J

S
S

ISSN 1615-9306 JSSCCJ 38 (12) 20072192 (2015) Vol. 38 No. 12 June 2015 D 10609

JOURNAL OF

SEPARATION
SCIENCE

Methods
Chromatography Electroseparation
Applications
Biomedicine Foods Environment

12 15

www.jss-journal.com

2024
Jessica Pandohee1
Brendan J. Holland2
Bingshan Li5
Takuya Tsuzuki3
Paul G. Stevenson2
Neil W. Barnett2
James R. Pearson4
Oliver A.H. Jones1
Xavier A. Conlan2
1 School

of Applied Sciences,
RMIT University, Melbourne,
Victoria, Australia
2 Centre for Chemistry and
Biotechnology, School of Life
and Environmental Sciences,
Deakin University, Geelong,
Victoria, Australia
3 Research School of Engineering,
Australian National University,
Canberra, Australia
4 Victoria Police Forensic Services
Department, Macleod, Victoria,
Australia
5 Institute for Frontier Materials,
Deakin University, Geelong,
Victoria, Australia
Received January 28, 2015
Revised March 25, 2015
Accepted March 27, 2015

J. Sep. Sci. 2015, 38, 20242032

Research Article

Screening of cannabinoids in
industrial-grade hemp using
two-dimensional liquid chromatography
coupled with acidic potassium
permanganate chemiluminescence detection
Widely known for its recreational use, the cannabis plant also has the potential to act as
an antibacterial agent in the medicinal field. The analysis of cannabis plants/products in
both pharmacological and forensic studies often requires the separation of compounds of
interest and/or accurate identification of the whole cannabinoid profile. In order to provide a complete separation and detection of cannabinoids, a new two-dimensional liquid
chromatography method has been developed using acidic potassium permanganate chemiluminescence detection, which has been shown to be selective for cannabinoids. This was
carried out using a Luna 100 CN column and a Poroshell 120 EC-C18 column in the
first and second dimensions, respectively. The method has utilized a large amount of the
available separation space with a spreading angle of 48.4 and a correlation of 0.66 allowing
the determination of more than 120 constituents and mass spectral identification of ten
cannabinoids in a single analytical run. The method has the potential to improve research
involved in the characterization of sensitive, complex matrices.
Keywords: Cannabis / Offline chromatography / Profiling / 2D chromatography
DOI 10.1002/jssc.201500088

Additional supporting information may be found in the online version of this article
at the publishers web-site

1 Introduction
Comprising over 525 compounds, cannabis is a complex plant
that includes three different species (Cannabis sativa, C. indica, and C. ruderalis) that has been of interest for its biologically active constituents for hundreds of years [1, 2]. It is well
known as a recreational drug due to its psychoactive effect, attributed to the cannabinoid 9 -tetrahydrocannabinol (THC)
[3]. Other cannabinoids of importance are cannabidiol (CBD),
cannabichromene (CBC), cannabigerol (CBG), and cannabinol (CBN), all of which have been shown to possess potent activity against a variety of methicillin-resistant bacterial strains
of current clinical relevance [4, 5]. Many other cannabinoids
have also been identified which have been associated with a
range of potential medicinal uses such as anti-inflammatory,
antibiotic, antifungal, analgesic, and antioxidant compounds

Correspondence: Dr. Xavier A. Conlan, Centre for Chemistry

and Biotechnology, School of Life and Environmental Sciences,
Deakin University, Pigdons Road, Geelong, Victoria, Australia
E-mail: xavier.conlan@deakin.edu.au

Abbreviations: CBC, cannabichromene; CBD, cannabidiol;

CBGA, cannabigerolic acid; CBG, cannabigerol; CBN, cannabinol; THC, 9 -tetrahydrocannabinol

