β-Carotene autoxidation: Oxygen copolymerization, non-vitamin A products, and immunological activity

Article abstract of DOI:10.1139/cjc-2013-0494

Carotenoids are reported to have immunological effects independent of vitamin A activity. Although antioxidant activity has been suggested as a basis of action, the ability of carotenoids to autoxidize to numerous non-vitamin A products with immunological activity is an alternative yet to be fully explored. We have undertaken a systematic study of β-carotene autoxidation and tested the product mixture for immunological activity. Autoxidation proceeds predominantly by oxygen copolymerization, leading to a defined, reproducible product corresponding to net uptake of almost 8 molar equivalents of oxygen. The product, termed OxC-beta, empirical formula C₄₀H₆₀O ₁₅ versus C₄₀H₅₆ for β-carotene, contains more than 30% oxygen (w/w) and 85% β-carotene oxygen copolymers (w/w) as well as minor amounts of many C8-C18 norisoprenoid compounds. No vitamin A or higher molecular weight norisoprenoids are present. The predominance of polymeric products has not been reported previously. The polymer appears to be a less polymerized form of sporopollenin, a biopolymer found in exines of spores and pollen. Autoxidations of lycopene and canthaxanthin show a similar predominance of polymeric products. OxC-beta exhibits immunological activity in a PCR gene expression array, indicating that carotenoid oxidation produces non-vitamin A products with immunomodulatory potential.

Full text of DOI:10.1139/cjc-2013-0494

305
ARTICLE
␤-Carotene autoxidation: oxygen copolymerization,
non-vitamin A products, and immunological activity
Graham W. Burton, Janusz Daroszewski, James G. Nickerson, James B. Johnston, Trevor J. Mogg,
and Grigory B. Nikiforov
Abstract: Carotenoids are reported to have immunological effects independent of vitamin A activity. Although antioxidant
activity has been suggested as a basis of action, the ability of carotenoids to autoxidize to numerous non-vitamin A products with
immunological activity is an alternative yet to be fully explored. We have undertaken a systematic study of ␤-carotene autoxi-
dation and tested the product mixture for immunological activity. Autoxidation proceeds predominantly by oxygen copolymer-
ization, leading to a defined, reproducible product corresponding to net uptake of almost 8 molar equivalents of oxygen. The
product, termed OxC-beta, empirical formula C40H60O15 versus C40H56 for ␤-carotene, contains more than 30% oxygen (w/w) and
85% ␤-carotene oxygen copolymers (w/w) as well as minor amounts of many C8−C18 norisoprenoid compounds. No vitamin A or
higher molecular weight norisoprenoids are present. The predominance of polymeric products has not been reported previ-
ously. The polymer appears to be a less polymerized form of sporopollenin, a biopolymer found in exines of spores and pollen.
Autoxidations of lycopene and canthaxanthin show a similar predominance of polymeric products. OxC-beta exhibits immu-
nological activity in a PCR gene expression array, indicating that carotenoid oxidation produces non-vitamin A products with
immunomodulatory potential.
Key words: ␤-carotene autoxidation, oxygen copolymer, non-vitamin A oxidation products, immune function, gene expression.
Résumé : Les caroténoïdes sembleraient avoir des effets immunologiques indépendants de l’activité de la vitamine A. Bien qu’il
soit suggéré que l’activité oxydante est a` l’origine de ces effets, une autre explication a` considérer attentivement est l’aptitude
des caroténoïdes a` s’auto-oxyder sous forme de multiples produits différents de la vitamine A et dotés d’une activité immu-
nologiques. Nous avons entrepris l’étude systématique de l’auto-oxydation du ␤-carotène analysé l’activité immunologique du
mélange de produits obtenus. L’auto-oxydation se produit essentiellement par copolymérisation de l’oxygène, conduit a` la
formation d’un produit déterminé et reproductible et absorbe une quantité nette de presque 8 équivalents d’oxygène. Ce
produit, appelé OxC-beta, de formule empirique C40H60O15 (et non C40H56, comme le ␤-carotène), contient plus de 30 % d’oxygène
(pourcentage massique) et 85 % de copolymères de ␤-carotène oxygénés (pourcentage massique), ainsi que des quantités minimes
de nombreux norisoprénoïdes (C8 a` C18). Il n’y a pas de vitamine A ni de norisoprénoïdes de poids moléculaire élevé. La
prédominance de produits polymériques n’a pas été constatée auparavant. The polymère obtenu semble être une forme peu
polymérisée de sporopollénine, un polymère naturel que l’on trouve dans l’exine des spores et des pollens. L’auto-oxydation du
lycopène et de la canthaxanthine montre une prédominance similaire de produits polymériques. L’OxC-beta présente une
activité immunologique dans une matrice d’expression génique de PCR, ce qui montre que l’oxydation d’un caroténoïde
entraîne la formation de produits différents de la vitamine A et pourvus d’un potentiel immunomodulateur. [Traduit par la
Rédaction]
Mots-clés : auto-oxydation du ␤-carotène, copolymère oxygéné, produits d’oxydation différents de la vitamine A, fonctions
immunitaires, expression génique.
