http://researchcommons.waikato.ac.

nz/

Research Commons at the University of Waikato

Copyright Statement:
The digital copy of this thesis is protected by the Copyright Act 1994 (New Zealand).

The thesis may be consulted by you, provided you comply with the provisions of the
Act and the following conditions of use:

 Any use you make of these documents or images must be for research or private
study purposes only, and you may not make them available to any other person.
 Authors control the copyright of their thesis. You will recognise the author’s right
to be identified as the author of the thesis, and due acknowledgement will be
made to the author where appropriate.
 You will obtain the author’s permission before publishing any material from the
thesis.
 THE ANTIBACTERIAL PROPERTIES

OF HONEY

A thesis
submitted in partial fulfilment
of the requirements for the Degree
of
Master of Science in Biological Sciences
at the
University of Waikato
by
KATHRYN MARY RUSSELL

University of Waikato
1983
 i

ABSTRACT

Honey has been used medicinally by man for thousands

of years. With the apparent misuse of many a�tibiotics
there has recently been renewed interest in such natural
antibacterial substances.
Honey is known to contain several natural antibiotic
systems but research so far has been largely confined to
the glucose oxidase (inhibine) area. Some past research�
however j has indicated the possible existence of anti­
bacterial substances other than hydrogen peroxide (inhibine)
and the osmotic potential of honey. The aim of this
project was to confirm the existence of these substances
and to isolate and characterise them if possible.
The antibacterial assay used to detect activity was
an agar diffusion technique incorporating stanhvlococcus
aureus.
Various honey samples were screened for non-peroxide
antibacterial activity and the most active honey (manuka)
selected for further investigation.
The activity was found to be completely heat-stable
(1 hr, 95 ° C) at acidic pH but less stable at neutral pH,
and the honey was less active at neutral pH.
Solvent extraction of the honey was carried out with
ethanol, and ether was added to precipitate most of the
sugar. All activity was recovered from the honey into
the extract in this way. Further isolation of the active
fractions was carried out using preparative thin-layer
chromatography.
 ii

Information on the fractions with activity that were

isolated was obtained by ultraviolet spectroscopy, infra­
red spectroscopy, nuclear magn�tic resonance and mass
spectrometry.
The compounds identified in the active fractions
were:
methyl 3,4,5-trimethoxybenzoate
(a-methyl sytingic acid methyl ester)
M. Wt. 226

methyl 4-hydroxy-3,5- dimethoxybe�zoate

(syringic acid methyl ester)
M. Wt. 212

3,4,5-trimethoxybenzoic acid
M. Wt. 212
 iii

.ACKNOWLEDGMENTS

I would like to thank my supervisor, Dr P.C. Molan, for

his guidance and constructive criticism_during both the
practical work and the writing of this thesis.

I would like also to thank Dr A.L. Wilkins for his

valuable advice during the practical work, interpretation
of spectrographic data and guidance during the writing
of Chapter 6.

I am also grateful to:

Murray Reid for obtaining the honey samples.

Ralph Thomson for the n. m.r. spectra9,.
Dr P. Holland and Dean McGaveston for the mass
spectral analysis of fractions.
Rex Julian for the photographs.
Ken Stewart for advice on illustrations.
Jacques Boube� for the translation of French
articles.
B.M. Moss for the typing.

Finally I would like to thank my family and friends for

their support, advice and encouragement throughout.
 iv

CONTENTS

ABSTRACT i

ACKNOWLEDGMENTS iii
TABLE OF CONTENTS iv
CHAPTER 1: INTRODUCTION 1
CHAPTER 2 : MATERIALS AND METHODS 18
a. Honey 18
b. Antibacterial as say 18
c. Buffers 20
d. Staining reagents 21
e. Artificial honey 23

f. Cata lase 23
g. H 2 o 2 determination 23

h. Ultrafiltration 25
i. Thin-layer chromatography 25
j. Ultra-violet analysis 26
k. Infra-red analysis 27
1. Nuclear magnetic resonance 27

m. Mass spec�rometry 27

CHAPTER 3: INTRINSIC PROPERTIES OF NATURAL HONEY 28

3.1 Developing an assay for

antibacterial activity 28
3. 2 S�reening honeys for activity 36
3.3 Heat stability 39
3.4 Effect of pH on the non-
peroxide activity 41
3.5 Combined effect of heating
an� neutralising honey 41
 V

3.6 Ultrafiltration
3.7 Summary 46
CHAPTER 4: SOLVENT EXTRACTION OF THE ANTIBACTERIAL
ACTIVITY FROM HONEY 47
4.1 Solvent extraction _ 47
4.2 E ffect of acid and alkali on
factor stability 48
4.3 E xtracts of inactive honeys 52
4.4 Thin-layer chromatography on
commercial plates 52
4.5 Further extraction 57
4.6 Summary 61
CHAPTER 5: ISOLATION OF THE NON-PEROXIDE
ANTIBACTERIAL ACTIVITY 62
5.1 Preliminary preparative TLC 62
5.2 Locating the antibacterial
activity 74
5.3 Further separation with
Preparative TLC 82
5.4 Locating antibacterial
compounds after
secondary chromatography 88
5.5 Tertiary chromatography of
Section B 101
5.6 Assessment of recovery of
activity from primary TLC.. 103
5.7 Summary 105
CHAPTER 6: CHARACTERISATION OF ANTIBACTERIAL
COMPOUNDS 106
6.1 Ultra violet spectroscopy 107
6.2 Infra-red spectroscopy 107
6.3 Nuclear magnetic resonance 112
6 .4 Ma�s spectromet�y of fraction
3 d 115
6.5 Mass spectrometry of other
active fractions 119
6.6 Conclusions 121
CHAPTER 7: DISCUSSION 124
REFERENCES 141
 1

CHAPTER ONE:

INTRODUCTION

Apiary products have been used to treat diverse

ailments for thousands of years. With the apparent misuse
of many antibiotics and discovery of harmful side-effects
from other laboratory-developed drugs there has been
renewed interest in recent times in such natural
therapeutic substances. Natural antibacterial compounds
have possibly been utilised by bees thBmselves for
10-20 million years as evidenced by the relatively sterile
environment of the hive. The active compounds present in
hive products can only have altered in as much as the bee
itself, or plant from which the materials were obtained,
has evolved. These hive products include propolis, bee
venom, honey and royal jelly.
Royal jelly is reported to be only weakly antibacterial
(Lavi�l960) but does contain 10-hydroxy-�Z-decenoic acid
which can prevent both bacterial and fungal growth. (Blum,
cited by Harman, 1983).
The pharmacologically active components of propolis
have been well researched. Propolis is a resinous, sticky
gum collected from the boughs, leaves and buds of various
trees, particularly poplar, birch, elm and alder� It is
used by bees to attach the comb to the roof of the hive,
to seal cracks and to surround the entrance to the hive
through which they must pass. There is a lower incidence
of bacteria and moulds inside the hive than out and it is
thought that the volatil? constituents of propolis reduce
 2

the aeroflora within the apiary (Ghisalberti, 1979). An

invader in the hive, such as a mouse, is usually killed and
then embalmed in propolis and may remain preserved for as
long as five years (Donald, 1982). It has been used by
man largely in dermatology: to treat bacterial and mycotic
infections, eczema and to aid tissue regeneration where it
is claimed to speed the remaking of-granule cells in wounds
(Cizmarik, 1979). It has also been used in the treatment
of rheumatism, gout, hyperlipidemia, ulcers, tooth and
gingival diseases, ear and respiratory tract infections,
some viral infections and to staunch blood loss, for
protection against radiation and as an analgesic.
Many of the pharmacologically active compounds in
propolis have already been identified. There is evidence
that some have come from plants unaltered and some have
\
been transformed by enzymes in the bees saliva (Ghisalberti,
1979). Flavonoids exist in rich variety and highly
concentrated form in propolis and exhibit a large number
of therapeutic effects. According to Donald (1982) their
major known effect is. beneficial action on the capillary
system, on the fragility and permeability of blood vessels
and upon the circulatory system in general as a vasodilator.
They also act as diuretics and increase bile production,
influence production of compounds from several endocrine
glands, including the thymus, thyroid, pancreas and
adrenal glands, and have antiviral, antiparasitic, anti-
bacterial and anticoagulant effects. In the review on
propolis by Ghisalberti (1979) the constituents shewn to
contribute towards its bacteriostatic or antibacterial
activity are galangin (a flavone), pinocembrin (a flavone)
 3

and the cinnamic acid derivatives: caffeic acid and ferulic

acid, which are active against both gram positive and gram
negative bacteria and show fungistatic activity. An
antimycotic effect is also shown by pinobanksin -3_-acetate,
pinocem1rin, p-coumaric acid benzyl ester and a caffeic
acid ester. Ferulic acid has a marked ftstringent effect
and a number of the flavones isolated have papaverine-like
spasmolytic activity. Stilbenes have also been isolated
from propolis and are known to possess fungicidal and
fungistatic properties toward wood-rotting fungi and
therefore may be responsible for some of the success with
mycotic infections. Flavonoid pigments also isolated
have antibiotic effects - some also have anti-inflammatory
effects on joints, skin and mucous membranes. The many
isolated compounds are cnly a proportion of the total
present and their effects requtre more clinical research
(Ghisalberti, 1973).
Bee venom also has various clinical effects owing to
its components. It has been used to treat the peripheral
nervous system (radiculitis, neuritis, neuralgia), joints
(arthritis), rheumatic and allergenic problems and to
decrease blood pressure in arteries. It is antibacterial
and reported to increase the radiation resistance of mice
(though it is not known what fraction achieves this),
(Orlov, 1979; Fennell et .al., 1968). Approximately 50% of
venom consists of mellitin, a polypetide which has
arrhythmic effects in non-toxic doses. This fraction is
also responsible for v�nom's antibacterial activity
(Fennell et al., 1968) (although Orlov (1979) claims venom
solutions are unstable, destroyed by digesting and
 4

oxidising enzymes and liable to bacterial infection and

alteration) and has a slightly higher (but insignificant)
effect than the venom itself. Various gram positive and
gram negative organisms are sensitive to it, including
penicillin-resistant S��tireus st�ain 80, although gram
positive organisms are more sensitive. - It was calculated
by one group of researchers that one sting had the anti-
bacterial potency of about 9 units of penicillin for
various gram positive organisms and a range of 9-170 units
when measured against a selected group of gram negative
organisms (Fennell et al., 1968). The analgesic effect of
venom is also attributed to mellitin and the impulse
inhibition in the central nervous system, vegetative
ganglions and peripheral nervous system. Other known
components of bee venom are: the polypeptide apamine which
has a specific excitation effect on spine marrow; MSD
peptide, which degranulates basophil c�lls; minimine;
phospholipose A, which will split lecithin to lysolecithin
and which degrades cell membranes; hyaluronidase, which
splits hyaluronic acid, a constituent of conjunctive tissue
and therefore favours the spread of active factors of
venom throughout the body. A polypeptide, newly
identified in bee venom also has one hundred times higher
anti-inflammatory effects than cydrocortisone. Venom also
stimulates the activity of the hypophysis and suprarenal
glands (shown by increased cortisone levels in the blood
and increased 17-keto-steroid in urine).
Honey may be one of the worlds oldest medicines. It
provided a wide variety of services to ancient and
mediaeval man, and is still widely used to treat colds,
 5

burns and other skin problems.� The oldest known written

reference to the medicinal use of honey is a Sumerian clay
tablet of several prescriptions incorporating it; found
in Iraq and dated at about 2,000 B.C. Thirty percent of
the 966 prescriptions in the ancient Egyptian Papyrus Ebers
(approximately 1500 B.C.) included honey. Hindu, Greek
and Roman medicine employed it and reference is made to it
in the Koran and the Talmud.
The value of different types of honey was recognised
also. Aristotle, in about 350 B.C. specified particular
types of honey for eye problems. "The Greek Herbal of
Dioscorides", written in the 1st century A.D. and on which
herbalists of many nations have drawn their inspiration for
15 centuries, prescribed honey for sores, ulcers, nits and
lice, throat and tonsil infections, eye infections and
even rabies. Dioscorides also described the best kind of
honey to use: honey from Attic� was the best and then
Sicilian. He said it should be sweet and sharp, fragrant,
pale yellow, glutinous and firm and which "in drawing does
leap back to the finger." (This may refer to either the
property of thixotropy which is due to a relatively high
content of certain proteins and is an isothermal gel-sol­
gel transformation induced by shearing and subsequent rest,
or to dilatancy, increased viscosity with increased rate
of shear). Dioscorides also specified that spring honey
was the best, and then summer honey, and that Sardinian
honey was very good for acne.
Many modern reports can be found in European literature
describing the value of honey in treating wounds, burns,
infections and urinary disorders. A report in 1982 in the
 6

American Bee Journal (vol. 122, no. 4, pg. 247) said that
doctors at several Israeli hospitals are now using it on
open wounds after surgery. An Israeli microbiologist
claims it prevents infection, speeds healing and that tests
show it contains components very similar to antibiotics
(no details of these substances were gi�en, however). It
was thought to work more quickly than many antibiotics
because of easy absorption into the blood stream. The
article also claimed it was good for such internal maladies
as ulcers by absorbing acids and lubricating the digestive
tract. Another reported use of honey on the public is by
an English physician (Blomfield, cited by Goulart, 1979)
who recommends it for both cleansing and healing. He was
using pure natural honey in an accident and emergency
department under dry dressings and claimed that it promotes
healing of ulcers and burns better than any other local
application. (Some of these properties may be partly due
to honey providing a good barrier).
In 1970 twelve patients with wound breakdown after (��()vn•CAL
surgery for carcinoma of the vulva were treated with hon�y,
and bacteriological wound sterility was attained in 3-6
days and within 8 weeks there was complete healing without
the need for any skin grafting. Bacteria cultured from
the wounds were killed in vitro by undiluted honey.
Other therapeutic uses of honey include: raising the
haemoglobin content in blood of anaemic childern, cardiac
therapy, treatment of stomach ulcers, speeding up alcohol
metabolism in drunks and use against sickness resulting
from radiation treat�ent (probably due to its high fructose
content), and treatment pf respiratory infections and
various digestive diseases.
 7

Some of the effects of honey may be ascribed to known

components which come from either plant or bee. It seems
likely that honey may inherit any curative qualities of
plants from which it is gathered because of the discrepancies
found between different researchers -one species of honey
bee is generally used domestically. H?wever, all honey
appears to contain several natural �ntibiotic systems:
The osmotic pressure in undiluted honey will be
important. Honey is �asically a superL/satura ted solution ( '1
(\

Q
of sugars. In one report the osmotic pressure ��--12_e more4
than 2 ,000 milliosmols. During the ripehing stag� honey
increases in density, energy and stability due to water
�'---"---

evaporation and an increase in simple sugars by the work

of inverting enzymes on sucrose.
The acidity of honey also prevents the growth of many
bacteria. The general pH range is between J.2 and 4.5
(average 3.9). The predominant acid is gluconic acid
which exists in equilibrium with gluconolactone. pH is
affected mainly by the mineral content, i.e. a higher
ash content will give a high pH.
The antibacterial property which has probably been
studied the most is the glucose oxidase system which
produces hydrogen peroxide.

Glucose+ o2 Gluconolactone+ H202

I�
Gluconic acid

H 2 o 2 accumulates during storage of honey, although there

are minor components in ,the honey which will destroy it-,

 8

such as vitamin C, catalase and other reducing substances.

The glucose oxidase system is reported by several
investigators to be acti ve only in diluted honey.

has been recognised as an effective antibiotic for many

years and was a major antibacterial component of some of
the early penicillin drugs such as "Not-atin" or Penicillin
A (found to be a glucose oxidase).
Some previous work has also indicated varying levels
of non-peroxide antibacterial activity due to other
materials in some honeys. Little research so far has
been done into these factors alone though, or even to
confirm their existence�
Pioneer work was done on honey by the Dutch scientist
Van Ketel (1892) who demonstrated the bactericidal effects
of honey and 1y Sackett (1919) who thought it was due to the
osmotic potential , but could not explain the phenomenon
in diluted honey. In 1937 Dold and CO'-workers, S;Uggested
the terni "inhibine" for the antibacterial activity
ubserved in diluted honey and found it to be heat and light
labile, being inactivated by heating at l00 ° C for 5 minutes,
80 ° C for 10 minutes or 56 ° c for 30 minutes. The absence of
inhibine was an indication that the honey had been heated
and in 1955 Dold and Witzenhausen reported an assay to
standardise inhibine assessment (Table 1). The assay
involved the innoculation of S.aureus into agar containing
known amounts of honey. After incubation for 24 hours
the plates were inspected visually for bacterial colony
development.
White and co-workers (1962, 1963) concluded that the
liberation and accumulation of hydrogen peroxide by ·way
 9

TABLE 1: Criteria for an inhibine value.

Dold and Witzenhausen, modified

after Adcock (1962).

Honey concentration (per cent) in medium Inhibine Value

25 20 15 10 5

f + t ..
t + 0
+ + + + 1
+ + + 2
+ + 3
+ L

+ growth of S.aureus
no growth
 10

of a honey glucose oxidase enzyme system could account for

all the known properties of honey inhibine. (Previous
work into the enzymic production of gluconic acid in honey
led to the discovery of the ho�ey glucose oxidase system).
They showed that inhibine was heat sensitive, but with a
wide variation in sensitivity to heat, qepending on floral
source. Of 29 samples heated at 70·o C for 10 minutes
about two-thirds of the samples lost more than 80% of their
activity, five lost 60-80%, two lost 40-60%and two were
more heat resistant. The heat resistance of their peroxide
accumulation systems varied 70-fold or more. They also
°
concluded that between 55 and 70 ° C the half-life of the
peroxide accumulation is a logarithmic function of
temperature. (Different researchers give conflicting
results for heat effect though e.g. Gubanski (1962) found
that honey boiled for an hour did not lose its anti­
bacterial properties especially in so far as intestinal
bacteria were concerned). No reasons were proposed for
the floral source influencing peroxide accumulation except
that the inhibine value is affected not only by glucose
oxidase content (which comes from the hypopharyngeal gland
of the tlee and should therefore be reasonably constant) rut
also by various minor peroxide-destroying cbmponents in the
honey from sources such as nectar, pollen, yeasts or by
enzymes such as catalase • . They proposed that it may also
be affected by handling, storage and processing of the
honey, and in some cases (again depending on floral source),
exposure to light. No explanation has been proposed for
effect of light exposure being related to floral source.
 11

White et al�also proposed that light-sensitive honeys

contained a sensitiser which when they were mixed with
unsensitive honey also rendered it light-sensitive
(particularly to ultra-violet radiation). (Table 2 ).
When White and S.ubers (l 96J) actually carried out a
chemical assay to relate inhibine number and peroxide
accumulation in various honeys they concluded that H 2 o 2

accounted for the major portion of the non-osmotic anti-

bacterial effect in diluted honey. Inhibine was measured
by the usual bioassay method of Dold and Witze:ll'hausen,
and H2 o 2 accumulation by a colourimetric method

incorporating s-dianisidine. The table they formed is

shown in Table 3. Of the 45 samples tested, however, 14
were anomalous - seven gave a higher peroxide accumulation
than would be expected for the inhibine numbers obtained
and seven gave lower readings than expected for peroxide
accumulation. Insufficient material was available to
determine the cause of the apparent anomalies.
Adcock (1962), working independently, also showed a
connection between inhibine value �nd·peroxide value,
which he measured iodometrically (Table 4). He could not
explain the lack of correlation between the two sets of
values - peroxide was eradicated with catalase and the
inhibine value was often only reduced. Adcock claimed
that in later tests increasing the amount of catalase could
completely destroy the inhibine values of the samples least
affected by catalase originally. He did not indicate,
however, whether he had increased the concentration of
catalase added or whether he just added a greater
 12

·TABLE 2: Inhibine assays of irradiated honeys (by the

method of Dold and Witzenhausen, 1955).
(White and Subers, 1964).

Sensitive Honey Inhibine No.