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

[6]. However, the use of cannabis as a medicine is still disputed because of socio-political pressure [1].
Hillig and Mahlberg [7] classified the cannabis plants
into three chemotypes depending on the THC/CBD ratio;
(i) drug-type plants with a high THC/CBD ratio (>>1:1),
(ii) intermediate type plants with a THC/CBD ratio close
to 1:1, and (iii) fiber-type plants with low THC/CBD ratios
(<<1:1). The latter form, also known as hemp is grown for
its seeds and fibers for agricultural and industrial purposes.
Hemp is known to be some of the best and most durable
fibers of natural origin and has been used to make paper,
banknotes, ropes, and sails for ships [1]. Hemp seeds are also
nutritious containing a high oil content, namely linoleic acid,
omega-3, and omega-6, essential fatty acids, and proteins;
they may be consumed in their natural form or as an oil, a
seed milk, or as a herbal infusion [8, 9].
Considering the complex nature and uses of the
cannabis/hemp plant, the need for separating its constituents
for further research is evident, but the degree of separation required varies with the purpose of the next study. For example,
forensic analyses for the identification/classification of fiber
or drug-type, and source tracing demand an accurate fingerprint of the entire cannabinoid profile [2]. In contrast, pharmacology often requires the isolation of a pure constituent(s)
of the plant extract [2]. Consequently, improvement in the

www.jss-journal.com

Liquid Chromatography

J. Sep. Sci. 2015, 38, 20242032

separation of such cannabinoids is essential to achieve baseline resolution for both isolation and profiling of the samples.
Currently, GC is the method of choice for creating
cannabis profiles and chemical fingerprints for attributing
the country of origin or conditions of cultivation, as it allows
identification of a large variety of cannabinoids with very high
resolution, particularly when coupled with MS [2]. However,
the high temperatures required for sample vaporization before injection causes the decarboxylation of the acidic forms
of cannabinoids and thermal degradation of the sample, thus
preventing analysis of cannabis in its natural state [2]. Furthermore, this thermal conversion of acidic cannabinoids seems
to be incomplete, resulting in a nonrepresentative analysis of
the sample [10]. The quantification of cannabinoids by GC
therefore requires a time-consuming derivatization before
analysis [10].
HPLC allows the simultaneous detection of both acidic
and neutral cannabinoids with no need of derivatization [11].
However, the majority of HPLC methods for cannabis analysis described in the literature either failed to separate all the
cannabinoids present and/or were not validated according to
appropriate guidelines [12]. The complex composition of the
material means that the separation of major cannabinoids is
not easily achieved and significant peak overlap occurs between CBD/CBG and CBN/cannabigerolic acid (CBGA) [5].
The use of MS coupled to HPLC may be a solution to resolve cannabinoids of interest in a single analytical run. This
method has been previously used to separate acidic and neutral cannabinoids allowing selective identification in hashish
samples with no derivatization or peak collection [13] and
determination of up to five cannabinoids in hemp [14]. Nevertheless, LCMS does not allow characterization of an entire
cannabis sample, but the determination of specific analytes.
However, for characterization and QC purposes, determination of cannabinoids in only one single run with high selectivity would be advantageous [14].
It is evident that 1D chromatographic techniques have
significant limitations, especially when chromatographic resolution of numerous compounds is desired [15]. A gain in
separation power/peak capacity can be achieved using 2D
chromatography [1618]. The technique involves combining
two dimensions of different separation mechanisms in series. Fractions from the first dimension are collected and
re-injected into the second dimension where they are further separated by an orthogonal (different) separation mechanism [16]. This provides a gain in peak capacity as the total
peak capacity of the 2D system is the product of the peak
capacity of each dimension [16].
The 2D LCUV/chemiluminescence (2DLCUV/CL)
methodology described here was developed to separate the
cannabinoids in industrial-grade cannabis efficiently. The orthogonal separation was then used to compare the chemical
profiles of samples of cannabis bark, hurd, and leaves of
both sexes of the plant. The acidic potassium permanganate
chemiluminescence detection system, provided an indication
of the chemical activity of compounds in the cannabis samples [19]. To the best of the authors knowledge, this is the

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

2025

first comprehensive 2D HPLC study of cannabis and has high

potential for future research.

2 Materials and methods

2.1 Chemicals
All chemicals were analytical grade reagents, unless otherwise stated. Potassium permanganate, ethyl acetate, sodium
sulfate (anhydrous powder) were obtained from ChemSupply (Gillman SA, Australia). Sodium polyphosphate
(crystals, +80 mesh, 96%), sodium thiosulfate (powder)
were purchased from Sigma-Aldrich (Castle Hill, NSW,
Australia). Methanol (Isocratic, HPLC grade) was acquired
from Scharlau Chemie (Gillman SA, Australia). Sulfuric acid
(98.0%) was sourced from Merck (Kilsyth, VIC, Australia).