ing an antioxidant effect under conditions of low partial oxygen
Introduction
pressure.4
There are reports in the literature of modulation of immune
Another possibility that has yet to be seriously explored relates
function by various members of the carotenoid family.1–3 As only
to the products that form when carotenoids spontaneously oxi-
several carotenoids have vitamin A activity, the immunological
dize, in situ or externally, that themselves may modulate immune
influence of non-provitamin A carotenoids must occur via alter-
activity. Surprisingly little attention has been paid to the biologi-
native pathways. One suggested possibility is an antioxidant
cal activities of carotenoid non-vitamin A oxidation products.
action of the intact carotenoid,3 perhaps acting in a relatively
Given the involvement of oxidative processes in certain aspects of
nonspecific manner by the intercept and removal of putative
harmful immune system generated reactive oxygen species. Spe-
cifically, the highly conjugated unsaturation of the carotenoid
can stabilize a carbon-centered free radical, potentially contribut-
immune function (e.g., macrophage activity, inflammation) and
the susceptibility of carotenoids to oxidation, we have been inves-
tigating links between carotenoid oxidation, non-vitamin A oxi-
dation products, and immune modulation.
Received 29 October 2013. Accepted 24 January 2014.
G.W. Burton, J. Daroszewski, T.J. Mogg, and G.B. Nikiforov.* Avivagen Inc., 100 Sussex Drive, Ottawa, ON K1A 0R6, Canada.
J.G. Nickerson. Avivagen Inc., 550 University Ave., Charlottetown, PE C1A 4P3, Canada.
J.B. Johnston. National Research Council of Canada, 550 University Ave., Charlottetown, PE C1A 4P3, Canada.
Corresponding author: Graham W. Burton (e-mail: g.burton@avivagen.com).
*Present address: K.K. Air Liquide Laboratories, 28 Wadai, Tsukuba, Ibaraki 300-4247, Japan.
Can. J. Chem. 92: 305–316 (2014) dx.doi.org/10.1139/cjc-2013-0494
Published at www.nrcresearchpress.com/cjc on 3 February 2014.

306
Can. J. Chem. Vol. 92, 2014
Fig. 1. Autoxidation of ␤-carotene at 30 °C in benzene solution
saturated with oxygen gas in a closed system, as determined by
oxygen uptake expressed as a ratio of moles of oxygen consumed to
moles of initial ␤-carotene. See supplementary material for
experimental details.
The first phase of our study examined ␤-carotene autoxidation.
We reported that there is substantial uptake of oxygen at a rate
that is dependent upon oxygen partial pressure.5,6 In air, the rate
at its maximum is half that for pure oxygen at 1 atm. The reaction
is inhibited by the antioxidants butylated hydroxytoluene (BHT)
and ␣-tocopherol and accelerated by the thermal free radical
initiator azobisisobutyronitrile (AIBN), consistent with a radical-
mediated oxidation. Products include epoxides, dihydrofurans,
carbonyl compounds, carbon dioxide, a small amount of some
presumed polymeric material, traces of alcohols, and possibly
carboxylic acids. Early on, the main products identified are 5,6-
and 5,8-epoxy-␤-carotene, 15,15=-epoxy-␤-carotene, diepoxides, and a
series of ␤-apo-carotenals and ␤-carotenones. As the oxidation pro-
ceeds, the putative polymeric material and the carbonyl com-
pounds become more evident and the epoxides are degraded. In
the final phase of the oxidation, the longer chain ␤-apo-carotenals
are themselves oxidized to shorter chain carbonyl compounds,
including ␤-apo-13-carotenone, ␤-ionone, ␤-ionone-5,6-epoxide,
dihydroactinidiolide, and probably carboxylic acids.