Controb unexposed 4
3 min. sun, original density 3
5 min. sun, original density 2
3 min. sun, and filter*, orig. density 4
1 hr. lab. lights, original density 2

5 min. sun, pH 3.9, dilute 2

5 min. sun, pH 6.5, dilute 5

Resistant Honey

Control, unexposed 5
5 min. sun, original density 5
5 min. sun, pH 6.5, diluted 5

Mixture

Control,unexposed 4
5 min. sun, original density 1
5 min. sun, pH 6.5, diluted 5

* Glass filter remo�ed plue-green light

 13

TABLE 3: Relation between inhibine number and

peroxide accumulation.
(White and Subers, 1963)

Inhibine No. Peroxide accumulation

0 < 3 .4
1 3.4 8.7
2 8.8 - 20.5
3 20.6 - 54.5
4 54.6 - 1 74
5 > 174

Peroxide accumulation given in micrograms

per gram of honey in 1 hr under assay
conditions.
 14

TABLE 4: Comparison of inhibine and peroxide values of

honeys before and after treatment with catalase.
(Adcock, 1962)

Country Inhibine value Peroxide value

of
Origin Before After Before After
catalase cata lase cata lase catalase
treatment treatment treatme nt treatment

5.0 1.5 2.48 0

4.0 1.0 2.08 0
2.0 0. 5 3.52 0
2.0 0 .5 3.76 0
2.0 1.0 2.96 0
1. 5 0 3.92 0
1.5 0.5 4.24 0
Australia 1.5 0.5 2.56 0
1.0 0 4.48 0
1.0 0 4.48 0
1.0 0 4.16 0
1.0 0 4.64 0
1.0 0.5 11.52 0
1.0 0. 5 2.40 0
0.5 0 2.72 0
4.0 1.5 9.92 0
Argentina 4.0 2.0 4.00 0
4.0 2.5 3.76 0
3.0 0. 5 1 o. 48 0
4.0 0.5 2.80 0
Belgium 2.5 1.0 2.90 0
0.5 0 2.32 0
Germany 3.5 1.0 4.00 0
2.0 0.5 2.24 0
Mexico 4.5 2.5 5.76 0
 l.�

quantity of the original concentration - in which case he

may have merely diluted away anj non-peroxide activity
which existed.
It does not appear that the above-mentioned workers
have carried out any further research on the antibacterial
properties of honey� Some others have suggested the
possibility of non-peroxide antibacterial factors
existing in diluted honey. These workers have attempted
to extract active fractions from honey with various solvents.
In 1951 Verge extracted antibacterially active
t (o �r.
')

fractions of honey with water, alcohol, ether or acetone. �

Schuler and Vogel (1956) extracted an active fraction of
honey in ether.
Lavie (1960) working with a mixed floral honey found
the acetone extract active but the ether extract inactive.
He also found that extracting honeydew would give an active
substance and that the degree of activity of the extracts
varied with sources. The activity was photolabile and
reduced by heat but some activity remained after 30 mins
at 80 ° c. Feeding sugar to bees, feeding the product
(honey) repeatedly to bees and testing the honey produced
each time showed that the activity was increased with each
bee "passage". Lavie concluded that the major portion of
the activity came from the bee - but not all however as
honeydew itself was active. (Plachy, 1944, also tested
honeydew honey and floral source honey, finding honeydew
honey more active. It might be possible that there are
antibacterial substances coming from the plant which are
more concentrated in form in the sap of the plant than in
the nectar). He also tested several non-nectar solutions
 16

from honeydew (of pine aphids), buds and needles and all
possessed activity.
Gannet and Lav.le (1960) working with a multifloral
honey containing lavender found an alcohol extract more
active than an acetone extract. The acetone extract of
this honey was ·unaffected by light or by neutralising the
honey. Heat reduced the activity (although it was still
active after autoclaving at 120 ° C for 15 minutezj but
heating and neutralising together eliminated it. The
activity returned after reacidifying however. Lavie (1963)
also found he was able to obtain a cold ether extract of
a honey sample which contained antibacterial substances
volatile at 95 ° c (a fraction collected at a higher
°
temperature than 95 c was inactive). Gannet and Laivie
concluded that besides antibacterial activity which
originated from the bee (H 2 o produced by glucose oxidase)
2
honey contained another group of antibacterial substances
(rrom the forage plants)which were volatile, photolabile,
extractable by solvents and more heat stable than inhibine.
Owing to administrative reasons no further work on honey
was done in that laboratory and therefore no attempt was
made to isolate and characterise the non-peroxide anti-
bacterial compounds. Dustmann (1978) also reported the
existance of an antibacterial effect from honey other than
the enzyme activity or sugar content but claimed it was
only a minor portion.
Papers have been published in recent years which still
ignore the possibility of non-osmotic non-peroxide anti-
bacterial effects, (James et al, 1972). Research continues
in Europe and Russia but no comprehensive papers appear to
 17

be available at present, i.e. papers containing experimental

details and characterisation of the responsible compounds.
For instance Mladenov (1974) concludes there are �nti­
bacterial substances with volatile, heavy volatile and
non-volatile character with bacteriostatic and bactericidal
. .
effect, but no more details of these compounds were given.
There is also much conflict in the literature about which
organisms are actually sensitive to honey's activity.
Possibly the influence of floral source on a honey's anti­
bacterial activity and mode of action is far more important
than has previously been reported. This may explain some
of the discrepancies in the literature about solvent
extraction, heat- and photo- lability, degree of activity
and the range of sensitive organisms.
Therefore, because of conflicting reports in the
literature about the existence and properties of non-peroxide
antibacterial activity in honey it was decided that a
valuable area of research would be to confirm its apparent
existence - to commence by screening various honey samples
for this activity; to extract and purify all potentially
antibacterial (non-peroxide) compounds from the most active
honey sample available and to characterise these compounds
if possible.
 18

CHAPTER TWO:

MATERIALS AND METHODS

a. Honey
Honey samples were obtained from various beekeepers by
Mr M. Reed, Apiary Advisory Officer, Ministry of Agriculture
and Fisheries, Hamilton. The age, origin and previous
treatment of eachfuney sample is given in Chapter 3.
The honey sample used for all experiments, other than
screening a range of samples for activity, was unextracted
manuka honey, 18 months old.
The honey was stored in air-tight plastic containers
at 4 ° C until required for use.

b. Antibacterial Assay
Antibacterial activity was tested using an agar well
diffusion technique.
Plates were prepared using a 24-48 hour bacterial
culture in nutrient broth (0.08 g in 10 ml distilled water).
The culture (10 ml) was added to 150 ml of sterile liquid
nutrient agar (3.45 g in 150 ml distilled water) at about
45 ° c. The plates were poured immediately in a Laminar
Flow Cabinet and stored at 4°c for a maximum of 4 weeks.
Holes, 8 mm in diameter, were cut in the agar and the
wells filled with test solution. The plates were
incubated at_37 ° C for 18 hours (Figure 1).
The relative extent of inhibition of growth was
measured with a ruler as the distance between the edge of
the well and the edge of the clearing. (The extent of
 19

FIGURE 1: Antibacterial assay by technique of

agar well diffusion.

4. honey + catalase (6255 units/ml) 3. 50% v/v honey

s elution
2. 1.01 mol/1 H2D + catalase
�
(6255 units/ml)
 20

clearing was checked under a dissecting microscope).

This assay method is semi-quantitative: the extent of
bacterial inhibition is not linearly related to the amount
of antibacterial activity present.
In order to find the most sensitive method of
detecting antibacterial compounds a method of detecting
activity with sensitivity discs was also tried. Discs of
8 mm diameter were cut from ashless filter paper. They
were soaked in water, ethanol, honey or honey extract for
one minute and �ried with a hairdryer. Following this
they were laid on a seeded agar plate for 2 hours at 4 0 C,
to allow infiltration of the activity into the agar before
bacterial growth could occur and "mask" its presence i.e.
so that the bacteria would not grow before the inhibitory
compound could potentially create an area of inhibition.
The plates were incubated for 18 hours at 37 ° C but no
inhibition of growth was seen around any of the discs.
Using a bacterial lawn instead of a seeded agar plate also
showed n� inhibition of growth.
Hence the agar diffusion technique was selected as
the most sensitive technique of antibiotic detection.

c. Buffers
Phosphate buffer (0.1 mol/1) was prepared by
dissolving 3.12 g of sodium dihydrogen phosphate in 200 ml
distilled water and adjusting to pH 7 with NaOH.
Citric acid buffer (0.1 mol/1) was prepared by
dissolving 5.2535 g citric acid in 250 ml distilled water.
 21

d. Staining Reagents
Lactate dehydrogenase detection reagent contained
Ht mg NAD'; 6 mg nitroblue tetrazolium, 15 mg lithium
lactate and 0.3 mg phenazine methosulphate dissolved in
2 ml distilled water. This was made up immediately before
use, the agar plate flooded with .it and_re-incubated for a
further 30 minutes. Living bacteria are stained purple
and other areas remain clear.
/'
Triphenyl tetrazolium chloride (Kirchner, 1978) is a
,- V
biological detection reagent. Innoculated nutrient agar
was mixed with 0.7 ml of a 5% solution of triphenyltetra­
zolium chloride in 50% methanol per 50 ml of medium. This
was poured gently onto the surface of the d�veloped
;_,�cu ( ct,y
chromatogram and when cool, a .,,,s;..oa-i' of sterile agar was
poured on top of this. Prior to incubation the plate was
placed in the refrigerator (4 °C) for 1 hour in order to
allow any inhibitory activity to diffuse into the agar
from the silica.
p-Aminophenol (Dawson et aL, 1969), the staining
reagent for sugars, was prepared fresh with 0.5 g
p-aminophenol and 2 g H3Po in 50 ml 96% ethanol. This
4
was sprayed (with a Shandon spray unit) onto the commercially
prepared TLC plate which was then heated for 5 minutes at
105-110 ° C. This stains:
glucose --.lo dark brown
fructose � lemon yellow
sucrose and maltose --i' brownish yellow
raffinose � light brown
 22

Sulphuric acid stain contained 25 ml 98% sulphuric

acid in 220 ml methanol i.e. a 10% solution. TLC plates
were sprayed with this and then incubated at 110 ° c for up
to 20 minutes. A 10% solution of aqueous sulphuric acid
was also used for staining commercial aluminium - foil­
backed TLC plates and heated in the same way.
Molybdophosphoric acid general· spray reagent (Dawson
�t al., 1969) consisted of a 10% solution of .molybd9phos-
phoric acid in 95% ethanol. Following spraying with this
the TLC plates were heated at 110 ° c for 10 minutes.
Ferric chloride stain (Kirchner, 1978) was a
saturated solution of anhydrous ferric chloride in
methanol, for detection of phenols and terpene phenols.
Anisaldehyde-sulphuric acid staining reagent (Stahl,
1969) was prepared fresh before use. Concentrated
H2 so (1 ml) was added to 0.5 ml anisaldehyde in 50 ml
4
acetic acid. Following spraying with this solution the
TLC plate was heated at l00 ° C for 10 minutes to detect
phenols, terpenes, sugars or steroids •
Ferric chloride-pociasium ferricyanide stain (Dawson

et al., 1969) was prepared with aqyeous 1% solutions of

each salt (A.R. quality) mixed equally. The solution had
to be orange-brown with no trace of blue. Reactive
phenols stained blue immediately without heating. (A
O.lN HCl solution was sprayed on top to give a permanent
record).
Vanillin-sulphuric acid stain (Randerath, 1966)
contained 1 g vanillin in 100 ml cone. H2 so . The TLC
4
plate was sprayed with this and then heated at 120 ° c for
10 minutes.
 23

e. Arti£icial hdrtey
An artificial honey solution, containing only sugars
rC
was used to test the resistance of bacteria to the sugar
content of honey. The solution consisted of sugarsand
water in the average proportions Thomson (1936) found in
21 New Zealand honey samples (Table 5), i.e. 17.5 ml water,
36.2 g glucose, 40 g fructose and 2.8 g sucrose.

f. Catalase
Catal�se solution was made up fresh when used and
consisted of 0.0278 g catalase powder (from horse�adish ,�/
L

origin, Sigma) in 10 ml distilled water, giving 6,255

units/ml activity.

g. li2�2 determination

An attempt to determine the H2 o produ�·tion of manuka

2
honey was carried out using the enzymic reaction:
+

peroxidase
In the course of the peroxidase reaction the hydrogen
donor 2-dianisidine (DH2) is oxidised to give a brown-

coloured product (D). Optical density was measured at

436 nm. The assay mixture consisted of:
2.47 ml �-dianisidine solution (6.6 mg dissolved
in 100 ml O.lM phosphate buffer and
neutralised with NaOH).
0�01 ml peroxidase suspension (4.93 mg pe�oxidase
with 73 units/mg activity in 1 ml
distilled water).
0_52 ml sample (honey solution or H2 0;z, standard).
 24

TABLE 5: The chemical composition of manuka honey

and the average for 21 various other
New Zealand samples (Thomson, 1936).

Manuka Average

water % 18.4 17.5

glucose % 34. 5 36.2
fructose % 42 40
sueras e % 2.8 2.8
ash % .15 .18
nitrogen al
/0 • 038 • 040
dextrins and C ; (undetermined) % 2.1
titrable acidity C.c 2.8 3.3
pH value 4. 05 3.81
formal titration C.c 1.2 1.3
specific gravity 1.407 1.423
glucose:fructose 1:1.22 1:1.11
 25

The spectrophotometer was zeroed on water, and the optical

density of the reagents alone, with no sample, was read to
compensate for their contribution to the coloured products.
The background colour of. the honey interfered too
much with the brown-coloured indicator. Hence o-tolidine
(0.04 g in 1 ml absolute ethanol plus 2 ml 30% acetic acid
l 200 f 1� in 100 ml phosphate buffer] ) was substituted for
o-dianisidine in the assay. This also produced a brown-
coloured product, however.

h. Ultrafiltration
Ultrafiltration of honey solution was attempted with
a Chemlab. Ultrafiltration Cell (Model C 50) with an
Amicon Diaflow UM-05 membrane (nominal retentivity 500
molecular weight). Pressure was applied with oxygen-free
nitrogen.

1. Thin-layer Chromatography
Commercial chromatograms used for preliminary work
were Eastman Chromagram Sheet 6060 with a plastic backing.
The silica contained a fluorescent indicator. Chromatogram
strips were developed in screw-top jars which had been
equilibrated with solvent for 1 hour beforehand.
Developed strips were allowed to air-dry in a fume
cupboard.
Preparative TLC plates were prepared using Merck
Kieselgel 60 PF 2.,-1+1"'360 (silica gel with a plaster of
. I

Paris binder and two fluorescent indicators).

Powder (100 g) was suspended in 240 ml distilled
water, swirled to mix and left to stand for 1 hour.
 26

Glass plates were c.-}eaned with an abrasive detergent,

rinsed, soaked in 3% Dec�n overnight, rinsed and dried.
They were then spread with the gel, using a Shandon
Unoplan Spreader (to a thickness of approximately 2 mm).
The plates were left to dTy overnight in a draughtless
room or cupboard and activated at 120 ° c for 2 hours.
Plates were developed in a m�tal chamber with a glass
plate across the top and the atmosphere saturated with the
aid of solvent�soaked filter papers down the sides.
Saturation of the atmosphere gives faster separation and
a straighter solvent front. It was found that developing
the plate a few centimetres, allowing it to dry standing
upright, and then continuing development, did not produc e
sharper bands. Also application size did not increase
linearly with the amount of material it contained. After
the preliminary experiments using preparative plates,
further plates were "cleared" of contaminants, such as
phthalate, before use by development (to the top of the
gel) in a slightly more polar solvent than the one to be
used.
Solvent svstems used to obtain final fractions:
I - ethanol:diethylether (60:40)
II - toluene:chloroform:acetone (40:25:35)
III - toluene:chloroform:acetone (40:20:40)

j. Ultr��Vidlet abSdrption sp��tros�opy

A quartz cell was used with distilled water as a
reference. Analysis was carried out with a Pye Unicam
SP8 - 500 u.v./vis.- spectrophotometer in the range
500 nm - 200 nm.
 27

k; Infra-red spectroscopy
This was carried out using a Shimadzu IR-27G
recording infra-red spectrophotometer in the range
4000-400 nm.
Potassium bromide discs were mad� of each freeze-
dried isolated sample. The anhydrous KBr and sample
were finely ground together with a pestle and mortar and
a 1 3 mm disc formed under 154 x 10 3 kP pressure under
vacuum for 1 minute.

1. Nuclear magnetic resonance

The samples to be analysed were dissolved in
deuterated chloroform and analysed using a Jeol FX 60 Q
n • m • r • machine •

m. Mass spectrometry
This analysis of samples was carried out on a Varian
Mat CHS mass spectrophotometer, which assigned masses.
,It had an associated data system, a 90 ° magnetic sector
instrument and samples were run at 24 ev.
 28

CHAPTER THREE:

INTRINSIC PROPERTIES OF NATURAL HONEY

3.1 Developing on Assay for Antibacterial Activity

An assay was developed in order to detect in honey
antibacterial activity other than that due to hydrogen
peroxide, osmolarity or acidity. The assay required a
test organism which was both insensitive to the high
/, �
\ .

osmolarity and acidity of honey and the growth of which

was inhibited by the non-peroxide antibacterial factors
present. It also required a means of eliminating
hydrogen peroxide from honey.
Previous work in this laboratory had demonstrated
that manuka honey possessed antibacterial activity in
addition to that due to hydrogen peroxide production
(Molan, personal communication). Therefore this floral
type of honey was used during development of the bioassay.
a. Estimating Hydrogen Peroxide Production
In order to devise a means of eliminating hydrogen
peroxide it was necessary to estimate how much might be
produced in honey during the incubation period of a
bioassay (approximately 18 hours).
An attempt was made to do this using the reaction:

where D represents the hydrogen -donor .�dianisidine,

which is oxidised to a brown-coloured product. The
colour produced from honey was compared with that
produced from a range of H2 o 2 ·standards, by measuring
absorption at 436 nm.
 29

The reaction mixture consisted of:

2.47 ml o-dianisidine solution (6.6 mg in 100 ml�
0.1 mol/1 phosphate buffer,
neutralised with NaOH)
0.01 ml peroxidase suspension (4.93 mg of enzyme,
activity 73 units/mg, in l ml water)
0.52 ml honey solution (incubated 18 hours), or
standard H2 o2 solution.

Dilutions of honey were used since the glucose

oxidase system is active only in diluted honey (White and
Subers, 1963). This method, however, was unsuccessful
as the background colour of the honey solutions interfered
with the absorbance readings. White and Subers (1963)
used o-dianisidine in their assay of peroxide accumulation
r
in various American honeys but their samples may have
contained fewer coloured compounds.
Therefore� peroxide was estimated by comparing the
relative antibacterial activity of a range of H2 o 2

standards with a range of honey dilutions. Activity was

measured using an agar well diffusion technique. The
agar was seeded with-itaphylococcus aureus, shown by
Dustmann (1979) to be highly sensitive to hydrogen peroxide
and used by Dold and Witzenhausen (1955) to determine
honey inhibine numbers.
A preliminary assay showed that a peroxide
concentration as low as 0.032 mol/1 would inhibit growth
of S.aureus.
The results are shown in Table 6.
Some of the inhibit�on by honey would have been due
 30

TABLE fu Inhibition of growth_ of S.�tireus on agar plates

by H o
2 2
standards and manuka honey.

HYDROGEN PEROXIDE STANDARDS

H2 o2 concentration Diameter of zone of

(mol/1) Inhibition (mm)

2.020 13
1.010 7.3
0.510 5.3
0.250 3.5
0.130 4.5
0 .. 063 2.5
o. 032 0.5

HONEY

Honey concentration Diameter of zone of

(% v/v) Inhibition (mm)

100 6.0
50 5.0
25 3.5
10 2.5
1 0.3
 31

to non-peroxide antibacterial action. Thus the relative

peroxide production of the honey was overestimated.
During the incubation _period the honey produced a clearance
approximating that of an H2 o concentration between the
2
range 0.51 mol/1 and l.�l mol/1 and it �as therefore
taken that the H2 o 2 content could not be greater than

1.01 mol/1.
b. Elimination of Hydrogen Peroxide with Catalase
To find the concentration of catalase required to
remove the H2 o 2 produced by honey, a range was tested with

1.01 mol/1 H2 o against growth of S.aureus.

