Histochemistry

Histochemistry (1990) 93:247 256

Springer-Verlag 1989

Romanowsky dyes and Romanowsky-Giemsa effect

5. Structural investigations of the purple DNA-AB-EY dye complexes
of Romanowsky-Giemsa staining
K. Friedrich, W. Seiffert, and H.W. Zimmermann
Institut fiir Physikalische Chemic der Universitfit Freiburg im Breisgau, Albertstrasse 21, D-7800 Freiburg,
Federal Republic of Germany
Received May 29, 1989 / Accepted September 1, 1989

Summary. A reproducible Romanowsky-Giemsa staining

(RGS) can be carried out with standardized staining solutions containing the two dyes azure B (AB) and eosin Y
(EY). After staining, cell nuclei have a purple coloration
generated by DNA-AB-EY complexes. The microspectra
of cell nuclei have a sharp and intense absorption band
at 18100 cm -1 (552 nm), the so called Romanowsky band
(RB), which is due to the EY chromophore of the dye complexes. Other absorption bands can be assigned to the
DNA-bound AB cations.
Artificial DNA-AB-EY complexes can be prepared outside the cell by subsequent staining of D N A with AB and
EY. In the first step of our staining experiments we prepared
thin films of blue DNA-AB complexes on microslides with
1:1 composition: each anionic phosphodiester residue of
the nucleic acid was occupied by one AB cation. Microspectrophotometric investigations of the dye preparations demonstrated that, besides monomers and dimers, mainly higher
AB aggregates are bound to D N A by electrostatic and hydrophobic interactions. These DNA-AB complexes are insoluble in water. Therefore it was possible to stain the
DNA-AB films with aqueous EY solutions and also to prepare insoluble DNA-AB-EY films in the second step of
the staining experiments. After the reaction with EY, thin
sites within the dye preparations were purple. The microspectra of the purple spots show a strong Romanowsky
band at 18100 era- 1. Using a special technique it was possible to estimate the composition of the purple dye complexes.
The ratio of the two dyes was approximately EY :AB ~ 1:3.
The EY anions are mainly bound by hydrophobic interaction to the AB framework of the electrical neutral DNA-AB
complexes. The EY absorption is red shifted by the interaction of EY with the AB framework of DNA-AB-EY. We
suppose that this red shift is caused by a dielectric polarization of the bound EY dianions.
The D N A chains in the DNA-AB complexes can mechanically be aligned in a preferred direction k. Highly orientated dye complexes prepared on microslides were birefringent and dichroic. The orientation is maintained during
subsequent staining with aqueous EY solutions. In this way
we also prepared highly orientated purple DNA-AB-EY
complexes on microslides. The light absorption of both
types of dye complexes was studied by means of a microspectrophotometer equipped with a polarizer and an analyser. The sites of best orientation within the dye preparations were selected under crossed nicols according to the
quality of birefringence. Subsequently, the absorption spec-

tra of the highly orientated dye complexes were measured

with plane polarized light. We found that the transition
moments, mAB, of the bound AB cations in DNA-AB and
DNA-AB-EY are orientated almost perpendicular to k, i.e.
mA~_l_k. On the contrary, the transition moments, mEv, of
the bound EY anions in DNA-AB-EY are polarized parallel
to k, i.e. mrv [Ik. The transition moments mA, and m,~v
lay in the direction of the long axes of the AB and EY
chromophores. For that reason, in both DNA-AB and
DNA-AB-EY the long molecular axes of the AB cations
are orientated approximately perpendicular to the D N A
chains, while the long molecular axes of the EY chromophores are polarized in the direction of the D N A chains.
Therefore, in DNA-AB-EY the long axes of AB and EY
are perpendicular to each other, mABZmry. This molecular
arrangement fully agrees with our quantitative measurements and with the theory of the absorption of plane polarized light by orientated dye complexes, which has been developed and discussed in detail.

Introduction
Romanowsky-Giemsa staining (RGS) can easily be performed with an aqueous solution of the two Romanowsky
dyes azure B(AB) and eosin Y(EY). After staining we observe the well known Romanowsky-Giemsa effect (RGE):
the cell nuclei show a purple coloration, which contrasts
with the bluish colour of the cytoplasm (Wittekind et al.
1976; Wittekind 1979; Galbraith et al. 1980; Marshall et al.
1981; Lapen 1982; Wittekind 1983, 1986). Neither AB nor
EY are purple. Therefore, the purple coloration of cell nuclei
is somewhat surprising and was the topic of numerous investigations (for references and discussions see e.g. Wittekind 1979, 1986; Galbraith et al. 1980; Marshall et al. 1981;
Horobin and Walter 1987). In a very extensive microspectrophotometric and physicochemical study we came to the
conclusion that the purple coloration of the cell nuclei is
generated by complexes of DNA, AB, and EY (Zipfel et al.
1981, 1982, 1984; Miiller-Walz and Zimmermann 1987). The
staining kinetics shows that there are two different steps
of dye binding. In the first step AB cations are bound to
the D N A polyanion by electrostatic and hydrophobic interactions. The nucleic acid acts as a chromotrope. Bound
monomers, 1, dimers, 2, and higher aggregated AB cations,
3, are formed, which can easily be distinguished by absorption spectroscopy. In the case of LM ceils, which have been
used as cell material in our model experiments (Zipfel et al.

248
~. [nm]

X [nm]
800 750 700

650

600

550

500

1.5

Z~50

RB

800 750 700

650

600

1.5

550

500

~50

18

20

22

RB

~5 1.0
o

1.0
(/)
.1:3
<~:

\\\
0.5

0.0

/ ~-

1/,

16

18
[

0.5

Xx
\\\\

20

22

103cm -1]

0.0

1~

16

~ [103cm-1]

Fig. l. a Absorption spectrum of cell nuclei of LM cells after Romanowsky Giemsa staining (RGS). The RNA of the nucleoli and
of the cytoplasm has been digested enzymatically to prevent disturbances by RNA-dye complexes (-). b Absorption spectrum of an
artificial DNA-AB-EY complex (-@ RB, Romanowsky band. (a
and b according to Zipfel et al. 1984)

Fig. 2. a Absorption spectrum of cell nuclei of LM cells after RGS

(according to Zipfel et al. 1984). b Absorption spectrum of AB-EY
precipitates. RB, Romanowsky band