2.2 Extraction of cannabinoids from hemp

Industrial-grade hemp samples, root hurd and leaf (Commins Stainless Manufacturing Farm, Whitton, NSW,
Australia) were washed under reverse osmosis water
(18.2 M), dried at 60C, ground into a powder, and passed
through a 368 m screen. Samples were prepared from 0.5 g
of the powder by solvent extraction with 300 mL of ethyl acetate for 1.5 h at 78C using a Soxhlet apparatus. The extract
was removed and the extraction repeated with another 300 mL
of solvent for 1 h. The two extracts were combined, evaporated
to dryness under vacuum at 40C, then reconstituted in ethyl
acetate (20 mL) and stored at <4C until use.

2.3 Preparation of acidic potassium permanganate

chemiluminescence reagent
The acidic potassium permanganate solution was prepared by
dissolving sodium polyphosphate (10 g/L) in distilled water,
adding potassium permanganate (1.9 mM) and adjusting the
pH to 2.50 with concentrated sulfuric acid followed by the
addition of sodium thiosulfate (1.0 mM) from a 0.1 M stock
solution.

2.4 Chromatography columns

The following columns were used for selectivity studies:
Eclipse XDB-C18 (Agilent, 4.60 mm 150 mm, 5 m particle diameter, Pd ), Luna 5u CN 100 (Phenomenex, 4.60 mm
150 mm, 5 m Pd ), Luna 5u HILIC 200 (Phenomenex,
4.60 mm 150 mm, 5 m Pd , Luna 5u NH2 100 (Phenomenex, 4.60 mm 150 mm, 5 m Pd ), Poroshell 120
EC-C18 (Agilent, 4.60 mm 50 mm, 2.7 m Pd ), Chromolith
RP18e (Merck, 4.60 mm 100 mm, 5 m Pd ).
Based on the data obtained from this study (discussed in
Sections 3.1 and 3.2), the Luna 5u CN 100 and Poroshell
www.jss-journal.com

2026

J. Pandohee et al.

J. Sep. Sci. 2015, 38, 20242032

Figure 1. Uni-dimensional separation of female leaf ethyl acetate extract on (A) HILIC, (B) NH2 , (C) CN, and (D) C18 showing selectivity of
each stationary phase toward the cannabinoids.

120 EC-C18 were used as first and second dimensions, respectively for the 2D analyses (see Section 2.7 for method
details).

2.5 Instrumentation
Chromatographic analysis was carried out on a 1200 Series HPLC system (Agilent Technologies, Mulgrave, VIC,
Australia), equipped with two quaternary pumps, solvent
degassers, an autosampler, and a continuously variablewavelength detector at 220 nm, followed by chemiluminescence detection. The two columns, a Luna 100 CN
(Phenomenex, 4.60 mm 150 mm, 5 m Pd ) for the first
dimension and Poroshell 120 EC-C18 (Agilent, 4.60 mm
50 mm, 2.7 m Pd ) for the second dimension were connected via an electronically controlled two-position ten-port
valve fitted with microelectric two-position valve actuators
that allowed alternate sampling of the eluent from the first dimension into the second dimension. Before use in the HPLC
system, all sample solutions and solvents (except HPLC-grade
solvents) were filtered through a 0.45 m nylon membrane.