In the next phase, reported here, a more comprehensive and
detailed study of the autoxidation has been carried out, especially
with regard to overall product composition, to establish the gen-
erality of the carotenoid autoxidation process and the potential
existence of immunologically active carotenoid oxidation prod-
ucts.
Fig. 2. GPC analysis of OxC-beta. UV absorbance was monitored at
220–400 nm. Vertical lines correspond to elution times of polyether
MW standards (selected markers shown). Column cutoff
approximately 10 000 Da.
Results
The kinetics of oxygen uptake by ␤-carotene in solution follows
an S-shaped curve, typical of an autocatalytic, free radical chain
reaction (Fig. 1). In the reaction in benzene under 1 atm of oxygen
at 30 °C, an endpoint is reached at 3–4 days when oxygen uptake
reaches a plateau and ␤-carotene is fully autoxidized. Oxygen up-
take is very high, with ␤-carotene consuming 7–8 molar equiva-
lents of molecular oxygen, resulting in a net increase in mass of
over 30% of the final oxidation product mixture, termed OxC-beta.
Oxidation continues beyond this point, albeit at a significantly
reduced rate and with only small changes to the composition of
the product.
As seen before, the reaction is accelerated by adding the ther-
mal free radical initiator AIBN5 and slows considerably in the
presence of 0.01–0.10 molar equivalents of either of the phenolic
antioxidants ␣-tocopherol or butylated hydroxyanisole (BHA). Ox-
ygen uptake during inhibition by either chain-breaking antioxi-
dant is linear and the rate is independent of the nature of the
antioxidant and its concentration (data not shown). Over a 6 day
period of inhibited oxidation, consumption of oxygen does not
exceed 10 mol% of ␤-carotene. Once the inhibitor is consumed, the
reaction follows its normal S-shaped kinetics.
For preparative purposes, the reaction is more conveniently
carried out in an open system by passing oxygen through a sus-
pension of ␤-carotene partially dissolved in ethyl acetate at room
temperature. The reaction is considerably slower, taking 10–14 days
to reach the endpoint, which is readily ascertained by an absor-
bance measurement at 380 nm.
Characterization of the chemically complex, fully autoxidized
product OxC-beta requires a combination of HPLC, GPC, 1H NMR,
FTIR, and UV-Vis techniques in conjunction with carbon, hydrogen,
and oxygen elemental analysis.
Elemental analysis gave values for carbon, hydrogen, and
oxygen of 61%, 7%, and 32%, respectively, in line with the extensive
uptake of oxygen and an overall increase in weight of more than
30% in the product. GPC analysis of OxC-beta indicated the exis-
tence of two broad product classes (Fig. 2) comprised respectively
of a large amount of higher molecular weight (MW) polymeric
compounds and low MW monomeric compounds. The chromato-
gram shows that OxC-beta has a broad, continuous MW distribu-
tion profile, ranging from less than 150 Da, based on identified
monomer compounds (see Fig. 9), to less than 10 000 Da, the cutoff
value of the GPC column. Intermediate MW calibrations were
provided using polyether GPC standards. The predominance of
higher MW compounds is attributed to the ease of formation of
oxygen copolymer compounds. The apex of the main, broad peak
corresponds to a MW estimated to be approximately 700–800 Da.
