978) also observed long chains

of Azospirillum in four- to five-day-old cultures. Umali-

Garcia (1978) observed aggregates of three to four cells

enclosed in a common coating, and Ruscoe et al. (1978)

reported large cyst-like structures in older and thicker

root segments where tissue was degraded.

Evidence of Nitrogen-Fixation in

Azospirillum has a typical nitrogenase which converts

N2 to NH3 (Burris et al., 1978). Addition of NH3 to an

azospirilla culture represses nitrogenase activity. The

activity of the nitrogenase in cell-free extracts of
2+ 2+
Azospirillum is found to be dependent on both Mn and Mg

a situation found elsewhere only in Rhodospirillum rubrum,

an organism whose activating factors are interchangeable with

those of Azospirillum (Burris et al., 1978; Okon et al.,


The acetylene reduction assay, whereby acetylene

(C2H2) is introduced into a nitrogen-fixing system, and

evolved ethylene (C2H4) is measured, is most widely used

for studying biological nitrogen fixation, since it is a

relatively inexpensive and rapid assay in comparison with

the Kjeldahl or 15N2 methods alternatively used. However,

as pointed out by Bouton (1977) and Benzion (1978),

acetylene reduction is a measure of nitrogenase activity at

the time of the assay only. If the bacteria are no longer

actively fixing nitrogen, dry weight produced before the

assay will not be accounted for by the assay and it will not

support yield parameters, as was observed by Smith et al.

(1978), Benzion (1978) and Bouton (1977).

Smith et al. (1976) planted 40 genotypes in the five

grass genera Digitaria, Panicum, Paspalum, Cynodon, and

Cenchrus. Yields of 'Transvala' digitgrass (Digitaria

decumbens) and guineagrass (Panicum maximum) inoculated with

Azospirillum brasilense strain 13t were 163% and 150%

greater respectively than the uninoculated controls, and

produced 61% and 80% more protein.

In 1975, Smith et al. (1977) selected for further

testing the nine genotypes which had given the best results.

The grasses were fertilized at varying N levels and inocu-

lated with either live or autoclaved A. brasilense strain 13t

in aqueous solution, at a rate of 8 X 106 cells/m of row,

and the inoculum was watered into the soil. There was no

response with Transvala digitgrass. Pearlmillet (Pennisetum

americanum), guineagrass, and buffelgrass (Cenchrus ciliaris)

showed a significant response to inoculation with live

bacteria when 30 to 80 kg of fertilizer nitrogen was added

initially. From these tests it was estimated that up to 40

kg N/ha could be replaced by inoculation. On digitgrasses

in Brazil, Schank et al. (1977) recovered 734 kg N/ha from

the best four digitgrass selections. Initial soil nitrogen

was estimated to be 346 kg, and 200 kg of N were applied,

leaving 188 kg N/ha contributed by N2-fixation or other

sources. Root isolations from adjacent plots showed a high

level of A. brasilense.

In a pearlmillet screening test with A. brasilense

strain 13t inoculant, Bouton (1977) observed a 32% increase

in dry weight and a 37% increase in total plant nitrogen

over the autoclaved inoculant control in the hybrid 'Gahi 3'.

Neither of the parents showed yield increases from inocula-


In a greenhouse experiment, Baltensperger et al. (1979)

inoculated eight turf-type Bermudagrass (Cynodon dactylon L.

Pers.) genotypes with a mixture of Azospirillum brasilense

and Azotobacter paspali. At low nitrogen fertility, inocu-

lation increased top growth by 17% and total nitrogen

content of top growth by 21%. In a similar greenhouse study

of bahiagrass (Paspalum notatum) genotypes, Benzion (1978)

inoculated with either Azospirillum brasilense or Azotobacter

paspali. Genotype P8, a diploid, and batatais, a tetraploid,

responded to Azospirillum brasilense; strain PT8, a

tetraploid isogenic to P8, responded to Azotobacter paspali

early in the experiment, but not overall; and cultivar

'Argentine', a diploid, showed a decrease in dry weight when

inoculated with A. paspali.

Ddbereiner et al. (1975) and Day (1977) observed

nitrogen-deficient leaves and low nitrogenase activity in

young sorghum and maize plants, but as the plants matured

the nitrogenase activity increased significantly and the

leaves turned dark green. Smith et al. (1978) found that

fertilizer nitrogen was required initially to produce a good

level of plant growth and photosynthesis, and for inocula-

tion response, in sandy soils. Even in highly efficient

Rhizobium-legume associations, addition of small amounts of

"starter" nitrogen has been found necessary under lowland

tropical conditions, in order to satisfy the plant's

nitrogen demand until the nodules are sufficiently developed

to fulfill the nitrogen need. This is necessary because

many tropical soils are low in available nitrogen, there is

a higher rate of nitrogen uptake under tropical conditions,

and there is a higher demand for nitrogen during early

plant growth (Kang et al., 1977). Balandreau et al. (1975)

observed that addition of up to 40 ppm (NH4)2S04 stimulated

nitrogen fixation, but higher applications caused a decrease.

However, rice plants fertilized with 40 to 60 kg N/ha were

more efficient nitrogen-fixers than those which were not

fertilized (Rao et al., 1978).

Burris et al. (1978) inoculated sterile maize seedlings

with A. brasilense and found that they evolved ethylene at

a low but measurable rate of 1 to 2 kg N/ha/growing season.

Excised roots fixed 10 kg N/ha/growing season, but, as

mentioned earlier, the authors felt that this was not

representative of field conditions since the system became

fermentative, producing organic acids on which the bacteria

thrived. At the termination of the experiment, the roots

were surface sterilized and crushed, and the recovered

bacteria evolved ethylene and had all the other character-

istics of A. brasilense (Burris et al., 1978).

Occurrence of Azospirillum

D6bereiner et al. (1976) found that approximately 60%

of soil and root samples collected between latitudes 150 N

and 230 S, excluding Kenya at altitudes above 1700 m,

contained Azospirillum spp. Panicum maximum roots showed

the most constant high incidence of Azospirillum, but the

bacteria were also found regularly on Brachiaria and

Pennisetum. Wheat, rye, maize and sorghum roots were also

consistently high (Dibereiner et al., 1976). The organism

has been found in several samples in subtropical southern

Brazil, 43 of 54 grass root samples from different sites in

Rio de Janeiro state, and 2 of 7 grass root samples from

the northern U.S.A. (Dobereiner et al., 1975; Dibereiner and

Day, 1976). Azospirillum brasilense has been isolated from

pearlmillet in Florida, from guineagrass in Ecuador, and

from maize in Venezuela (Smith et al., 1978; Schank et al.,

1976). Dobereiner recovered azospirilla from uninoculated

Wisconsin soils in 1975 and Burris et al. (1978) repeated

the recovery in 1976, demonstrating the bacteria's ability

to overwinter in soil under severe climatic conditions. In

Brazil, Dobereiner has isolated more than 150 pure cultures

of Azospirillum from roots and soils collected in Brazil,

U.S.A., Africa and Asia, which fix nitrogen at a rate com-

parable to Azotobacter (Dobereiner, 1978). Other researchers

in India, Germany, U.S.A., and Australia also report isola-

tion of this organism. Tyler et al. (1979), working with

samples from South America, Florida and western South Africa,

found a significantly lower incidence of Azospirillum than

Dibereiner reported. Nevertheless they isolated the

organism from the rhizosphere of more than 40 grass genera,

finding percentages no higher in tropical soils than under

subtropical/temperate conditions.