2
0.5 ml catalase suspension (in the strength range
6,255 �igma units/ml to 100,080 �igma units/ml) was shaken
with 0.5 ml 1.01 mol/1 H2 o 2 and the mixture assayed for

antibacterial activity as above.

alone as a control.
It was found that the lowest concentration of catalase
used, 6,255 ;_igma units/ml (2.78 g/1), totally eliminated
the antibacterial activity of 1.01 mo�/1H2 o 2• This

concentration of catalase was therefore used during anti­

bacterial assays to eliminate hydrogen peroxide when
studying the other antibacterial factors in honey.
c. Selecting a Test Orgartism
A variety of bacterial species was screened in an agar
diffusion test fo� resistance to the sugar content of honey
and for sensitivity to non-peroxide antibacterial factors.
It was then demonstrated by neutralising the honey in a
bioassay that the non-peroxide activity was not due to the
 32

acid pH of honey.
In order to test the osmotic effect alone, a solution
of sugars was mixed in similar proportions to the
constitution of honey (Thomson., 1936. Table 7). The
sugar solution contained:
17.5 ml water
36.2 g glucose
40.0 g fructose
2 .8 g sucrose
All organisms were tested with the following
solutions:
1.01 mol/1 H o 2 (control)
2
0.5 ml 1.01 mol/1 H 2 o 2 + 0.5 ml catalase (control)

100% (v/v) honey solution

50% (v/v) hone.y solution
2 5% (v/v) honey solution
10% (v/v) honey s elution
1% (v/v) h·oney solution

100% (v/v) sugar solution

50% (v/v) sugar solution
25% (v/v) sugar solution
10% (v/v) sugar solution
1% (v/v) sugar s elution

0.5 ml 100% (v/v) honey solution + 0.5 ml catalase

0.5 ml 50% (v/v) honey solution + 0.5 ml catalase
0.5 ml 25% (v/v) honey solution + 0.5 ml catalase
0.5 ml 10% (v/v) honey, solution + 0.5 ml catalase
o .. 5 ml 1% (v/v) honey solution + 0.5 ml ca talase
The results are shown in Table 8.
 33

TABLE 7: Extract from n Chemical Composition

of New Zealand Honey", R.H.K. Thomson.

- -
-s::
E
•rl
Cfl -
�·rl tu) E Q)
Cfl tu) (1) � rl
•rl rl rl E
E H cu � H
'°s
'° '°s::
� 0 0-i Q)

t
•rl 0 '"d 0-i'"d
rl s:: � Cfl s::
Cfl

.£1
H� •rl �

ror
0 +:> rl .1:: +:> rl
s:: s:: s::�
Q)�
:> � 0-i�
O H� s:: Q) C) s::
i-1 E-1 s:: Q) s <� Q)
ro.-=l ::3
C) Q) �
:> 0
0 � P-i � < <

water % 16.6 .18.0 18.4 17.5

glucose % 38.4 37.1 34.5 36.2
fructose % 39.7 38.9 42.0 40.0
sucrose % 1.9 2.3 2.8 2.8
ash % 0.050 0.370 0.150 0.180
nitrogen % o. 029 0.056 o. 03 8 0.040
dextrins, and c, (undetermined) % 3.3 3.2 2.1
titrable acidity C.c 2.8 4.6 2.8 3.3
pH value 3.42 3.70 4.05 3. 81
formal titration C.c 1.2 1.8 1.2 1.3
specific gravity 1.421 1.41E 1.407 1.42.3
glucose:fructose ·1:1.03 1:1.05 11:1.22 1:1.11

••• uncalculated
TABLE ___§: Inhibition of growth of various bacteria x no inhibition
by manuka honey and sugar solutions / slight inhibition
...// inhibition> 1 mm
not tested
Sugar solution Honey solution Honey+ Catalase H202 H2 °
% (v/v) % (v/v) % (v/v) 2
catala.se
I'
Test Organism 1 00 50 25 10 1 100 50 25 10 1 100 50 25 10 1

Alcaligenes viscolactis* - - - - - - - - - - ... ... ... ... ... ... ...

Proteus vulgaris I J J J X
./J .IJ ./J ,/J X ... ... ... ... ... ... ...
Citrohacter frendii I ./ ../ ./ X ./) JJ .J./ X ... ... ... ... ... ... ...
... ... ... ... ... ... ...
jj

Shigella flexneri /J ././ JJ ./J X J./ JI J./ jj X

Salmonella typhimurium ./ ,/ ./ ./ X ././ .;/ ././ ./ X /./ ...� / X X // X

Pseudomonas aeruginosa X X X X X X X X X X ... ... ... ... ... ... ...

Serra�tia marcescens X X X X X J J ./ X X ./ ./ X X X ././ X

Escherichia coli Kl2 F- X X X X X JI ./,/ ./,/ ./ X /I X X X X ././ ·x

Escherichia coli Kl2 HFr ... ... ... ... ... ... ...
... ... ... ... ... ... ...
I ./ / / X J./ JI ././ /.! X

Esche!:_ichia coli Kl2 A I.I ././ ././ .// X ././ J./ .;/ .I./ X
Escherichia coli CSH34 I./ Jj I./ X X ./ ./ ./ X X ... ... ... ... ... ... ...
Staphylococcus aureus /.I //
... ... ... ... ... ... ...
X X X X X J./ ././ X .// ././ ././ X X v'./ X

Micrococcus luteus X X X X X X X X X X .

Streptococcus faccili� X X X X X JI ,I./ ,/ X X ./ X X X X I/ X

... .... ... ... ...

. . . . . . . . . . . . .. . . . . . . .
.S.a..r.Q.1 n a l utea .I ./ X X X X X X X X .I./' X

Bacill� circulans X X X X X X X X X X
'

�� Cul tnre did not grow

\.,.)
�
 35

Alcaligenes viscdlactis did not grow during the 18

hour incubation period and was therefore not considered a
possible test organism. It was found that Citrobacter
frendii grew during storage at 4 C and hence it also was
0

not considered.
The results show that four of the organisms tested
were both resistant to the high osmolarity of honey and
sensitive to the non-peroxide fact�rs. T·hese were
Serratia marsescens, Esch�tichia coli Kl2 F -,
Streptococcus faecilis and Sta�hylocbccus aureus.
Serratia marsescens was discarded because inhibition of
growth by all concentrations of honey was only slight and
therefore difficult to see. Staphylococcus aureus was
chosen as the test organism for all antibacterial assays
because of its greater sensitivity to the non-peroxide
activity and its sensitivity to lower concentrations of ...,

this factor. ( E. coli Kl2 F- and Streptococcus fe.e·cilis

were irihibfted·by non-peroxide activity only in undiluted
honey).
d. Effect of Honey Acidity on S.aureus
The effect of the low pH of manuka honey on growth of
S.aureus was assayed. Honey solutions were buffered with
0.1 mol/1 citric acid and the pH altered with sodium
hydroxide solution. Artificial sugar solution (as used to
screen bacteria, 3.lc) was treated in the same way and used
as a control.
The following solutions were assayed against S.aureus:
 36

1.01 mol/1 H 20 2

0.5 ml 1.01 mol/1 H 2 o 2 + 0.5 ml catalase

0.5 ml honey+ 1.5 ml water, pH 3.53

0.5 ml honey+ 0.5 ml water, pH 3.53+ 1 ml catalase

0.5 ml honey+ 0.5 ml citric acid, pH 4 + 1 ml water

0.5 ml honey+ 0.5 ml citric acid, pH 4 + 1 ml catalase

0.5 ml honey+ 0.5 ml citric acid, pH 7 + 1 ml water

0.5 ml honey+ 0.5 ml citric acid, pH 7 + 1 ml catalase

0.5 ml sugar+ 0.5 ml citric acid, pH 4 + 1 ml water

0.5 ml sugar+ 0.5 ml citric acid, pH 7 + 1 ml water

The results are given in Table 9.

The acidic pH of manuka honey had a slight effect on
the growth of S.aureus. A 2 5% (v/v) neutralised honey
solution produced a clear zone of inhibition however.
3.2 Screening Honeys fo� Activity
Honeys of different floral origin and age were
compared for non-peroxide antibacterial activity.
Samples diluted to 50% (v/v_) and 2 5% (v/v) were assayed,
both with and without catalase. The types assayed were:
 37

TABLE 9: The effect of pH on growth of S.aureus

in an agar diffusion. test.
(Concentration of all sugar and honey
solutions: 25% v/v)

Solution assaved Diameter of zone of

inhibiti on mm

H2 0 2 11

H2 o 2 + catalase 0

Natural honey (pH 3.53) 3.0

Natural honey (pH 3.53) + catalase 2. 0

Honey (pH 4) 3.0

Honey (pH 4) + catalase 3.0

Honey (pH 7) 2. 0
Horuey (pH 7) + catalase 2.0

Sugar (pH 4) 0.3

Sugar (pH 7) 0
 38

Floral T:y:·pe Origin State Age

Ling Heather Urewera National (a) comb 2 mths

(.Calluna vulgaris) Park (b) extracted 2 mths
Manuka (a) comb 18 mths
(Leptospermum
-�.-_.......-- sp.) (b) comb 18 mths
•.... ...�-
(c) comb 6 mths
(d) extracted 11 years
and boiled
Thyme Central Otago extracted 2 years
(Thymus vulgaris)
Penny Royal Waikato extracted 1 year
(Mentha pulegium)
Thistle-Nodding extracted 4 years
or California
Astelia extracted 4 years
(hanks ii or,, nervosa)
Cabbage tre·e Putaruru extracted 1 year
(Cordyline ind�yisa)
Clover (a) - extracted 1 year
(Trifolium sp.) (b ) Kihikihi extracted 1 year
Rewarewa (a) comb 6 mths
(Knightia excelsa) extracted: 4 years
(b) clear top
layer
( C) granular lower
layer
Towai extracted:
(Weimannia silvicola) (a) clear top 4 years
layer
(b) granular lower
layer
Various Wairoa extracted 1 year
(Sample pollen count: White clover 43%, lotus 31%,
salix 9%, reworewa 4%, briar 3%, dandelion 1%, gum 1%,
grass 1%, cabbage tree 1%, manuka 1%).

Extracted honeys had been heated at approximately 45 ° c

during the extraction process.

assayed on each agar plate to check that the catalase was

 39

active. The results are shown in Table lC.

Four types assayed had antibacterial activity in
addition to that from hydrogen peroxide. This activity
was strong iri all manuka honey. samples tested. It was
also present in rewarewa honey, and penny royal honey
possessed slight activity. The activ�ty in the honey
labelled "various" may indicate that the factor in rewarewa
or manuka honey is very active even in low concentrations,
or it may be due to one of the other floral components not
tested in isolation.
It was also concluded from these results that the
non-peroxide activity being investigated originated in the
plant and not the bee. These honey samples were all
collected from small commercial producers and it can be
assumed that ill were produced by the honeybee Apis mellifera.
Therefore existence of non-peroxide activity appears to
depend on the floral source of the honey.
The manuka honey whic.h was 11 years old and boiled had
reduced activity in comparison to the three other manuka
samples. It waB not possible to tell whether this was due
to prolonged heating at high temperatures or its age.
Manuka honey produced in 1980 was selected for further
experimental work and extraction because it had the
strongest antibacterial activity and of the active samples
was available in the largest quantity. After 1 year of
n
storage, in the dark at 4-c, it did not appear to have any
reduction in its activity against S.aureus.
3.3 Heat Stability
The neat stability of the non-peroxide antibacterial
factor was determined b� subjecting raw manuka honey to a
 40

TABLE lC: Activity of various New Zealand honeys,

with and without catalase, against growth
cf S.aureus.

�, inhibition of growth
x, no inhibition

Honey Concentration (v/v)

Floral T :£12 e Alone Including Catalase

50% 25% 50% 25%
Heather (a) X X X X

(b) ./ X X X

Manuka (a) ./ ./ ,/ ./

(b) ./ ./ ./ ./

( C) ./ ./ ./ /
(d) ./ I ./ /
Thyme X X X X

Penny Royal ../ X sl • sl.

Thistle ./ X X X

Astelia ./ X X X

Cabbage Tree X X X X

Clover ( a) X X X X

(b) X X X X

Rewarewa (a ) / X ../ X

(b) ./ X I
./ X

( C) ./ X sl. X

Towai (a ) X X X X

(b) ./ X X X

Various � ./ ./
 41

range of temperatures for up to l hour and then assaying

its activity.
A 50% (v/v) solution of honey was divided into thirty
1 m1 a1.iquots and heated at temper,atures of 45 ° , 55 ° , 6-5 ° ,
75 ° , 85 ° or 95 ° C in water baths for lO, 15, 30, 45 or 60
minutes. The solutions were then allo�ed to cool to room
temperature, and 1 ml catalase solution was added to each
becore assaying for antibacterial activity.

mol/1) and H o + catalase were also tested to check that

2 2
the catalase was effective. The results are given in
Table 11.
The factor retained all its activity after heating at
95 ° C for l hour. All test solutions (25% v/v) gave
approximately 4 mm diameter zones of clearance.
3.4 Effect of pH on the Non-peroxide Activity
The effect of pH was determined in order to see if the
factor required an acidic environment to be active.
A 50% (v/v) solution of honey in 0.1 mol/1 citric acid
was divided into eight 1 ml aliquots and the pH of pairs of
aliquots brought to 4, 5, 6 or 7 with sodium hydroxide
solution. 1 ml of catalase or distilled water was then
added to each and the solutions assayed. The experimental
sugar solution was treated in the same way to gauge the
effe6t of pH alone on S.au��us. The zones of clearing
measured are shown in Table 12.
Increasing the pH of the honey solution decreased the
activity of the factor but this was partially due to a
direct effect of low pH on bacterial growth.
3.5 Co�bin�d eff�ct of. H�ating and Neutralisin� Hortev
The combined effect' of pH and heat on the non-peroxide
 42

TABLE 11: Inhibition of growth of S.aur�us, as an

indication of heat stability of the non­
peroxide antibacterial factor.

Tem2erature Time (minutes)

o
( C) 10 15 30 45 60

45 j j ./ ../ ../

55 j ../ ./ ../ .I
65 .J J ./ ../ ../

75 ./ I ./ I I
85 ,/ ./ ./ ../ .J
95 I ../ ../ ./ I

J , inhibition of growth S.aureus

 43

TABLEJ2: The effect of pH on the antibacterial

factor, shown as inhibition of S.aureus.

Diamete� of Zo�e of Inhibition (mm)

Solutions DH 4 pH 5 pH 6 nH 7

Buffered honey+ water 3.0 3.0 2.0 2.0

buffered honey+ catalase 3.5 3.0 2.0 2.0
Buffered sugar + water 0.3 0 0 0 I
 44

antibacterial factor was investigated and compared with the

individual effects of heat and neutralising. Gannet and
Lavie (196 0 ) experimented with multifloral honeys and
found that neutralising their honey extracts gave no
decrease in antibacterial activity, and heating for 30 mins
at so 0 c produced only a 50% decrease in_ activity. However
when they combined the treatments (heating at 80° C for
30 mins at pH 8.4) the activity was lost, reg�rdless of
wJ:uch treatment was carried out first. They found that
activity returned as the pH was returned to the natural
pH of the honey. At pH 7 it was still very slight.
5 0% solutions of honey'·in water or citric acid were
neutralised or heated at 9 0° c for 30 minutes or both.
Catalase was added to half of them and the solutions were
assayed for antibacterial activity. Untreated honey was
also assayed for comparison.
The test solutions contained:

1. 01 mol/1 H2 02 (control)

1 ml 1. 01 mol/1 H 2 02 + 1 ml catalase (control)

Unheated
0 .5 ml honey + 1.5 ml water
0.5 ml honey + 0.5 ml water + 1 ml catalase
0 .5 ml honey + 0 .,5 ml citric acid, pH 7 + 1 ml water
0.5 ml honey + 0.5 ml citric acid, pH 7 + l ml catalas e
Heated at 9 0° c for 3 0 mins
r,
v.)
,..,
ml honey + 1.5 ml water
0.5 ml· h·oifey + 0.5 ml water + 1 ml catalase
0.5 ml honey + 0 .5 ml citric acid, pH 7 + 1 ml water
0.5 ml honey + 0.5 ml citric acid, pH 7 + 1 ml catalase

The zones of clearance measured are shown in Table 13.

 45

TABLE B: Inhibition of growth of S.au�eus by

manuka honey that has been heated,
neutralised or both.·
(All soluti�ns 25% v/v honey)

Solution Diameter of zone of

inhibition mm

Honey 4.5
Honey + catalase 6.0

Neutralised honey 3.5

Neutralised honey+ catalase 3.5

Heated honey 4.5

Heated honey+ catalase 4.5

Heated, neut. honey 2.5

Heated, neut. honey+ catalase 1.5
 46

The results demonstrated that heating at low pH did

not decrease the antibacterial activity in manuka honey;
neutralising produced a slight decrease in activity, and
combining the treatments reduced the activity by about half.
No explanation could be found for the apparent increase in
activity of untreated honey when catalase was added or the
decrease when catalase was added to·heated, neutralised honey.
3.6 Ultr�filtration
Ultrafiltration with an Amicon UM 05 membrane was
attempted to obtain an idea of the size of the anti­
bacterial molecule(s) involved. The membrane had a
nominal retentivity of 500 daltons.
A 25% v/v honey solution was used and pressure was
applied for two hours with continuous stirring. At the
end of this time however nothing had passed through the
membrane. It was concluded that the honey solution was
too viscous for ultrafiltration due to the high
concentration of sugar present.
3.7 Summary
Non-peroxide antibacterial activity was found in
manuka, rewarewa and penny royal honey. The activity in
manuka honey was further studied. S.�ureus was the most
sensitive osmotolerant organism and was used to assay the
activity. The active compound(s) was heat-stable at an
acidic pH but less stable at neutral pH, and was less
active at neutral pH.
 47

CHAPTER FOUR:

SOLVENT EXTRACTION OF THE ANTIBACTERIAL

ACTIVITY FROM HONEY

Gonnet and Lavie (1960) obtained an antibacterial

fraction from several multifloral honeys by extraction
with cold acetone. This extract c6ntained only a portion
of the activity of the whole honey, whereas alcohol J

extracted the activity totally. None of the activity

would pass into ether. In further research Lavie (1963)
obtained an ether extract of a hot alcohol extract. He
was also able to extract this honey (unspecified type):, in
cold ether and when the extract was redissolved in water
and distilled he found the active fraction was volatile
at 95 ° c.
A solvent or solvent mixture was required for manuka
honey which would extract all the non-peroxide anti­
bacterial activity, and would leave behind most of the
sugar present. It was necessary to eliminate most of
the sugar from the extract because it would become more
concentrated when dried on a chromatogram and may appear
as an area of inhibition during a bioassay for activity.
4.1 Solvent Extraction
Honey (10 ml) was homogenised wi�n absolute
ethanol (20 ml) in a Waring blender for about 2 minutes.
The solution was allowed to cool to room temperature
(about 15 minutes) and then centrifuged at 8,000 r.p.m.
U)• • ,: ,.,;
for 15 minutes. The supernatant was rotary evaporated
 48

and freeze-dried tQ remove all the alcohol. The yellow

viscous produce was redissolved in 10 ml distilled water
and assayed for antibacterial activity.
The test solutions were:
0.5 ml extract+ 0.5 ml water+ 1 ml catalase
0.5 ml extract+ 1.5 ml water+ 2 ml catalase

0.5 ml honey+ 0.5 ml water+ 1 ml catalase

0.5 ml honey+ 1.5 ml water+ 2 ml catalase

(control)

1 ml 1.01 mol/1 H o + 1 ml catalase (control)

2 2
The results are given in Table 14.
All non-peroxide antibacterial activity could be
extracted ,in absolute ethanol.
The above procedure was repeated using diethyl ether
instead of ethanol but no activity was extracted. After
the rotary evaporation stage a waxy white precipitate was
left around the flask. This would not dissolve into
water and it was difficult to redissolve into ether.
An attempt to extract the dried alcohol extract with
diethyl ether gave the same result.
4.2 Effect 6f A�id artd Alkali dn Factdr St�bilitv
An attempt was made to extract the active factor into
("-(1 \U l I• (' I..

ether from an i aci�ic and an alkaline solution of the

alcohol extract in-wa-t-er.
An alcohol extract of 10 ml honey redissolved in 10 ml
water was diYided into two 5 ml aliquots; one aliquot was
acidified to pH J with HCl and the other taken to pH 10
with NaOH. They were then each vigorously shaken with
 49

TABLE 14.: Comparison of inhibition of growth of

-s�au��us by raw manuka honey and an
ethanol extract of it.

Test Solution Diameter of zone of inhibition

(containing 50% v/v catalase) (mm)

Extract (25% v/v) 5

Extract (12.25% v/v) 2

Honey (25% v/v) 5

Honey (12.25% v/v) 2

H202 (no catalase) 16

H2 o2 + catalase 0
 50

15 ml ether in separating funnels and left to settle.