The monomer compounds are represented by the three peaks
superimposed on the low MW side of the main, broad peak. The
HPLC chromatogram of OxC-beta (Fig. 3) illustrates the chemical
complexity of the monomer fraction. Discrete peaks representing
numerous monomer compounds are superimposed on a large,
very broad asymmetric rise in the baseline that is attributed to
the polymer compounds. The baseline is similarly affected in the
HPLC of an isolated pure polymer fraction (data not shown). The
FTIR spectrum (Fig. 4) is notable for the strong absorptions at
approximately 3500 and 1700 cm−1, consistent with the presence
of hydroxyl and carbonyl groups, respectively, and at approxi-
mately 2900 and the 700–1500 cm−1 region, corresponding to C−H
Published by NRC Research Press

Burton et al.
307
Fig. 3. Reverse-phase HPLC of OxC-beta. UV absorbance was
monitored at 220–400 nm. Numbered peaks refer to norisoprenoid
compounds described in Fig. 9.
Fig. 5. 1H NMR of OxC-beta in CDCl3 (approximately 10 mg mL−1
)
with TMS as an internal reference.
Fig. 6. UV-Vis spectrum of a solution of OxC-beta in acetonitrile
(approximately 1 mg mL−1) using a quartz cell with 1 mm path
length.
Fig. 4. FTIR spectrum of OxC-beta obtained using NaCl disks and a
film cast from a CHCl3 solution (one drop of approximately
50 mg mL−1).
Of particular note, a 100-fold variation in the initial ␤-carotene con-
centration, from 0.2 to 20 mmol L−1, had no significant effect on the
product composition.
Only a few solvents are suitable for the reaction, either because
of the low solubility of ␤-carotene in most solvents or because of
the solvent itself chemically participating in the reaction. The
reaction has been carried out in benzene, carbon tetrachloride,
and ethyl acetate. Although autoxidation in carbon tetrachloride
yields results very similar to those obtained in benzene, elemental
analysis indicated that chlorine was present in the product. For
preparative purposes, ethyl acetate is an acceptable compromise.
Although the reaction takes significantly longer because of
␤-carotene’s limited solubility in this solvent, laboratory-scale
amounts of product (5–25 g) can be prepared with excellent repro-
ducibility.
When first prepared, OxC-beta is a yellow, glass-like metastable
solid that during storage at room temperature, over weeks
and months, gradually transforms into a dark brown, resin-like
form. However, there are only slight changes observed in the
HPLC, reflecting minor changes in a few compounds in the mono-
mer fraction. OxC-beta initially has a fruity smell that becomes
more pungent, resembling dried dates as it transforms into the
resinous form. OxC-beta is readily soluble in polar solvents (e.g.,
lower alcohols, lower ketones, lower esters, tetrahydrofuran, diox-
stretching and bending frequencies, respectively. The 1H NMR
spectrum (Fig. 5) is especially rich in aliphatic hydrogen absorp-
tions that provide distinctive fingerprint detail for helping char-
acterize the product. The UV-Vis spectrum (Fig. 6) is consistent
with the loss of much of the conjugated system of 11 double bonds
in ␤-carotene, being totally devoid of any absorption in the region
that is characteristic of ␤-carotene, namely the broad, strong ab-
sorption band at approximately 370–540 nm.
Each spectroscopic and chromatographic technique provides
distinctive and complementary information that, when com-
bined, makes it possible to determine the dependence of the
OxC-beta chemical composition on various reaction conditions,
including solvent, closed versus open reactor, reaction time, ini-
tial ␤-carotene concentration and purity, shaking, stirring, and
ambient light. In this way, it was found that the endpoint for
formation of OxC-beta is robust, in terms of a highly reproducible
product composition. The distinctive patterns of absorptions in
the 1H NMR and FTIR spectra were essentially invariant. GPC con-
firmed the predominance of the polymeric fraction and HPLC,
essentially a fingerprint representation of the many monomer
compounds, showed only minor variations under some conditions.
Published by NRC Research Press

308
Can. J. Chem. Vol. 92, 2014
Table 1. Proportions and elemental composition of fractions obtained by successive solvent precipitations of
OxC-beta.