1984), the different molecular species have absorption peaks,

A, at A1 15400 c m - 1 (649 nm), A2 16800 c m - 1 (595 c m - 1),
and A 3 17 800 cm-1 (562 nm), respectively. With increasing
aggregation the peaks are metachromatically shifted to
shorter wavelength. Under the usual conditions of RGS
the bound AB cations form besides monomers and dimers
mainly higher dye aggregates on the DNA. The concentration of bound molecules is very high. Binding isotherms
of AB to calfthymus D N A as chromotrope demonstrate
that the AB cations completely compensate or even overcompensate the negative charge of the phosphodiester residues of the nucleic acid (Miiller-Walz and Zimmermann
1987). Electrically neutral of positively charged DNA-AB
complexes are formed, which are the target of the EY anions
in the second step of the dye binding in RGS. It can be
shown experimentally that EY binding does not occur unless AB cations are bound and compensate the negative
charge of the DNA, but not vice versa (Zipfel et al. 1984).
Thus purple DNA-AB-EY complexes are generated in the
cell nuclei by hydrophobic and electrostatic interactions.
Their absorption spectrum shows a distinct and intense absorption band at 18100cm" 1 (552 nm), the so called Romanowsky band (RB) (Fig. 1). RB is due to the EY part
of the DNA-AB-EY complexes. The EY absorption is
shifted to a longer wavelength by the interaction of EY
with the DNA-AB framework of the dye complexes. This
red shift is the main reason for the purple coloration of
the cell nuclei (Zipfel et al. 1984). RB overlaps with the very
broad absorption of the DNA-bound higher AB aggregates,
but can easily be distinguished from A 3 spectroscopically.
Enzymatic digestion experiments also demonstrate the
participation of D N A in the purple dye complexes. After
D N A digestion with highly active DNAse the nuclear chromatin does not stain under RGS conditions while the
cytoplasm is stained as usual. On the contrary, enzymatic
R N A digestion completely prevents the staining of the cytoplasm and the nucleoli, but the purple coloration of the
cell nuclei still remains. These experiments clearly demon-

strate the significance of D N A and R N A in the staining

of the nuclear chromatin and the cytoplasm (Zipfel et al.
1984).
The participation of DNA-AB-EY complexes in the
purple coloration of cell nuclei has been investigated by
artificially preparing such complexes outside the cell using
a special experimental technique. Successive staining of
D N A with AB and EY yields purple DNA-AB-EY complexes, whose absorption spectra are pratically identical
with the spectrum of Romanowsky-Giemsa stained cell nuclei (Fig. 1) (Zipfel et al. 1984). This result confirms that
D N A acts as a chromotrope in RGS of nuclear chromatin.
Obviously, the dye complexes have several "spheres". The
inner sphere consists of the D N A helix, surrounded by AB
cations in the next sphere and finally EY anions in the
outer one. The aim of the following investigations is to
find out the geometrical structure of these dye complexes.
Recently Horobin and Walter (1987) suggested that ABEY precipitates may be the origin of the purple coloration
of the nuclear chromatin. Such precipitates are well known
and frequently occur in relatively concentrated azure-eosin
staining solutions. Obviously, these dye solutions are oversaturated and not in thermodynamic equilibrium with their
surroundings. We analysed such AB-EY precipitates and
found a molar composition AB:EY of 2:1; two AB monocations are bound to one EY dianion by simple electrostatic
interactions. The absorption spectrum of the solid AB-EY
precipitates has been measured and compared with the spectrum of Romanowsky-Giemsa stained cell nuclei (Fig. 2),
The two spectra are completely different. Therefore, the precipitates mentioned above cannot be the origin of the purple
coloration of nuclear chromatin under usual RGS conditions, although at very long staining periods and high dye
concentrations the AB-EY precipitates may be superimposed on ordinary RGS.
The concepts of precipitates and DNA-AB-EY dye complexes can be combined. The DNA-AB comlexes, which
are formed in the first step of dye binding under RGS condi-

249
tions, are insoluble in water. In these precipitates the DNA
is saturated or even oversaturated with bound AB cations.
Each anionic phosphodiester residue of the D N A is occupied by AB cations (Mfiller-Walz and Zimmermann 1987).
The insoluble DNA-AB precipitates stabilize the stained
nuclear structures prior to binding of EY and formation
of the DNA-AB-EY complexes. The latter are also insoluble
in aqueous medium, which is one of the reason why RGS
is stable over a long period. Therefore, dye precipitates play
an important part in RGS.
D N A and also other polyanions, for example chondroitin sulfate (CHS), act as a chromotrope in RGS. We studied
the binding of AB cations to CHS in aqueous solutions
and in the solid state by absorption spectroscopy (Hfiglin
et al. 1986). In aqueous solutions three different dye species
with absorption bands at A1 15500cm -1 (646nm), A2
16800 cm -1 (597nm), and A3 18000 cm -~ (555 nm) can
be distinguished. Again these are assigned to monomers,
1, dimers, 2, and higher aggregated AB cations, 3, respectively, which are bound to CHS according to the increase
of the dye concentration in solution. Finally, the biopolymer
is saturated, and the metachromatic CHS-AB complex with
absorption at 18000 cm -~ has 1:1 composition; each anionic SO~- and COOe-binding site of CHS binds one dye
cation. The AB cations are planar molecules. The distance
between adjacent binding sites in CHS is small enough to
allow the molecular planes of the AB cations to touch and
the ~-electrons to interact. This interaction is due to the
blue shift of the long wavelength absorption of the bound
dye aggregates.
Using a special dialytic technique it was also possible
to prepare solid metachromatic CHS-AB complexes with
1:1 composition (Hfiglin et al. 1986; Friedrich et al. 1989).
These have a very broad and intense absorption band with
its main peak at about A3 17900 cm -1 (560 nm). It corresponds with the maximum of the metachromatic dye complexes in aqueous solutions. In our context it is now of
interest that these solid 1 : 1 CHS-AB complexes additionally bind EY dianions by hydrophobic interactions forming
purple CHS-AB-EY complexes with an intense and sharp
absorption band at 18100 c m - * (552 nm) (Friedrich et al.
1989). This band can clearly be distinguished from the broad
absorption of the metachromatically bound AB cations, 3.
It is generated by the EY chromophore, whose absorption
is shifted to a longer wavelength by interaction with the
CHS-AB framework. The sharp and intense absorption
peak at 18100 cm -* behaves completely like the Romanowsky band (RB) of the DNA-AB-EY complexes at
18100 cm -1. Therefore we concluded that it has the same
physical basis, and that the purple CHS-AB-EY complexes
are models for the DNA-AB-EY complexes of RGS.
In order to investigate the structure of the dye complexes
we mechanically aligned the CHS chains of CHS-AB and
CHS-AB-EY in a preferred direction, k. Thin films of highly
orientated dye complexes were prepared on microslides and
studied with a microspectrophotometer equipped with a polarizer and an analyser. They were birefringent and dichroic
- the more birefringent, the better the mechanical orientation. We selected the sites of best orientation within the
film according to their birefringence and measured the absorption of these regions with plane polarized light. The
light absorption has a maximum on the condition that the
transition moment, m, of the dye molecules and the vector,
%, of the polarizer are parallel. By setting ep paralel (ll)

