Key words
Echinacea purpurea
- Asteraceae - macrophage activation - lipopolysaccharide - bacteria - bacterial load
- PCR method
Introduction
Clinical trials testing Echinacea ’s (Asteraceae) potential for preventing and/or treating the common cold are difficult
to interpret because they have used a wide variety of chemically ill-defined preparations.
A review [1 ] of clinical trials using Echinacea found that while many studies have reported some benefit in the treatment of common
colds, others have failed to show any efficacy. At present, one of the most problematic
factors for conducting such trials is in the selection of Echinacea material or extract type. This is due to the incomplete identification of therapeutically
relevant components that would be used to standardize these products. In addition,
despite decades of research, it is not known if Echinacea ’s main therapeutic action on colds and the flu is immunostimulatory, anti-inflammatory,
or a combination of both. In vitro [2 ], [3 ], [4 ] and animal research [5 ], [6 ], [7 ] indicates that this botanical exerts both therapeutic actions, and Echinacea alkylamides represent the major anti-inflammatory component. With respect to the
immune-enhancing potential of Echinacea , our research [8 ] suggests that bacterial Braun-type lipoproteins and lipopolysaccharides (LPS) were
responsible for over 97 % of the in vitro macrophage activation exhibited by extracts of Echinacea and seven other botanicals traditionally used to enhance immune function. The contribution
of bacterial components within extracts to macrophage activation was assessed using
agents that targeted bacterial Braun-type lipoproteins (lipoprotein lipase) and LPS
(polymyxin B). These biochemical approaches were also used to determine that variations
in the content of these two bacterial components were responsible for the up to a
200-fold difference in in vitro macrophage/monocyte activation potential exhibited by commercially diverse E. purpurea and E. angustifolia bulk material obtained from six North American suppliers [4 ]. It is our hypothesis that differences in bacterial content and/or bacterial type
are responsible for the substantial variation in the innate immune-enhancing activity
exhibited by this botanical.
The objective of the present study was to directly assess total bacterial load within
E. purpurea root and herb (aerial) samples, and to determine if differences in bacterial load
correlate with in vitro macrophage activity of the plant material. To estimate the total bacterial load,
we developed a PCR-based quantification method that circumvents the problems associated
with nonviable/nonculturable cells (which precludes using plate counts) or the coamplification
of plant mitochondrial or chloroplast DNA with the use of universal bacterial primers
(which precludes the use of qPCR).
Materials and Methods
Echinacea purpurea (L.) Moench plant material
Bulk root and herb (aerial) material for E. purpurea were obtained from the following six commercial suppliers: Frontier Natural Products
Co-op, lot number 50811.2331, Gaia Herbs, lot numbers 00033874 and 00033507, Glenbrook
Farms Herbs & Such, lot numbers not available, Mountain Rose Herbs, lot numbers 11987
and 12066, Richters, lot numbers 21283 and 21580, and Trout Lake Farm LLC, lot numbers
EPR-K6041-BCP and EPH-S2051-E3P.
Individual plants of E. purpurea were cultivated at the University of Mississippi. Root and herb parts were harvested
by washing extensively to remove soil and then immediately freeze-dried to prevent
postharvest growth or introduction of bacteria.
Voucher specimens for all E. purpurea plant samples were deposited in the NCNPR repository at the University of Mississippi
(voucher numbers NP1019, NP1048, NP2001–NP2006, NP2008–NP2012).
Alfalfa sprout germination and ampicillin treatment
Organic alfalfa (Medicago sativa L., Fabaceae) seeds (Johnnyʼs Selected Seeds, lot #25089) were surface sterilized
by soaking in 10 % chlorox/0.1 % Tween-20 for 12 min followed by four rinses with
sterile water. One hundred seeds were then transferred to each of two flasks containing
either 10 mL sterile water alone or 10 mL sterile water containing ampicillin (30 µg/mL).
Seeds were aseptically germinated for 7 days at ambient temperature under natural
light and, on days 2, 4, and 6, treatment solutions were replaced with fresh solutions.
Sprouts were freeze-dried after removal of hulls and unsprouted seeds.