The ether fractions were dried by rotary evaporation, 1 ml
distilled water was added to each and the solutions
assayed. The remaining aqueous phase from each separating
funnel was neutralised and assayed. The results are
shown in Table 15.
Lowering the pH for the extraction did not make the
factor more soluble in ethe�. The high pH inactivated the
factor completeTy. It was therefore not possible to see
which fraction it had entered. Hence ·the above procedure
for extracting from alkaline solution was repeated but both
phases were acidified before neutralisi�g and assaying to
see if the activity would return. When the ether fraction
had been dried by rotary evaporation acidified water was
added to the flask, shaken, then neutralised and assayed.
As a control the second 5 ml aliquot of extract was ·checked
for antibacterial activity with no further treatment.
The results obtained were the same as above. The
alcohol extract itself was active however. Alkali again
destroyed the antibacterial activity and this was not
restored by reacidifying before neutralising. The same
results were also obtained when the� procedure was_ repeated
in the cold room (at 4 ° c) with chilled ether.
The effect of taking the aqueous solution to pH 8 was
then investigated.
An aqueous solution of the alcohol extract was
adjusted to pH 8 with NaOH. The solution was left to
stand for approximately the same time (about 15 minutes)
as the factor had teen exposed to pH 10 during the ether
extraction attempt in the separating funnel. The solution
 51

TABLE 15: Activity of ether extracts of manuka

honey measured against growth of S.aureus.

TREATMENT Acid Alkali

PHASE water ether water ether

INHIBITION inhibition no no no
inhibition inhibition inhibi tion

Al l test solutions 50% v/v honey with catalase.

Control

Test Solution Inhibition

inhibition

no inhibition
 52

was then neutralised and assayed. The results are given

in Table 16 •·
The extract retained only slight antibacterial
activity. It was concluded tpat increasing the pH of the
active solution brought a concomittant decrease in
antibacterial activity.
4.3 Extracts of Inacti�e Honeys
The alcohol extracts of two honey types, which were
previously thought to have no non-peroxide antibacterial
activity, were assayed to test if any components were lost
during extraction which inhibited their activity while raw.
The extracts of heather and c�over honeys were
obtained by the same procedure as for manuka honey and
assayed by an agar diffusion test against S.aureus.
The results are shown in Table'l7.
It was concluded that these contained no non-peroxide
antibacterial activity.
4.4 Thin-layer Chromatography on commercial plates
Some preliminary chromatography was carried out with
,,
the extract of manuka honey using commercial thin-layer
;

chromatograms to o�tain information on how the non-peroxide

activity migrated in various solvent mixtures of ethanol
and diethyl ether. A �ioassay was carried but to find
its approximate position, by laying the used chromatogram
on seeded agar plates for 2 hours, then removing it and
incubating the plates overnight. Areas of clearing of
S.aureus ser ved to indicate the position to which the activity
had migrated during chromatography. It was thought to be
unnecessary to incorporate catalase into this assay
because any glucose oxidase not precipitated during
extraction (White et al. 1962) would be denatured and
 53

TABLE 16: Effect of pH 8 on antibacterial activity of

an alcohol extract of manuka honey, measured
as inhibition of growth of S.�ureus.

Test Solution Inhibition of growth

(50% v/v)

Extract inhibition
Extract+ catalase inhibition
Extract - neutralised inhibition
Extract - neut. + catalase inhibition

Treated extract - neut. slight inhibition

Treated extract neut. + catalase slight inhibition

H2 02 inhibition

H2 o 2 + catalase no inhibition

sl. inhibition = < 1 mm diameter zone of inhibition.

 54

TABLE 17: Activity of extracts of two inactive raw

honeys, against growth of s;�ureus.

Honey samples: Heather-(b)

Clover (b)

Solution Activity
(50% v/v) Heather Clover

Honey slight inhibition inhibition

Honey+ catalase no inhibition no inhibition
Extract+ catalase no inhibition no inhibition

H2 o 2 inhibition

H2 o2 + catalase no inhibition

slight inhibition = < 1mm diameter zone of clearance

 55

therefore inactivated by the high temperatures attained by

the hairdryer used to dry applications on the chromatogram.
In order to find the loading necessary for this assay
method a solution of honey extract in alcohol or twice its
equivalent concentration in raw honey was applied to a thin
layer chromatogram in 5, 7 and 10)11 amounts. The assay
procedure was carried out without developing the chromatogram
in solvent. The agar plate was incubated overnight and
then flooded with a freshly-made solution of lactate
dhydragenase detection reagent and incubated for another
30 minutes in order to make clearings more visible in the
disturbed agar. Clearings were observed. at the three
points where material from the extract had diffused into
the agar from the chromatogram.
As the leadings appeared to be appropriate it was thus
repeated with the same quantities of extract applied and
run in a solvent mixture of 80% ethanol - 20% diethyl ether.\
Lines were etched along the chromatograms to ensure the
extract applications migrated in straight lines. The
appearance of the seeded agar plate is illustrated in
Figure 2.
This indicated that to some extent the areas of no
growth were artefacts due to the thickness of the agar and
surface effects. No explanation could be found for the
shape of the clearing in the centre of the plate or for
the 10 f 1 paint of application producing a smaller clearing
than the 7 fl point of application. The experiment was
repeated. The results of the second assay showed little
bacterial growth on the top half of the plate (the agar
was also thinner here) and on the lower half there was a
 56

etched lines
'I/

'\l -.) .}�r.,"'J

� '.,JI.

j
�solvent front

), �

.,
�
'
� ?;S

�
� �AREA OF CLEARING
,. y

�-
.. �_. ----�- .. --. -- -----.. -- �application
�
line
A
�
�AREA OF CLEARING
U'w
AA A

1' t t
5pl 7 fl lOpl
alcohol extract
applied

BJ stained purple i.e. normal growth of S.aureus

D no staining i.e. inhibition of growth of S.aureus

Figure 2: Appearance of seeded agar plate after

superimposing thin-layer chromatogram,
incubation and staining with lactate
dehydrogenase detection reagent.
 57

clearing over the position of the application line again.

This time the area of clearing was lazger with increasing
volume of extract applied.
To test whether this consistent clearing over the
application area might be due to a high concentration of
sugar, a mixture of sugars was made up �n the same
proportions as honey (as in Chapter 3) and extracted with
alcohol as was the honey. The sugar ex�ract was applied
to a chromatogram in 5, 7 and lOJAl quantities, developed
in alcohol:ether (80:20.). solvent and assayed as above.
A clearing corresponding to each application was observed
on the agar plate with the largest clearing where 10?1 of
sugar extract was applied.
A second sugar chromatogram was run in the same way
and stained for sugar with p-Aminophencl instead of
assaying for antibacterial activity. The appearance of
the stained chromatogram is illustrated in Figure 3.
It was concluded that there was still a very high
percentage of sugar in the extract and that this was
probably producing the clearing corresponding to the
application point on the thin-layer chromatogram.
4.5 Further Extraction
The activity due to the sugar could have been masking
another antibacterial component. It was therefore
necessary to modif� the extraction process to eliminate
the sugar. A solvent mixture was required which was
sufficiently non-polar to dissolve only a little of the
sugar but polar enough for the antibacterial acitvity to
be soluble in it. Solvent mixtures of absolute ethanol
and diethyl ether were tnvestigated.
 58

strips of silica removed

�

-solvent front

Charring - fading ... ..

from dark brown :::
up to brownish�
yellow

charred -- application line

dark brown
5pl lOpl
sugar extra-ct
applied

Figure 3: Appearance of thin-layer chromatogram with

sugar extract applied. It was run in ethanol:
ether (80:20), dried, sprayed with p-Aminophenol
and charred at 110° c, 5 mins.
 59

An alcohol extract of 10 ml of honey was divided into

five aliquots and ethanol and ether added in the quantities:
4 ml+ 1 ml, 3 ml+ 2 ml, 2 ml+ 2 ml, 2 ml+ 3 ml and
1 ml+ 4 ml.
Each mixture was shaken on a vortex mixer for 2 minutes
to aid sugar precipitation and centrifuged at 8000 r.p.m.
for 10 minutes. The supernatants were removed and dried
by rotary evaporation. Distilled water (1 ml) was added
to all dried fractions and precipitates and the resulting
solutions assayed for antibacterial activity. The results
are shewn in Table 18:.
The activity was soluble in all solvent mixtures
tested. The precipitate from the 80% ether solution may
have caused inhibition of growth of S.aureus by its .high
sugar concentration only but a 60% ether solution was
selected for use in further extraction because some of the
inhibition may have been due to the presence of a little
of the antibacterial factor. The loss of this from the
supernatant may have been so small as to fall within the
marginal error for measuring activity by zones of clearance.
The modified extraction procedure subsequently
adopted was thus:
(1) - extract 10 ml honey in 20 ml ethanol and centrifuge
(2) - rotary evaporate supernatant to remove alcohol
(3) - freeze dry to eliminate as much water as possible
(4) redissolve in 20 ml ethanol and add 30 ml diethyl
ether
(5) - agitate on vortex mixer until sugar (white)
precipitate forms
(6) - centrifuge to remov� sugar
(7) - concentrate supernatant by rotary evaporation
 60

TABLE lB: Solubility of antibacterial activity in

solvent mixtures of different polarity,
shown as inhibition of growth of S.aureus.

Solvent mixture Fraction

(ethanol:ether) precipitate supernatant

80 :20 no inhibition slight inhibition

60: 40 no inhibition slight inhibition
50:50 no inhibition inhibition
40:60 no inhibition inhibition
20:80 inhibition inhibition

slight inhibition - <. 1 mm diameter zone of inhibition

 61

An extract obtained � the modified procedure was

applied in 10, 20 and 30fl quantities to two thin-layer
chromatograms and developed in solvent as before. One was
stained with p-Aminophenol and- the other was assayed for
antibacterial activity. Application and dryi�g of extract
was found to be quicker and easier with- less sugar present.
The agar plate on which one chromatograph was
superimposed showed a clearing which extended across the
tracks from the lOpl and 20fl applications, ,R·f = 0.68.

This may have been an artefact as it was continuous across

the area between the two tracks where no extract had been
applied. There was a clearing again over the area
corresponding to the application points of the chromatogram,
which was continuous and increased in size and intensity
with the larger extract applications. p-Aminophenol
stain, however, indicated little sugar present.
4.6 Summary
All the non-peroxide antibacterial activity of manuka
honey could be extracted into absolute ethanol, but none
would dissolve in ether.
Ether was added (60%) to the ethanol extract to
precipitate most of the sugar from the honey but maintained
the solubility of the activity.
The solubility and behaviour on thin-layer
chromatography indicated that the activity was fairly polar.
Acidifying the extra6ted activity·did ncit make it more
soluble in ether. Adding alkali to pH 10 eliminated the
activity, making it undetectable.
Extracts of inactive raw honey were also inactive when
assayed for activity.
 62
CHAPTER FIVE:

ISOLATION OF THE NON-PEROXIDE

ANTIBACTERIAL ACTIVITY

Thin-layer chromatography (TLC) is a useful technique

for separating the compounds within a mixture because it is
rapid, convenient, sensitive and gives good resolution. With
preparative thin-layer chromatography heavier loading is
possible and the various fractions can subsequently be eluted
and evaluated. Hence preparative TLC was selected to aid
isolation of the non-peroxide antibacterial factor or factors.
Migration of the different substances in the honey extract
was followed by various visual methods after each chromato­
graphic run, as R�I values vary in preparative TLC.

of a compound can vary with size of application, length of

chromatogram, activity of adsorbant, sealing and dimensions
of the chamber, atmospheric conditions (temperature and
humidity), differences in the solvent mixture and with
thickness of the layer.
Simple solvent mixtures were used to begin with as they
give better reproducibility.
The ascending method was used.
5.1 Preliminary P��par�tive ·TLC
Some preliminary TLC was carried out in order to find a
solvent mixture which would separate the activity from the
other components of the extract, particularly the sugars.
This was necessary as, for reasons stated above, migration on
a preparative plate would not be the same as that on the
Eastman Chromagram Sheet u9ed previously. Therefore the
 63

location of the active components had to be retraced.

Plate preparation is described in Chapter 2. After
activation and cooling of the plates an alcohol-ether
extract (5 ml) of 10 ml honey was applied to a line marked
15 mm from the adsorbant edge, and dried with a hairdryer.
The high temperatures reached by the hatrdryer would also
inactivate any glucose oxidase enzyme (and any other
proteins) present should any have remained soluble in the
ethanol. During these preliminary investigations, plates
of size 50 x 200 mm were used and development was carried
out in a glass chamber sealed with a silicone-greased glass
plate. Bioassays, to localise activity, were carried out
after chromatography by superimposing the air-dried TLC
plates on agar plates, of the same size, for two hours in
a sealed plastic container (lined with damp strips of
filter paper). The agar plates were then incubated at
37 ° c overnight and later stained for lactate dehydragenase,
as described previously.
Agar plates were prepared in the cold room (about 4 ° c),
the warm, seeded agar being poured onto a tray of glass
plates which had been chilled (4 ° c) for several hours
beforehand, to obtain a thinner agar. It was necessary for
the agar plates to be thin so that any activity would diffuse
through the whole depth of agar and would not be masked,
after staining, cy- bacterial growth beyond the limiting
dilution of activity.
The general detection reagents sulphuric acid and
molybdophosphoric acid were used to visualise the position
of the separated compounds on the TLC plates which were
not bioassayed.
 64

a. Testing General Reagents and Bioassay

Initially three chromatograms were run in the same
solvent in order to check the method of detecting the
activity describ�d above and to see if the sulphuric acid
and molybdophosphoric acid stains detected the active
factor(s) and other compounds present. - These preliminary
runs were also used to check that a heavier loading of
extract did not significantly alter the migration pattern
of compounds and that sugar was not affecting the bioassay.
Three preparative TLC plates were activated, co�led
and each given an application of 5 0 f 1 honey extract, 5 0 }t 1
sugar s elution extract and 75 p 1 honey extract. The
plates were developed at the same time in 80% ethanol - 20%
ether in a sealed glass tank until the solvent front was
10-20 mm below the top edge of the silica. After removal
from the chamber the plates were allowed to air dry and
then one was used for the bioassay, as described above.
One plate was sprayed with sulphuric acid and the other
with molybdophosphoric acid and then both were incubated at
100-110 ° C for 30 minutes to complete staining.
After staining with lactate dehydrogenase detection
reagent, the agar plate showed two green "spots" of
different size (corresponding to applied honey extract)
which represented zones of growth inhibition of S.aureus.
These were approximately just below tl'e�position of the
solvent front. An area of clearing (green) was also
present about 30 mm above the origin in the line of
migrati on of the 75 }' 1 of honey extract. Unlike the
clearings near the solvent front, this one did not
correspond in position tp any compounds stained with the
 65

general reagents on the other two plates. The stained

plates are represented in Figure 4.
The time taken for stained spots and streaks to
appear with heating was about 2-10 minutes. They became
darker and more distinct with further heating.
Comparing all three plates as closely as possible it was
concluded that the upper areas of bacterjal inhibition
corresponded to areas somewhere within the upper 20 mm of
development. The purple colour of compounds stained with
sulphuric acid in this region indicated that they were
possibly phenolic derivatives.
Sugar from the honey extract was l�cated by comparison
with the migration of the extract of the sugar solution,
but the honey had produced a larger and more darkly charred
spot. This indicated that other compounds in the honey
extract had moved to the same position as the �ugar. As
there was no inhibition on the bioassay plate corresponding
to these areas it was concluded that there was no need to
attempt further separation and isolation of these compounds.
Comparing the migration patterns of the 50 f 1 and 75;,i 1
honey extract spots showed that a heavier loading did not
significantly alter the chromatographic behaviou� and
hence the position of separated fractions.
Further preliminary TLC work was necessary to
elucidate certain possible artefacts before attempting to
isolate active factor(s). The clearing produced 30 mm
above the application line could have been an artefact as
with previous TLC, or a compound with strong antibacterial
activity which was in such a small quantity as to be
undetectable with a spray reagent. This could be confirmed
 66

KEY TO FIGURES:

very dark
•
•

.. . . :. :. .. . very faint

x = no inhibition of growth

s1 = �1 mm inhibiti on

,/ = �1 mm inhibition
 67
sulphuric molybdophosphoric

'silica edge/

purple
20 mm

lime­
green
black-grey� back-
�groun<
with
dark
blue
spots

I
....:...
::.:

white
background

grey---,=, --e--@-­ �application-=f . --@--@--

(purple line
centres)
f!o-����--����-�silica edge-+--�------�--------...J
i
50�1
i
50Jl
t
75yl
i
50pl
i
50f1-
f
75y.l
honey sugar honey honey sugar honey

FIGURE 4: Staining of preparative TLC plates, following

honey and sugar extract application and
development in 80% ethanol:20% ether.
 68

by repeating chromatography with the same solvent mixture

and others, less polar and more polar, in order to see if
the position of the area of clearing was changed in the
path of migration. Also both stained TLC plates
indicated an artefactual staining was produced along the
solvent front, which was also visibly yellow before
spraying and charring. Hence the ·darker stained compounds
on the solvent front, which occurred in line with both
sugar and honey extract origins, may also have been arte­
facts due to the use of bulk solvents (less than 95% pure)
during these �reliminary investigations.
b. Effect of Different Solvertt Mixtures on Comoound
Separation
A solvent mixture for chromatogram development was
required which would give a good separation distance
between components, particularly between sugar and anti­
bacterial factor(s), and sharp migration patterns, i.e.
limited tailing and spreading of spot size.
Two solvent mixtures were investigated other than the
80% ethanol-20% ether previously used; a more polar
solvent, 100% ethanol and a less polar mixture 60% ethanol-
40% ether. Three chromatograms were developed in each
solvent and treated as previously with either sulphuric
acid spray, molybdophosphoric acid spray or bioassayed for
antibacterial activity.
Honey extract (75pl) was applied to each plate and
the migration pattern compared with that.from 75?1 sugar
extract and a control �f .75�1 of 40% alcohol-60% ether
solution, in order to see which stained spots consisted of
sugar and which were artefacts jue to the solvents used.
·,·
The results are shown in Figures 5, 6 and 7.
 69
sulphuric acid molybdophosphoric acid bioassay

brown-yell w

brown

-@- --
@·. . ..·.·.
,"f-f•. - - _. _. - -

"' -if"
...,.. -1' -1'- � ,f"' '1'
75�1 75fl 75;,tl 75r1 75J-Al 75.)Al 75_µ 1 75 ,tt 1 7 5 "'i'
.,ul
sugar honey solvent sugar honey solvent sugar honey solvent

FIGURE 5: Chromatogram developed in 100% ethanol

 70
molybdophosphoric
sulphuric acid acid bi oassay

. .. .
:·. .
brown
. ..
.... .-
.... ..
.: ..

75p 1 75JJ1 75_µ1 75]-A 1 75 ;;l 75 µ 1 75�1 75�1 75µ1

sugar honey solvent sugar honey solvent sugar honey solven�

FIGURE 6: Chroma to gram de \eloped in 80% ethanol: 20% ether

 71
molybdophosphoric
sulphuric acid acid bi oassay

. . ·.

No bacterial
growth on plate
- agar too thin
·brow

.. .

... ·.·.·.·
brown
..

----------- -- --
'j 1' 1' 1' 1'- 1" /j
75)'{ 1 75µ1 75µ1 75pl 75pl 75�1 75}{ 1 75�1 75 _.,u l
sugar honey extract sugar honey extract sugar honey extract

FIGURE 7: Chromatogram developed in 60% ethanol:40% ether

 72

The stained and charred plates indicated that 60%

ethanol-40% ether solvent produced the greatest separation
of fractions. Ethanol, alone, carried most of the honey
extract into the upper half of the chromatogram, with
little distance between different compounds. These plates
also confirmed that artefactual S'taining was produced at
the front from the solvent alone. Therefore it was
concluded that, prior to loading and developing
chromatograms for qualitative or quantitative analysis,
they would have to be cleared of impurities, such as
phthalate, by development in a more polar solvent to the
end of the adsorbant and then reactivated prior to develop-
ment with extracts. This would have to be halted several
centimetres before the end of the silica to avoid the band
of impurities.
Variability in Rf values was also clearly demonstrated.