Fraction
1
2
3
4
1–4a
Weight %
37
17
14
32
100
C
58.5
58.7
61.2
65.3
61.1
H
7.2
7.0
7.6
8.7
7.7
O
34.3
C40H58.7O17.6
C90H132O40
34.3
C40H56.8O17.5
C90H128O40
31.2
C40H59.2O15.3
C90H133O34
26.0
C40H63.5O12
C90H143O27
31.2
C40H60.1O15.3
C90H135O35
Empirical formulab
aEstimated by combining fractions 1–4.
bExpressed relative to C40H56 for ␤-carotene and the arbitrary C90 basis that has been used for sporopollenin.7,8
ane, dimethyl sulfoxide (DMSO)), partially soluble in hydrocarbons,
and insoluble in water.
The addition of almost eight molecules of oxygen to ␤-carotene
with its 11 conjugated double bonds implies that addition occurs
at most of the double bonds to form new C−O bonds. The associ-
ated disruption of the conjugated double bond system is reflected
in the UV-Vis absorption spectrum of OxC-beta (Fig. 6) compared
with ␤-carotene, showing a maximum absorption at a much lower
wavelength (approximately 240 nm with a shoulder at 280 nm).
The isolated polymeric material, like OxC-beta, is soluble in
numerous organic solvents, including methanol, acetone, acetoni-
trile, and tetrahydrofuran, but is only partially soluble in hy-
drocarbons and is insoluble in water. It can be stored indefinitely
at −20 °C.
Even after only partial reaction, it was evident by GPC that
the polymer was already a substantial part of the product. A
semiquantitative time course study of polymer formation was
conducted by GPC analysis of the reaction in benzene solvent.
Figure 7 shows that polymer was already present at the first time
point of 5.6 h, corresponding approximately to 13% of total oxygen
uptake, as estimated from the uptake curve in Fig. 1. The presence
of a significant amount of unreacted ␤-carotene was also evident,
as shown by the sharp GPC peak at approximately 8 min and
confirmed by UV-Vis spectroscopy. However, ␤-carotene was ab-
sent at subsequent time points, with the polymer clearly being
dominant and the presence of monomers becoming increasingly
apparent.
Modest changes occur in the composition of the monomer frac-
tion of OxC-beta samples incubated for 3 months at 40 °C. How-
ever, these changes are apparently inconsequential because the
biological efficacy of samples subjected to up to 9 months of ac-
celerated aging at 40 °C and 75% relative humidity is unaffected
(results to be reported elsewhere). OxC-beta appears to be stable
indefinitely when stored at −20 °C.
The polymer fraction
An initial, approximate estimate of the total amount of poly-
meric material formed was obtained using successive solvent pre-
cipitations, taking advantage of its relative insolubility in hexane.
The solid residues recovered from the first three precipitations
after solvent evaporation and washing and the residue obtained
from the evaporation of the combined supernatant and washings
of fraction 3 were each weighed and subjected to GPC and elemen-
tal analyses. The results in Table 1 indicate that from the amount
of precipitate obtained in fractions 1–3 alone, the polymer ac-
counts for at least 68% of the total product. GPC analysis of the
fractions indicated that fractions 1 and 2 were practically entirely
polymeric (data not shown), containing over 34% oxygen by weight
(Table 1), whereas fractions 3 and 4 contained increasing amounts
of monomer compounds, with an associated corresponding de-
cline in oxygen content.
The direct GPC−UV absorbance method for estimating the poly-
mer and monomer proportions in OxC-beta was used to estimate
the polymer as a fraction of the overall product at each time point.
(It was assumed that extinction coefficients of the polymer and
monomer fractions did not change significantly as the reaction
progressed.) The results, shown in Fig. 8, support the existence
of substantial polymer formation. The value of approximately
95% polymer at 5.6 h is artificially high because unreacted ␤-carotene
(MW 537) is inadvertently included in the polymer peak area cal-
culated for material exceeding the 300 Da MW cutoff. However,
␤-carotene is absent at subsequent time points and initially the
polymer value range is approximately 85%–88% over the period of
22–54 h, corresponding to approximately 67%–93% of total oxygen
uptake in Fig. 1. This value is comparable with the polymer con-
tent found for OxC-beta itself. Although the polymer remains
dominant throughout, it does gradually decline to approximately
80% at 70 and 77 h and eventually to approximately 72% at 166 h,
well past the endpoint of approximately 3 days. We attribute this
effect to a slow, induced decomposition of the polymer by oxida-
tion, with possible evolution of carbon dioxide, carbon monoxide,
and diacetyl (see Discussion).