Extraction of plant material for analysis of activity and content of LPS
Finely ground plant material (50 mg for E. purpurea samples or 20 mg for alfalfa samples) was extracted four times with 95 % ethanol
(1.0–1.5 mL fresh solvent added and incubated at 75 °C for 30–45 minutes for each
extraction) to remove anti-inflammatory components. Ethanol extracted plant material
was dried at 50–55 °C and then extracted with 0.5 mL of water containing 4 % SDS at
98 °C for 1 hour. Following removal of SDS using SDS-out reagent (Pierce) in the presence
of 1 % octylglucoside, crude extracts were assessed for activity and endotoxin levels.
Monocyte activation assay and limulus amebocyte lysate (LAL) assay
The THP-1 human monocyte cell line (American Type Culture Collection) was transfected
with a luciferase reporter gene construct containing two copies of NF-κB motif from
HIV/IgK, and samples were evaluated as described previously [9 ]. Monocyte activation exhibited by Echinacea extracts is reported as an EC50 value that represents the concentration of plant material (µg/mL) required to activate
cells to 50 % of maximal activation of NF-κB by LPS (10 µg/mL, E. coli , serotype 026:B6; Sigma).
The amount of LPS (bacterial endotoxin) in the extracts was determined using a limulus
amebocyte lysate (LAL) assay [Chromo-LAL test kit with Glucashield® (1 → 3)-β -D-Glucan Inhibiting Buffer] from Associates of Cape Cod, Inc. Data is reported as
endotoxin units (EU) per g of dried plant and represents the average of duplicate
determinations.
Determination of plant total bacterial load
DNA was extracted from a known mass (100 mg for E. purpurea samples or 20 mg for alfalfa samples) of ground plant material using PowerSoil DNA
isolation kits (Mo Bio), followed by a clean-up procedure (PowerClean; Mo Bio) to
remove potential PCR inhibitors. A portion of the bacterial 16S rRNA gene was amplified
from each purified sample using the primer pair 799 f (5′-AACMGGATTAGATACCCKG-3′)
and 1492 r (5′-GGTTACCTTGTTACGACTT-3′) described by Chelius and Triplett [10 ]. These primers exclude the coamplification of chloroplast DNA, and potentially yield
a bacterial product (approximately 735 bp) and a larger (1090 bp) mitochondrial product
when used to amplify DNA extracted from the plant material [10 ]. Amplifications occurred in 50 µL reactions comprised of a 2 mM MgCl2 PCR buffer, 0.2 mM deoxyribonucleoside triphosphates, 0.4 µM of each primer, 1.0 U
Taq polymerase, as well as 2 µL of sample DNA. Reactions consisted of 3 min at 95 °C,
followed by 30 cycles of 94 °C (20 s), 50 °C (40 s), 72 °C (40 s), and a final extension
of 72 °C (7 min). Amplification products (5 µL) were visualized on 1.2 % agarose gels
and the intensity of the bacterial band was determined using a Kodak Gel Logic 200
imaging system running Molecular Imaging Software 4.0 (Eastman Kodak).
A strain of Pseudomonas (NPC16) that we have previously isolated from freshly harvested E. purpurea (data not shown) was used as a reference bacterium to relate band intensity to bacterial
load. Pseudomonas NPC16 was cultured in trypticase soy broth at 22 °C and a sample was taken after
20 h (a time determined by plate counts to correspond to the late exponential phase).
The sample was serially diluted, and the bacterial load in the culture was determined
as the viable count in cells/mL. A second 20-h sample was taken, serially diluted,
and each dilution centrifuged (6000 g, 10 min) to pellet cells, and the pellets were
frozen for later DNA extraction. Results from the viable count allowed us to determine
the bacterial load in each of these cell pellets, which ranged from 1.0 × 102 to 1.0 × 109 cells. DNA was extracted and amplified from each pellet using the same procedure
as for the plant material. Intensity of each of the resulting amplification products
in agarose gels was similarly determined to develop a standard curve-relating band
intensity to bacterial load in cells.
Statistics
Simple linear regressions were used to relate bacterial load to PCR product band intensity
for the standard curve samples. These regressions were then used to determine bacterial
load in plant extracts based on their PCR product band intensity. Relationships between
bacterial load, monocyte activation (as EC50 ), and the amount of LPS (endotoxin) in E. purpurea samples were examined by pairwise linear regressions. Because the values for each
variable spanned a range of 2–5 orders of magnitude and were not normally distributed,
data for each variable were log10 transformed prior to regressions [11 ]. All transformations and regressions were conducted in Microsoft Excel 2007.