Two of the bioassay plates gave results. In both,

large clearings were produced just behind the solvent front.
The antibacteri�l activity here appeared to be quite
potent, as in both cases it extended across the possible
path of sugar extract, where no compounds were detected
with either spray reagent. It was concluded that an
active fraction did migrate to near the solvent front as
it was visualised in all three bioassays which had worked,
but that this method of bioassay was not precise enough
to indicate the exact position of the fraction. The
second clearing which appeared in all cases on the lower
halt of the agar plates was concluded to be an artefact
of the bioassay procedure. Its ."migration" only
approximated that of hJney extract components and did not
 73

correlate with the polarity of solvent mixture used i.e.

with development in 80%. ethanol it did not migrate to the
same position both times. Each time, too, it was
associated with a "fault" in the agar.
It was decided to discard this method of bioassay
because of its inaccuracy due to .the ease of diffusion of
the antibacterial factor(s) and the difficulty in making
agar plates of adequate and constant depth. Pouring
plates in the cold to allow quicker setting and therefore
thinner layers caused uneven layers as overlapping of agar
occurred, without mixing. Hence while superimposing
chromatograms·· small areas were not in contact with the
silica and very thin areas of agar werB often completely
absorbed into it, or would not support growth of S.aureus.
An attempt was made to impro ve the plates by pouring
them in individual metal dishes. It was difficult to
obtain metal dishes with a flat bottom however and this did
not eliminate the production 0£ dark veins and spots
throughout the agar after staining. These were covered by
silica adsorbing to the agar during superimposing. They
could not be washed off until after incubation and growth
of S.aureus by which time they were completely embedded in
the agar.
In order to avoid the artefacts mentioned above, a
biological detection method (Ki�ohner, 1978) was tried
which involved coating the thin-lay chromatogram itself
with seeded agar. After air-drying the developed
chromatogram ,inoculated agar, containing 0.7 ml of a 5%
solution of·triphenyltetrazolium chloride in 50% methanol
per 50 ml of media was poured gently onto its surface.
 74

After cooling this layer was prcrtected � pouring a thin

coating of sterile agar on top. When this was set the
plate was kept in a closed container in the refrigerator
(o 0 c) for one hour to allow the antibacterial compounds to
diffuse into the agar. After incubation (37 ° C overnight)
bright yellow spots of inhibition should have appeared in
a red-brown background. However, when warm agar was
poured onto the TLC plate areas of silica bubbled up and
became pitted and after incubation the area of agar in
between remained colourless.
5.2 Locating the Antibacterial Activity
Several methods of detection were attempted in order
to find out more accurately which compounds were anti-
bacct erial and to discover, if possibl e , . to which chemica1
group they belonged. Also a detection method was
required which was non-destructive to enable qualitative
chemical analysis of the antibacterial factor(s) once
located.
a. Biological Detection of Antibacterial Factor(s)
Anti bacterial activity was assayed using an agar well
diffusion technique.
Honey extract (1 ml) wa� applied as a band to each of
two activated TLC plates. The chromatograms were developed
in 60% ethanol-40% ether (Solvent I) and air-dried. One
chromatogram was then divided into thirty horizontal strips
(approximately 5 mm in width) across the path of migration
of the honey extract. Each strip of silica was removed
from the glass plate, mixed with 200-300? 1 distilled water
(to form a wet sludg�) and added to a well in a seeded agar
plate, which was then incubated. The second plate was
sprayed with sulphuric acid to indicate the location of
 75

extract fractions. The results are given in Figure 8.

Repeating this with sugar extract, instead of honey extract,
produced no bands of inhibitory activity.
Following this a 300 mm p�eparative TLC plate was
loaded with a band of 6 ml honey extract and developed in
Solvent I. Ascending developmerit �as.. carried out in a
metal chamber with filter papers lining the sides to maintain
a solvent-satura�ed atmosphere. A glass plate was laid on
top to seal the cha�ber. The plate was air-dried and
vertical strips of 50 mm width were used in the following
investigations. This enabled a more consistent comparison
of the different detection methods because the antibacterial
fraction(s) would be located at the same migration distance
in each test.
b. Detection Under Ultra-Violet Illumination
Ultra-violet illumination of the developed chromatogram
was investigated as a non-destructive method of locating
components of the honey. A section of the chromatogram
was visualised under U.V. illumination of 254 nm and 350 nm
wave length and the patterns recorded. The same section
was then assayed for antibacterial activity as described
above. The degree of inhibition of bacterial growth was
given a relative visual assessment only - the zone of
clearing was not measured. The results are shown in
Figure 9.
The inhibition of growth shown 1:y the strips numbered
11 and 29 was only slight and therefore assumed to be
artefacts. This was confirmed when this procedure was
repeated with a 50 mm TLC plate, where only fractions
corresponding to strips J2-36 produced inhibition of growth.
 76

front

., __ 30
.; 29
..I 28
sl 27
��������--'l__.26
X 25
X 24
X 23
X 22
black 21
X
X
X
X
X
X
X
X
X
X 2
sl 1
X 0
sl 9
sl 8
X 7
X 6
X 5
X 4
X 3
>-------------- ·
--------- ,:-- 1
2

Sprayed with Assayed against growth

sulphuric reagent of S.aureus

FIGURE 8: Comparison of 2 chromatograms of 1 ml honey

extract, developed in Solvent I.
One plate was stained with sulphuric acid and
the other assayed by agar well diffusion for
antibacterial activity in the strips removed
from the plate.
 77

Key to Figure 9

Absorbance

Strong absorbance
 78

254 inm 3 5 Q IfiID B1.oaa:say

,--.

�
. . ..
�- - ..
.././ 37�
..j,// 36
././ ./ 35
j// 34
// 33
/./ �?
y 11
X '3 0
./ 29
X 28
X 27
X 26
X 2 c;
V ? i.
X 23
V ??
21

il[ l!i 1' !i[il'!l li l [il! l' li !i[l! !li!'l �--------

X
20
1
X
1 X 19
X 18
X 17
V 1h
V 1 c;
X 14
X
�-:-:,:,:-·.:-:-:-:-:-:-:-:,;;.r------ 11
X 12
1 1
X 10
X g
�r Q
X 7
X 6
X c;

: ::: : ::
/,

:
: ::::::::: ::::: : : ::: ::: : :::::::::::::::::::::::::?::::::/:::::
V
............................... � ... . ...
appl ication line
'3
� 2
------- - - ----� --- ._._.,_ � - ----- -� -- ---- ------
y 1

FIGURE 9: Location of antibacterial fraction(s') by

comparing appearance under u.v. illumination

with assay against S.aureus.

 79

Strip 37 (the front) was also concluded to be an artefact.

Inhibition by No •. 37 was probably due to a high concentration
of phthalate and other impurities in the Keiselgel. The
,solvent in which the 300 Illiil TLC plate was developed had
been allowed to ascentl to the top of the adsorbant and the
plate had not been 11 cleared" prior _to this development.
A chromatogram, applied with sugar extract and
developed in SolventI, when viewed under ultra-violet
illumination of 350 �m wavelength showed that sugar had
move approximately to the first 1:and of absorbance (assay
bands numbered 14-20, Figure 9).
c. Spray Reagents
Spray reagents can be used to give an indication of
what type of com.pounds are in the separated fractions on a
chromatogram. The sulphuric acid reagent used to locate
compounds gave a purple colour in the region where anti­
bacterial activity was detected. This reagen� was similar
in composi, tion to a spray of concentrated H So 4 - water -
2
95% ethanol (1:1:8) which colours phenolic derivatives
purple or blue (Elliot et al.,1969). All artefacts of
solvent development, on the solvent front, also coloured
purple·after staining and heating. Therefore, an attempt
was made to confirm whether the active fraction(s) was/were
phenolic by using other spray reagents.
Five spray reagents were used to stain 50 mm sections
side by side on the 300 mm preparative plate. Aluminium
foil was used to protect other sections during spraying.
 80

Sulphuric Acid
The general spray reagent used previously appeared to
stain purple only the solvent front i.e. kieselgel and
solvent impurities. The band� of activity just behind
this were not stained. An area half-way up the plate was
charred black and it was concluded that this was sugar.
Ferric Chloride
A section was sprayed with a saturated solution of
anhydrous ferric chloride in methanol. Kirchner (1978)
lists this spray for terpene phenols. Elliot et al�,(1969)
lists a similar reagent:- a saturated solution in anhydrous
diox�n or chloroform to stain most phenolics red-violet.
They also claim that phenols having two or more vicinal
hydroxy groups, or ortho (or peri) hydroxycarbonyl groups, �- {-cJ
give coloured chelates with ferric salts.
In this case however, none of the antibacterial bands
were stained.
Anisaldehyde-Sulphuric Acid
Concentrated H2 so 4 (1 ml) was added to 0.5 ml anis-

aldehyde in 50 ml acetic acid. The chromatogram was

sprayed immediately after the solution was mixed, and heated
°
to l00 C for 5-10 minutes (Stahl, 1969). Lichen
constituents, phenols, terpenes, sugars and steroids yield
violet, blue, red, grey or green products (Stahl, 1969).
This was found to stain all compounds which had been
r
visualised previously with a general reagent. The area of
antibacterial activity was stained bright pink-mauve.
This was repeated with a small TLC plate spot-applied with
honey extract and the active area was stairied pink-orange
when sprayed.
 81

Fe�rid Chl��id�-Potassium Ferricyanide

This stain is fairly specific for phenolic compounds.
Aqueous 1% solutions of each salt (A.R. quality) were mixed
in equal volumes to form an orange-brown solution, with no
trace of blue (Elliott �t �1.,1969). Reactive phenols
should give blue spots on a pale blue background immediately,
without heating,
The two active bands corresponding to strong absorbance
under 254 pm u.v. illumination appeared dark blue immediately
after staining. The two bands of less strong absorbance
above each of these also appeared blue, but with less
intensity. This result was confirmed with a chromatogram
run from a spot application. However only one elongated
blue spot appeared on staining. This plate was treated
with ferric stain immediately after air-drying whereas the
larger plate had been left a few days before staining.
Partial oxidation of the activity may have occurred on
standing and only then have given the appearance of two
separate bands of u.v. absorbance (Figure 9).
Vanillin-Sulphuric Acid
Owing to the inconclusive result obtained with the
anisaldehyde stain and the possible plant origin of the
antibacterial activity, a reagent which detects terpene�
and essential oils was tested. A section of chr�matogram
was sprayed with a solution of 1 g vanillin in 100 ml 95%
The solvent front and sugar

fraction only became stained.

 82

Con�lusions
Although staining the chromatogram with sulphuric acid
or anisaldehyde gave inconclusive results about the chemical
nature of the antibacterial activity, the ferric chloride­
potassium ferricyanide spray indicated that it was probably
phenolic. The negative results obtain�d with vanillin or
ferric chloride stains tend to eliminate the possibility of
it being a terpene. The lack of detection with ferric
chloride also limits the possible structure somewhat, i.e.
it does not have two or more vicinal hydroxy groups.
5.3 Furth�r Separ�tion with Preparative TLC
On primary TLC of honey extract two antibacterial bands
appeared to migrate just behind the solvent front and were
visualised as dark absorbing bands under ultra-violet
illumination. Two lighter absorbing antibacterial bands
were immediately behind each one of these (Figure 9).
Further separation of the active fractions was necessary in
order to see if the lighter bands were different compounds
or tailings of the darker bands. Sharper separation of all
active bands was desirable before qualitative analysis, to
avoid overlap of fractions. Secondary development on a
chromatogram would also give an indication of how pure these
bands were.
A 400 mm preparative plate was loaded with the alcohol
extract of 40 ml of honey and developed in Solvent I. It
was air-dried for two days, the u.v. patterns recorded and
a 50 mm vertical strip was divided into th±rty 5 mm strips,
cross�wise. These were removed and assayed for anti­
bacterial activity and the results are shown in Figure 10.
Under ultra-violet,illumination of 254 pm wavelength
 83

visible light 254 _ mm 350 mm bioassay

(Brown
represents a
mixture of
purple and
----------- -------�---
green)

./'II"./ 28
././,/
Fraction 2� ././

X
X
X
X 21
X 20
X 19
X 18

X 16
X 15
X 1
13
X 12

X 8
y
X 6
X 5
X
application lin X
------ J, ____ ----------- -----�---2
X 1

J relative inhibition of S.aureus growth

x no inhibition

FIGURE 10: Preliminary TLC of honey extract with Solvent I

viewed under u.v. illumination and assayed for
antibacterial activity.
 84

the remaining active area (labelled Fractions 1 and 2 in

Figure 10) was scraped from the plate. It was added to
30 ml absolute ethanol and ground up in a pestle and mortar
to allow all compounds adsorbed.to the silica to re-dissolve
into the alcohol. The alcohol and sludge were centrifuged
at 8000 r.p.m. for 15 minutes and the supernatant reduced
to 14 ml by rotary evaporation. Seven 50 mm preparative
TLC plates were loaded with 2 ml each of the concentrated
supernatant and developed in one of the following solvents:
chloroform
n-hexane-ethylacetate (95:5)
n-hexane-ethylacetate (72:29)
ethyl acetate - petroleum ether (3:1)
10% acetic acid in chloroform
diethyl ether - petroleum ether (1:1)
toluene-cploroform-acetone (40:25:35)
Each chromatogram was then air-dried for two days and the
band patterns viewed under u.v. illumination (254 a� and
350 �m wavelength) and recorded. They were then sub­
divided into 5 mm strips across their width and assayed
for activity.
All plates which had been developed in solvents
containing ethyl acetate or acetic acid indicated varying
degrees of inhibition in all strips assayed from them.
It was concluded that these plates still contained acetic
acid bound to the silica (although it was not detectable
by smell) and that these would not be convenient solvents
to use for secondary development.
The chromatograms which had been developed in the
other solvent mixtures demonstrated no bands of bahterial
inhibition, although under u.v. illumination they appeared
 85

to have bands of varying absorbance and fluorescence. It

was considered that either the solvents used were reacting
with the antibacterial compounds and rendering them inactive
or the activity was not being removed from the silica of
the first plate.
In order to test this, 5 ml of honey extract was
applied to a 100 mm preparative plate and chromatographed
in Solvent I. After drying, one half of the plate was
assayed for antibacterial activity as a positive control.
The active bands were scraped from the second half and
ground up in 10 ml absolute ethanol with a pestle and mortar.
After centrifuging, the supernatant was concentrated and
applied to a 50 mm TLC plate and developed again in Solvent
I. The plate was then air-dried, viewed under u.v.
illumination and the bands seen assayed for activity. The
results are shown in Figure 11.
It can be seen that the loss of activity was not due
to the use of a s�cond solvent mixture. It appeared from
the band patterns, seen under u.v. illumination, that the
compounds were r�dissolving into the alcohol and being
applied to the second chromatogram, although they did not
show up as strongly and their migration patterns seemed to
altered slightly. It was concluded that grinding up the
active compounds with silica in alcohol was rendering them
inactive, or that the active component(s) was not being
eluted.
A method of removing the active fractions nrom the
silica of the primary TLC plate by elution was investigated.
Alcohol extract of 40 ml of honey was ap�lied to�a 400 mm
preparative plate and c�romatographed with Solvent I.
 86
visible lie:ht 254 rim 35 0 rim bioassay

X 10
X

FIGURE 11: Re-chromatograph of honey extract

with Solvent I.
 87

When dry the two fractions of antibacterial activity

(Figure 10) were scraped from the plate separately.
They were put in separate glass columns with packed
cottonwool in the bottom and eluted with twice their
volume of Solvent I. The eluted extracts were then
concentrated to 20 ml. One millilitre of each was rotary
evaporated to dryness and 200pl distilled water added and
shaken vigorously. The brown coating inside each flask
was only partially soluble in water but the· aqueous
solutions of fractions l and 2 were both assayed for anti­
bacterial activity.
Fraction 1 produced a 1.5 mm diameter zone of inhibition
and Fraction 2 4 mm. This indicated that the elution
method of removing active fractions from a TLC plate was
successful and activity was being recovered. The anti­
bacterial activity of the two fractions appeared to be
relatively less than the estimation of what it should have
been. This, and the_ laGk of sol�bility of�the eluted
material, indicated that activity may have been trapped in
water-insoluble material. It also indicated that the two
fractions were still very impure and needed further
separation. This was confirmed when pooled samples of
Fraction 2 were analysed by nuclear magnetic resonance.
Many small peaks were obtained, some possibly indicating
aromatic substances, and a very large peak indicating
aliphatic substance.
A secondary solvent system was investigated for
separating the compounds in Fractions 1 and 2, and to give
an indication of how many compounds were in the mixtures.
A solvent mixture commoµly used for chromatography of
 88

phenols was used. A 50.fl sample of Fraction 2, dissolved

in deuterated chloroform,·was spot-applied to a commercial
aluminium foil-backed silica plate. The plate was
developed in toluene:diox,an:acetic acid (180:50:8) in a
sealed jar, dried, sprayed with aqueous 10% sulphuric acid
and charred at 100 ° c for 30 minutes. When viewed under
ultra-violet illuminati.on of 254 fil� wavelength the
chromatogram appeared to have approximately eight separated
compounds, with a large spot of hydrocarbon material
(identified by colour and position) just behind the solvent
front.
Because of the difficulty, found previously, in
removing acetic acid from thin-layer chromatograms a solvent
mixture of similar elutive power, but containing no acetic
acid was tested. This solvent contained toluene:chloroform:
acetone (40:25:35) and produced a resolution at least as
good. The appearance of both stained and charred
chromatograms is represented in Figure 12. This solvent
mixture was subsequently used for all secondary chromato­
graphy and designated Solvent II.
5.4 Locating Antibact�rial Cdmpounds after Secondary
Chromatography
In order to find the position of antibacterial bands
after secondary chromatography two 400 mm preparative
plates were cleared in Solvent I and re-activated. One
was loaded with alcohol extract cf 40 ml or honey and
developed in Solvent I. A 300 mm width of the active
area (Fractions 1 and 2, together) was then removed and
the material eluted and applied to 300 mm of the second
plate. This was developed in Solvent II. The bands
were then located by viewing under u.v. illumination and
 89

Solvent: tol uene:dioxff�:acetic Solvent: toluene:chloro ­

acid (180:50:8) (40:25:35) form:ace t on�

front
-----. -------�
solvent

\�------- ---- --
0
Q
0
c::;:::;
0
0 c:,
CD
C>
0

0 0

I,)

i- - - --- - -- - - - �applica t ion--)- ----- - - - -- -­

l ine

FIGURE 12: Appearanc e of TLC plate s loaded with

Fraction 2, deve lop e d with the sol ve nt s

s how n and s tained with aqueous 10%

su l phuric acid.
 90

a 50 x 5 mm strip of each one removed, eluted in Solvent II,

dried and assayed (in 200pl water) for antibacterial
activity. A 50 mm wide strip of the active area from the
first plate was eluted in Solvent I and assayed to see if
any activity was lost during secondary development (any
insoluble material was also scraped int� the agar well).
The �ctive area from the first plate produced a 4-5 mm
diameter zone of inhibition. The results for the second
plate are given in Figure 13.
Zones of inhibition cannot be added to give total
inhibition but it was clear from the assay of the second
plate that either activity was being lost between plates or
the compounds work synergistically. Therefore elution and
assaying was carried out with 50 mm strips of ·bands
designated 6, 8, 9 and 11 (Figure 13) from the second plate,
a pooled mixture of 6, 8, 9 and 11, a 50 mm wide strip from
the total length of the second plate (i.e. including all
separated compounds), and a 50 mm wide strip of active area
from the first plate for comparison. The results are
given in Table 19.
It was concluded that the loss of activity was not due
to a synergistic requirement of the active compounds and
that activity was being lost after application to the second
plate. The results also indicated that not all activity
was 1 ocated in bands 6, 8 , 9 and 11 . The :f efore assaying
of bands in total, rather than 5 mm strips from in them,
was carried out and the area from which strip no. 12 had
been taken was found to contain slight antibacterial
activity.
Antibacterial activtty after secondary development was
 'j .L

Diameter of zone

--r
of inhibition
visible 254 nm 350 nm
solvent front t brown =
)
urple
----- __ _4..._ - '�il.ica e dge; +_g reen ____ -----------

1 8

. -;, X

-- - - - _x__l4.
----t- - - --
a blicati on lin

FIGURE 13: Secondary chromatography of honey extract

in Solvent II. Viewed under u.v. illumination
and assayed for antibacterial activity.
 92

TABLE 11: Inhibition of growth of S.aureus by various

eluted chromatographic fractions.