Under normal conditions microorganisms may be observed

under a light microscope, but it is rarely possible to

recognize them since microorganism-like artifacts may

obscure the cells. In addition, different kinds of micro-

organisms are usually indistinguishable without extensive

biochemical and physiological testing. The fluorescent

antibody technique, introduced by Coons et al. (1942),

provides a highly specific marker permitting the identifica-

tion of specific microorganisms in their natural habitats

(Schmidt, 1974; Davis et al., 1973). In this method,

fluorescein isothyocyanate (FITC) is conjugated to anti-

bodies which specifically bind to the bacteria in smears and

tissue section, after which the slide is examined using a

light microscope equipped with an ultraviolet source and

filters. The antibody-coated cells appear bright apple-

green on a dark blue or black background. A modification,

using the orange-red fluorescing rhodamine B isothyocyanate,

allows examination of different antibodies in the same cell

(Davis et al., 1973).

Animals such as rabbits, monkeys, chickens, cows,

horses, dogs, goats, sheep, and man have been used to pre-

pare the antiserum for immunological use (Coons, 1958). The

immunogens are usually injected into the skin or muscle, but

intravenous injections are sometimes used. Gels of alum,

aluminum hydroxide or aluminum phosphate, known as adjuvants,

are often injected along with the immunogen, in order to

promote the maintenance of low, effective antigen levels in

the tissue (Davis et al., 1973). The most effective

adjuvant is Freund's complete adjuvant, a water-in-oil

emulsion which contains living or dead mycobacteria. After

a single injection with Freund's, antibody formation is

detected within four to five days and continues for eight

to nine months. Immunogen dosage depends on the method.

An intravenous injection may require 100 pg, but sub-

cutaneously, using Freund's adjuvant, as little as 1-10 pg

may be required. A smaller amount is sometimes used to

"prime" the animal for a subsequent secondary response

(Davis et al., 1973).

The first encounter with an injected antigen evokes a

primary response, and antibodies may be detected in 1 to 30

days, but usually within 10-14 days when the antigen is

bacteria. A subsequent encounter with the same antigen

usually evokes a secondary response, characterized by a

lower threshold dose requirement, a shorter lag phase, a

higher rate and longer persistence of antibody synthesis,

and higher titers (Davis et al., 1973).

When titers indicate that antibody formation has

reached its peak, the animal is bled to recover the antibody-

containing serum. The serum albumen is discarded, so the

final conjugated antibody solution will be as concentrated

as possible. This fractionation is usually performed by

means of 50% saturation with ammonium sulfate, then dialysis

with buffered saline solution (Coons, 1958). Use of whole

serum or of the ammonium sulfate fractions of gamma globulins

(GG) results in high backgrounds, probably caused by soluble

aggregates. The DEAE-cellulose purified IgG or antibodies

prepared by immunoadsorption produce lower nonspecific back-

grounds (Kraehenbuhl and Jamieson, 1973).

When an unlabelled specific antiserum prepared in

rabbit is layered over a tissue section containing the

antigen, the antigenic material present will react specif-

ically with the antibody and minute amounts are deposited in

the areas of a tissue section where the antigen is present.

The FITC-labelled antirabbit GG produced in goats is reacted

with the rabbit antigens and the microdeposit of fluorescent

antibody is visible under the fluorescent microscope (Coons,


Specificity of Antibodies

Only antisera which are specific to a strain or a

species are of value. Eren and Pramer (1966) used fragments

of the fungus Anthrobotrys conoides, which captures and kills

nematodes, to prepare antibodies which were conjugated to

fluorescein. Anthrobotrys conoides fluoresced more intensely

than the 17 other hyphal species, but the antibodies were

cross-reactive with other fungi. There was less affinity

for older hyphae and spores. Bohlool and Schmidt (1970)

examined soil microorganisms from 12 Minnesota soils looking

for two strains of Rhizobium occurring naturally in these

soils. Immunofluorescence was used on contact slides which

had been placed in the soil and in cultures grown from soil

suspensions. Only one organism was isolated with strong

cross-reactivity--an actinomycete, whose morphology was

definitely distinguishable. They also found bacteria

similar to the two strains in 11 of the 12 soils. These

strains were cross-reactive with the antiserum, but on

further testing were found to be in the same serogroup, but

not identical in antigenic or other properties.

Enumerating Nitrobacter in soils is usually estimated

using most probable numbers. This method is time consuming,

lacks precision, and is of unknown validity due to the

absence of alternative methods. Rennie and Schmidt (1977)

prepared fluorescent antibodies against two

principal Nitrobacter strains and found that the antiserum

was highly specific and capable of staining all chemo-

auxotrophic nitrate-oxidizing isolates tested. When the two

methods were compared, the FA counts were found to be rapid,

precise, and give 10- to 100-fold higher counts than the MPN.

Antibodies against two Nitrobacter strains, N. agilis and N.

winogradskyi, were specific with only slight cross-

reactivity (Fliermans et al., 1974). All isolates from

natural habitats fluoresced with one antiserum or the other,

but none of the 332 anaerobic or 336 aerobic bacteria tested

were more than slightly cross-reactive. An FITC-labelled

antiserum against isolate BC of Beijerinckia was found to be

highly specific, since it was nonreactive with 54 species

tested, including 6 species of Azotobacter and 4 species

of Beijerinckia (Diem et al., 1978b).

A cytological examination of buffelgrass roots revealed

spiral-shaped and rod-shaped bacteria present in root cortical

cells (Schank et al., 1976). The spiral cells reacted with

antiserum prepared against Azospirillum brasilense, strain

13t, but the smaller rod-shaped bacteria did not. Coelho et

al. (1978) found fluorescent antibody against A. brasilense

Sp 7 to be specific for Azospirillum, but cross-reactive

with all tested strains of A. brasilense and A. lipoferum.

Dazzo and Milam (1976) prepared antiserum against A.

brasilense 13t. The antiserum did not agglutinate

Escherichia coli, Enterobacter aerogenes, Bacillus cereus,

B. megaterium, Pseudomonas fluorescens, Proteus vulgaris, or

Micrococcus luteus, but did agglutinate to varying degrees

several unidentified grass root bacterial isolates which

were culturally similar to A. brasilense and demonstrated

nitrogen-fixing potential. Of eight Azospirillum strains

tested, only strain 84 was completely negative, although the

degree of agglutination was best with 13t and Sp 7, and at

1:100 dilutions, only four of the eight were positive. In

addition, seven-day-old cultures had a three- to four-times

higher agglutination titer than did three-day-old cells

(Dazzo and Milam, 1976). The antiserum against strain 13t

also was found to be specific for sterile grass roots,

suggesting that the mode of attachment to roots and the

mechanism of plant host selection may be similar to the

mechanism believed to exist between Rhizobium and legumes,

by specific binding between the cross-reactive surface

antigens of the bacteria and the host.