mAB

N
AB

CH3

CH3
mEy
Br

Br

@0"In
Br

I%
T

Br
COOE~

Fig. 3. Azure B cation (AB) and Eosin Y dianion (EY) with transition moments mAn and mFy. In EY the planes of the xanthene
chromophore and of the phenyl residue are nearly perpendicular
to each other
or perpendicular (L) to k, we found that the transition moment, mAB, of the long wavelength absorption of AB in
the CHS-AB and CHS-AB-EY complexes is polarized almost perpendicular to the preferred direction k, i.e. mABZk.
But the transition moment, m~v, of EY in CHS-AB-EY
is polarized parallel to k, i.e. mEy IIk. The long wavelength
absorptions of AB and EY are ~ --+~* transitions. The transition moments, mAB and mEy, lay in the molecular plane
in the direction of the long axes of the AB and EY chromophores, respectively (Fig. 3). Therefore, in both CHS-AB
and CHS-AB-EY the long axes of the AB molecules are
approximately perpendicular to the CHS chains; but in
CHS-AB-EY the long axes of the EY chromophore are
parallel to the chains of the biopolymer. This structure is
somewhat surprising. In the purple CHS-AB-EY dye complexes the chromophores of AB and EY are not parallel
but perpendicular to each other, mAnLm~v.
The same experimental procedure has also been employed for investigating the structure of the purple DNAAB-EY complexes of RGS. We used calfthymus D N A as
chromotrope in our model experiments. Unfortunately
DNA is mechanically not as stable as CHS. Therefore it
was very difficult to mechanically align the chains of the
biopolymer. However under cautious experimentation it
was possible to produce highly orientated dye complexes
with D N A as chromotrope, which have been used to measure the absorption spectra with polarized light. Due to
the unfavourable experimental conditions, the sites of best
orientation within the preparations were very small. Therefore, to get good results, we again used the microspectrophotometer for investigating the linear dichroism of the orientated dye complexes.
Material and methods

Substances. Azure B chloride (AB) and eosin Y(EY) were prepared

and purified according to Zipfel et al. (1984, 1982). EY was used
as dye acid, dissolved in water. In aqueous solutions eosin acid
dissociates forming dianions (Fig. 3). Calfthymus DNA was used
in our model experiments. It was purchased from Worthington

250
and employed without further purification. The content of the nucleic acid was determined by phosphate analysis (Gerlach and Deuticke 1963).

DNA-AB complexes. The dye complexes were prepared according

to dialytic methods described elsewhere (Hfiglin et al. 1986; Friedrich et al. 1989; Friedrich 1989). In aqueous medium DNA-AB
precipitates are formed. To determine the AB content, ethanolic
solutions of the dye complexes were prepared and analysed by
absorption spectroscopy (Friedrich et al. 1989; Friedrich 1989). In
our microspectrophotometric experiments we used only DNA-AB
complexes with 1 : 1 stoichiometry. In these complexes each anionic
phosphodiester bindig site of the DNA was occupied by one AB
cation. The dye complexes can mechanically be aligned on microslides (Friedrich 1989). After air drying the preparations were ready
for mierophotometry. Mounted with Entellan (Merck, Darmstadt,
RFG) they were stable for months.

DNA-AB-EYcompIexes. Air dried DNA-AB precipitates mechanically orientated on microslides were treated for 30 rain with a saturated aqueous EY solution. Afterwards they were carefully rinsed
with water to remove a surplus of eosin acid. Very thin spots of
the dye preparations are now purple and consist of orientated
DNA-AB-EY complexes, while thicker ones are violet. Obviously,
the diffusion of the eosin anions into the DNA-AB precipitates
is retarded and the purple DNA-AB-EY complexes emerge very
slowly. Since the preparations of the orientated dye complexes are
unhomogeneous, it is difficult to determine the stoichiometry of
the purple dye complexes. After air drying the dye preparations
can be used for microscopical investigations. Mounted with Entellan they were stable for some days (Friedrich et al. 1989; Friedrich
1989).

Microspectrophotometry. The microspectra of the orientated dye

complexes were measured with a home-built, completely computerized microspectrophotometer, equipped with a polarizer and an
analyser. The spectrometer has been described in detail by Ohmes
et al. (1980).
Results

Absorption spectra and linear dichroism

of the DNA-AB complexes
The same dialytic technique has been employed to prepare
solid CHS-AB and D N A - A B complexes. They had 1:1
composition and the anionic binding sites of the biopolymers were completely occupied by AB cations. Nevertheless,
CHS and D N A behave quite differently in dye binding.
CHS forms 1 : 1 dye complexes, which only consist of bound
higher AB aggregates, 3, with their characteristic metachromatic absorption A 3 at 17900 cm-1 (560 nm) (Hiiglin et al.
1986). D N A also forms 1:1 dye complexes, but different
molecular species can spectroscopically be distinguished.
The spectrum shows a very broad and intense absorption
in the range of about 1 6 0 0 0 - 1 9 0 0 0 c m -1 (625-526nm),
which is generated by a superposition of bands A 2 and
A 3 of bound dimers and higher AB aggregates (Fig. 4). A
weak long wavelength shoulder at 15200 cm -1 (657 nm)
can be assigned to the absorption of bound monomers A 1.
According to the binding study of Mfiller-Walz and Zimmermann (1987) the appropriate absorption bands are expected at A1 1 5 2 0 0 c m -1 (658nm), A 2 1 7 1 0 0 c m -1
(585 nm), and A 3 18000 cm -1 (556 nm).
The dissimilarities of CHS and D N A in dye binding
are due to differences in the distances of the binding sites
and in the flexibilities of the polymer chains. In order to
get metachromatic absorption only, the molecular planes

800 750 700 650

X [nm]
600
550

500

~.50

20

22

RB

1.0

~ 0.5
0.0

14

16

18
[103cm-~]