2.7 2D configuration
Data of the selectivity study led to a linear gradient condition
being employed on both columns, starting from an initial mobile phase composition of 30% methanol, running to a final

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

mobile phase composition of 100% methanol at a rate of 8.6%

per min and initial mobile phase composition of 100% water,
running to a final phase composition of 100% methanol at a
rate of 33% per min for the first and second columns, respectively. The injection size was 100 L in the first dimension
and a flow rate of 1 mL/min was used for both dimensions.
The 2D-HPLC analysis involved transferring the eluent flow
stream from the first dimension (200 L) into the second
dimension (using a heart-cutting process) after the second
dimension separation was completed a second sample was
injected into the first dimension. The heart-cutting process
was repeated, each time sampling every second 200 L portion of the eluent from the first dimension 33 times, this
resulted in a total analysis time of 12 h.
Postcolumn acidic potassium permanganate chemiluminescence was generated using the custom-built manifold
previously described [20]. The chemiluminescence reagent,
propelled at a flow rate of 1.2 mL/min merged with the HPLC
eluent at a T-piece junction and the light emitted from the reacting mixture was detected with a custom-built flow-through
luminometer, which comprised a coiled flow cell, mounted
flush against the window of an electron tube photomultiplier
tube (model 9828SB, ETP) set at a constant voltage of 900 V
from a stable power supply (PM20D, ETP) by a voltage divider (C611, ETP). An analog to digital interface box (Agilent
Technologies) was used to convert the signal from the chemiluminescence detector. 2D data plotting and peak picking
were undertaken using custom written code in Mathematica
(version 8, Wolfram Research, Champaign, IL, USA).
www.jss-journal.com

J. Sep. Sci. 2015, 38, 20242032

Liquid Chromatography

2027

Figure 2. Optimized uni-dimensional separations of female leaf extract on (A) CN column with vertical gridlines representing fraction cuts
(B) Poroshell EC-C18 .

2.8 MS
An HPLC, electrospray time-of-flight (HPLCESI-TOF) mass
spectrometer (Agilent 6210 MSD, Agilent Technologies) was
used in a positive mode to identify the cannabinoids under the
following conditions: drying gas: nitrogen (7 mL/min, 350C);
nebulizer gas: nitrogen (16 psi); capillary voltage 4.0 kV; vaporizer temperature 350C, and cone voltage 60 V. An injection volume of 10 L was used. Reference masses used were
the ESI tuning mix G2412A (Agilent Technologies).

3 Results and discussion

3.1 Selectivity testing First dimension method
development
Selectivity testing was performed using HILIC, amino (NH2 ),
cyano (CN), and C18 columns to determine which stationary
phase was most suited for the analysis. (Fig. 1 panels AD).
The HILIC column (Fig. 1A displayed a limited number of
resolved peaks and the majority of the components eluted by
8 min. The NH2 column (Fig. 1B) utilized a large amount of
separation space; however, there were issues with the broad
peaks observed in particular with the variations in the noise
in the detection system. Nevertheless, judgements on the column efficiency could still be made. The CN column (Fig. 1C)
performed similarly to the NH2 column in terms of use of
separation space but better resolution of the components is
present (this takes into account the signal-to-noise differences
between these two runs). Separation was found to be improved by utilizing the unused space from 3 to 7 min. The
C18 (d) displayed the best use of separation.
Based on these findings, the CN column was deemed to
offer the best separation for the first dimension separation
that would be coupled with the C18 . Nonetheless, to use the
separation space of the CN column efficiently and reduce the
separation time, a series of additional tests were undertaken.
First, each consecutive separation involved an increment of
10% methanol in the initial mobile phase composition. The

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

initial mobile phase content range investigated was from 0 to

70% methanol over a gradient of 12 min followed by 3 min of
100% methanol. The separation starting with 30% methanol
(at a rate of 5.8% per min) was found to be the most appropriate in gaining in usage of the separation space while
simultaneously maintaining a fair resolution. Then, a series
of experiments with increasing speed of the previously optimized gradient was performed with their separation time
being reduced from 12 to 6 min Electronic Supporting Information (ESI1) It is important to note that at some point the
trade-off between separation effectiveness and speed of the
analysis needs to be carried out. In this case, the separation
with a gradient increasing at a rate of 8.75% methanol per
minute was chosen as the first dimension protocol as it offered a reasonable separation in a practical length of time, the
minor loss in resolution at this point is trivial since further
separations on the second dimension allowed a significant
gain in peak capacity.