Schank et al. (1979) prepared antisera against

Azotobacter paspali and against A. brasilense, strains 13t,

51e, 84, and JM125A2. They observed cross-reactivity within

A. brasilense, although strains 51e, 75, 84 and 76 were not

highly reactive with strains 13t and JM125A2, and isolates

from Florida, Venezuela and Ecuador were mostly nonreactive

with other strains. Antiserum to JM125A2 was highly

specific and did not react with one-day-old or seven-day-

old cells of Azotobacter chroococcum, Bacillus cereus,

Pseudomonas aeruginosa, or colonies of 50 unidentified soil

or root bacteria. Fungal spores occasionally reacted with

low fluorescence, but were easily distinguished from the

bacteria based on their size, round shape and thicker cell

walls (Schank et al., 1979). In equal mixtures of strains

13t and JM125A2, each strain reacted only with its homolo-

gous antiserum. There was some cross-reactivity with

isolates from Florida, Venezuela and Ecuador to the anti-

sera of strains 51e and 84.

Immunocytochemical Methods for
Electron Microscopy

With the advent of electron microscopy in the late

1940's and early 1950's, biologists acquired a means of

verifying and identifying minute particles, organelles, and

organisms. At the higher resolutions possible on the

electron microscope, too much similarity exists among

bacterial cells to identify them positively as to genus or

species. In order for an antibody label to be useful in

electron microscopy, it must have the ability to scatter

electrons; it must be attached to the antibody molecules

by means of stable linkages without altering the nature of

the antibody; it must be small enough that it will not

interfere with the mobility of the labelled antibody in

inter- and intracellular spaces; and it must be stable under

the adverse physical and chemical conditions presented by

the EM (Nakane, 1973). Ferritin was an early choice for a

sensitive antibody marker for the EM because of its high

iron content in concentrated electron-dense micelles, but

ferritin-labelled antibodies have low contrast and are not

detectable at low resolutions. Mercury, uranium, osmium

and ferrocene were also used, but metal-labelled antibodies

escape detection at high resolutions (Sternberger, 1973).

Nakane (1973) postulated that if histochemically

demonstrable enzymes could be conjugated to antibodies, the

antigens could be localized in both light and electron

microscopy. Enzyme markers can be seen at both low and

high resolutions on the EM. They may be covalently linked

to antibodies, or bound to specific antibodies by

immunologic processes (Sternberger, 1973). The enzyme must

be available in pure form, there must be an absence in the

antigenic material of endogenous enzymes similar to the

chosen enzyme, and there must be a cytochemical method for

detecting the enzyme reaction product with the light and

electron microscopes. The first enzyme to fulfill these

criteria was acid phosphatase. However, the use of

peroxidase has been surpassed by few cytochemical methods,

because of its reliability and resolution. Horseradish

peroxidase appeared to be an ideal marker for immunoglobulin

since its histochemical reaction product is visible with

both light and electron microscopes, and since it is a

relatively small globular glycoprotein (Nakane, 1973).

In the indirect method, which is used almost exclusively

(Sternberger, 1973), the tissue is first reacted with

unconjugated antibodies to the antigen which is to be

localized, then it is washed, and finally reacted with the

enzyme which has been conjugated to the antiimmunoglobulin.

This conjugation takes place in a mixture containing

bifunctional amino acid reagents which usually possess two

identical reactive groups which crosslink proteins. The

best method of coupling peroxidase to the antirabbit

immunoglobulin utilizes the carbohydrate chains located on

the surface of the peroxidase (Nakane, 1973). First, alpha

and epsilon amino groups are blocked with fluorodinitro-

benzene, then aldehyde groups are formed on these

carbohydrate chains by periodate oxidation, and the

peroxidase aldehyde can couple to proteins that have avail-

able amino groups. This method results in 99% of the

immunoglobulin becoming labelled with at least one

peroxidase when they are reacted in the molar ratio of 3

peroxidase to one immunoglobulin, with retention of 98% of

peroxidase activity and 92% of immunological activity.

Immunocytological staining for the EM can be carried

out either on the tissue block prior to embedding, or

directly on ultrathin sections. However, standard fixation

of the material may interfere with the ability of the

antigens to interact with their antibodies, so a fixation

must be chosen that will adequately preserve the anti-

genicity and at the same time preserve the fine structure

of the cells (Sternberger, 1973; Kraehenbuhl and Jamieson,

1973). For example, membranous structures which would be

preserved by osmium tetroxide postfixation are not always

visible when cells are fixed in aldehyde only (Moriarty and

Halmi, 1972).

For detection of antigen "X," the tissue is first

reacted with a specific antiserum to "X," produced in a

suitable species such as rabbit. After washing to remove

the nonspecific proteins, the tissue is reacted with an

antirabbit IgG produced in another species such as sheep.

The terminal FAb portion of the anti-IgG reacts as an anti-

body with the localized Fc portion of the rabbit anti-"X"

antigen. The sheep anti-IgG is used in excess so that only

one FAb reacts, leaving the other FAb free, and providing

an unreacted combining site specific for rabbit IgG. In

this step, several sheep antirabbit IgG molecules react

with each molecule of rabbit anti-"X," thereby providing a

greater number of reaction sites in future steps, and

leading eventually to higher electron density, In the third

step purified rabbit antiperoxidase (anti-PO) is applied,

reacting as an antigen with the free FAb site of the anti-

IgG, via antigenic determinants on the Fc region. Next, the

PO is applied. This reacts as an antigen with the specific

combining sites of anti-PO at the termini of the FAb portions.

Finally, the bound enzyme is stained with 3, 3'

diaminobenzidine (DAB), followed by osmium tetroxide (Os04)

(Sternberger, 1973). The separate applications of the anti-

PO and the PO have now been combined into a single step,

with the anti-PO serving as the antigen which reacts with

the sheep antirabbit IgG, and the PO moiety providing the

immunohistochemically stainable enzyme. There are advantages

in combining the anti-PO and the PO into a single peroxidase-

antiperoxidase molecule (PAP). First, when excess free PO

is used, there is nonspecific binding to some of the tissue

components, but not when PAP is used. Secondly, preparation

of purified anti-PO is a low yield procedure, but prepara-

tion of PAP gives high yields (Sternberger, 1973).

The PAP complex is homogenous and contains peroxidase

and antiperoxidase in a 3:2 ratio. It is usually a flattened

ring or pentagon 200 to 300 A in diameter (Sternberger,

1973; Moriarty and Halmi, 1972). Corners one and three of

the pentagon are the Fc fragments of anti-PO, while corners

two, four and five are PO. Undiluted PAP may be stored

frozen for over 18 months without a change in its physical

or staining properties, but it is not stable beyond a few

weeks at 5C, and dilute PAP is unstable (Sternberger,


According to the immunohistochemical standards set by

Coons, there are many method and specificity controls. For

method controls, omission of the goat antirabbit serum, DAB,

or PAP results in no staining or nonspecific staining. In

specificity controls, omission of the antiserum specific to

the antigen, substitution of a normal serum, or reacting

tissues devoid of the specific antigen with the specific

antiserum should result in no staining or nonspecific

staining (Sternberger, 1973; Moriarty and Halmi, 1972).

When specific antiserum, IgG serum, PAP and the enzyme

stain DAB were first sequentially applied to ultrathin

sections prepared for the EM (Sternberger, 1973), localiza-

tion of electron-dense deposits were nonspecific, including

even parts of the sections devoid of tissue. This was not

observed when the enzyme-conjugated antibody method was used.

It was found that tissues contained components which could

react nonspecifically with the applied proteins and, in

addition, each of the protein reagents used in the unlabelled

antibody enzyme method attached nonspecifically to embedding

medium. Normal sheep or goat serum was unreactive with any

of these reagents; therefore, when the section was pre-

treated with normal serum, nonspecific absorption by

subsequent proteins was prevented (Sternberger, 1973).