The high percentage of oxygen in the first three fractions, rep-
resenting the majority of the product, is consistent with the di-
rectly measured high uptake of oxygen by ␤-carotene and the
more than 30% net increase in weight of the OxC-beta product. By
combining the elemental analysis results for fractions 1–4, the
overall carbon, hydrogen, and oxygen composition was deter-
mined to be 61%, 8%, and 31%, respectively, in good agreement
with the values of 61%, 7%, and 32%, respectively, obtained directly
for OxC-beta.
By expressing in Table 1 the elemental analysis results for each
fraction relative to the molecular formula of C40H56 for ␤-carotene, it
is evident that the amount of oxygen incorporated closely corrob-
orates the directly measured uptake of almost eight molecules of
oxygen per ␤-carotene molecule and supports oxygen copolymer-
ization being the predominant reaction.
The total amount of polymer present in OxC-beta was deter-
mined more precisely using preparative GPC to remove residual
polymer from a polymer-depleted fraction obtained after removal
of the bulk of the polymer by solvent precipitation and then
weighing, after evaporation, the residue of the separated fraction
containing only the monomer compounds. The contribution of
the monomer fraction to the total weight of OxC-beta was found
to be 14% 2%, indicating that the polymer represents approxi-
mately 86% of OxC-beta. Direct determination of the polymeric
and monomer fractions from relative GPC−UV absorption peak
areas of three samples taken from a different OxC-beta batch,
each run in duplicate, gave values of 85.3 1.3 and 14.7 1.3,
respectively, in close agreement with the gravimetric result.
Monomer products
The monomeric fraction and several of its most abundant mem-
bers were isolated using preparative HPLC. Analysis of prep-HPLC
fractions by GC−MS indicated that numerous compounds were
present over a range of concentrations. Further purification of
compounds was carried out where possible by silica gel chroma-
tography or HPLC to allow NMR analysis. The compounds were
Published by NRC Research Press

Burton et al.
309
Fig. 7. GPC−UV analyses at 255 nm of samples taken during
autoxidation of ␤-carotene with 100% oxygen in benzene at room
temperature. Each plot was analyzed to determine the relative peak
areas of polymer and monomer fractions to estimate the degree of
polymer formation at each time point (see Fig. 8).
Fig. 9. Norisoprenoid compounds identified in OxC-beta.
1: R1 = OH
12: R3 = O
O
R1
R3
Ac
2: R1 = OMe
13: R3
=
O
OH
O
3: R1
=
O
14: R4 = H2, R5 = Ac
R5
O
4a/b
O
15: R4 = H2, R5 = OAc
16: R4 = H2, R5 = CO2H
R4
17: R4 = O, R5 = Ac
R2
O
5: R2
6: R2
=
=
O
H
, R5 = Ac
18: R4
=
OH
O
O
O
19: R2 = CHO, R4 = H2
20: R2 = CHO, R4 = O
21: R2 = CO2H, R4 = H2
R2
O
O
7: R2
8: R2
=
=
R4
O
O
O
22
CO2H
9: R2
=
10: R2
11: R2
=
=
O
O
O
Fig. 8. Plot of estimated percent polymer formation (w/w) as a
function of extent of reaction, as determined by relative peak area
in corresponding GPC analyses (Fig. 7). Note that the value of the
first data point at 5.6 h is significantly overestimated because of
unavoidable inclusion of unreacted ␤-carotene in the polymer
fraction peak area. Also shown is a plot of the production of geronic
acid (1) during the reaction, expressed as mole percent of the initial
amount of ␤-carotene after adjusting by a factor of 2 for the
possibility that each ␤-carotene molecule can generate two
molecules of geronic acid.