3.2 Second dimension method development

A comprehensive offline approach, where the first dimension
is repeatedly performed for fraction collection and followed
by subsequent injection in the second dimension, allows high
separation power to be achieved [21]. However, a very long
analysis time is a cost for the substantial gain in peak capacity.
Consequently, a series of C18 columns was tested to identify
the best type of column for this separation (considering both
peak capacity and time constraint).
The performance of a Poroshell EC-C18 and a monolithic column were compared using a linear gradient at a
flow rate of 1 mL/min, starting with 0% methanol to 100%
methanol in 3 min followed by 3 min of 100% methanol. The
chromatograms obtained are shown in (ESI1). The Poroshell
EC-C18 was found to have a shorter analysis time and more
importantly a better resolution power than the monolith. An
advantage on the short column is that it provides faster analyses and in this case the resolution power was noteworthy
compared to the monolith. Further separations were carried
www.jss-journal.com

2028

J. Pandohee et al.

J. Sep. Sci. 2015, 38, 20242032

Figure 3. (A) 2D plot showing separation space used in a male leaf ethyl acetate separation (B) Geometric approach to factor analysis
results of five cannabis samples.

out to reduce the analysis time on the C18 but a negligible

gain in the usage of the separation was achieved. Therefore,
the original method, starting from 0% methanol to 100%
methanol, was chosen for the 2D-HPLC runs.
After the columns had been chosen from the selectivity
study, each separation was studied to ensure the columns
could be coupled in the 2D system in the most effective manner. The stationary phase is the most important consideration
of any separation, followed by the mobile phase composition and the effect it may have on separation selectivity. This
consideration is most important for sample transfer modulation to suppress band broadening and achieve peak focusing,
peak capacity, orthogonality, and sampling frequency [22].
The main challenge remains to maintain efficient coupling of
the columns while preserving the mobile phase/column compatibility. Limited solvent miscibility and/or solvent strength
problems may significantly affect the retention and selectivity
but also peak shapes and band broadening in the second dimension. When coupling two dimensions for 2D-HPLC analyses, the use of two completely different column-retention
mechanisms is highly beneficial for a gain in orthogonality,
hence maximizing the increase in peak capacity [16]. However, a major issue faced by several research groups [23, 24]
is solvent incompatibility of the nonpolar normal-phase solvents and the polar reversed-phase solvents. To remedy to
this, the same solvents (water/methanol gradient) were used
in both columns as previously described [25, 26].
In Fig. 2, the optimized separations of the CN first dimension (A) and the C18 second dimension (B) can be observed.
The chromatogram generated on the CN column shows the
complexity of the hemp extract and that accurate chemical


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

characterization of the sample cannot be obtained from the

analysis at this stage. Moreover, it is evident that the peak
capacity has been exceeded. The vertical gridlines on ESI1
represent the time at which fractions were collected and injected in the C18 second dimension. Although the C18 was
selective and showed some degree of resolution, the continuum of sample from 3 to 5 min shows that the peak capacity
has again been exceeded. ESI2 illustrates a heart cut fraction
from the first dimension at 2.2 min of female (A) and male
(B) leaves. It is very interesting to observe the strong similarity between both separations with the peak at 3 min as the
sole difference. These fraction-cuts also show the number of
unresolved peaks from the first dimension and the need for
2D-HPLC.
To assess the separation space occupied by the sample
components and orthogonality of the system [27, 28], a geometric approach to factor analysis (GAFA) was carried out
on each of the five cannabis sample. The 2D plot in Fig. 3
shows that the separation partially occupies the available
space, which is indicated by the spreading angle of 48.4.
The larger the spreading angles the more separation space
will be used by the sample components and a spreading angle of 90 would indicate full usage of the separation space.
While most spreading angles obtained for the cannabis separations were < 50, correlation coefficients also indicated that
the dimensions were correlated with a high correlation coefficient of up to 0.80 with the female leaf extract, which would
probably be explained by the use of RP separations in both
dimensions. It is of note that achieving a separation that uses
the 2D space effectively can be very challenging and involves
the usage of columns that are as dissimilar as possible. The

www.jss-journal.com

J. Sep. Sci. 2015, 38, 20242032

Liquid Chromatography

2029

Figure 4. (A) Typical 2D plot of hemp with identified cannabinoids a-CBV b-(CBCV,CBDV,CBLV), c-CBGV, d-CBN, e-(CBC,CBD, CBL), f-CBG,
g-CBE, h-CBT, i-CBNA, j-(CBCA,CBDA,CBLA), k-CBGA.