Kraehenbuhl and Jamieson (1973) localized secretary

proteins in bovine exocrine pancreatic cells which were

present in high concentration in zymogen granules and found

that approximately 23% of the granule content was

trypsinogen A. A monospecific antiserum against the

trypsinogen A was prepared and found to be nonreactive to

chymotrypsinogen. The antigen was then localized in

granules, in the acinar lumin, in the Golgi apparatus, and

in condensing vacuoles, using surface localization

techniques. However, low sensitivity, one of the limita-

tions of this technique, was indicated, since labelling of

the cisternae of the rough endoplasmic reticulum was weak.

Moriarty and Halmi (1972) used unlabelled antibody

and PAP complex to identify the adrenocorticotropin secreting

cell in the anterior pituitary lobe of the rat, and to

localize the adrenocorticotropin thyroid hormone (ACTH) in

ultrathin sections. Faint nonspecific staining was observed

at low dilutions of antisera (1:10 or 1:20), but was not

observed if the dilution was 1:100 or higher. Attempts to

circumvent this nonspecific staining by increasing normal

goat serum or by washing in 1% normal goat serum were only


partially successful. Nakane (1973) also found that the

reaction products had a tendency to deposit nonspecifically

on the surfaces of the ultrathin sections when he localized

a growth hormone, prolactin, the ACTH and luteinizing

hormone from the anterior pituitary gland of the rat. He

found nonspecific labelling was reduced when the reaction

was carried out in a constant flow of the substrate.


Bacteria and root specimens were collected from the

following four sources for fixation and embedding in epoxy

in preparation for viewing on the TEM:

Source 1) Azospirillum brasilense 48-hour-old pure

cultures of strains 13t and JM125A2 were

obtained from Dr. Max E. Tyler, Dept. of

Microbiology, University of Florida,

Gainesville. Cultures were fixed in 3%

glutaraldehyde and in 1.5:1 acrolein and

glutaraldehyde for 2-1/2 hours. Bacteria

were centrifuged and pellets were resus-

pended in hot 2.5% agar. After rinsing,

the centrifuged pellets of agar containing

bacteria were dehydrated in ethanol and

acetone. Bacteria were embedded in Spurr's

epoxy (Spurr, 1969) or Luft's araldite

(Luft, 1961).

Source 2) Root samples were collected from a green-

house experiment in which turf-type

bermudagrass genotypes were inoculated with

autoclaved or live Azospirillum brasilense

Source 3)

Source 4)

(Baltensperger et al., 1979). Samples were

cut into 1 cm segments, fixed, dehydrated,

and embedded in Spurr's epoxy.

Root samples were collected from an experi-

ment conducted by Schank (unpublished),

in which the hybrid pearlmillet, Gahi 3, was

grown in an axenic system and inoculated

with autoclaved or live Azospirillum

brasilense. All of the systems received

both 13t and JM125A2 strains. The samples

were cut into 1 cm segments, fixed,

dehydrated, and embedded in Luft's araldite.

Pearlmillet root samples were collected

from a field experiment conducted by Smith

et al. (1976) in which there were eight

replicates of eight treatments in a com-

pletely randomized design (Fig. 1).

Treatments number 1 and 5 received no

nitrogen fertilizer, treatments 2 and 6

received 20 kg N/ha, treatments 3 and 17

received 40 kg N/ha, and treatments 4 and 8

received 80 kg N/ha. In addition, treat-

ments 1 through 4 were inoculated with A.

brasilense strain 13t which had been

autoclaved, while treatments 5-8 received

live inoculum. Each plot consisted of three

101-7 301-8 501-2 701-4

102-5 302-3 502-6 702-5

103-1 303-4 503-1 703-2

104-8 304-5 504-5 704-8

105-6 305-1 505-3 705-3

106-3 306-7 506-8 706-1

107-2 307-6 507-4 707-6

108-4 308-2 508-7 708-7

Range Range Range Range
1 2 3 4

201-4 401-2 601-5 801-7

202-2 402-7 602-1 802-6

203-7 403-1 603-3 803-8

204-6 404-8 604-4 804-5

205-5 405-4 605-6 805-4

206-3 406-3 606-8 806-2

207-1 407-5 607-2 807-3

208-8 408-6 608-7 808-1

Figure 1. Field experiment. Completely randomized design.
Each plot contained 3 rows of pearlmillet. Row
1 was a border. Row 2 and Row 3 received
treatment. Treatments #1 + 5 = no nitrogen.
Treatments #2 + 6 = 20 kg N/ha. Treatments
#3 + 7 = 40 kg N/ha. Treatments #4 + 8 = 80 kg
N/ha. Treatments #1-4 received live Azospirillum
brasilense, strain 13t. Treatments #5-8 received
autoclaved Azospirillum brasilense, strain 13t
(Smith et al., 1976).

rows of pearlmillet; the first row was a

border row, and the second and third rows

received the treatment. This field was

planted for three successive years with the

same design, i.e., uninoculated plots

(those receiving autoclaved inoculum) were

always located in the same place. In order

to avoid variation due to plant or environ-

ment, all samples were taken from the fourth

plant in plots 302-3 (40 kg N, autoclaved

inoculum) and 306-7 (40 kg N, live inoculum).

The samples were collected 4, 6, 12, 24,

and 96 days after inoculation, and designated

300 (uninoculated) or 700 (inoculated) plus

days after inoculation, as, 396 and 796.

The samples were fixed in 1.5:1 acrolein

and glutaraldehyde (Hess, 1966), dehydrated,

and embedded in Spurr's epoxy or Luft's


Fixation of Roots for TEM

In order to test the effectiveness of various fixatives

on both bacteria and root material, approximately 1 cm root

segments were collected from Source 2 and treated as follows:

1) Two root samples were placed in 5% glutaraldehyde

in 0.1 M Na cacodylate buffer, pH 7.2-7.5, with

three drops/ml H202 added, for two hours at room

temperature (Peracchia and Mittler, 1972),

followed by one hour in 5% glutaraldehyde.

2) Two root samples were placed in a dish with Os04

fumes for five minutes (Aldrich;* personal communi-

cation), then into a Karnovsky's fixative

(Karnovsky, 1965) which was modified to contain

2-1/2% glutaraldehyde, 2-1/2% paraformaldehyde,

and 0.1 M Na cacodylate buffer, pH 7.2-7.5, for

two hours.

3) Two root samples were placed in a mixture of 3%

acrolein, 2% glutaraldehyde, and 0.1 M Na cacodylate

buffer for two hours (Hess, 1966).

4) Two root samples were placed in a mixture of 1.5%

acrolein, 1% glutaraldehyde, and 0.1 M Na

cacodylate buffer for two hours, a modification of

Hess (1966).

5) Three root samples were placed in the modified

Karnovsky's fixative, containing 2-1/2%

glutaraldehyde, 2-1/2% paraformaldehyde, and 0.1

M Na cacodylate buffer, for two hours.

6) Two uninoculated, sterile root samples were placed

in the modified Karnovsky's fixative for two hours.

*Aldrich, H.C., Professor, Dept. of Microbiology and
Cell Science, University of Florida, Gainesville.