Fig. 4. Absorption spectrum of (a) 1 : 1 DNA-AB and (b) DNA-ABEY complexes. The absorption spectra represent the average of
measurements at ten different sites within the dye preparation. RB,
Romanowsky band
of the bound AB cations must touch one another and form
higher dye aggregates on the biopolymer. In general it cannot be expected that neighbouring binding sites of biopolymers have exactly the right distances for allowing the molecular planes to touch and the molecules to interact. However
a single-stranded biopolymer like CHS is flexible enough
to bring the binding sites in favourable positions and therefore allows the formation of bound higher AB aggregates
along the polymer chain by electrostatic and hydrophobic
interactions. On the contrary, the double-stranded D N A
is much less flexible than CHS, and the formation of higher
aggregates is hindered. Therefore, we observe besides higher
AB associates also bound dimers and even monomers. At
high dye concentrations in the staining solutions a surplus
of AB cations can be bound and the monomers and dimers
change into higher aggregates (Miiller-Walz and Zimmermann 1987).
Previously we suggested that the electrically uncharged
1:1 CHS-AB complexes bind EY anions by hydrophobic
interactions (Friedrich et al. 1989). To compare the EY
binding capacity of CHS-AB with D N A - A B we also used
1:1 D N A - A B complexes in the following binding experiments. Moreover, in ordinary R G S monomers and dimers
are bound to nuclear D N A as well as higher aggregated
AB cations (Zipfel et al. 1984). Therefore, the artificial 1:1
D N A - A B complexes are realistic models for studying the
EY binding in RGS.
The D N A - A B complexes have been mechanically
aligned on microslides in the preferred direction, k. The
sites of best orientation within the preparations were selected using the birefringence of the aligned D N A - A B fibres
and their absorption measured with plane polarized light.
The microspectra at various angles e between k and the
vector, %, of the polarizer are given in Fig. 5. These represent the average of measurements at ten different sites within
the dye preparations. The maximum of light absorption
is observed at e = 90 , the minimum at e = 0 . With growing
the intensities of A1, A2, and A 3 strongly increase. On
the condition that the polymer chains are preferentially orientated in the direction k and that the transition moments,

251

800 750 700 650

k [nm]
600
550

X [nm]
500

450

800 750 700

650

600

550

500

&50

20

22

1.5
1.5

oc~ 1.0

i3

~ I.0

u0
LI

0.5

0.5
2

0.0

1/~

16

18

2'0

'

22

[103cm-ll

0.0

l&

16

18
[ 103crn-I]

Fig. 5. Microspectra of mechanically aligned 1 : 1 DNA-AB complexes at various angles e between the preferred direction k of
the DNA chains and the vector ep of the polarizer, c~= 0, ep [Ik
(spectrum 1); 30 (2); 60 (3); 90, %k (4). The absorption spectra
represent the average of measurements at ten different sites within
the dye preparations

Fig. 6. Microspectra of mechanically orientated purple DNA-ABEY complexes at various angles e between the preferred direction
k of the DNA chains and the vector % of the polarizer, c~= 0,
% IIk (spectrum I); 30 (2); 60 (3); 90, epA_k (4). The absorption
spectra represent the average of measurements at ten different purple sites within the dye preparation. RB, Romanowsky band

mAB, are polarized in the long axes of the AB cations, mAs

and k are than about perpendicular to each other, i.e.
mAak. The same molecular arrangement has been found
earlier in case of the CHS-AB complexes (Hfiglin et al. 1986;
Friedrich et al. 1989).
In the spectrum (1) of Fig. 5, ep[[k (e=0), two weak
absorptions at T 1 15 400 and T 2 16 800 c m - 1 (649, 595 nm)
can be observed with a polarization in the direction k of
the polymer chain. They may not be confused with the absorptions A1 and A 2 of bound monomers and dimers at
15 200 and 17100 cm-1, which are polarized almost perpendicular to k. Unfortunately both types of bands overlap.
Probably T 1 and T2 are caused by light-induced intermolecular electronic transitions between adjacent AB cations
in the bound dye aggregates. Similar transitions have also
been observed in the 1 : 1 CHS-AB complexes (Htiglin et al.
1986).

preparations were again measured with plane polarized

light at various angles e between ep and k. The spectra
in Fig. 6 are the average of measurements at ten different
purple-coloured sites within the dye preparations.
A comparison of the corresponding spectra in Figs. 5
and 6 of the D N A - A B and D N A - A B - E Y complexes at various angles, e, is very instructive. The perpendicular polarized spectra (4) of both complexes, %A_k (~ = 90), are nearly
identical. The absorptions have been assigned to the 7c~ re*
transitions of the AB cations. Accordingly, the transition
moments and the long molecular axes of the AB cations
(Fig. 3) are orientated almost perpendicular to the polymer
chains, mABlk, and the AB molecules have more or less
the same molecular arrangements in both dye complexes.
The parallel polarized spectra (1), ep l[k (c~= 0), of the
two dye complexes are completely different. In the spectrum
of the D N A - A B complex the two transitions T1 and T2
can only be observed, while in the case of D N A - A B - E Y
besides T 1 and T2 the strong R o m a n o w s k y band, RB, with
its absorption at 18100 cm -1 particularly dominates the
spectrum. According to this we came to the conclusion that
the transition moments and the long molecular axes of the
EY chromophores (Fig. 3) are mainly orientated parallel
to the polymer chains, mEy ][k. Therefore, in D N A - A B - E Y
the transition moments of AB and EY are predominantly
perpendicular to each other, mABim~.v. Similar results have
been described for CHS-AB and CHS-AB-EY (Friedrich
et al. 1989).