Results
Using molecular methods (PCR, qPCR) to estimate total bacterial cell load is problematic
since the use of universal 16S ribosomal bacterial primers is complicated by the coamplification
of chloroplast and mitochondrial DNA. To overcome this problem, we used bacterial
16S rDNA primers that do not amplify chloroplast DNA [10 ] to develop a PCR-based quantification method. DNA is extracted from dried E. purpurea plant material, followed by a secondary sample clean-up step. The clean-up step is
necessary to remove plant components such as polyphenols, polysaccharides, and humic
substances that interfere with DNA amplification (data not shown). Bacterial PCR products
(735 bp) are separated from mitochondrial products (1090 bp) on agarose gels and the
intensity of the bacterial band is then compared to band intensities of a standard
curve generated using PCR products from known bacterial numbers. [Fig. 1 a ] and [b ] provide a representative standard curve illustrating linearity (R2 = 0.9938) between 1.1 × 104 and 5.7 × 106 bacterial cells (concentration range of over two orders of magnitude). [Fig. 1 c ] shows an example that demonstrates the usefulness of this method to determine differences
in total bacterial load between bulk commercial E. purpurea plant samples that exhibit “low” and “high” monocyte stimulatory activity. An example
is also provided that illustrates detection of total bacterial load in E. purpurea herb (aerial) and root samples from plant material cultivated at the University of
Mississippi. These plant parts were freshly harvested, extensively washed to remove
soil and surface bacteria, and then immediately freeze-dried to prevent postharvest
growth or introduction of bacteria.
Fig. 1 An example illustrating bacterial load determination in dried Echinacea plant samples. A representative standard curve is presented in a and b . DNA was extracted from six known quantities of a bacterial isolate (Pseudomonas NPC16) and amplified with bacterial 16S rDNA primers; PCR products were visualized
on an agarose gel (a ). Band intensities were quantified (log10 , y-axis) and plotted against a known bacterial number (log10 , x-axis) (b ). c E. purpurea root and herb (aerial) plant parts were obtained as bulk raw material from commercial
growers and an individual plant cultivated at the University of Mississippi. “Low”
and “High” indicates activity of extracts from commercial plant material as measured
by NF-κB activation in THP-1 cells. DNA was extracted from dried plant material and
amplified with bacterial 16S rDNA primers. PCR products are visualized after separation
of bacterial bands “B” from mitochondrial bands “M” on an agarose gel.
To allow accurate detection of monocyte stimulatory components, the plant material
was first extracted four times with 95 % ethanol to remove alkylamides and other anti-inflammatory
substances that are inhibitors of monocyte activation. Ethanol extracted plant material
was then further extracted with 4 % SDS (98 °C, 1 h) to obtain the immune-enhancing
components. E. purpurea root and herb (aerial) bulk commercial plant material used in our previous study
[4 ] was also included in this research since they exhibit a wide range of in vitro monocyte/macrophage activity. The total bacterial load in plant samples was estimated
using the PCR-based quantification method described above, and levels ranged from
6.4 × 106 to 3.3 × 108 bacteria/g of dry plant material. The results presented in [Fig. 2 a ] show that the activity (NF-κB activation in THP-1 cells) exhibited by Echinacea extracts was correlated with the total load of bacterial cells in this plant material
(R2 = 0.54, p = 0.004). Likewise, [Fig. 2 b ] demonstrates that the content of LPS as determined by the LAL assay was also correlated
with the total bacterial load (R2 = 0.53, p = 0.005). In support of our previous research [4 ], [Fig. 2 c ] shows that there is a very strong dependence of monocyte stimulatory activity on
LPS concentration in E. purpurea plant extracts (R2 = 0.88, p = 0.000003).