Samples were rinsed three times each in 0.2 M Na

cacodylate buffer, placed in 2% Os04 in 0.1 M Na cacodylate,

and refrigerated overnight. Samples were rinsed in 0.2 M Na

cacodylate, then in distilled H20, and dehydrated in ethanol

for 10-15 minutes each in steps increasing by 10%. After

45 minutes in 100% acetone, the samples were polymerized in

Spurr's epoxy in a 600C oven for eight hours. Sections

were cut with a Dupont diamond knife mounted on an LKB

Huxley microtome, placed on copper grids, and poststained

with lead citrate and uranyl acetate prior to viewing on the

Hitachi HU 11E (75kV).

Embedding Procedures and Materials

Basically two procedures of fixation and dehydration

were used:

1) For TEM morphological studies, the samples were

post-fixed with 1% Os4. Uranyl acetate was

included, during the dehydration, in the 75%

ethanol. After dehydration was completed,

including 45 minutes in 100% acetone, samples were

placed in 30% Spurr's epoxy and 70% acetone for

one hour, then 70% Spurr's epoxy and 30% acetone

for one hour, followed by one hour in 100% Spurr's

before polymerization for 8-12 hours in Beem

capsules in a 600C oven. Ultrathin sections were

cut, using a Dupont diamond knife, on an LKB

Huxley microtome, and viewed on a Hitachi HU 11C

or HU 11E (75kV) transmission microscope after

poststaining in lead citrate and uranyl acetate

in order to improve contrast. Identification of

bacterial cells with antigenic methods was not

possible using the above procedure.

2) For samples that were to receive cytochemical

treatment post-fixation in 1% Os04 was omitted,

as was the uranyl acetate step during dehydration.

After dehydration was completed, including 45

minutes in 100% acetone, samples were placed in a

mixture of 30% Luft's araldite and 70% acetone

for one hour, then 70% Luft's araldite and 30%

acetone for one hour, followed by one hour in 100%

Luft's araldite, a mixture of 27 ml araldite 502,

23 ml DDSA, and 1 ml DMP-30, before polymerization

in Beem capsules for three days in a 45C oven.

Ultrathin sections were cut using a Dupont diamond

knife mounted on an LKB 8800 Ultrotome III, and

placed on nickel grids in preparation for running

the immunocytochemical peroxidase-antiperoxidase

test before viewing on the Hitachi HU 11C or HU

11E (75kV) or the Jeol Co. JEM 100 C/X SEG (80kV)

transmission microscopes.

Peroxidase-Antiperoxidase Labelling for TEM

Grid holders were especially designed for ease in

manipulation of the specimens throughout the procedure.

Approximately 10 cm lengths of monofilament nylon fishing

line were threaded through the edges of 1 cm segments

of clear Tygon tubing, od 8 mm. Slits were cut in the edge

of the Tygon ring with a razorblade, deep enough to hold

the nickel grids securely, but not deep enough to cover the


Immediately before treatment with PAP, bacteria or

root samples, previously embedded in Luft's araldite, were

sectioned on an LKB 8800 Ultrotome III, placed on nickel

grids, and mounted on the grid holders. After etching for

30 minutes on 10% H202 the samples were rinsed in distilled

H20 for 30 minutes. Next they were rinsed in a jet of

saline-Tris buffer, pH 7.6, containing 0.05% gelatin

(Sternberger, 1973), and placed in normal goat serum

(Sternberger, 1973) diluted 1:25 with buffer. After

rinsing in a jet of buffer, the grids were rinsed in a 100

ml beaker half-full of buffer on a magnetic stirrer for 5

minutes. This rinsing procedure was used after each of the

treatments. Next the sections were placed in various

dilutions of anti-Azospirillum serum, prepared in rabbits

against strain 13t or strain JM125A2 (Schank et al., 1979)

for 5 minutes at 37C. The sections were rinsed and then

immersed in a 1:10 dilution of antirabbit serum for 5

minutes, followed by a thorough rinsing and a 5 minute

immersion in peroxidase-antiperoxidase diluted 1:10 with

buffer. After rinsing, the sections were suspended and

continuously agitated in 100 ml DAB (Moriarty and Halmi,

1972; S. Coleman, personal communication*). Then the

sections were rinsed 5 minutes in buffer, followed by two

changes of distilled H20 for 5 minutes. The sections were

immersed in 1% aqueous Os04 for 30 minutes under a hood,

then rinsed in three to four changes of distilled H20 on a

magnetic stirrer over a period of 15 to 20 minutes. Each

grid was rinsed with a jet of approximately 20 ml H20 as it

was removed from the grid holder. One section from each

treatment was poststained with lead citrate and uranyl

acetate in order to improve cellular detail (Moriarty and

Halmi, 1972).

Methodology controls included omission of antirabbit

serum, PAP, and DAB. Controls for specificity included

omission of either 13t or JM125A2 antiserum or substitution

of Azotobacter antiserum for the 13t or JM125A2 antiserum.

Peroxidase-Antiperoxidase Labelling for SEM

A soil core near the crown was taken from Source 4

(Fig. 1), plot 204-6. Root samples were cut into one-half

*Coleman, Sylvia, Research Scientist, Veterans
Administration Hospital, Gainesville.

to three-quarter inch segments and placed in glutaraldehyde

and tris buffer for one hour. Roots were rinsed three

times, 10 minutes each, then placed in a 1:40 mixture of

antiserum against Azospirillum strains 13t and JM125A2 for

10 minutes, followed by a 10 minute rinse in three changes

of buffer. Next the samples were treated for 10 minutes in

1:10 antirabbit serum, followed by 10 minutes in 1:10

peroxidase-antiperoxidase. The roots were wrapped in cheese-

cloth and suspended in DAB for 3 minutes on the magnetic

stirrer, rinsed in buffer, rinsed in distilled water, and

placed in 1% aqueous Os04 for one hour. Following the

dehydration schedule to 100% ETOH, the roots were dried in

a Pelco Model "H" Critical Point Dryer and gold-coated in an

Eiko 1B-2 Ion Coater in preparation for viewing on the

Hitachi S-450 scanning electron microscope.


Fixation and Embedding of Roots for TEM

Mixtures of acrolein and glutaraldehyde, both 3:2 and

1.5:1, gave good to excellent fixation of both bacteria and

cell contents (Figs. 2, 3 and 4). Since all of the other

fixatives tried gave poor fixation, with deformed or missing

cortical cells, inferior bacterial fixation, and disrupted

cellular contents, all subsequent samples were fixed in

1.5:1 mixtures of acrolein and glutaraldehyde.

The practice of embedding three to six root segments

in each Beem capsule allowed the screening of many sections,

thereby diminishing the difficulty encountered when the

viscous araldite epoxy did not completely impregnate the

hard roots. Longer periods of impregnation did not seem to

increase the degree of infiltration, but greater numbers of

roots per section yielded root sections with satisfactory

fixation and penetration.

Peroxidase-Antiperoxidase Labelling for TEM

Bacterial cells of Azospirillum brasilense, both

strains 13t and JM125A2, were successfully labelled with PAP

and identified using the electron microscope. Bacteria

which were heavily outlined with an electron-dense deposit

of osmium were labelled specifically and positively as

Azospirillum (Figs. 5, 9, 11, 13, 17, 20 and 21). Bacteria

which were not heavily outlined were considered negative

(Figs. 6-8, 10-12, 14-16, 18-20, and 22-24). When root

sections from Source 3 were treated with the PAP, all

bacteria did not show a positive reaction to a mixed anti-

serum against strains 13t and JM125A2 (Figs. 13 and 25-27).