Test Fraction Diameter of Zone

of Inhibition mm)

Active bands of first run (with 4-5

Solvent I) (eluted with Solvent I)
50 mm-wide strip of total length of 2
re-run active bands (eluted in
Solvent II)
The four active bands of re-run 1
(pooled, eluted in Solvent II)
Band 6 from re-run (eluted in ..( 0. 5
Solvent II)
Band 8 from re-run (eluted in 1
Solvent II)
Band 9 from re-run (eluted in 1.5
Solvent II)
Band 11 from re-run (eluted in 1
Solvent II)
 93

not being totally recovered, either because of chemical

instability -on exposure to oxygen and the secondard solvent,
or inefficient removal from silica by Solvent II during
elution. To test these possibilities ·the following
solutions from 50 mm wide strips of preparative plates were
assayed for activity:
(1) active area of first run (with Solvent I) eluted with
Solvent I.
(2) active area of first run, (with Solvent I) eluted with
Solvent I, dried by rotary evaporation, 20 ma Solvent
II added, poured into 90 mm diameter petrL dish and
left uncovered in fume cupboard for 2 hours, dried and
assayed.
(3) total length of a second run (with Solvent II) plate,
eluted in Solvent I.
(4) the four most active bands of the secondary plate, i.e.
6, 8, 9 and 11 (Figure 13), eluted with Solvent I.
The diameters of zones of inhibition of S.aureus, for these
test solutions are given in Table 20.
The results obtained (Tables 19 and 20) showed that
loss of activity was not due to inefficient elution by
Solvent II but was due to inactivation by exposure to oxygen
and Solvent II.
In order to find out which component of Solvent II was
oxidising the antibacterial activity a 250 mm width of the
active area of a first run (with Solvent I) was eluted with
Solvent I and divided evenly between five 50 ml flasks.
The solutions were then dried by rotary evaporation and
treated by one of the following methods:
 94

TABLE .20: Inhibition of S.aureus by antibacterial

fractions treated in various ways.

Test Solution Diameter of Zone

of Inhibition (mm)

Pla.te 1 acttvity 3
Plate 1 activity - Solvent II �0.5
treated
Plate 2 in total - eluted in 0.5-1
Solvent I
Plate 2 - 4 major active bands, 1
eluted in Solvent I
 95

Fla.sk No. Treatment

(1) control 200}" 1 distilled water added, shaken
vigorously at 30 minute intervals over
3 hours and the water-soluble fraction
(plus insoluble material) assayed for
activity.
(2) 15 ml toluene added, shaken, left one hour;
dried and treated as flask (1).
(3) 15 ml chloroform added, shaken, left one
hour; dried and treated as flask (1).

( 4) 15 ml acetone added, shaken, left one

hour; dried and treated as flask (1).
( 5) 15 ml Solvent II added, shaken, left one
hour; dried and treated as flask (1).
The results of the bioassay are shown in Table 2il.
Acetone was found to be the solv9nt component which was
reducing the antibacterial activity. Exposure to oxygen
was also found to be an important factor: when treated
with Solvent II and exposed to a small amount of oxygen
there was no inactivation of the antibacterial activity.
Bulk acetone had been used in Solvent II up to this
point and it was thought likely that iron from the metal
storage drums could have been catalysing the oxidation of
the active compounds. In order to test this, the effect
of reagent grade acetone on activity from primary TLC was
compared with that of bulk acetone. A 200 mm width of
primary plate activity was eluted with Solvent I and
divided equally between four 50 ml flasks •. The solutions
were dried by rotary evaporation and the residues treated
in one of the following ways:
 96

TABLE .2r: In.hibition of S. aureus by active fraction

from primary TLC, treated with various
solvents.

Solvent treatment Dia�eter cif Inhibition Zone (mm)

control 5
toluene 5
chloroform 6
acetone �l
Solvent II 5
 97

Flask No. Trs�tmsnt

(1) - control - 200 _µ 1 distilled water added, shaken at
30 minute intervals over 3 hours and the
water-soluble fraction assayed for activity.
(2) 15 ml Solvent II (containing reagent grade
acetone) added, poured into a 90 mm petri
dish and left one ho�r, dried and treated
as flask (1).
(3) 15 ml reagent grade acetone added, dried
after one hour and treated as flask (1).
(4) 15 ml bulk acetone added, dried after one
hour and treated as flask (1).
The results of the bioassay are given in Table 22: It
was concluded that loss of activity was due to the use of
bulk acetone in Solvent II, used for developing the
secondary chromatogram. It was also concluded that it
was not necessary to exclude oxygen from the system when
reagent grade acetone was used.
Location of antibacterial bands was attempted again
but with a chromatogram developed in Solvent II containing
reagent grade acetone. The results are shown in Figure
14. (The full extent of the bands in the direction of
migration was taken for assay). This secondary
chromatographic development produced three areas containing
antibacterial activity. The first, marked A, occurred in
the upper third of the plate and had two bands of slight
activity. The second occurred just above the middle of
the plate (B) and contained three antibacterial bands which
appeared under u.v. illumination of 350 nm wavelength as
a purple band of strong absorbance, a band of medium
absorbance and a band of green fluorescence. The medium
 98

TABLE iz: Inhibition of S.�ureus by activity from

primary TLC subsequently treated with bulk
acetone or reagent grade acetone.

Test Solution Diamet�� of Inhibition Zone mm)

Control 6
Solvent II (with reagent 6
grade acetone)
Reagent grade acetone 6
Bulk acetone 1
 99
visual 254 rnm 3 5 0 nJ!l

X 1
X 2

r------:-�-����-i-��-t----11
X
0.5 4
t---------t,....;-----:.:.......;.;..__;;;.;:.::....;_��.::.;.;;,:,;,�----------'A
�----t;;-����..;;;.___---+-------:---, 0.5
X
5
6
f
X 7
0.7 8
(
1------t����-:--r�����--.::...:._::.__..2_�
1.0 9 B
�����-:;-:--��-+--��-If
1.0 10

5 0.5 11
6 0.5 12

7 C

1.0 13

-- - - .o.� -.l4- --1

1'
Active Barid No

FIGURE 14: Secondary TLC of honey extract in Solvent II.

Viewed under u.v. illumination and assayed for
antibacterial activity.
 100

purple band and the green band appeared to overlap slightly

and the division between the two purple bands was indistinct.
It was decided therefore that the bands in area B required
further separation before thei� analysis was attempted.
The third area, C, occupied the lower half of the
plate. This antibacterial area was m�inly colourless,
with some absorbing material correiponding to activity on
the origin and a band (No. 6) which was antibacterial and
appeared only on some secondary plates, and showed slight
absorbance under u.v. illumination of 254.�m. The
a�iivity of the colourless area cotild have been due to a
honey co mponent, or components present without u.v.
absorbance or fluorescence, or an antibacterial residue
from the silica gel or solvent which is not concentrated
enough to show up in other assayed bands (with which
shorter lengths of the plate were taken). In order to
test this the following 60 x 50 mm preparative chromatogram
strips were scraped and eluted with Solvent I and arranged
for activity:
(1) A blank strip run with Solvent I and reactivated, to
check if activity was due to something in the silica.
(2) A blank strip run with Solvent I, reactivated and
developed in Solvent II (no extract also), to see if
inhibition was due to a residue from Solvent II.
Neither of these inhibited the growth of S.aureus and it
was concluded that antibacterial material from the honey
extract was causing inhibition from the colourless area.
The next step was to see if this activity could be located
in any particular position in the colourless area.
A 150 mm secondary plate was developed and dried.

This time BO coloured band corresponding to No. 6 of

 101

Figure 14 was present. A 100 mm wide strip of the plate

was used instead of 50 mm in order that·any activity might
be more easily detected. The length of the colourless
area between the application line and oran�e band
(unavoidably transferred from the primary plate) was 85 mm.
The area in the 100 mm width correspond�ng to region C was
divided into three equal lengths of.approximately 28 mm each.
The remaining 50 mm width of the TLC plate was divided into
eight lengths of 10 mm and one of 5 mm which was removed
from just below region B. All fractions were ind�vidually
eluted in Solvent I, dried, ?OO,,.ul water added and assayed
for antibacterial activity. The results are given in
Table 23,
It appeared that the colourless area may have contained
some antibacterial material which was dissolving slowly off
the origin and was not concentrated enough to be detected
in sections of 10 mm. The material may have reen carried
slowly � the solvent, only migrating to higher areas when
lower ones were saturated with it, in which case the blank
area would be chemically homogenous, or it was being oxidised
as it was migrating and therefore would not be chemically
homogeneous. It is possible that seyeral unrelated
compounds (not visible under u.v. illumination), were
present, the most active being coincidentally closest to
the origin but this is less likely. It was not possible
to find the correct explanation until various sections of
the colourless area had been chemically analysed.
5.5 Tertiary Chrbmatography of Sedtiort B
Better separation of the three antibacterially active
bands of section D was necessary before characterisation
of the bands could be attempted. It would also serve to
 102

TABLE 23: Antibacterial activi ty of sections of area C

from secondary developmen t of honey extract.

Section from 50 mm wid th Diame ter of Zone of Inhibi tion

of pla te (mm)

1 - top 5 mm of area C 0
(next to B)
2 - next 10 mm 0
3 - next 10 mm 0
4 - next 10 mm 0
5 - next 10 mm 0
6 - next 10 mm 0.15
7 - next 10 mm 0.3
8 - next 10 mm 0.3
9 - next 10 mm 0.5 - 1.0

Section from 100 mm

wid th of plate

1 - top 28 mm of area C 1.3

2 - middle 28 mm of area C 1.3
3 - bo t tom 28 mm of area C 1.5
 103

indicate whether the medium purple band was a tailing of

the dark purple band (as it was always directly behind it
with no intervening space) or a different compound
altogether. An indication or how much �ctivity was due
to overlap with area C compounds might also be given.
A 100 mm-wide strip of area_ B was _eluted in Solvent I
and then applied to a cleared 50 mm preparative plate and
developed. in a different toluene:chloroform;acetone
mixture (40:20:40), designated Solvent III. Bands were
viewed und�r u.v. illumination, and eluted and assayed
against S.aureus. The results are shown in Figure 15.
They indicated that the band of medium absorbance is a
different compound to that contained in the band of dark
absorbance, which is only very slightly active. Some of
the inhibition previously attributed to these two bands
was concluded to be due to overlap with activity in area C.
This showed up as a light coloured mixture of purple and
green under u.v. illumination of 350 mm wavelength on the
tertiary TLC plate, as did most of area-C on the secondary
plate. Both the absorbing bands (3a and 3d in Figure 15)
must possess some activity of their own, other than that
due to the possible overlap mentioned, as they both
extended about 10 mm, which was too little to display
activity when compared with strips of 10 mm extent from
region C (Table 23).
5.6 Ass�ssment of Recdvery df Activity from Primary TLC
Recovery of activity from the second TLC plate was
estimated in order to see if all antibacterial activity
had been located.
A 50 mm wide strip from a primary plate was eluted
and assayed and the diameter of the zone of inhibition
 104
visible Bioassay (Inhibition

--- --- -- - \<-.solvent

not assayed front

0.7
:::. --·- -
- "l- - .....,;;:- ... . b 1.3

not assayed

L. 0. 3

3 ( C) 1.0

-----------� �rigin

FIGURE 15: Chromatography wi t h toluene:chloroform:

acetone (40:20:40) of ar ea B from secondary
TLC, i.e. t erti ary chromatography� Viewed
under 350 mm wav elength u.v. illumination
and assayed for antibact erial activity.
 105

compared with that of the pooled areas A, Band C eluted

from a 50 mm secondary plate. Bot� plates �ere.dev�loped
with the same set of extracts (i.e. half of material
eluted from primary plate �as used to load secondary plate)�
The results showed that the pooled fractions �ere nearly
O. 5 mm less active. This was conclude� to be within
experimental error after development on two chromatograms
and two elutions.
5.7 Summary
Separation of honey extract fractions required three
chromatographic developments befoie characterisation of the
compounds was attempted. Primary development was with
60% ethanol-40% ether (Solvent I). All activity was
localised just behind the solvent front. It was eluted
from the silica in columns with the same solvent and
developed on a second chromatogram with toluene:chloroform:
acetone (40:25:35, Solvent II). The areas of activity
were separated, one of which required further separation
(toluene:chloroform:acetone, 40:20:40; Solvent III).
Reagent grade solvents were used as oxidation of the
antibacterial activity occurred with bulk acetone.
 106

CHAPTER SIX:

CHARACTERISATION OF ANTIBACTERIAL

COMPOUNDS

Characterisation of the various antibacterial

substances was attempted by obtaining structural information
from ultra-violet and infra-red analysis, nuclear magnetic
resonance and mass spectroscopy.
Alcohol extracts of 160 ml of honey were obtained and
chromatographed in Solvent I. Secondary chromatography
was carried out in Solvent II and the active bands, labelled
1, 2, 3, 5, 6, 7 and 8 in Figure 14, were removed from the
TLC plates separately. Bands labelled 3 were removed
together and further chromatographed in Solvent III and the
bands, labelled as in Figure 15, were then removed. A
sample (3 e) was also removed from the tertiary TLC plate
which had appeared as small aligned spots of medium
absorbance under u.v. illumination directly behind 3 d.
This may have �en tailing of 3 d due to the humid
conditions during chromatography. Humid conditions can
Of+en gi·ve -..0- neno,..1. s � di·,... .1. or .1. -d 1 �+=ea1 --e�
I.I --
1
...._ ,:i L, L, � / ;:, ..; 4.. .n... U. �.;,.......,.-:i + i,..,'Y'I
W..J... b .!. U. '.J - '...I.!..!.
-na++ ,_.n
l-'
0
v '-' ,.., .J. - e

Each sample was removed from the silica by elution with

Solvent I, which was subsequently removed from the samples
by rotary evaporation. Then, to help obtain pure samples,
1 ml distilled water was added to each and shaken vigorously
at intervals of 30 minut�s over a three. hour period. The
 107

solutions of the fractions were then filtered with Whatman

No. 1 filter paper and the filtrate made up to 3 ml with
water.

The ultraviolet absorption spectrum of samples was

measured with a Pye Unicam SPB-500 spectrophotometer with
quartz sample cells and distilled water as a reference.
The scan range was 200 nm-500 nm. The results are shown in
Figures 16 and 17 and a summary of the results is given in
Table 24.
Carbonyl groups show absorbance in the range.275 nm-
295 nm (Dyer, 1965). Assuming a 10 nm margin of error,
samples 1, 3 a, 3 c, 3 d, 3 e and 8 showed absorbance
within this range. Samples 5 and 7 required better
resolution.
6.2 Infrared Spectroscopy
The samples used for obtaining the u.v. spectra were
freeze-dried to remove water prior to infrared analysis.
Samples were ground with potassium bromide and formed
into discs under 154 x 103 kP pressure under vacuum for
1 minute. Spectroscopy was carried out with a Shimadzu
IR-27G spectrophotometer in the range 4000 nm-400 nm.
Most of the fractions produced infrared spectra with
only broad absorption peaks which had little diagnostic
value and indicated heterogeneity in the samples. Band 3 d,
however, displayed relatively sharp absorption peaks at
1695, 1085 and 1100 cm-1• This indicated the presence of
a conjugated acid, carbonyl and -c-o- groups respectively.
A broad absorption peak at approximately 3400 cm-1 also
indicated the presence of one or more hydroxyl groups.
It was therefore decided'to carry out further analysis on
band 3 d.
108
Figure 16 and 17: Ultra violet absorption spectra
of antibacterial compounds from
manuka honey.
 tt 109

.,��
l
2
,: 3 a
,:
,:.
:
I:
\.
\:.\:
'\
\:
,:
\�
\\
"....---- ... ,
\�

\
\
\
\
\
\
\
\
\

''
\
\

''
······ ",.......
��····· .
'� ... ...... ····· ........... .... . .....
nm-
. ..... ..
200 250 300 350 400 450 500

''
'
l
\
3 b
\
\
3 C
\
\
\
3 d
\
\
\
\ :
\
\
\
\
\
\

..
\
\

.
\
\
\
\
\
\
\
\
\

-- ...'
\
\

\... _,,,_,,
\

nm
• :""': ..... -- ••••• '" ••••••• flll'I'..-... � --.

200 250 300 350 450 500

Figure 16
 ' \
110

''
\
I
J e
''\

--.-.-- 4
''
' \
.......... 5
\
I

'
\
\

\
\
\
\
\
\
\
\

--,,
\\

', ....... , ___ -,

.....
···.... · \

· ····· .. \
\
·· . . . . ... \

. .. ·····
\
\

nm ...... . .
•

200 250 300 350 400 450 500

\
'
\
\

''
I
\

' 6
\
\ 7
\

'
\
l
8
\
' 9
'·
\
\
\
\
\ \\
\ \
\ \
\ \
\
\ \

\ \\ . . ./"\
\ \
·t..)._ ./ '
\ \ / -.,: \
\ .... \.I' •••••• \

' ..... ..,'....··..

. . �, ....4\

,,_ .........
', ·,.,....... . __ ..... ---..,.
'•,

............... · ·· . · .
........ ,, .-... ···· · ··· · '
. . ·-.......
·�·----�. ' , · ..= : -,..�. .
.... ,, ···r-P-1...-.:,·.:-:-::-:-":.�
·.-:-:":�:-:-

nm
200 250 300 350 400 450 500

Figure 17
 111

TABLE 14: Peaks/plateaux of ultraviolet absorbance of

antibacterial substances from manuka honey.

Wavelength (nm)
Sam le no. Peak s Plateau x) Notes

1 243,289
2 256
.3 a 268
249
3 C 271
3 d 272.5
)
3 e 270
5 323 Possib1y three Jila-tB'aux-�
on further dilution
6 239,258,311
7 Possibly three plateaux
on further dilution
8 265
 112

The infrared spectrum of sample 3 d is shown in

Figure 18-.
6.3 Nuclear Ma�netic Resonance (n.m.r)
In order to obtain further information on the
structure of 3 d, proton n.m.r analysis was carried out
using a Jeol FX 60 Q instrument.
A sample of 3 d was dissolved in deuterated
chloroform and its proton n.m.r spectrum measured.
Absorption signals were present at 5 7.32, 3.94 and 3.89
ppm in the approximate ratios of 2:6:3. From the
combination of I.R. and n.m.r data it was therefore
concluded that 3 d was possibly a tri-substituted aromatic
compound of the type:

(x)

b 3�89 is consistent with the possibility of a COOCH

3
group here but does not exclude other possibilities.
The ratios of peak intensity of the above chemical
shifts indicated one aromatic hydrogen environment and
one methoxyl group environment, which further suggested
that the molecule possessed a plane of symmetry.
The proton n.m.r spectrum of fraction 3 d is given in
Figure 19.
 Figure 18: IR spectrum of fraction 3 d

10

20

30

40

50

60

70

80

90
1-1
1-1
\J.)

���--.--�r----.--�,�--.------,.----,,----,,----,,--�r----,-�����.---�r--�.-�-y----�..,---·�..-�.-----;-\100
1
400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2400 2800 3200 3600 4000
(cm-1)
 Figu��· 19: Proton n.m.r. spectrum of fraction 3 d

0...,
;I
'

" .s
,,,.
�
�
.. .t
\,) "'i

","'­ :-
-i ;t ,;
�

I--'
���� ��}II�� �V'�� I--'
.p-..
 115

6�4 M�Ss Sp�ctr�m�t�y 0£ Fradtion 3 d

Mass spectral analysis of 3 d was carried out on a
Varian Mat CH 5 mass spectrometer. This revealed M+ to
have m/e (mass to charge ratic·) 212 and that this peak had
M + 1 and M + 2 isotope peaki corresponding to a formula
This data also showed peaks indicating loss

of methyl and methoxy groups from the molecular ion.

It the compound was fully saturated a formula of
c 10H22 would be expected.

found indicated a deficit of ten hydrogens and therefore

the presence of five double b�nds and/or ring(s). This
is consistent with the aromatic - conjugated ester
suggested by the I.R. and n.m.r gnalyses.

Hence the remaining substituent (x) of the aromatic

compound, 3 d, was concluded to be an -OH group (Table ?5).
Proof of this was obtained by methylation and further mass
spectrometry. Fraction 3 d was reacted with diazomethane
at room temperature, in a sealed ��ssel (Sch1enk and
,Gellerman, 1960, modified by P. Holland, personal
communication). This gave a permethyl compound of m/e
226, indicating the replacement of a single -OH group by a
-OCH group.
3
 116

TABLE 25: Deduction of substituent X, of the aromatic

compound 3 d, utilising mass spectral analysis.

Chemical Group Evidertce Molecul�r Weight

proton n.m.r signals, 72

u.v, I.R and mass
(skeleton) spectral data

I.R. and proton n.m.r 62

(COO-) I.R. and proton n.m.r 44

proton n.m.r 15
189

3 d m/e = 212

molecular weight of Xis 13

and, molecular weight of -OH is 13
 117

The foregoing combination of proton n.m.r and mass

spectral d�ta were consistent with four possible
structural formulae. They were:

OCH
3
HO �-C�
O

. � "'-OCH3
OCH3
(I) (II)

CH30

HO -Q-c��CH
3
CH 0
3
(III) (IV)
(syringic acid methyl ester) (trimethoxy benzoic acid)

In order to reduce the number cif possible formulae

hydrolysis of 3 d was attempted by heating it at 130 0 C,
overnight in methanol with 1 mol/1 NaOH, in a sealed vessel.
The solution was then acidified with H2 so 4, extracted into

diethyl ether and analysed. The mass spectrum indicated

the fdrmation of a hydrolysed product of m/e 198 (M + ).
This was consistent with the conversion:

�o
C - C,, - OH
- --..OCH
3 0

arid hence structures (II� and (IV) could be eliminated.