Fig. 2 Correlations between bacterial load, THP-1 monocyte stimulatory activity, and LPS
levels in extracts from Echinacea. E. purpurea root and herb (aerial) plant parts were obtained as bulk raw material from six commercial
growers (11 samples) and an individual plant cultivated at the University of Mississippi
(herb and root). The plant material was extracted with 4 % SDS and the crude extracts
were assessed for activity in THP-1 monocytes transfected with an NF-κB luciferase
reporter plasmid (EC50 values of extracts are expressed in µg of plant material/mL, ranging between 3 and
1940). The LPS content of each extract, expressed as EU/g of dried plant material
(levels ranged between 10 and 121 160), was determined using the Chromo-LAL assay
with Glucashield® (1 → 3)-β -D-glucan inhibiting buffer. The total bacterial load in dried plant samples was estimated
as described in [Fig. 1 ]. Relationships are shown as pairwise linear regressions between macrophage stimulatory
extract activity and total bacterial load of plant material (a ), extract LPS levels and total bacterial load of plant material (b ), and macrophage stimulatory activity and LPS levels in plant extracts (c ). Regression analyses were performed on log10 transformed data using Microsoft Excel 2007.
We have previously used [7 ], [8 ] alfalfa sprouts as a model system since extracts of this plant material exhibit
high monocyte stimulatory activity and germination/harvest conditions are easily controllable.
High levels of activity are observed in extracts of sprouts germinated under aseptic
conditions using surface sterilized seeds and the appearance of this activity is suppressed
by treatment with ampicillin and other antibiotics [8 ]. The results presented in [Fig. 3 ] provide further evidence that the activity detected in this system is of bacterial
origin. In agreement with our previous research [8 ], extracts from sprouts grown in the presence of ampicillin for seven days exhibited
undetectable activity at all concentrations tested as compared to control sprouts.
[Fig. 3 ] insert shows that ampicillin treatment also substantially suppresses total bacterial load
during the germination of alfalfa seeds. Total bacterial load estimates in cells/g
dry weight of control sprouts versus ampicillin-treated sprouts were 7.7 × 108 and 6.3 × 106 , respectively (a difference of 122-fold). The content of LPS was undetectable in
both control sprouts and ampicillin-treated sprouts. Since only gram-negative bacteria
produce LPS, this result suggests that most of the bacteria present were gram-positive.
Ampicillin treatment did not influence the final biomass or outward appearance of
the sprouts (data not shown).
Fig. 3 Determination of total bacterial load and monocyte stimulatory activity of alfalfa
sprouts treated with ampicillin during germination. Alfalfa seeds were surface sterilized
and germinated in aseptic conditions in the presence (ctl + amp) or absence (ctl)
of 30 µg/mL ampicillin for 7 days. Freeze-dried sprouts were extracted with 4 % SDS
at 98 °C for 1 hr and the extracts were assessed for activity in THP-1 monocytes transfected
with an NF-κB luciferase reporter plasmid. Values are the average of duplicate determinations
± range. Extracts of unsprouted alfalfa seeds exhibited undetectable activity at all
concentrations tested (data not shown). Maximal inducing concentrations of LPS gives
100 RLA. (Insert ) DNA was extracted from freeze-dried sprouts, amplified with bacterial 16S rDNA primers,
and PCR products were visualized on an agarose gel (“B” and “M” signify bacterial
and mitochondrial bands, respectively).
Discussion
In the present study, we have developed a method to assess total bacterial load (culturable
and nonculturable/nonviable bacteria) within both freshly harvested plants and commercially
diverse dried bulk material. Using this method, we show that the activity (NF-κB activation
in THP-1 cells) and content of LPS exhibited by Echinacea extracts is strongly correlated with the estimated total bacterial load of these
plant samples.
In our initial experiments to estimate total bacterial load in Echinacea plant material, we used culture-dependent techniques. However, we found that less
than 1 % of the activity exhibited by extracts of the plant material was accounted
for by the activity of extracts from the culturable bacteria (data not shown). This
suggested that a culture-dependent method was problematic due to nonviable/nonculturable
bacteria and/or potential cell loss during processing (filtering of homogenate before
plating). Plant-derived immunostimulatory agents could not account for the observed
discrepancy in activity since we have shown that the majority of the activity from
this plant material is due to bacterial components [8 ]. Further experiments attempting to determine bacterial load using flow cytometry
were also ineffective because of plant mitochondria interference (data not shown).