However, methodology controls, including omission of PAP

(Figs, 8, 16 and 18), omission of DAB (Figs. 12 and 19),

and omission of antirabbit serum (Figs. 7 and 24), were

entirely negative.

In spite of extensive washing procedures and the use

of normal goat serum initially to circumvent nonspecific

staining of the sections, this was still a problem. By

careful elimination of antirabbit serum, PAP, and DAB

individually, the most likely cause of the light grey to

dark background was considered to be osmium (Fig. 28).

Additionally, there was frequently scattered residue with a

black net-like appearance (Fig. 29) which was missing on

sections when DAB was omitted, and so was attributed to non-

specific adherence of the DAB.

Pleomorphic Forms of Azospirillum brasilense

Pleomorphic forms of Azospirillum brasilense were

observed in the axenic root culture (Figs. 26 and 30-33).

The encapsulated forms were much larger than the vibrioid

forms, and contained one, two, or more cells completely

surrounded by a capsule. Both vibrioids and encapsulated

forms reacted when antiserum against either 13t (Fig. 33)

or JM125A2 (Fig. 31) was used, indicating that both types

were actually Azospirillum. Since field-grown samples were

more complex and contained a large variety of microorganisms,

an attempt was not made at this time to identify encapsu-

lated forms in the field-grown pearlmillet roots.

Invasion of Host Roots by Azospirillum

Azospirillum brasilense, strains 13t and JM125A2, were

immunologically identified intercellularly and intra-

cellularly in roots of field-grown pearlmillet (Figs. 35,

44 and 47). Small roots appeared to have intact cortical

regions with few microorganisms present (Figs. 38, 40 and

42). Large roots contained many microorganisms in the

cortical region (Figs. 34, 36, 37, 39 and 43), and they

were sometimes in the endodermal (Fig. 45) and stelar (Fig.

46) cells. Where microorganisms and soil particles were

abundant, cells were broken or sloughed (Figs. 34, 41 and


Bacterial Longevity and Migration

Both Azospirillum brasilense strains, 13t and JM125A2,

were immunologically identified in field-grown root sec-

tions which had been inoculated with autoclaved strain 13t.

Strain JM125A2 was originally isolated from Florida soil

in 1976; therefore, its presence in treated samples sug-

gested that the strain was able to overwinter and survive

at least three years in north Florida soils. Strain 13t is

not indigenous and it was concluded that this strain could

occur in uninoculated plots only if it had migrated from

plots which had been inoculated with live Azospirillum

brasilense 13t.

Scanning Electron Microscopy

Bacterial cells were clearly seen on the surface of

plant roots (Figs. 48, 49 and 50). However, positive

identification using immunological techniques was not

accomplished using the SEM. If osmium were deposited

specifically on the bacterial cell wall, as PAP methods

previously demonstrated with TEM, then X-ray analysis should

show a higher concentration of osmium on bacteria than on

plant tissue. Osmium counts were scattered (Figs. 50 and

51) using X-ray analysis of SEM sections, indicating a non-

specific deposition of osmium.

Figure 2.

Figure 3.

Azospirillum brasilense. Fixation in
3.0 acrolein: 2.0 glutaraldehyde. Uranyl
acetate and lead citrate poststain.
Ph = polyphosphate bodies; PHB = poly-B-
Mag. X 60,000.

Azospirillum brasilense. Fixation in
3.0 acrolein: 2.0 glutaraldehyde. Uranyl
acetate and lead citrate poststain.
Mag. X 67,500.

Figure 4. From sample 396. Fixation in 1.5 acrolein:
1.0 glutaraldehyde. Uranyl acetate and
lead citrate poststain. Note cytoplasmic
contents in stele and intact cortical cells.
Mag. X 2,600.




t" 3

I d


1 '





Figure 5. A pure culture, JM125A2, treated with
JM125A2 antiserum and PAP. Heavy outlines
(double arrows) denote positive reaction.
Mag. X 38,800.

Figure 6.

Figure 7.

Figure 8.

A pure culture, JM125A2, without treatment
or poststain. No reaction.
Mag. X 35,000.

A pure culture, JM125A2, antirabbit serum
omitted. No reaction (bacteria denoted
by 'n').
Mag. X 47,000.

A pure culture, JM125A2, PAP omitted. No
reaction (bacteria denoted by 'n').
Mag. X 47,100.

- S ,,

4 a-



; n

n1 N

p~' i-

Figure 9.

A pure culture, JM125A2, treated with
JM125A2 antiserum and PAP. Heavy outline
(arrows) denotes positive reaction.
Mag. X 55,000.

Figure 10. A pure culture of JM125A2, Os04 omitted.
No reaction (bacteria denoted by 'n').
Mag. X 47,000.

Figure 11.

Figure 12.

Axenic root culture, treated with 13t
antiserum and PAP. No reaction denoted by
'n', positive reaction denoted by 'p'
Mag. X 42,800.

Axenic root culture, DAB omitted. No
reaction (bacteria denoted by 'n').
Mag. X 24,000.

r; .-*
I-r I

1% ':





4 r


1 -

* i


~' '~:~ "~1~;

Figure 13.

Figure 14.

Figure 15.

Figure 16.

Axenic root culture, treated with JM125A2
antiserum and PAP. No reaction denoted by
'n', positive reaction denoted by 'p'.
Mag. X 30,000.

Axenic root culture, treated with 13t
antiserum and PAP. No reaction (bacteria
denoted by 'n'). Soil particles denoted
by 'S'.
Mag. X 49,600.

Axenic root culture, without treatment or
poststain. No reaction (bacteria denoted
by 'n'). Soil particles denoted by 'S'.
Mag. X 32,000.

Axenic root culture, PAP omitted. No
reaction (bacteria denoted by 'n'). Soil
particles denoted by 'S'.
Mag. X 30,000.

-~~ ~E



r S





rC A

LIuE ;t

S .-4


* -4



J,; n


Figure 17.

Figure 18.

Figure 19.

Figure 20.

Sample 396, treated with 13t antiserum and
PAP. Positive reaction denoted by 'p';
'CW' denotes cell wall in root cortex.
Mag. X 42,000.

Sample 396, PAP omitted. No reaction
(bacteria denoted by 'n'); 'S' denotes soil
Mag. X 24,500.

Sample 396, DAB omitted. No reaction
(bacteria denoted by 'n'); 'S' denotes soil
particles; 'cw' denotes root cortical cell
Mag. X 35,000.

Sample 396, treated with 1:40 mixture of
13t and JM125A2 antisera. No reaction
denoted by 'n'; positive reaction denoted
by 'p'
Mag. X 32,300.








.. *Y-


. *


+ l

4 t






; ~ f


0,- *
* ^ .

Figure 21.

Figure 22.

Figure 23.

Figure 24.

Sample 396, treated with 1:20 13t antiserum
and PAP. Positive reaction denoted by 'p'
Mag. X 87,500.

Sample 396, poststained with OsO only. No
reaction (bacteria denoted by 'n ); 'S'
denotes soil particles; 'CW' denotes root
cortical cell wall.
Mag. X 16,300.

Sample 396, anti-Azospirillum serum omitted.
No reaction (bacteria denoted by 'n');
'S' denotes soil particles; 'CW' denotes
root cortical cell wall.
Mag. X 35,000.