Absorption spectra and linear dichroism

of the purple DNA-AB-E Ycomplexes
Fine fibres of solid 1 : 1 D N A - A B complexes on microslides
were treated with aqueous EY solutions (see the Materials
and methods). After the reaction with EY very thin spots
of the dye preparations have a purple coloration, while
thicker ones are violet. The purple sites consist of D N A - A B EY complexes with a sharp RB at 18100cm -1 (552 nm)
(Fig. 4). The EY dianions are bound to the electrically uncharged 1 : 1 D N A - A B complex mainly by hydrophobic interactions. RB is caused by the EY chromphore of the dye
complex; its absorption is 1200 cm -1 red shifted by the
binding of the dye from aqueous solution to the D N A - A B
framework. This red shift generates the purple coloration
of the nuclear chromatin in RGS.
Highly orientated D N A - A B - E Y complexes can be prepared by using aligned 1:1 D N A - A B fibres in the above
staining experiments with EY. They still were birefringent
and dichroic. The microspectra of purple areas of the dye

Quantitative treatment of the linear dichroism

of the orientated dye complexes
The spectroscopic investigations with polarized light gave
us a qualitative impression of the molecular arrangements
of the dye molecules within the orientated dye complexes.
To verify these structures, quantitative considerations are
necessary. Particularly, the absorbance, A, of the dye preparations at different angles, c~, between ep and k must be

252

in agreement with the theory of light absorption. However

some difficulties arise. The incident light changes within
the orientated dye preparations from plane to elliptical polarization. Moreover, the ellipsis rotates and the ratio of
the two semiaxes alters when the light passes through the
microscopical specimen (Ohmes et al. 1980). In our case the
conditions are somewhat simpler. In general, one axis of
the ellipsis is much longer than the other one, and therefore
the elliptical polarized light behaves approximately like
plane polarized light. The plane of polarization rotates by
passing the light through the dye preparation, which affects
the light absorption of the orientated dye complexes. This
effect must be taken into account in calculating the absorbance of the orientated dye preparations at various angles,
The absorbance of orientated polymer fibres loaded with
dye molecules has been calculated in general (Ohmes et al.
1980). In the following considerations two types of molecules shall be distinguished with transition moments m parallel and perpendicular to k. In the case of an ideal alignment of the polymer chains and perfect orientation of the
transition moments the theoretical absorbances E are:
(1 a)
Ell = 2 K C O S 2
E = K ( 1 --cos 2 c0,

(1 b)

where K = const. [rnl 2 c x = ? x, c is the dye concentration, x

is the thickness of the dye preparation, 7 is the absorption
coefficient characterizing the dye preparation, and ~ is the
angle between % and k.
Deviations from the ideal orientation frequently occure
in practice. Then, instead of Eqs. (la, 1 b) we have (2a, 2b),
which describe the absorbance E of dye preparations with
a preferred - but not ideal - orientation of the transition
moments parallel and perpendicular to k (Ohmes et al.
1980):
(2 a)

Ell = 2 y x ( a + b COS2 (~),

E = 7 x ( p - q cos 2 e),

a<b
p>q.

(2b)

The constants a, b, p, q > 0 describe the statistical deviations

from the ideal alignment and orientation. In the ideal case
the ratios of these constants are a/b = O, p/q = 1.
Eqs. (la, lb, 2a, 2b) are true for a constant angle c~
between % and k. But the plane of polarisation rotates when
the light passes through the stained specimen. We assume
that the angle of rotation A c~is proportional to the thickness
of the specimen:

A~z=cox.

(3)

co has the physical meaning of a specific rotation of the

plane of polarization along the distance x = 1. Passing the
light through a specimen of thickness x, the angle of polarization changes from e to/3:
/ 3 = e + A c~= c~+cox.

(4)

/3 is the angle between e, and k, e, is the direction of the

analyser at maximum light transmission.
A c~ and accordingly co varies with the angle ct and the
wave number g of the incident light. In our experiments
we used angles in the range of 0_< c~_<90. In this range
A 7 was negative, A c~_<0,/3_< e. For example, the maximum
value of about A e = --40 has been determined in case of

1.5,

o~ 1.0
g
g~0.5

0.0
0.0

0.5

1.0
1.5
2.0
F
Fig. 7. Absorbance AAB of orientated DNA-AB complexes vs. the
function F, Eq. (7), at various wave numbers g. g=14000 cm -1
(v); 15000 cm -1 (o); 16000cm -1 ([]); 17000cm -1 (zx);
18000cm -1 (v); 19000 cm -~ (e); 20000cm -1 (m). The plots of
AAB VS. F represent the average of measurements at ten different
sites within the dye preparations

a D N A - A B preparation at c~=45 and at the maximum

of the absorption band g = 18000 cm 1. This value of Ae,
of course, is to high to be ignored in the following considerations.
Taking the rotation of the plane of polarization into
account the absorbance shall be labelled A instead of E.
A changes by dA when the light passes through an infinitesimal layer of thickness d x at distance x. Using Eqs. (2a,
2b, 4) we immediately get:
dA II = 2 ~ [a + b cos 2 (c~+ cox)] d x

(5 a)

dA = ~ [p-- q cos 2 (~ -[- COX)] d x.

(5 b)

By integration of Eqs. (5 a, 5b) from the distance x = 0 to

the total thickness of the dye preparation x = x, we get the
absorbance A with regard to the rotation of the plane of
polarization from ~ to fi, A ~ = f i - ~, (in radians):

All =2?x(a+bF/2)
A =yx(p--qF/2)
F = 1 + (1/2 A e) [sin 2 (c~+ A e) - sin 2 c~].

(6a)
(6b)
(7)

In case of A e ~ 0 we have
lim F = 2 cos 2 c~,

(8)

Ae~0

and the equations change from (6a, 6b) to (2a, 2b).

Let us consider the D N A - A B complexes with transition
moments mAB of the bound AB cations almost perpendicular to k. Figure 7 shows the absorbance Aa = AAB Vs. the
function F, Eq. (7), at various wave numbers, "~. The calculations were carried out with data from the spectra in Fig. 5.
In agreement with Eq. (6b) we get straight lines with the
negative slope -(1/2) qTaBX, which changes with ~.
N o w let us consider the D N A - A B - E Y complexes, which
are composed of two different dye species EY and AB with
polarizations almost parallel and perpendicular to k, respectively. The total absorbance, A, is the superposition of the
absorbances All = A E y and A=AAB, Equations (6a, 6b),
which are weighted according to the absorption coefficients
YEv and ~/AB:

A=AEy+AAB=X[EaTEy+p~AB+(bTEy--q~AB/2)F].