Since molecular methods (PCR, qPCR) using standard universal bacterial primers are
complicated by the coamplification of chloroplast and mitochondrial DNA, we developed
a PCR quantification method using 16S rDNA primers that do not amplify chloroplast
DNA. Bacterial PCR products are separated from mitochondrial products on agarose gels
and the intensity of the bacterial band is then compared to band intensities of a
standard curve using known bacterial numbers (PCR amplified using the same primers).
This simple method has not been previously described and can be used to estimate total
bacterial load in both dried and freshly harvested plant material ([Fig. 1 ]).
We have previously reported [4 ] that the LAL-determined LPS content of Echinacea extracts was correlated with in vitro macrophage stimulatory activity. In the present study, these extracts were reanalyzed
using the Chromo-LAL test kit with Glucashield®. Glucashield® reagent blocks the contribution
of (1 → 3)-β -D-glucans in the LAL reaction. The use of this reagent is crucial during the analysis
of plant extracts since trace levels of glucan are present from cellulosic material
and from fungal/bacterial origin. When the Echinacea extracts were rerun in the presence of Glucashield®, endotoxin units decreased by
several folds for high LPS samples, whereas these values decreased by several orders
of magnitude for samples with low levels of LPS. Regression analysis using these new
endotoxin values resulted in a much stronger correlation ([Fig. 2 c ]) between monocyte stimulatory activity and LPS content of Echinacea extracts.
Our previous research [8 ] demonstrated that the majority of the in vitro innate immune cell activation potential of Echinacea plant material was derived from bacterial Braun-type lipoproteins and LPS. Differences
in the levels of these two bacterial components were also reported [4 ] to be responsible for the substantial variation (up to 200-fold) in activity from
E. purpurea material sourced from six major growers/commercial suppliers in North America. On
the basis of this data, we hypothesized that differences in bacterial content and/or
bacterial type are responsible for the substantial variation observed in the macrophage
activation potential exhibited by this plant material. In the current study, we show
that the differences in total bacterial load between plant samples are strongly correlated
with both the variation in activity ([Fig. 2 a ]) and the content of LPS ([Fig. 2 b ]) within the extracts. However, the R-squared values (0.53 and 0.54) for these regressions
indicate that total bacterial load does not fully account for the variation in content
of LPS and activity. This suggests that additional factors, such as type of bacteria,
are also important. We have begun to isolate and identify endophytic bacteria within
freshly harvested plants and have preliminary data showing that there is substantial
variation in the activity (more than 10 000-fold) from extracts of different bacterial
isolates (unpublished data). Future research is required to determine the extent that
the type of bacteria in addition to its prevalence within E. purpurea contributes to this plantʼs immune-enhancing action. It is unlikely that plant-derived
components contribute substantially to the observed variation in activity of the samples
analyzed in the current study since we have previously shown that over 97 % of the
activity for in vitro macrophage activation exhibited by extracts of Echinacea was due to bacterial components [8 ].
In ways that are analogous to the commensal bacteria naturally associated with animals,
most plants are colonized by bacterial endophytes that are typically present within
vascular tissue and intercellular spaces [12 ], [13 ]. These bacteria originate from soil around the roots or from the leaf surface [13 ], [14 ]. The isolation of bacteria from virtually every plant studied suggests that it is
likely that all plant species are colonized by endophytes [14 ]. Although it is our hypothesis that the bacterial load detected in E. purpurea plant material is predominantly from endophytes, it is possible that the high bacterial
load detected in some of the commercial bulk plant material samples in [Fig. 2 ] could originate from postharvest events. However, we have previously reported [4 ] that different postharvest drying conditions do not significantly influence the
content/activity of bacterial components within Echinacea plant samples. Furthermore, the current study indicates that it is possible to obtain
a high bacterial load (1.2 × 107 and 3.1 × 107 bacterial cells/g of dried root and herb, respectively) from a plant harvested under
controlled conditions at the University of Mississippi. This plant was washed extensively
and then immediately freeze-dried to prevent postharvest growth or introduction of
bacteria. Since washing removes most epiphytic bacteria (reviewed in [15 ]), it is likely that endophytes are predominantly responsible for the detected bacterial
load in this plant material.