Sample 396, antirabbit serum omitted. No
reaction (bacteria denoted by 'n'); 'CW'
denotes root cortical cell wall.
Mag. X 35,000.

ri luje*






- MR


Figure 25.

Figure 26.

Figure 27.

Axenic root culture, treated with 1:100
dilution of mixture of 13t and JM125A2
antisera. No reaction denoted by 'n';
positive reaction denoted by 'p'; 'S'
denotes soil particles.
Mag. X 9,500.

Axenic root culture, treated with 1:100
dilution of mixture of 13t and JM125A2
antisera. No reaction denoted by 'n',
positive reaction denoted by 'p'; 'E'
denotes encapsulated bacteria.
Mag. X 9,300.

Axenic root culture, treated with 1:100
dilution of mixture of 13t and JM125A2
antisera. No reaction denoted by 'n';
positive reaction denoted by 'p'; 'rh'
indicates root hair; 's' denotes soil
Mago X 9,500.


q; .T. I



1: J rh
^ rt

.a L



4p -






Figure 28. Sample 396, treated with 1:40 dilution of
mixture of 13t and JM125A2 antisera. Light
grey to dark grey spots probably OsO4 (small
arrows). Positive reaction denoted
by 'p'; 's' denotes soil particles.
Mag. X 8,000.

Figure 29. Sample 396, treated with 1:40 dilution of
13t and JM125A2 antisera mixture. Light
grey to dark grey spots probably OsO4. Net-
like black area attributed to DAB residue
(double arrows). Positive reaction denoted
by 'p'; bacteria with no reaction denoted
by 'n'.
Mag. X 9,800.

,CAV4 "R'! :

T, T J.


T k "F' V1,r' `

J4 r r51

Figure 30. Axenic root culture, treated with 1:100
dilution of 13t and JM125A2 antisera
mixture. Pleomorphic forms, 'E' denotes
encapsulated forms; 'V' denotes vibrioid
forms; 'S' denotes soil particles.
Mag. X 7,900.

Figure 31. Axenic root culture, treated with 1:20
JM125A2 antiserum and PAP. Pleomorphic
forms. Positive reaction denoted by 'p';
's' denotes soil particles.
Mag. X 11,000.




A 'I g.,


V 4 31

Figure 32. Axenic root culture, poststained with
uranyl acetate and lead citrate. 'E'
denotes encapsulated forms; 'V' denotes
vibrioid forms; 'S' denotes soil particles.
Mag. X 7,900.

Figure 33. Axenic root culture, treated with 13t
antiserum and PAP. Positive reaction
denoted by 'p'; 'n' denotes no reaction;
's' denotes soil particles.
Mag. X 25,500.


t^il ,



- I

* ,
4- 4



Figure 34.

Figure 35.

Sample 396, poststained with uranyl
acetate and lead citrate. Large numbers
of microorganisms in cortical cells.
Bacteria denoted by 'b'; 'cw' denotes
cell wall; 's' denotes soil particles.
Mag. X 2,700.

Sample 396, treated with 1:4 dilution of
13t and JM125A2 antisera mixture and PAP.
Positive reaction denoted by 'p'; 'n'
denotes no reaction; 'cw' denotes root
cortical cell wall.
Mag. X 7,400.


"tC WW
b .'


* I

k b b-- h-

- 7 -
Ib- ..4


q*,c c

r '-?*
it i
"* "." ''- .,
,- ^ \- --.^ -



Figure 36.

Figure 37.

Sample 396, poststained with uranyl
acetate and lead citrate. Bacteria in
cortical cell and cell wall. 'F' indicates
fungal spore; 'b' indicates bacteria; 'cw'
denotes cell wall.
Mag. X 12,000.

Sample 396, treated with 1:20 dilution of
13t antiserum and PAP. Positive bacteria
denoted by 'p'; 'n' indicates no reaction;
'cw' indicates cell wall.
Mag. X 9,600.

. 4.1







13 lI "* *

* *,



* S



-. *

Figure 38.

Figure 39.

Figure 40.

Sample 396, poststained with uranyl acetate
and lead citrate. Young root, containing
few microorganisms. Whole cortical cells
without cytoplasm. Cytoplasm present in
stele. 'S' indicates stele; 'E' denotes
endodermis; 'C' denotes cortex; 'cw' denotes
cell wall.
Mag. X 2,000.

Sample 396, poststained with uranyl acetate
and lead citrate. Old root. Microorganisms
in cell wall (cw) and in cells. Bacteria
indicated by 'b'
Mag. X 1,500.

Sample 396, poststained with uranyl acetate
and lead citrate. Young root. Cortex intact,
no cytoplasm. Stele (S) contains cytoplasm.
Few microorganisms. 'E' indicates endodermis;
'b' denotes bacteria; 'N' denotes nucleus.
Mag. X 2,600.




- 9r .

S ...

- -
S lC

Figure 41.

Figure 42.

Sample 396, poststained with uranyl acetate
and lead citrate. Sloughed cortical cells,
many microorganisms. 'F' indicates fungal
spores or hyphae; 'cw' denotes root cell wall;
'b' indicates bacteria; 's' denotes soil
Mag. X 3,400.

Sample 396, poststained with uranyl acetate
and lead citrate. Young root. Few micro-
organisms; cortex (C) intact; 'E' indicates
endodermis; 'S' indicates stele' 'b' denotes
Mag. X 2,000.

Figure 43. Sample 396, poststained with uranyl acetate
and lead citrate. Old root. Presumed
endomycorrhiza inside cells (F); bacteria (b)
in cells. Soil particles denoted by 's'.
Mag. X 2,400.

t 'a-. Yi



Figure 44.

Figure 45.

Sample 396, treated with 1:20 dilution of
13t antiserum and PAP. 'CW' indicates
cell wall; 'p' denotes positive reaction;
'n' denotes no reaction.
Mag. X 9,000.

Sample 396, treated with uranyl acetate
and lead citrate. Invasion of endodermis
by soil particles (s); 'cw' denotes cell
Mag. X 4,500.


e .-n




* A




a. I_"
.~ < \

j'" &4




* b

I "

o. <* t

'Ii V ,








Figure 46. Sample 396, poststained with uranyl acetate
and lead citrate. Invasion of stelar cell
by bacteria (B); cell wall denoted by 'cw'.
Mag. X 33,000.

Figure 47. Sample 396, treated with 1:40 dilution of
13t and JM125A2 antisera mixture. Bacteria
in cell wall (CW). Positive reaction
denoted by 'p'; 'n' denotes no reaction.
Mag. X 36,800.


ac I

.I ,

, -,
* ,

- -.

** ..' ..a,,."
*C '
*ria q.

At '



'' *,--

5 '.

b *

-.r I

S --


0 ~C W -3-

- 4WQ i.*.



'-, -~14


:~ ~


* 9-


& -a



- .


Figure 48. SEM micrograph of field root section showing
root hairs, soil debris and microorganisms.
Mag. X 320.

Figure 49. SEM micrograph of blowup of area (inset) from
Figure 48. Cell junction with bacteria.
Mag. X 5,300.

Figure 50. SEM micrograph of heavy population of spiral-
shaped bacteria on root surface, treated with
antiserum against 13t and JM125A2, and PAP.
Mag. X 3,600.