(9)

253
1.5

1.5
.<

taJ
~,
1.0
u

g to

d~

< 0.5
0.0

0.0

~. 0s
,

0.5

1.0

1.5

0.0
0.0

2.0

0.5

1.0

COS 2 C(,

Fig. 8. Absorbance A=AAB+AEy of orientated purple DNA-ABEY complexes vs. the function F, Eq. (7), at various wave numbers
~7. g=14000cm -1 (A); 16000cm-1 (o); 17000 cm-1 (I);
18000cm 1 (ix); 19000cm -1 (o); 20000cm -1 (D). The plots of
A vs. F represent the average of measurements at ten different
purple sites within the dye preparations

Fig. 9. Calculated absorbance EAB of 1:1 DNA-AB complexes at

18000 cm -1 vs. cos 2 e. c~is the angle between k and %. The wave
number is related to the maximum of absorption of the higher
aggregated AB cations A3

Once more, the diagram A vs. F shows straight lines with

a negative slope at various wave numbers g (Fig. 8; d a t a
from Fig. 6). E q u a t i o n (9) is fulfilled, and the negative slope
--(1/2)(qTAB--2bTEy)X indicates, that the AB anions dominate over the EY anions, q TAB> 2 b 7EY.
Now, we eliminate the influence of the rotation of the
plane of polarization by calculating the theoretical absorbance E from the difference between E and the experimental
absorbance A. F o r example, to consider the D N A - A B complexes, the difference EAB--AAB leads us in combination
with Eqs. (2 a, 6 a) to the relation:

now E is a superposition of EEy and EAB. Our aim is to

separate the total absorption spectrum E into the two parts
Emc and EAB.
To determine the a b s o r p t i o n spectrum EEy we subtract
the absorption EAB of the AB part of the D N A - A B - E Y
complex from the total a b s o r p t i o n E:

EAB = AAB + q TABX (F/2

COS200.

(12)

EAB is given by Eq. (2 b):

EAB = TABx ( p - - q COS2 ~X).

(2b)

(10)

In Eq. (10) qTAB x(F/2-c0s2 Orefers to the rotation of the

plane of polarization and corrects the experimental absorba n c e AAB. In the limit A ~ 0
and F = 2 c o s 2 ~ , Eq. (10)
becomes EAB=AAB . The factor qTABx can be determined
from the slope of the straight lines in Fig. 7. Finally, the
absorbance EAB w a s calculated at various wave numbers
and angles ~. EAB = E(~, ~)AB is the theoretical absorption
spectrum at constant angle c~, which is no longer disturbed
by the rotation of the plane of polarization. In Fig. 9 the
calculated absorbances EAB at 1 8 0 0 0 c m 1 are plotted
against cos 2 cc The wave number is related to the m a x i m u m
of the a b s o r p t i o n spectrum of the D N A - A B complex. In
full agreement with Eq. (2b) we get a straight line with
a negative slope, which can be used to determine the ratio
p/q = 1,24. It is only slightly higher than the ratio p/q = 1,00,
which is the limit in the case of an ideal polymer alignment
and a perfect dye orientation. This result demonstrates the
excellent mechanical orientation of the D N A - A B complexes
used in our experiments.
The above procedure can be extended to D N A - A B - E Y
complexes. N o w the differences between the total absorbances E - - A = EEy + EAB -- AEy -- AAB must be taken into account. As before, we get in combination with Eqs. (2a, 2b,
6a, 6b):
E = A + x (q TAB-- 2 b 7~Y)(F/2 -- cos2 c0"

EEy=E--EAB.

(11)

The absorbance A can be taken from the experiments and

the factor x(qTAB--2bTEy) from the slopes of the straight
lines in Fig. 8 at various wave numbers g. In this way the
total absorption spectra E(~, c~) have been calculated; but

In principle EAB is k n o w n from the investigations of the

D N A - A B complexes. The results, which are related to these
experiments shall be labelled by an apostrophe:
E~,B = 7~ B x'(p--q COS2 C0.

(13)

As will be shown later, the ratios of the parameters p and

q are nearly identical in both types of dye complexes. Therefore, in a first approximation, it is not necessary to lable
p and q in Eq. (13). The thickness x of the dye preparations
and the AB concentrations c, which enter into the absorption coefficient 7as, m a y be different in D N A - A B and D N A AB-EY. Therefore, the a b s o r p t i o n spectra of both types
of dye complexes must be a d a p t e d to one another. This
can easily be done by multiplying Eq. (13) with a constant
factor 6. In this way the complete A B spectrum can be
expanded or diminished until both spectra Eqs. (12, 13)
agree within a spectral range, which is only determined by
the AB part of the D N A - A B - E Y complex and which is
not affected by the EY absorption. In the range
< 1 6 5 0 0 c m -1 ( 2 > 6 0 6 n m ) only the AB cations absorb,
E=EAB. The difference AE between the absorbance EAB
and EkB of b o t h dye complexes shall be minimized for all
wave numbers g < 16 500 c m - 1 and all angles c by a p r o p e r
factor 6:

AE = EAB-- Es = 0,

(14)

and it follows immediately:

~AB .X: = ~ ]?:kB X'.

(15)

254
X [nrnl
800 750 700

650

1.5

600

550

500

~.50

800 750 700

RB

650

k [nm]
600
550

500

450

20

22

1.5

RB

1.0

1.0
c
c
cl
d::l

to
J=l

0.5

< 0.5

0.0

14

16

18

20

0.0

22

1/.

Fig. 10. Adapted and corrected absorption spectra of DNA-AB

(a, EAB) and DNA-AB-EY (b, E=EAn+EEy) at ~=0 , %Ilk. RB,
Romanowsky band

Additionally, small band shifts between the AB parts of

the D N A - A B and D N A - A B - E Y spectra must be taken into
account, which are caused by the interaction of EY with
the AB framework of the D N A - A B - E Y complex. To minimize the difference AE between the adapted spectra, Eq.
(14), a 200 cm -1 blue shift of EAB was introduced, which
corrects this effect primarily in the long wavelength region
of the spectrum. But it is difficult to say whether the same
correction is also sufficient in the short wavelength range,
where EY additionally absorbs. Fortunately, in our strutural investigation of the D N A - A B - E Y complex, we are only
interested in the c~-dependency of the absorbances EEy and
EAB of the bound EY and AB ions at one proper chosen
analytical wave number. We choose the maximum of RB
at 18100 c m - 1 which is close to the maximum of the AB
absorption, A3, at 1 8 0 0 0 c m -1. The latter is very broad,
and therefore small band shifts scarcely affect the ~-dependency of A 3 at this analytical wave number.
As an example, two adapted D N A - A B and D N A - A B EY spectra are shown in Fig. 10, ~ = 0 . Notice the excellent
agreement in the range 7 < 16500 cm-1. Figure 11 presents
the absorption spectra of EY in D N A - A B - E Y at different
angles e, which have been generated by the subtraction of
absorption spectra as described above. The R o m a n o w s k y
band, RB, which is due to the EY part of the dye complex,
has its maximum absorbance at c~=0 , its minimum at
= 90 .
Finally, the absorbances of EY and AB in D N A - A B - E Y
at 18 i00 c m - i , which have been separated from the total
absorbance E, are given in Fig. 12. EEy and EAB are plotted
against cos 2 ~. In full agreement with Eqs. (2a, 2b) we get
straight lines with opposite slopes. With increasing c~, and
therefore with shrinking cos 2 ~, RB decreases, while the AB
absorption increases. These results clearly demonstrate that
RB is polarized almost in the direction k and AB perpendicular to k, mEymAB. F r o m the straight lines in Fig. 12 the
ratios a/b = 0.05 and p/q = 1,27 have been determined. These
are only slightly different from the values of ideal orientated
polymers and dye molecules parallel (a/b = 0,00) and perpendicular (p/q = 1,00) to k. This proves the excellent mechani-