In agreement with our previous research [8 ], we show that ampicillin suppresses the appearance of activity (NF-κB activation
of THP-1 cells) during the aseptic germination of alfalfa seeds. Furthermore, we show
([Fig. 3 insert ]) that ampicillin treatment dramatically decreases total bacterial load in alfalfa
sprouts during the 7-day germination period. These results demonstrate that endophytic
bacteria, residing within the seeds, are able to proliferate (by over 100 times) during
germination to give rise to the detected activity. The total bacterial load estimated
in the germinated control alfalfa seeds is similar to the levels detected in some
of the E. purpurea plant samples tested in [Fig. 2 ].
The current study, together with our previously published research [4 ], [8 ], provides strong evidence that bacteria within E. purpurea are a major source of immune-enhancing components (Braun-type lipoproteins and LPS)
in this popular botanical. Cells of the innate immune system detect Braun-type lipoproteins
and LPS through Toll-like receptor 2 (TLR2) and TLR4-dependent pathways, respectively
[16 ]. Research from our lab and others have demonstrated that oral consumption of TLR2
and TLR4 agonists can impact immune parameters in animal and human clinical studies.
Mice orally fed an extract from Spirulina platensis enriched for Braun-type lipoproteins exhibited augmented ex vivo production of IL-6 and IgA from Peyerʼs patch cells and IFN-γ from spleen cells [17 ]. In two separate human clinical trials, dietary supplementation with this extract
for one week enhanced NK cell activity an average of 40 and 54 % [18 ]. Additional research by Akao et al. reported that oral consumption of a hot water
Spirulina extract reduced tumor growth in mice through enhancement of NK cell activity via
a TLR2- and TLR4-dependent process [19 ]. The high levels of TLR2 (Braun-type lipoproteins) and TLR4 (LPS) agonists in Echinacea may contribute to the therapeutic action of this plant material. For example, these
bacterial components may explain the enhanced NK cell activity observed in mice by
dietary administration of Echinacea [6 ], [20 ].
Oral ingestion of Echinacea plant material containing a high bacterial load may have effects similar to those
reported in studies using probiotic bacteria. A recent study using human subjects
evaluated the in vivo intestinal mucosal gene expression profile six hours after oral administration of
heat-killed Lactobacillus plantarum [21 ]. Ingestion of dead bacteria was found to induce gene expression mainly involved
in innate and adaptive immune responses. These effects may be most pertinent to the
potential effect of bacteria within Echinacea since the majority of bacteria within dried plant material are nonviable.
The average bacterial load within the Echinacea plant material evaluated in this study is comparable to the daily therapeutic dose
of live bacteria used in studies evaluating the effect of probiotics on disease resistance.
We found that the average bacterial load in E. purpurea was 7 × 107 bacterial cells/g of dried plant material. Consuming a typical recommended dose [22 ] of 2.7–3.0 g Echinacea plant material/day, containing this average bacterial load would result in a daily
dose of about 2 × 108 bacterial cells. This bacterial load within Echinacea is comparable to the daily dose of probiotics (between 5 × 107 and 2 × 1010 CFU) that have been used in studies reporting statistically significant effects on
various parameters related to the common cold and flu infections (reviewed in [23 ]). For example, a recent double blind, placebo-controlled study on the consumption
of probiotics in 326 children (3–5 yrs old) reported statistically significant reductions
over placebo during a six-month period with respect to incidence of fever (73 %),
coughing (62 %), rhinorrhea (59 %), and antibiotic use (84 %) [24 ]. Additional research is required to evaluate whether the bacteria associated with
Echinacea plant material can impart similar therapeutic effects on colds and flu infections
to that observed in studies using oral consumption of live probiotics.
Acknowledgements
This research was partly funded by Grant Number R01AT007042 from the National Center
for Complementary and Alternative Medicines (NCCAM) and the Office of Dietary Supplements
(ODS). The contents of this manuscript are solely the responsibility of the authors
and do not necessarily represent the official views of the NCCAM, ODS, or the National
Institutes of Health. Additional funding of this research was also provided by a grant
from the USDA, Agricultural Research Service Specific Cooperative Agreement Nos. 58–6408–6–067
and 58–6408–1–603.