Figure 51. X-ray analysis of nonspecific osmium scatter
on root and bacterial area shown in Figure 50.
Mag. X 3,600.




r h:.
.~ ~ ~ ~~~ '.. ..' r:


Scanning electron microscope (SEM) studies have rarely

used immunological methods for specific SEM identification

of microorganisms. However, Molday (1976), Linthicum (1975)

and LoBuglio (1972) have successfully used the SEM for

antigen detection. Two attempts to specifically identify

Azospirillum on field-grown pearlmillet roots using the SEM

and PAP were not successful. There appeared to be a large

amount of nonspecific osmium staining, with no differences

between osmium counts on bacteria and on background tissue

(Figs. 50 and 51). Since this experiment was not conducted

on axenically-grown roots, it could not be determined whether

the bacterial cells tested were indeed Azospirillum which

had not reacted specifically, or if they were of a different


Because of its low viscosity, Spurr's epoxy (Spurr,

1969) is highly favored for embedding of hard material such

as roots, but it is not a commonly used embedding medium for

cytochemical techniques. Although contrasts were greater

when specimens were embedded in Spurr's medium rather than

in araldite, results with Spurr's epoxy were unpredictable,

at times eliciting no response to the antiserum and PAP, at

other times producing a reaction. However, the reaction

product with Spurr's epoxy was never as well defined as

when the reaction was carried out on araldite; therefore

it became necessary to use the higher viscosity araldite

embedding medium described by Luft (1961).

Improved techniques were necessary since the treated

sections were extremely fragile under the intensity of the

electron beam and shattered easily. It was necessary to
cut rich gold sections, 1000 A thick, to give more strength

than silver sections. Grid holders were devised, in order

to obtain all of the treatments and controls needed for TEM

observation. These holders permitted treatment of up to

four grids at one time, giving additional opportunities to

experimentally observe the desired reaction products.

It was found that lowering the gelatin content to

0.05% reduced the nonspecific staining caused by osmium.

The DAB was found to leave a black net-like residue, but

this did not interfere with the observations as much as the

osmium residue.

In the immunologically-treated axenic cultures, it was

possible to label bacterial cells for EM observation. All

of the bacteria did not show a reaction product, even though

they were known to be Azospirillum (Figs. 25 and 27). A

positive reaction is strong evidence that the organism not

only is Azospirillum, but also is of the tested strain,

either 13t or JM125A2. A negative reaction, however, does

not eliminate the possibility that the bacterial cell could

be an Azospirillum. Dazzo and Milam (1976) reported that

two- to three-day-old Azospirillum were less reactive to

the antisera than five- to seven-day-old cells, so

possibly the younger bacterial cells were not reacting with

PAP. Because of lack of control over the age of the

bacterial antigens in field-grown roots, an assumption was

made that more Azospirillum were present than actually could

be identified from the reaction product.

Pleomorphic forms of A. brasilense were observed in

the axenic root cultures (Figs. 26 and 30-33). These

enlarged structures, called cyst-like by Ruscoe et al. (1978)

and Umali-Garcia (1978), and encapsulated by Berg* (personal

communication) were observed to react with both the 13t and

the JM125A2 antisera. Berg suggests that these may be

specialized structures for nitrogenase protection from 02,

but as suggested by Umali-Garcia (1978) they could also be

a specialized form for survival under adverse climatic


Since field-grown pearlmillet root samples used in this

study were collected from soil cores taken near the crown

of the plants, no attempt was made to determine the age of

the various root segments. However, von BUlow and Dibereiner

(1975) reported that 3 to 5 mm maize roots with many

*Berg, R.H., Post Doctorate Fellow, Dept. of Botany,
University of Florida, Gainesville.

laterals were usually the most active, so an attempt was

made to stay within this size range.

Root samples from inoculated and uninoculated plots

were viewed on the TEM before selection was made for

immunocytochemical studies, and only those with high

bacterial populations were selected. Of all the collections

made, the greatest number of bacteria were observed in

sample #396, the uninoculated plot, 96 days after inocula-

tion with autoclaved 13t Azospirillum brasilense (Figs. 34-

37). This does not infer that specimen 396 had a greater

microbial population in association with its root system,

but only that, of the sections viewed, this particular

segment showed greater activity.

Smaller, and presumably younger, roots were observed

to be relatively free of microorganisms. These younger

roots also had intact cortical and epidermal cells. No

mucigel and few soil particles were evident (Figs. 38, 40

and 42). Although the stelar region contained functional

cells, many of the cortical cells were devoid of cytoplasmic

contents or contained small amounts of dark inclusions, as

was reported by Foster and Rovira (1976) and Holden (1975)

in wheat and barley roots.

Extensive microbial populations were usually observed

in and around larger roots. In most cases, the epidermal

cells were ruptured or sloughed (Figs. 39, 41 and 43), and

in many cases the cortical cells were ruptured, invaded, or

entirely missing (Figs. 41 and 44). There was occasionally

evidence of endodermal cell rupture and invasion (Fig. 45),

and one case where a bacterium was determined to be in a

cell within the stele (Fig. 46). Since this sample had been

post-fixed in osmium and embedded in Spurr's epoxy, it was

not possible to specifically identify this organism. How-

ever, it is likely that this was a chance parasitic invader,

since it was not a normally encountered phenomenon.

Okon et al. (1977) indicated that more Azospirillum

are located outside the roots than are inside, and Umali-

Garcia (1978) and Umali-Garcia et al. (1978) observed that

Azospirillum colonized slime on the root. By four weeks,

the middle lamella of the second layer of cortical cells

had been invaded, but invasion of the cells themselves was

reported only in older, dead cells (Umali-Garcia, 1978).

It should be pointed out here that most of Umali-Garcia's

studies were conducted in liquid media or agar, and there-

fore, were not representative of field conditions.

Dibereiner and Day (1976) and Smith et al. (1977) observed

Azospirillum localized within the cortical cells of roots

using tetrazoleum reduction and immunofluorescence.

Observations of field-grown pearlmillet roots treated

with PAP immunology led to the conclusion that Azospirillum,

being closely associated with the soil particles, fungi,

and mucigel surrounding the root, are found within ruptured

cells of the epidermis, cortex and sometimes endodermis of

field-grown pearlmillet roots (Figs. 35, 37, 44 and 47).

There was no indication that the bacteria were actively

invading living cells, although bacteria which reacted to

the antisera as well as bacteria which did not react were

observed in the middle lamella (Figs. 29, 35 and 47).

Umali-Garcia (1978) also reported seeing numerous inter-

cellular Azospirillum. However, the necessity for invasion,

either to obtain nutrients in the form of exudates, or to

obtain protection from 02 for the nitrogenase system, does

not exist under field conditions where soil particles and

fungal hyphae can rupture cells, or where mucigel can pro-

vide both nutrients and protection.

Since both A. brasilense strains JM125A2 and 13t were

found in uninoculated root sections this is consistent with

the fluorescent antibody studies (Smith et al., 1978) on

the organism's ability to overwinter in the north Florida

soils, and its ability to migrate from inoculated to

uninoculated plots.

Nitrogen-fixing symbiosis probably arose by an associa-

tion between green plants with excess carbohydrates and a

heterotrophic microorganism which was able to reduce

dinitrogen. This evolved by stages from a free-living

rhizosphere association, to a cortical association, to the

organized nodule of the more advanced nitrogen-fixers

(Parker, 1968). Since my observations showed bacterial cells

within plant cell walls, I think Azospirillum could be