CN column comprises nitrile groups bonded to silica-based

column and has a high affinity for polar compounds. The C18
is a superficially porous micro-particulate column packing
with a solid silica core and a porous silica outer layer. Unlike
the CN column, the C18 reduces strong adsorption of basic
and highly polar compounds and these were the two most
differing column chemistries.

3.3 Tentative identification of the cannabinoids

using LC-MS
The known masses of the eight neutral cannabinoid classes,
together with that of the varin (cannabinoids with a propyl
group) and acidic cannabinoids (Fig. 3) were used to obtain
extracted ion chromatograms (EIC) from each sample. The
retention times of the assigned cannabinoids peaks from the
EIC of both the CN and short C18 chromatograms were subsequently determined and utilized as coordinates to estimate
the retention position of the cannabinoids in the 2D contour
plots, as shown in Fig. 4. The overall separation has been
divided in three regions: region 1 containing compounds not
highly retained on the CN column, region 2 containing compounds highly retained by the Poroshell EC-C18 , and region
3 compounds highly retained on both columns. Cannabinoids of interest were found to elute in region 2. Although
THC, CBC, CBD, and CBL have the same molecular weight
of 314.46 amu, identification of each component was made

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

with the support of previous LC-MS results [29].The sample being fiber-type cannabis would contain CBD as its most
abundant cannabinoid (assigned to peak with highest intensity) and the presence of THC would be negligible. CBC was
shown to elute after CBD, so it was assigned to the peak with
a retention time of 5 min, leaving CBL as the first peak of the
red EIC. It is not surprising to see a higher content of CBL
compared to CBC, as the latter degrades into CBL indicating
that the sample may have undergone some degradation. This
can also be established by the presence of only two of the
acid cannabinoids (CBCA/CBDA/CBLA). It is evident that
CBD has a major role in the antibacterial properties of hemp;
the presence of other cannabinoids such as CBL, CBC, and
perhaps CBGA and CBG suggests synergistic activity.
One of the key outcomes of this technology is the ability to generate in-depth chemical fingerprints of the complex
hemp extracts, the power of this can be observed through the
2D chromatograms of extracts from the major parts of the
hemp plant from both sexes. Figure 5 shows the 2D plots of
the chromatographic separations of male bark (A), male leaf
(C), female leaf (D), male hurd (E), female hurd (F) hemp
extract using chemiluminescence detection, and female leaf
(B) using UV absorbance detection. The white points represent peaks detected while the assigned letters correspond to
the respective cannabinoids as per Fig. 4.
At first sight, the general distribution of the cannabis
sample has a surprisingly high resemblance with that of
previously analyzed coffee samples by Mnatsakanyan and
www.jss-journal.com

2030

J. Pandohee et al.

J. Sep. Sci. 2015, 38, 20242032

Figure 5. 2D chemiluminescence contour plots with CN

first dimension and short C18
(50 mm) second dimension
(A) Male Bark; (B) Female Leaf
UV; (C) Male Leaf; (D) Female
Leaf; (E) Male Hurd; (F) Female
Hurd ESI1 Separation of female
leaf extract on (a) Monolithic
and (b) Short C18 column. Both
gradient increasing from 0 to
100% methanol in 3 min and
holding the organic phase for
the next 3 min. ESI2 Heart cut
of (a) female leaf and (b) male
leaf extracts at 2.2 min from CN
column to short C18 column.

co-workers [25]. In their analysis, the separation obtained

from a CN first dimension and 150 mm C18 second dimension were examined as three primary zones. The same approach was performed here with the three zones listed as
region 1, region 2, and region 3 on Fig. 4 within the 2D
separation plane. The two optimized dimensions showed distinct retention behavior within specific regions of the separation space yielding separation of compounds, which were