16

18
[ I03cm-I ]

[I03cm-I]

Fig. 11. Calculated absorption spectra of EY in DNA-AB-EY complexes at different angles e between k and %. e = 0 , % IIk (spectrum
1); 30 (2); 60 (3); 90, % l k (4). RB, Romanowsky band

1.5

~ B

bA

31.0
~5
k, 0.5

X3

00,

0.0

0.5

1.0

C 0 S 2 (21,

Fig. 12. Calculated absorbances EAB (ZX)and Ezy (O) in DNA-ABEY complexes at 18100cm -1 vs. cos2 cc ~ angle between k and
%. The wave number is related to the maximum of RB
cal orientation of the D N A - A B - E Y preparations. Similar
results have been found in the case of orientated CHS-ABEY complexes.

Discussion
D N A - A B - E Y complexes with different compositions can
be prepared by the treatment of solid 1:1 D N A - A B complexes with aqueous EY solutions. Using mechanically
aligned D N A - A B comlexes, the D N A - A B - E Y preparations
generated are orientiated, dichroic, and birefringent. Very
thin sites are purple and show a sharp and intense R o m a n owsky band, RB, at 18100 cm -1, which is caused by the
EY chromophore. F r o m the absorption spectra we concluded that the composition of the purple D N A - A B - E Y
complexes only varies within narrow limits. Unfortunately,
the purple sites of the dye prepartions were very small, and
it was impossible to analyse their composition exactly.
With the binding of a dye molecule from aqueous solution to a chromotrope like D N A only small changes of
the integrated absorption of the long wavelength band occur. Of course, on account of band shifts and splittings the
integration must be extended over the entire absorption

255
band. The integrated absorptions of EY and AB can be
determined from the measurements of Zipfel et al. (1981,
1982), and can be compared with the integrated absorptions
of both dyes in DNA-AB-EY, which have been estimated
from the spectra of the orientated dye complexes measured
with polarized light parallel and perpendicular to the direction k of the polymer chains. In this way we determined
the ratio of the dye molecules in the purple dye complexes
as EY:AB,~ 1:3. The same ratio has been estimated for
the CHS-AB-EY complexes (Friedrich et al. 1989). In both
dye complexes the AB cations predominate over the EY
dianions.
The binding of AB to D N A has recently been analysed
in detail (M/iller-Walz and Zimmermann 1987). Several
kinds of bound dye species must be distinguished. At low
dye concentrations intercalated (1) and pre-intercalatively
(2) bound dye molecules were observed, binding constants
K1 = 2.7 x 103 M - i, K2 = 1.9 x 105 M 1. With increasing
dye concentrations co-operative dye binding occurs. The
constants of nucleation (n) and aggregation (q) are K ,
= 1.5 x 103 M - 1, Kq= 1.2 x 105 M -1, co-operativity parameter q = 80. The two binding mechanisms at low and
high dye concentrations are completely different and are
competitive with each other. Therefore, at high dye concentrations, which are frequently used in RGS, co-operative
binding favours the formation of bound higher AB aggregates.
The binding constant of AB intercalation is very small
compared with the constants of other intercalating dye cations (Zimmermann 1986a, b). Therefore, under conventional RGS conditions only few AB cations intercalate. Nevertheless, we tried to find out whether intercalated dye molecules are able to bind EY and to generate the purple dye
complex. With a special technique, which has been described elsewhere (Ohmes et al. 1980), we prepared orientated DNA-AB complexes with intercalated AB cations only.
The transition moments mAB of these dye molecules were
orientated almost perpendicular to k, which is in agreement
with the concept of intercalation. On no account has it
been possible to bind EY anions to these intercalated AB
cations and to generate the purple dye complex. The intercalated dye molecules are shielded by the base pairs of the
nucleic acid, and therefore, the interaction with EY is prevented. Thus intercalation plays a minor part in RGS.
The target of the EY anions in RGS are the externally
bound AB aggregates, which are formed by co-operative
binding. They can be displaced by small metal cations,
especially by divalent cations such as Mg 2 . etc., which affect RGS. Therefore, in order to prevent failures of staining,
the staining solutions must be carefully composed.
The structure of the 1:1 DNA-AB complexes is comparable with a winding stairway. The central pillar is the D N A
double helix with its anionic phosphodiester binding sites,
and the steps are the AB cations, whose long axes are orientated perpendicular to the direction k of the polymer chains.
The molecular planes of adjacent dye molecules are in van
der Waals contact, and the ~-electrons interact. The blue
shift of the long wavelength absorption of the AB aggregates
would otherwise not be comprehensible.
The 1:1 DNA-AB complex is uncharged. Therefore it
seems likely that EY binding predominantly occurs by hydrophobic interactions. We expected that in this case the
chromophores of EY and AB would be parallel. But on
the contrary, they are perpendicular to each other,