 118

Methyl groups in the other positions around the aromatic

ring would not have been susceptible to hydrolysis owing
to the effect of -,-{-bonding orbitals around them.
Prior to the hydrolysis at 130 ° c, this conversion
was attempted by the same method, but at room temperature.
No hydrolysis was observed. This is aonsistent with a
conjugated "deactivated" ester, as suggested by the I.R.
stretching frequency (1695 cm -1), also due to the effect
of 11"" -bonding orbitals around the aromatic ring. By
contrast unconjugated methyl esters typi�ally absorb at
V max 1720 cm-1• (I) and (III) are conjugated "deactivated"
ester str�ctures (functinnal group - COOCH ).
3
The possible products of hydrolysis were therefore:

CH o �
HO
3

CH 0
0COOH or, HO o:�OH

3 OCH
3
(3,5-dimethoxy-4-hydroxy (2,6-dimethoxy-4-hydroxy
benz oic acid) benzoic acid)

Compound (III) is a methyl ester of syringic acid, a

well known plant metabolite (and constituent of lignini).

COOH syringic acid

Therefore (III) was, further investigated and

·Subsequently estahlished as the structure of 3 d.
 119

A sample of (IV) (3,4,5-trimethoxy benzoic acid) was

available in the laboratory. It was methylated with
diazomethane,

methylation
OH���������
diazometha,ne

(IV) (V)

and analysed by mass spectrometry. The mass spectrum

obtained w�s identical with that produced by the permethyl-
ation of 3 d. Since (IV) was not considered a possible
structure for 3 d because of the evidence from hydrolysis
and mass spectrometry and insolubility in chloroform, but
produced the same methylation product, then compound (III)
(methyl 4-hydroxy - 3,5-dimethoxy-benzoate or syringic acid
methyl ester) was concluded to be the structure of 3 d
(i.e. both .(III) and (IV) had methyl groups substituted at
positions 3 and 5 on the aromatic ring).
6.'5 Mass Spectrometry of Other Active Fractions
Mass spectrometry was carried out on the other
isolated active fractions from manuka honey in order to see
if any additional evidence could be obtained on their
structures.
Fraction 1 was found to have an m/e of 226 and was
hence concluded to be methyl 3,4,5-trimethoxy-benzoate
(a-methyl syringic acid methyl ester), which had the same
structure as (III) or (IV) after methylation i.e:

C - OCH3
\\
0
 120

Fractions 2 and 3 a contained mostly a compound of

m/e 212 which was· concluded to be the same as 3 d (methyl
4-hydroxy - 3,5-dimethoxybenzoate).
Fraction 3 b contained mainly a compound of m/e 226,
concluded to be methyl 3,4,5-trimethoxybenzoate.
Fractions 3 c and 3 e were also shown to 6on��in
mainly methyl 4-hydroxy - 3,5-dimethoxy benzoate.
Fractions 6, 7 and 8 (from the application line) also
showed mainly m/e 212. Owing to their low Rf values tp.ey

were thought to be the free benzoic acid of methyl

3,4,5-trimethoxyb�nz�ate (i.e. 3,4,5-trimethoxybenzoic atid),
rather than methyl 4-hydroxy - 3,5-dimethoxybenzoate.
Elution is a displacement process and firmness of phenol
binding depends on the number and position of hydroxyl
groups. The possible structure of 6,7 and 8 woufrd
therefore be:

CH3 0
CH
3
o-9)- COOH
CH 0 3,4,5-trimethoxybenzoic acid
3

Fraction 5 was very impure and gave peaks of molecular

weight 212, 213, 206, 207, 181 and 182. Signals 206 and
possibly 207 would have been due to the presence of
fraction 4 (Figure 14) which was also analysed by mass
spectrometry, although not itself antibacterial.
Fraction 4 was concluded to be an involatile compound
giving no obvious molecuiar ion and a strong peak at m/e
 121

206. Fraction 5 was concluded to be mainly either

3,4,5-trimethoxybenzoic acid or methyl 4-hydroxy -
3,5-dimethoxybenzoate.
6.6 Conclusions
The major antibacterial components of manuka honey,
separated by thin-layer chromatography,- were concluded to
be:

CH o
3

r
methyl 4-hydroxy -
HoO-coocH 3,5-dimethoxybenzoate
3
CH
3 (syringic acid methyl ester)
(molecular weight 212)

and,

CH 0
3
CHJ� COOCHJ methyl 3,4,5-trimethoxy­

CH�O benzoate (0-methyl syringic

(molecular weight 226) acid methyl ester)

Fractions 1 and 3 b were mainly methyl 3,4,5-trimethoxy­

benzoate.
Either methyl 4-hydroxy - 3,5-dimethoxybenzoate was
dissolving slowly off the application line, extending up
the TLC plate to include fractions 2, 3 a, 3 d, 3 c and 3 e
and the lower fractions 5, 6, 7 and 8, or it extended across
the upper fractions only and the fractions with lower Rf

(5, 6, 7 and 8) were 3,4,5-trimethoxybenzoic acid which

would bind more firmly to the adsorbant.

3,4,5-trimethoxybenzoic
acid
 122

The impression from the band patterns (seen under

ultraviolet illumination) that there were possibly eleven
different fractions may have been due to the presence of
the impurities. There were actually two or three only.
Table 2� shows the probable structural groupings and
evidence for these.
TABLE 2£: Probable structural groups of antibacterial .fractions from manuka honey and evidence for
· these groupings.

Sample J U.V. absorbancel I.R. absorbance spectrum I M.S. data Proposed structure
No. spectrum and name

A I
1 I Plateau at
289 nm
High peak at 1100 cm-1 '
relatively low peak at
1400 cm-1, few peaks
I Molecular weight 226,
fairly pure
CH3n
CH 0 · ,g CXXlCH3
3
- ... - - - _, - - - - •.. - - - � ·-
- below -
- - 800 -.
-1
- cm- - - - - - -+ - - - - - - - - .. .. - I•
CH30
methyl 3,4,5-trimethoxy­
3 b Plateau at "Noisier" peak pattern Molecular weight 226, benzoate
250 nm than sample 1 - with some 212 (0-methyl syringic
probably due to presence acid methyl ester)
of some material of
212 m.wt.
B 3 a Peak or plateau Peaks at 1100 cm --.l and Molecular weight 212, CH3
3 d within range -1 fairly pure
�(Y'\

} c 268-273 nm 1400-cm. of relatively H O � vVJCf-13

equal height. Fe�3
3 e
- - - - - - - - :1- - • - - -
peaks below 800 cm CH 0
�thyl 4�hydroxy-
2 I Plateau at Peak at 1100 cm high Molecular weight 212, 3, 5-dimethoxybenz oate
256 nm relative to peak at with some 226 (syringic icid methyl
1400 cm-1 probably due ester)
to presence of material 1
of m.wt 226.
C I Three plateaux Peaks at 1100 cm -.L and Molecular weight 212, Mix�ure of group Band D
6
1400 cm-1 of relatively fairly pure
equal height.
o-1 Three plateaux High peak at 1100 cm -.1, Molecular weight 212,
relatively low peak at fairly pure
7 1 4 00 cm-1 • "Noisy peak 3,4,5-trimethoxy-benzoic
8
pattern below 800 cm -1 • acid CH 3_0 n j-J

-0r 0H
CH 01�- l\)

��
\..,J

V
o
"j
 124

CHAPTER SEVEN;

DISCUSSION

Antibacterial properties have been found in a diverse

range of natural materials, many of which have been used in
medicine for centuries. Synthetic substances or the
purified active components of natural material are preferred
because they are more concentrated and it is easier to
measure an accurate dose. However, it must first be found
which compound(s) in a natural antibacterial material is
responsible for the antibacterial activity and find its
structure and mode of action before synthetic imitations
or related compounds can be manufactured. Many of the
antibiotics used today are relatively short-lived due to
their misuse and over-use and there has been much inte�est
shown in recent years in returning to the use of substances
derived from natural sources.
Hive products, such as propolis and honey, ha iB been
utilised by bees for 10�20 million years and are still
effective for their needs. The pharmacologically active
compounds present in these substances can only have altered
in as much as the bee itself or plant from which the raw
materials were obtained has evolved. Much conclusive
research has been carried out on the active components of
propolis and bee venom but there are many contradictions
in the literature resulting from previous honey research.
In this work an attempt was made to isolate and
characterise the compounds which gave rnanuka honey its
apparent non-peroxide antibacterial activity.
All non-peroxide a�tibacterial activity in rnanuka
 125

honey could be extracted into absolute alcohol. A 100%

v/v honey solution i.e. undiluted honey,or alcohol extract
of equivalent concentration produced a zone of inhibition of
4 mm diameter in agar diffusion tests with S.�ureus. In
cam�arison though, 50% of the total activity eluted from
5 ml of undiluted honey run on a prima�y thin-layer
chromatogram (i.e. eluent was dried, redissolved in 0.2 ml
distilled water, 0.1 ml of which was assayed for activity)
produced a 6 mm diameter zone of clearing. Assuming an
average plate agar depth of 2 mm, the clearing given by the
honey would have been produced by only 0.1 ml of undiluted
honey, whereas the agar well measuring inhibition eluted
from the chromatogram should have contained the equivalent
of 2.5 ml of undiluted honey antibacterial activity.
Hence more than half of the antibacterial activity was
being lost during attempted isolation of these compounds.
However the extent of bacterial inhibition is not linearly
related to the amount of antibacterial activity present in
agar di·ffusion tests and it was therefore not possible to
concluJe from this that 0.1 ml of undiluted honey run on a
primary TLC plate and eluted would only have produced a
0.24 mm diameter zone of clearing (equivalent to 6% recovery
of activity during the attempted isolation of activity).
Also some activity may have been trapped in the water­
insoluble compounds still present at this stage (i.e.
elution of a primary TLC plate) and some may have adhered
to the equipment used. Assay of the total material eluted
from the rest of the silica of a primary plate developed
with extract of 5 ml of honey (and excluding the active
area already assayed) produced a clearing of diameter
 126

0.7 mm. Therefore tailing of the activity over the rest

of the plate could not account for the remainder of the
activity loss. As only the major antibacterial components
were being investigated in thLs thesis, this slight amount
of activity which was spread over the rest of the primary
TLC plate was not investigated further�
It was concluded that most of the activity not
accounted for was therefore being lost during application
of the alcohol extract to the first plate, during
chromatography or during subsequent assay. This activity
may have been chemically different to the compounds
characterised (Chapter 6) and possibly more volatile.
Although zones of inhibition are not additive it appeared
that a11· the activity recovered from primary chromatography
was also being recovered from secondary chromatography.
A-proportion of the activity may therefore have been
oxidised during primary chromatography, or have evaporated
or been altered by drying with the hairdryer i.e. the loss
did not occur during elution or assay of the material.
This would have to be investigated in future work on this
topic. Lavie (1963) found an antibacterial fraction from
honey which was extractable into cold ether and volatile at
°
95 C. The temperatures reached by the hairdryer were never
measured but probably would have exceeded 100 ° c to be able
to dry the extract applied (which still contained a percent-
'
age of sugarJ.·
A further �roblem was that the proportion of the
activity recovered after primary chromatography was found
to be sensitive to c·ertain handling techniques. The anti-
bacterial activity of the manuka honey extracts was lost by
 127

oxidation when exposed to oxygen in an impure solvent

mixture, probably containing iron. Species with no anti-
bacterial activity were also produced when the compounds
adsorbed to silica after chromatography were ground up ·in
ethanol with a pestle and mortar.
Isolation of the non-peroxide antibacterial components
was also difficult because originally little was known about
the class of compounds being dealt with. The finding that
the presence of the activity in honey d?pended on the plant
source suggested it was not a protein. Antibacterial
substances isolated from animals are usually p�ptides or
proteins whereas plant antibacterial and antifungal
substances are usually phenolic compounds, terpenes or
flavones (Booker et al,,1961; Walker, 1975).
The probability that the antibacterial compounds were
of non-protein nature was supported by their heat stability.
Heating raw manuka honey to 95 ° c for up to 60 minutes did
not reduce its non-peroxide antibacterial activity.
Solvent solubility and the development patterns also
suggested that the compounds were not peptides. An idea
of the polarity of the active substances was given by their
apparent solubility in ethanol and ether mixtures after
alcohol extraction of manuka honey. They were concluded
to be fairly soluble beca�se they were extractable into
alcohol but not ether and remained soluble in a mixture of
20% ethanol - 80% ether. This was supported by their
migration behaviour during thin-layer chromatography.
The subsequent identification by mass spectrometry of
the isolated antibacterial compounds as methyl
3,4,5-trimethoxybenzoate (a-methyl syringic acid methyl
ester), methyl 4-hydro�y-3,5-dimethexypenzoate (syringic
 128

acid methyl ester) and trimethoxy benzoic acid complied with

the observations on total non-peroxide activity obtained
earlier.
A phenol substituted �ith- one or more hydroxy groups
could explain loss of activity when alkali was added to the
honey or its extract, though phenols arB usually re-hydrolys-
able. This reaction may have been occurring:

OOH + NaOH

Reacidifying should have restored the activit� though it

did not appear to. Methyl 4-hydroxy-3,5-dimethoxy benzoate
would have to be re�cidified to approximately pH 3 to
return its activity (P. Holland, personal communication).
During attempted extraction with ether the solution was
acidified only slightly and therefore not enough acid would
have been added to rehydrolyse the molecule. Alkali could
also cause loss of activity from 3,4,5-trimethoxybenzoic
acid by formation of water by the hydroxyl group of the
alkali with the acid group of the phenol.
It is not known if alkali would react with methyl
3,4,5-trimethoxybenzoate as this was not investigated.
It is possible that this compound retained its activity but
was not detected by bioassay (after addition of alkali to
the extract) as it was only a minor proportion of the anti­
bacterially active compounds present.
It is not known how loss of activity was occurring when
the compounds, still bound to silica, were ground in
ethanol with a pestle and mortar. This reaction, with
loss of water was possiblv� occurring: '
o,s./o,
0/
·,o, ........-OH···'
'Si
,.,.,,., "
 129

Evidence of the phenolic nature of the antibacterial

compounds was also obtained when spray reagents were
applied to developed thin-layer chromatograms of honey
extract. The presence of phenols was shown by a stain of
ferric chloride - potassium ferricyanide. A ferric
chloride stain indicated that they probably did not contain
two or more vicinal hydroxy groups or ortho (or peri)
hydroxy-carbonyl groups.
Phenols tail badly during thin-layer chromatography
(A. Wilkins, personal communication) and this is also
consistent with the migration behaviour of the antibacterial
compounds during isolation.
Confirmation of the identification of the antibacterial
activity as the compounds mentioned above is still required
however. This should be obtained by testing these
compounds (authentic standards) for antibacterial activity
by the assay method used throughout this work. It is
possible that the compounds isolated and identified by mass
spectrometry were present with another substance, or
several others, which were not detected because they were
in too small a quantity or because they were non-volatile,
but which were responsible for the antibacterial activity.
Another aspect of this study to be investigated
further is the true origin and synthesis of the compounds
which were isolated and cha-racterised. A possible
precursor of them is syringic acid which occurs in vast
amou·nts in li.gnin ,· in combined forms and may be liberated
by alkaline hydrolysis.
 130

Booker, Co�bie and Cooper (1981), however, did not

find these compounds or syringic acid in manuka. They
isolated from the bark of manuka the diketone, leptosp­
ermone; triterpene acids; ur-solic acid acetate; m�nnitol;
the dicoumarinJellagic acid and its a-methyl ethers; and
various terpenes. They purified amGl �haracterised the
compounds which were present_ in a reasonable quantity in
the bark and it is possible that the compounds found in
this study may have been present only in the nectar of the
manuka plant, or they occurred only in minute quantities
which somehow became concentrated in honey. It is also
possible that the compounds may exist only in the pollen
(and leach out into the honey). It would be of interest
to analyse the composition of manuka nectar and pollen.
The compounds do not appear to have been isolated from other
native plants either. The pollen, nectar and sap of these
could also be investigated for antibacterial compounds.
Similar structures might be expected if the origin of the
compounds in this study is syringic acid or lignin.
The apparent nectar origin of the non-peroxide anti­
bacterial factors in manuka honey implies that the anti­
bacterial properties of a honey will depend on its floral
source. This could account for some of the discrepancies
in the findings of previous research concerning honey.
Classification of honey floral type in New Zealand is an
approximate procedure only and may be carried out through
taste, aroma, pollen analysis, or a combination of these.
The honey is labelled according to the apparent major
nectar source. For export purposes, however, the law
requires at least 51% of,a specific floral type to be
 131

present to be labelled as that floral origin. These laws

vary in other countries, and the antibacterial activity will
depend on its total floral composition and not just its
major floral source •. This was clearly demonstrated when
a honey labelled "various" was assayed for activity
(Chapter 3). This honey contain�d nbn-peroxide anti-
bacterial activity although its major source (43%) was
white clover according to a pollen analysis and it would
have been marketed in New Zealand as clover honey.
However other "clover n honeys assayed for activity gave
negative results. A honey containing only a small
percentage of an antibacterially active floral source may
still show strong activity depending on the content and
activity of the compound(s) characteristic to that nectar
source.
The discrepancies between various studies on the anti­
bacterial properties of honey can also be attributed to
some extent to the assay method used. James et aL. (1972)
investigated the antibacterial properties of Jamaican honey
using streak plates (t�st cultures streaked on agar plates
containing 2, 10 o� 20% honey dilutions), ditch plates
(test organism streaked on agar plates containing ditches
of 10% honey in agar), sensitivity disc method (2 or 10%
honey-saturated filter discs applied to agar plates
previously flooded with test culture and dried), and an
agar well method (the same procedure as used in this study,
with 5, 10 and 20% dilutions of honey). The disc method
and well method proved to be very insensitive. None of
the honey samples produced any inhibition in the disc
method or the well method. However it was noted that the
higher concentrations oi honey often caused a mu�doid-
 132

appearance of the growth immediately around the well.