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

co-eluting beforehand. For example, the compounds in region 1 were poorly separated on the first dimension but have
had much greater retention and subsequent separation in
the second dimension. Region 2 corresponds to the broad
band that was observed in the original separations. While little separation of those compounds was observed on the C18
dimension; they were selectively separated on the CN phase.
It is important to note that the cannabinoids of interest were

www.jss-journal.com

Liquid Chromatography

J. Sep. Sci. 2015, 38, 20242032

2031

GC and uni-dimensional HPLC. The combination of 2D separation with both UV absorbance and enhanced acidic potassium permanganate chemiluminescence detection is a big
leap forward in profiling technologies for cannabinoids, offering comprehensive information about complex matrix characterization. This versatile separation technique can also be
used to address current uncertainties about complex cannabinoid chemistry for diverse purposes in a range of areas including taxonomical species identification, and in forensic
classification (fiber/drug type), identification, and individualization (source tracing). Further application of this procedure
to phytochemical and pharmacological studies may bring the
selectivity needed for the accurate identification of cannabinoid profiles.

present in the top right hand portion of this region (Fig. 4).
It is not surprising to observe most of the cannabinoids eluting in the same region as they essentially have the same
C21 terpenophenolic structure with minor changes in their
functional groups. Finally, the compounds in region 3 were
strongly retained on both phases.
Although the overall distribution of the cannabis constituents in each sample is similar, there were subtle differences in their chemical composition. The female hurd
had the least number of components; lower signal intensity was also noted. As expected, the leaves were found to
contain the greatest number of components with 125 peaks
detected in the female leaf and 123 peaks in the male leaf
using UV absorbance, 11 of which have been identified. Another distinct feature of the leaves profile is the presence of
CBCV/CBDV/CBLV (peak b), while CBC/CBD/CBL (peak e)
and CBN (peak d) were found in all the extracts.
A comparison between Fig. 5 (b) and (d) indicates substantial differences in the UV absorbance and chemiluminescence detection of the female leave separations. A major
difference between UV and CL detection is the presence of
high intensity peaks in the bottom left corner of the UV plot.
Although present in a large quantity in the cannabis sample,
those peaks are not one of the 17 cannabinoids presented in
Fig. 3(A) and are not of specific interest to this study. However it is important to note that this area of separation is commonly of importance in 2D-HPLC separations and would be
of significance in a chemical fingerprinting study of cannabis
samples.
The use of chemiluminescence detection offers detection
selectivity toward the reducing agents present in the extract
and allows a screening of the cannabinoids. We have previously shown that the signal intensity generated by an analyte
with acidic potassium permanganate chemiluminescence is
related to the antioxidant activity of the molecules [19]. The
initial driver for this proposal was to identify the key bioactive
constituents in the hemp extract and their ability to act as antibacterial agents. Many antioxidants and cannabinoids have
been shown to offer antibacterial activity and it is possible
that the cannabinoids that have been selectively determined
with the chemiluminescence reagent here may be directly displaying this bioactivity [30, 31]. The detection of fewer peaks
with chemiluminescence may thus provide the potential determination of chemically active compounds in the sample.
To fully realize this hypothesis however, a detailed study on
the structurefunction relationship and chemiluminescence
signal intensity would need to be performed in conjunction
with a cellular bioactivity test. This 2D LC technique is an
original approach to overcome frequently faced issues during cannabinoids analysis in both GC and uni-dimensional
HPLC [2].

[13] Rustichelli, C., Ferioli, V., Vezzalini, F., Rossi, M. C., Gamberini, G., Chromatographia 1996, 43, 129134.

4 Concluding remarks

[14] Stolker, A. A. M., Schoonhoven, J. V., Vries, A. J. D.,

Bobeldijk-Pastorova, I., Vaes, W. H. J., Berg, R. V. D.,
J Chromatogr A 2004, 1058, 143151.

X.C., N.B., and P.S. would like to acknowledge the support of Australian Research Council Linkage Project Funding
LP130100681. J.P. thanks RMIT University and the Commonwealth Scientific and Industrial Research Organisation (CSIRO)
for the award of a PhD scholarship (grant number 13/03634).
The authors have declared no conflict of interest.