mE_I_mAB. This shall be explained by at least one plausible

structure. In EY the planes of the xanthene chromophore
and the phenyl residue are twisted by steric hindrance and
are nearly perpendicular to each other. Thus, we can construct a molecular arrangement with parallel planes of the
AB cations and the phenyl residues of the EY anions. This
structure would agree with the linear dichroism of the orientated DNA-AB-EY complexes, and would be stabilized by
hydrophobic interaction. However, it is highly speculativ
and must be verified by further experiments.
However, hydrophobic interaction cannot explain the
red shift of the EY absorption by the dye binding to the
DNA-AB complex. It has been pointed out that the phenyl
residue and the xanthene chromophore of EY are nearly
perpendicular to each other. The interaction between the
~-electrons of these two orthogonal parts of the EY molecule can be approximately neglected, and the long wavelength absorption is almost unaffected by changes in the
surroundings of the phenyl residue. In order to explain the
red shift, other effects must be taken into account. EY is
a dianion with a strong dielectric polarizability. Therefore,
changes in the dielectric environment of the chromophore
can easily affect its light absorption. This effect is well
known and is used in analytical chemistry. For example,
in the titration of Br e with Ag e EY is employed to indicate
the equivalence point. At this point the AgBr precipitates
adsorbate EY, which is accompanied by a 1200 cm-1 red
shift of the long wavelength absorption of the bound dye
molecules (Fajans and Hassel 1923; Zipfel et al. 1984). This
is the same direction and of the same order of magnitude
as the red shift of the EY absorption in RGS. Therefore,
it seems likely that in our staining experiments the red shift
is caused by the dielectric polarization of the bound EY
dianions. The close relation between the dye adsorbation
on one hand and the dye binding in RGS on the other
is obvious. In usual RGS the binding of EY anions is additionally favoured by the surplus of AB cations, which are
bound to D N A and generate positively charged DNA-AB
complexes by co-operative binding effects (Mtiller-Walz and
Zimmermann 1987).

Acknowledgement. We are indebted to the Bundesministerium fiir

Forschung und Technologic (01Z08501/6) and to the Fonds der
Chemischen Industrie for generously supporting this study. We
are also grateful to Dr. D. Hfiglin for valuable comments and thank
Mr. H. Geiger for technical assistance and Mrs. C. Stein for writing
the manuscript.
References

Fajans H, Hassel O (1923) Eine neue Methode zur Tritation yon

Silber- und Halogenionen mit organischen Farbstoffindikatoren. Z Elektrochem 29:495 500
Friedrich K (1989) Mikrospektralphotometrische und polarisationsoptische Untersuchungen zur Struktur des purpurnen
Farbstoffkomplexes der Romanowsky-Giemsa-F/irbung. Dissertation Universit/it Freiburg i. Br., Institut f/it Physikalische
Chemic
Friedrich K, Hfiglin D, Seiffert W, Zimmermann HW (1989) Modelluntersuchungen zur Struktur des purpurnen Farbstoffkomplexes der Giemsa-F/irbung. Histochemistry 9l:257-262
Galbraith W, Marshall PN, Bacus IW (1980) Microspectrophotometric studies of Romanowsky stained blood cells. I. Subtraction analysis of a standardized procedure. J Microsc 119:313
330
Gerlach D, Deuticke B (1963) Eine einfache Methode zur Mikro-

256
bestimmung von Phosphat in der Papierchromatographie. Biochem Z 337:477-479
Horobin RW, Walter KJ (1987) Understanding Romanowsky
staining. I: the Romanowsky-Giemsa effect in blood smears.
Histochemistry 86:331 336
Hfiglin D, Seiffert W, Zimmermann HW (1986) Spektroskopische
und thermodynamische Untersuchungen zur Bindung von
Azure B an Chondroitinsulfat und zur Bindungsgeometrie
des metachromatischen Farbstoffkomplexes. Histochemistry
86:71-82
Lapen D (1982) A standarized differential stain for haematology.
Cytometry 2:309-315
Marshall PN, Galbraith W, Navarro EF, Bacus IW (1981) Microspectrophotometric studies of Romanowsky stained blood cells.
II. ComparisOn of the performance of two standardized stains.
J Microsc 124:197-210
Mfiller-Walz R, Zimmermann HW (1987) Uber RomanowskyFarbstoffe und den Romanowsky-Giemsa-Effekt. 4. Mitteilung:
Bindung yon Azur B an DNA. Histochemistry 87:15%172
Ohmes E, Pauluhn J, Weidner JU, Zimmermann HW (1980) Polarisationsoptische, mikrospektralphotometrische Untersuchung
zur Bindungsgeometrie yon intercaliertem Ethidiumbromid.
Theorie ffir die Lichtabsorption orientierter doppelbrechender
F/iden. Bet Bunsenges Phys Chem 84:23-36
Wittekind D (1979) On the nature of Romanowsky dyes and the
Romanowsky Giemsa effect. Clin Lab Haematol 1 :247-262
Wittekind DH (1983) On the nature of the Romanowsky Giemsa
staining and its significance for cytochemistry and histochemistry: an overall view. Histochem J 15:1029-1047
Wittekind D (1986) On the nature of the Romanowsky Giemsa

staining and the Romanowsky Giemsa effect. In: Boon ME,

Kok LP (eds) Standardization and quantitation of diagnostic
staining in cytology. Coulomb Press, Leyden, pp 27-40, 118119
Wittekind D, Kretschmer V, L6hr W (1976) Kann Azur B Eosin
die May-Gri.inwald-GiemsaF/irbung ersetzen. Blut 32:71-78
Zirnmermann HW (1986a) Physikochemische und cytochemische
Untersuchung zur Bindung von Ethidium- und Acridinfarbstoffen an DNA und an Organellen in lebenden Zellen. Angew
Chem 98:115-131
Zimmermann HW (1986b) Physicochemical and cytochemical investigations on the binding of ethidium and acridine dyes to
DNA and to organelles in living cells. Angew Chem Int (Engl)
25:115-130
Zipfel E, Grezes JR, Seiffert W, Zimmermann HW (1981) Uber
Romanowsky-Farbstoffe und den Romanowsky-Giemsa-Effekt. 1. Mitteilung: Azur B, Reinheit und Gehalt von Farbstoffen, Assoziation. Histochemistry 72:279-290
Zipfel E, Grezes JR, Seiffert W, Zimmermann HW (1982) Uber
Romanowsky-Farbstoffe und den Romanowsky-Giemsa-Effekt. 2. Mitteilung: Eosin Y, Erythrosin B, Tetrachlorfluorescein. Spektroskopische Charakterisierung der reinen Farbstoffe,
Assoziation yon Eosin Y. Histochemistry 75:539-555
Zipfel E, Grezes JR, Naujok A, Seiffert W, Wittekind DW, Zimmermann HW (1984) Llber Romanowsky-Farbstoffe und den Romanowsky-Giemsa-Effekt, 3. Mitteilung: Mikrospektralphotometrische Untersuchung der Romanowsky-Giemsa-F/irbung.
Spektroskopischer Nachweis eines DNA - Azur B - Eosin Y
- Komplexes, der den Romanowsky-Giemsa-Effekt verursacht.
Histochemistry 81 : 337-351