Inhibitory action was best seen when honey was incorporated
in agar plates. (The method of Dold and Witzenhousen,
1955). Therefore the method used in the present study
was relatively insensitive compared to that used in some
previous research.
Much of the past research into the antibacterial
properties of honey has focused on inhibine (hydrogen
peroxide) and has n ot quantitatively taken into account
the presence of other antibacterial compounds. The inhibine
number of a particular honey was measured using a solutihn
of whole raw honey, not an extract. Therefore total
inhibine number was measured which would have included
inhibition due to other factors present as well as H2 o •
2
For instance White and Subers (1963) measured inhibine by
the procedure of Dold andWitz��h?usen and measured H 2 o 2 by

a colourimetric method using o-dianisidine. They formed

a table relating inhibine number to peroxide accumulation
from these results but of the 45 samples tested more than
one quarter of them (14) gave anomalou§ results. Seven
produced more peroxide than accounted for by their inhibine
number and seven prod·uce:Cl� "less. Their use of o-dianisidine

2 o2
to measure H may also have produced overestimations of

H o (Chapter 3) and therefore the presence of other anti-

2 2
bacterial factors would have been 11 masked".
In 1964 White and Subers measured the effect of heat
on honey inhibine and found that on heating at 70 °C for
ten minutes nineteen·samples lost over 80% of their activity
against S.Aureus, five lost 60-80%, two lost 40-60% and two
were more heat resistant. During the present study it was
 133

found that heating an a lcohol extract of manuka honey at 95 ° c

for up to 1 hour did not reduce its antibacterial activity
at all. This assay measured non-peroxide activity only
whereas that of White and Subers would have measured all
antibacterial activity present. Adcock (1962) found that
heating a 50% honey solution at 90° c fo� 15 minutes completely
destroyed its peroxide value. Only one experiment was
carried out however and the inhibine valrie (i.e. total anti­
bacterial effect) was not measured, and the peroxide system
may have been unstable at even lower temperatures. Hence
the antibacterial activity �fter heating, as measured by
White and Subers, may have been due entirely to non-peroxide
factors.
Gonnet and Lavie (1960), also using alcohol extracts
of various ·honeys (and therefore assaying non-peroxide
antibacterial factors), found that heat only slightly
decreased activity, depending on the type of honey. All
retained some activity.after autoclaving at 120° c for
15 minutes. They also found that heating (80° C, 30 minutes)
and neutralising honey caused a complete loss of activity
but re-acidfying restored it. The present study with
manuka honey gave similar results. It is to be expected
that the non-peroxide antibacterial factors of other
honeys will have different properties depending on their
origin.
The composition of a honey _and its pollen content will
vary according to the flowers in ·the vicinity of the hive.
The composition of bee venom, royal jelly and propolis is
less variable however and it is therefore unlikely that
the antibacterial compo�nds present in manuka and certain
other honeys will have come from any of these. Bee venom
 134

has only slight antibacterial activity (Lavie, 1960),

although melittin (the largest single component, by w�ight)
has a stronger effect and penicillin-resistant S.aureus
st�airt 80 is sensitive to it, {Fennell et al� 1968). It
also contains phospholipase A (Orl�v,· 1979) which can
degrade cell membranes. Bbth these compounds are proteins
and not related to the structures isolated in this study
and it would be unlikely for any substance in bee venom to
enter honey. The acid fraction of royal jelly is weakly
antibacterial, is soluble in water, alcohol and ether, but
is very unstable (Lavie, 1960). It is also unlikely that
royal jelly would come into contact with the honey in the
hive. Propolis is strongly antibacterial and is soluble
in water and alcohol (Lavie, 1960). During its collection
some of the constituents may come into contact with nectar
or pollen, also being collected� Propolis is mainly
collected from the boughs, leaves and buds of poplar, birch,
elm, alder, beech and horse chestnut trees (Ghisalberti, 1979).
Therefore, if this was the origin of honey 1 s antibacterial
compounds it would. be expected that most honey samples
assayed, irrespective of floral origin, would possess some
antibacterial activity. Also the compounds which have
been shown to contribute toward propolis' antibacterial
action are galangin, pinocembrin (both flavones), caffeic
acid and �rulic acid (both cinnamic acid d�rivatives)
(Ghisalberti, 1979); which were not detected in this study.
Antibacterial activity, soluble in water and alcohol,
has also been found ·in some types of pollen, increasing
with time and temperature during storage (Lavie, 1960).
Hence the possibility of the antibacterial compounds in
 135

manuka honey coming from its pollen content requires further

investigation. The bees wax in th� hive also contains
thermostable, acetone - and water-soluble antibacterial
activity (Lavie, 1960) and is in constant contact with the
honey. Compounds may therefore leach from the wax into
the honey and be present after extract�on. However Lavie
(1960) reports that it has no activity against Salmonella
(manuka honey does) and its composition is unlikely to vary
greatly, irrespective of honey type (Harman, 1983).
It is not yet known if the isolated compounds are
synthesised in the manuka plant and are present in nectar
(or pollen) in the same form as they are in honey, or
whether precursors of them exist in nectar (or pollen) which
are converted to these antibacterial compounds by reaction
�
in the bees saliva (i.e. converted to more or less complex/
'
compounds during the ripening of nectar into honey). The
acidic environment of the honey (also· effected by bee
salivary enzymes) may also be involved in the conversio�.
This is thought to occur with some of the compounds present
in propolis (Cannon et aL, 1973, cited by Ghisalberti,
197 9). A sample of Western Australian propolis was found
to contain pterostilbene which had not been reported in
Eucalyptus. It was thought possible that bees transform
naturally-occurring stilbenes by methylation or dem·EthylatiDn
to pterostilbene. Benzyl alcohol was also in the propolis
and may be a degradation product of the pterostilbene.
Several glands are connected to the oral cavity in the
head and thorax of the workar honeybee. The pharyngeal
(which contain antib.iotic activity of their own; Lavie,
1960) or hypopharyngeal,glands,_ the post cerebral, thoracic
and labial glands (salivary glands) and possibly also the
 136

mandibular glands are the important ones for the processing

and ripening of honey. The hypopharyngeal glands secrete
bee milk (part of the food for brood rearing), and also
produce a secretion rich in di�stase, invertase and glucose
oxidase, used in elaborating honey. The salivary glands
are believed by many to provide the liq�id needed to dissolve
solid food but produce no enzymes to break down carbohydrate.
One person however (Inglesent, 1940; cited by Crane, 1975)
has claimed that they produce diastase and invertase. The
mandibular glands mostly contribute to the "brood-food" but
their secretions have been discovered in the honey sac
contents so it is possible that they may contribute toward
the rtpening of honey. The composition of these gland
secretions could be more closely investigated and the
individual components, particularly the enzymes, could be
tested for their possible effect on the antibacterial
compounds of this study and their potential precursors.
Inactive precursors of manuka compounds may enter
unripened honey as more simple unsubstituted molecules
(such as syringic acid which may exist free in the plant as
a lignin precursor) or as more complex molec�les which
become degrade� as the result of enzyme action (e.g. enzymes
produced by white rot fungi growing on sawdust will produce
various compou�s by oxidation and demethoxylation,
including syringic acid and syringaldehyde from lignin
(Walker, 1975; Schubert, edited by Miller, 1973)). The
decreasing pH of ripening honey may be an important
influence. For instance, it seems likely that the content
of syringic acid methyl ester (methyl 4-hydroxy - 3,5-
dimethoxybenzoate will increase as the manuka honey becomes
more acid and therefore the antibacterial potential of
 137

manuka honey should increase. (Activity is restored to

this compound after alkali-inactivation by decreasing the
pH to approximately pH 3). It is possible therefore that
this compound will not exist at all in unripened honey.
0-methyl syringic acid methyl ester (methyl 3,4,5-trimethoxy
benzoate) may be produced in localised areas of unripened
honey where methylation is possible by substitution of the
syringic acid methyl ester molecule after alkaline hydrolysis.
This compound may originally exist in this form as phloem
sap is usually slightly alkaline:

DiscoveYy of the origin and synthesis of these antibacterial

compounds would be aideq by complete analysis of manuka
plant nectar, contents of the bees honey sac and unripened I
/\
honey samples.
The mode of action of the isolated compounds from
manuka honey should also be investigated. Little work has
been done on the actual.mode of action of antibacterial
factors in honey against bacteria. Mohrig and Messner
(1968) found honey had a bacteri-0lytic component which
acted on most gram po�itive organisms and had a generally
bacteriostatic effect. They claimed that all the
characteristic features of inhibines in honey described in
the literature could be explained by the lysozyme. James
et al./1972), found the samples of honey they tested had a
bacteriostatic effect only. It caused .increased pleomorphism
and variability in gram reaction and indistinct cell margins
in both gram negative and gram positive or�anisms.
 138

The mode of action of the non-peroxide antibacterial

factors in this study may be related �o their phenolic
nature. Phenolic hydroxyl groups �E.?-E,mit stable cross-links
with proteins and this may be the way in which syringic
..
acid methyl ester and possibly 3�4,5-trimethoxybenzoic acid
inhibit S.aureus. ·They could bind to the cell membrane
proteins, inhibiting their biochemical functions, such as
ATP production and potassium uptake; or simply cause
disruption and therefore lea�age by binding to anionic
sites on the cell wall, as do cationic detergents. They
may also be transported across the cell membrane owing to
their small size, and interrupt cellular functions.
It will be necessary to investigate whether the anti­
bacterial compounds are bacteriostatic or bactericidal.
This can be carried out by growing cultures of S.aureus in
broth tubes containing different concentrations of honey
or an isolated factor. After an incubation period loopfuls
of the test organisms could be spread on nutrient agar
plates or transferred to fresh nutrient broth to check for
viability. Lo_wer concentrations may be bacteriostatic and
higher concentrations may be bactericidal.
Further work will also be required to then find the
mode of action of the compounds. Information on this
could be obtained by turbidity tests to measure the release
of cell constituents (indicating disruption of the cell
membrane) or by electron microscopy. Disruption of the
cell membrane could also be investigated by measuring
release of radioactively-labelled cell components. Radio-
active labelling of the antibacterial substances could be
used to find if they bind to or enter the bacterial cell.
 139

Specific assays can be used to determine which biochemical

functions of the cell are inhibited by the antibacterial
factors.
The isolated antibacterial compounds should be re­
tested against a range of bacteria in different concentra-
tions. Their mode of action against e�ch, and whether they
are bacteriostatic or bactericidal also requires
in vestigation •
The mode of action of the compounds and their spectrum
of action will indicate if they may have further use as
antibiotics. For instance if they have a wide spectrum of
action, especially against such spe�ies as penicillin­
resistant S.aureus strain- 80, and theh"' mode of action is
such that they will not have adverse effects on the
eucaryotic cell (as phenols usually do) the� they may have
potential.
Other compounds separated from manuka honey and not
thought to be antibacterial could be investigat�d for
inhibitory activity when in higher concentrations.
Further work could also be done on investigating other
compounds from manuka honey which may be antibacterially
active. The active compound(s),apparently volatile at the
temperatures reached by the hairdryer used in this study
have yet to be isolated - and may constitute the major
portion of the non-peroxide manuka activity. Other active
compounds may be present too which were not detected by mass
spectroscopy because they were involatile or were present in
minute quantities (though very acti ve) and therefore may
have only appeared as contiminants in the samples. Seas anal
var�ations in the manuka honey composition could also be
 140

investigated and the method of high pressure liquid

chromatography (�PLC) may be more useful for this.
Another interesting area of research which could be
investigated is the isolation qf antibacterial compounds
from other plant nectar sources, or the honey, if this is
where the compounds are in active or more concentrated
form. The honey or nectar of medicinal herbs may prove a
particularly interesting and useful area of research.
 141

REFERENCES

Adcock, D. (1962). The effect of catalase on the inhibine

and peroxide values of various honeys. J. Anicultural
Res.1:38-40.
Arunachalam, K. (1980). Antimicrobial activity of garlic,
onion and honey. Geobios. 7·:46�47.
Atkinson, N, and Brice, H.E. ( 1955) • Antibacteria 1
substances produced by flowering plants. 2. The
antibacterial action of essential oils from some
Australian plants. Austral, J. exp. Biol. 33:547-554�
·Barrass, R. (1978). Scientists Must Write. Chapman and
Hall Ltd., London, 176 pp.
Boehringer, (1973). Biochemica Information. Boehringer
Mannheim GmbH, Biochemica, p. 97.
Briggs, L.H. (1947). Plant products of New Zealand.
J. Roy. Soc. N.S.W. 80:151-177.
B_ooker, S.G., Cambie, R.C., Cooper, R.C. (1981).. New
Zealand Medicinal Plants, Heinemann, Auckland,
pp. 69-74.
Brown, S.B. (ed.) (1980). An introduction to spectroscopy
for biochemists. Academie Press, London, pp. 1-65,
235-261.
Burgett, M. (1978). Antibiotic systems in honey, nectar
and pollen. In Honey bee pests, predators and
diseases. Morse, R�A. (ed) .. Cornell University
Press, N.Y. pp. 298-308.
Cain, B.F. (1963). The water-soluble phenolics of bark
extracts of Leptospermum scoparium Forst. N.Z. J.
Sci. 6:264-8.
 142

Chu, F. (1979). A study of the therapeutic effects of

propolis. Apiacta. XIV (1): 23-25.
V
Cizmarik, J. (1979). Utilisation of propolis in human
medicine. Apiacta XIV (1): 16-22.
Crane, E. (ed) • ( 1975) • Hciriey, ·Heinemann, London, 608 pp.
Daghie, V, Cirnu, I. (1973). Contributions to the research
on the bactericidal and bacteriostatie action of honey.
26 (2) :13-16.
Dawson, R.M.C, Elliott, D.C, Elliott, W.H, Jones, K.M. (eds).
(1969). D�t� fci�_bibchemic�l �ese��ch, 2nd ed. Oxford
University Press, pp. 509-593.
Dold, H and Witzenhausen, R (1955). Ein Verfahren zur
Beurteilung der ortlichen inhibitorischen (keimvermehrung­
stemmerden) Wirkung von Honigsorten verschiedenenr Herkunft.
Zeit. Hyg. Tnfektionskrankh. 141: 333-337.
Donald, A. _( 19 82) • Propolis -- natures antibiotic. Bes tways •
. ..•
(Jan�. pp. 51�54:
Dustmann, J.H (1979). Antibacterial effect of honey.
Apiactta, XIV (1): 7-11.
Dyer, J.R (1965). Applications of abscirption spectrosco py of
organic compounds. Prentice-H�ll Inc., N.J. 147 pp.
Fennell, J.F, Shipman, W.H, Cole, L.J (1968). Antibacterial
action of melittin, a polypeptide from bee venom.
Proc. Soc. Biol. Med. (United States), 127 (3): 707-710.
Ghisalberti, E.L (1979}. Propolis: a review. Bee World,
60 (2): 59-84.
Gannet, M and Lavie, P (1960). Influence du chauffage sur le
facteur antibiotique present d�ns les miels. Ann. Abeille,
3 (4): 349-364.
Gannet M and Lavie, P and Louveaux, J. (1964). Le pasturisation
des miels. Ann. ·Abeil1e· •. 7 ( 2): 81-102.
 143

Gonnet, M., Lavie, P and Nogueira - Neto, P. (1964).

Etude de quelques caracteristiques des miels
recottes par certains meliponines bresiliens.
c •R •· Acad • sc • . Paris , 258: 31 o 7-31 o 9 •
Gonnet, M. (1969), Sur l'origine des substances phyto­
inhibititrices presentes dans les miels de lavande.
C.'R.· Acad ... Sc. Paris, 268: 8"59-861.
Gannet, M. ( 19 81). Facteurs antibiatiques naturels
presents dans le miel. Re�ue Francaise d'apiculture.
399 (7): 27-30.
Goulart, F. S. (1979). First aid from the hive:- how
honey heals. Am. Bee J. 119 (11): 792.
Gubanski, M. (1962). Tobacco mosaic virus inhibitors from
bees and h.e.e products. Rezerat ·Bukowy Wiaczyn.
I (13): 93-98.
Gunther, R.T. (1934). The Greek Herbal of Dioscorides.
Hafner, New York. pp. 124-126.
Harman, A.W. (1983). Hive Products for therapeutic use.
A ni • Bee J • 12 3 ( 1 ) : 39 - 4 4 •
Heftmann, E. (ed) (1975). Chro-niato graphy. A laboratory
handbook of chromatographic· and ·electrophoretic
methods. (3rd ed.). Van Nostrand Reinhold Co.,
N.Y. 969 pp.
Izdebska, K and Gubanski, M (1962). Badania had antybak-
teryjnyni wtasciwasciama naturalnego miodu pszczelego.
Zesz; natik; Uni�; Iodz. 2 (12): 21-26 e
Jackman, L.M and Stemmnell, S (1969). Applications of
riticlear magnetic resonance spectrb�copy in organic
chemistry. Vol. 5 of International series of mono-
graphs in organic chemistry. Barton, D.H.R and
Doering, W. (Gen. eds) (2nd ed.). Pergamon Press,
London, pp 1-17.
 144

James, O.B.O'L, Segroe, W and Ventura, A.K (1972). Some

antibacterial properties o-f Jamaican honey. W.I.
Med. J. XXI (7): 7-17.
Kirchner, J.G (1978). Thiri_;la ver chromatography, Vol.
XIV T�chriigu�s of dhemistry. (2nd ed.). John Wiley
and Sons Inc., N.Y. 1137 pp.
Knowles� P.F, Marsh, D, Rattle, H.W.E (i976). Magnetic
reS�riaride ·df bio��lectiles. John Wiley and Sons Ltd,
London. pp. 1-29.
Kotova, G (1981). Apiary products are important in Soviet
medicine. Ani·. -Bee J. 121 (12): 850-851.
Lavie, P (1960). Les substances antibadt�riennes dans la
coldnie d'AbeilleS (Api& mell±fic� L.). Thesis
Station experimentale d'Apiculture, Montfavet, France.
299 pp.
Lavie, P (1963) • Sur l'identification des substances
antibacteriennes presentes dans le miel. C.R Acad.
Sci. Fr. 256: 1858-1860.
Lisboa, B.P (1964). Characterisation of �4 -3-oxo-c21-

steroids on thin-layer chromatograms by in situ"

colour reactions. J� Chromatog. 16: 136-151.
Mladenov, S (1974). Present problems of apitherapy. In,
Hive urod·udts:· fodd; health arid beauty. Proceedings
of the International symposium on apitherapy, Madrid,
Spain. Apimondia Publishing House, pp. 135-136.
Mohrig, W and Messner� B (1968). Lysozym als anti­
bakterielles Agens im Bierenhonig und Bienengift.
Adta •. bidl. nied. germ. 2li 55-95.
Nicholas, H.J (1973). Terpenes, Miscellaneous volatile
plant products Tri Phytodheniistrv, Vol� II, Organic
Metabolites. Miiler, L.P, (ed.). Van Nostrand
Reinhold, Co., N.Y pp. 254 -310, 381-401.
 145

Orlov, B.N (1979). Theoretical basis for therapeutical

use of bee venom. Aoiasta XIV (1): 12-15.
Popescu, V, Paunescu, T, Ghitescu, I, Velescu, G,
Maftei, I, Iliesau, I (1944). First results
obtained in treatment of Actinomycosis by bee products
and vegetal extracts. In, Hive p�cidticts: f6od,
health artd ti�atity. Proceedings of the Inter-national
symposium on aiptherapy, Madrid, Spain. Apimondia
Publishing House, pp. 123-127.
Pryor, L.D and Bryant, L.H (1958). Inheritance of oil
characters in Eucalyptus. Proceedings of Linnean
Society of New South Wales, 83:55-64.
Randerath, K (1966). Thin�layer Chromatography. (2nd ed.)
Verlog Chemie, GmbH, 285 pp.
Revathy, V and Banerji, S.A (1980). A preliminary study
of antibacterial properties of Indian honey.
Indi an J • Bi ocherri • Bi oph ys • 17, Aug • 198 0 -
supplement no. 242.
Rodriguez-Navarro, A and Ruiz-Argueso, T (1970).
Ripening honey bacteria. Lebensm-Wiss, u. Technol.
3 (6): 118.
Ruiz-Argueso, T and Rodriguez-Navarro, A (1973). Gluconic
acid-producing bacteria from honey bees and
ripening honey. J. Gen. Microbial. 76: 211-216.
Ruiz-Argileso, T and Rodriguez-Navarro, A (1978). Micro-
biology of ripening honey. Aoolied Microbiology.
3 (6): 893-896.
Schlenk, H and Gellerman, J.L (1960). Esterification of
fatty acids with diazomethane on a small scale.
Anal; ·chem. 32: 1412-1414.
 146

Schubert, W.J (1973). Lignin. In, PhytDchemistry, Vol.

chemicals. Miller, L.P (ed). Van Nostrand

Reinhold Co., N.Y. pp. 132-154.
Schuler, R and Vogel, R (1956). Workstoffe des Bienenhonigs.
Arzneiniitt.· Forsch. 6: 661-668.
Smith, M.R, Mccaughey, W.F and Kemmerer, A.R (1969).
Biological effects of honey. J. Apic. Res. 8 (2):
99-110.
Stahl, E (1969) • Thin-layer Chr6mat6gr�phy. 2nd ed.
Springer-Verl�g, New York. 1041 pp.
Thomson, R.H.K (1936). Chemical composition of New Zealand
honey. Cawthron Institute Publ. No. lJ. Extra.
from N.Z. Journal of Sc�ence and Technology, XVIII
(2): 124-131.
Tomlinson, A (1982). Hospitals using honey as a fast new
antibiotic. Am. Bee J. 122 (4): 247.
Tysset, C, Durand, C, Taliercio, Y.P (1970). Contribution
i l' �tude du microbisme et de l'hygi�ne des miels du
commerce. Rec. Me'd. Vet. CXLVl: 1471-1491.
Tysset, C, Rousseau, M, Durard, C (1980). Microbism and
wholesomeness of commercial honey. Apia ta XV (2) :
51-60.
Walker, J.R.L (1975). Studies in Biology, No. 54. The
Biol6gy of Plant Phendlics. Edward Arnold Ltd,
London. 57 PP•
Wells, F.B (1976). Hive product uses - honey. Am. Bee J.
116: 469-471.
White, J.W, Subers, M�H and Schepartz, A (1962). The
identification of inhibine. · Am. Bee J. 102 (11):
430-431.
 147

White, J.W and Subers, M.H (1963). Studies on honey

inhibine. 2. A chemical assay. J. Api c. Res • 2:
93-100.
White, J.W and Subers, M.H (1904). Studies on honey inhibine.
3. Effect of heat. J. Apit� Res. 3 (1): 45-50.
White, J.W (1966). Inhibine and gluco�e oxid�se in
honey - a review. Arri. Bee J� 106 (6).