5 References
[1] Hazekamp, A., Fischedick, J. T., De, M. L., Lubbe, A.,
Ruhaak, R. L., Comprehensive Natural Products II 2010,
3, 10331084.
[2] Raharjo, T. J., Verpoorte, R., Phytochem. Anal. 2004, 15,
7994.
[3] Elsohly, M. A., Slade, D., Life Sci. 2005, 78, 539548.
[4] Mechoulam, R., Ben-Zvi, Z., J. Chem. Soc. D 1969, 7,
343344.
[5] Turner, C. E., ElSohly, M. A., Boeren, E. G., J. Nat. Prod.
1980, 43, 169234.
[6] ElSohly, M. A., Marijuana and the Cannabinoids, Humana Press, New Jersey 2007.
[7] Hillig, K. W., Mahlberg, P. G., Am. J. Bot. 2004, 91, 966.
[8] Hazekamp, A., Proefschrift Universiteit Leiden 2007.
[9] Callaway, J. C., Euphytica 2004, 140, 6572.
[10] Dussy, F. E., Hamberg, C., Luginbuhl, M., Schwerzmann,
T., Briellmann, T. A., Forensic Sci. Int. 2005, 149, 310.
[11] Hazekamp, A., Peltenburg, A., Verpoorte, R., Giroud, C.,
J. Liq. Chromatogr. Relat. Technol. 2005, 28, 23612382.
[12] De Backer, B., Debrus, B., Lebrun, P., Theunis, L., Dubois,
N., Decock, L., Verstraete, A., Hubert, P., Charlier, C.,
J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2009,
877, 41154124.

[15] Guiochon, G., J Chromatogr A 2006, 1126, 649.

This 2D-HPLC technique is an original approach to overcome

frequently faced issues during cannabinoids analysis in both

C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

[16] Cohen, S. A., Schure, M. R., Multidimensional Liquid

Chromatography John Wiley & Sons, Hoboken 2008.

www.jss-journal.com

2032

J. Pandohee et al.

J. Sep. Sci. 2015, 38, 20242032

[17] Bassenasse, D.N., Conlan, X. A., Barnett, N. W., Stevenson, P. G., J. Sep. Sci. 2015, DOI:10.1002/jssc.201500054

[24] Jandera, P., Guiochon, G., J. Chromatogr. A 1991, 588,

114.

[18] Stevenson, P. G., Bassenasse, D.N., Barnett, N. W., Conlan, X. A., J. Sep. Sci. 2013, 36, 35033510

[25] Mnatsakanyan, M., Stevenson, P. G., Conlan, X. A., Francis, P. S., Goodie, T. A., McDermott, G. P., Barnett, N. W.,
Shalliker, R. A., Talanta 2010, 82, 13581363.

[19] Conlan, X. A., Stupka, N., McDermott, G. P., Barnett, N. W., Francis, P. S., Anal. Methods 2010, 2,
171173.

[26] Gray, M. J., Dennis, G. R., Slonecker, P. J., Shalliker, R.

A., J. Chromatogr. A 2004, 1041, 101110.

[20] Terry, J. M., Adcock, J. L., Olson, D. C., Wolcott, D. K.,

Schwanger, C., Hill, L. A., Barnett, N. W., Francis, P. S.,
Anal. Chem. 2008, 80, 98179821.

[27] Liu, Z. D. G. P., H., Anal. Chem. 1995, 67, 38403845.

[21] Fairchild, J. N., Horvath, K., Gooding, J. R., Campagna,

S. R., Guiochon, G., J. Chromatogr. A 2010, 1217, 8161
8166.

[29] Holland, B. J., Deakin University, Waurn Ponds 2011, p.

58.

[28] Stevenson, P. G. G., G., J. Chromatogr. A 2012, 1247,

5762.

[22] Cohen, S. A., Schure M. R., Multidimensional Liquid

Chromatography, Wiley & Sons, Hoboken 2008.

[30] Appendino, G., Gibbons, S., Giana, A., Pagani, A., Grassi,
G., Stavri, M., Smith, E., Rahman, M. M., J. Nat. Prod.
2008, 71, 14271430.

[23] Shalliker, R. A., Kavanagh, P. E., Russell, I. M., J. Chromatogr. A 1991, 558, 440445.

[31] Caturla, N., Vera-Samper, E., Villalain, J., Mateo, C. R.,

Micol, V., Free Radical Biol. Med. 2003, 34, 648662.


C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.jss-journal.com