OH
23
24
O
O
HPLC showed that compound 4 had a relatively large UV absorp-
tion compared with other monomer compounds (Fig. 3). It was
isolated from OxC-beta by prep-HPLC as a mixture of 2E,4E (4a)
and 2Z,4E (4b) isomers in the ratio of 7:3, as determined by
1H NMR. The geometry of the C4–C5 bond of both isomers was
determined by the trans coupling constants of H4 and H5 (J = 15.5
and 15.4 Hz, respectively). Nuclear Overhauser effect (NOE) spec-
troscopy analysis of the aldehyde protons was used to assign the
C2−C3 bonds. In 4a, an NOE was observed between H1 and H3,
whereas in 4b, the NOE was between H1 and H4. 4a has been
reported in the literature,11–13 but to the best of our knowledge, 4b
has not (the 2E,4Z isomer has been reported14). GC−MS of the
4a/4b mixture gave one peak that was identical to a reference
sample of 4a.
Compounds 8, 9, and 13 have not previously been identified as
␤-carotene autoxidation products and 8 and 13 appear to be new.
␤-Apo-13-carotenone-5,6-epoxide (11) was isolated from OxC-
1
beta and characterized by H NMR and GC−MS. There are two
reports of 11 being found as a product of ␤-carotene oxidation.15,16
Wu et al.15 tentatively identified 11, or its 5,8-epoxy analog, by
HPLC−MS, while Caris-Veyrat et al.16 identified it by HPLC−UV and
HPLC−MS.
No detectable amounts of any form of vitamin A or other cleav-
age compounds of the same or higher MW were found, consistent
with the very extensive degree of oxidation occurring in the au-
toxidation.
Compounds 1, 4, 5, 12, and 14 were the monomer products most
readily detectable by UV-Vis or LC−MS. Deuterated versions of
each compound served as internal standards to provide accurate
measures of the amounts in OxC-beta. The results, reported in
Table 2, show that geronic acid (1) is the most abundant monomer
product at 2.4% w/w of OxC-beta.
Geronic acid is extensively oxidized, containing 28% oxygen by
weight as a result of oxidative cleavage of two C−C bonds. It is of
interest to know if this compound, as a representative monomer,
is formed continuously during the reaction or at a later stage.
Figure 8 shows that there is a continuous release of 1 throughout
the reaction, slowly tapering off to a value of approximately 4% of
the initial ␤-carotene at 166 h, which is similar to the level present
in OxC-beta (Table 2). Expressed relative to the degree of reaction,
1
identified using GC−MS and (or) H NMR and almost all were
confirmed by reference to authentic samples obtained either
commercially or by independent synthesis. Figure 9 shows the
24 compounds that were isolated and identified. Of particular
note are compounds 1 (geronic acid) and 4a/4b (2-methyl-6-oxo-
2,4-heptadienal), which have not previously been identified as
␤-carotene autoxidation products. Compound 1 previously was
identified as a major product of the reaction of ␤-carotene with
ozone9 and later in a photo-oxygenation reaction.10 Compound 4
has been identified as a minor product of the autoxidation of
retinoic acid.11
Geronic acid (1) was identified by GC−MS and 1H NMR data and
confirmed with a reference sample prepared by ozonolysis of
␤-cyclocitral (19) according to the method of Strain.9
Published by NRC Research Press

310
Can. J. Chem. Vol. 92, 2014
Table 2. Amounts of selected cleavage compounds in OxC-beta as
measured by GC−MS using deuterated internal standards.
Fig. 10. FTIR spectra of OxC-lyc and OxC-beta. For comparison, plots
are provided of synthetic sporopollenin and sporopollenin from
Lycopodium clavatum obtained by transcribing the data from the plots
reported by Geisert et al.24 and replotting on the same wavenumber
scale.
Mol %
Weight %
of initial
Compound (m/z)a
of OxC-beta ␤-caroteneb
1
Geronic acid (154, 160)c
2.4 0.13
4.9
0.6
4a/4b 2-Methyl-6-oxo-2,4-heptadienald 0.23 0.03
(138, 142)
5
12
14
␤-Ionone-5,6-epoxide (123, 129)
Dihydroactinidiolide (180, 186)
␤-Ionone (177, 183)
1.2 0.02
1.1 0.01
0.80 0.02
5.7
2.0
2.1
1.5
11.1
Total
aValues in parentheses of m/z for corresponding nondeuterated and tetra- or
hexadeuterated ions, respectively.
bAdjusted for the fact that ␤-carotene can in principle yield two copies of each
of these cleavage products.
cIons of corresponding methyl esters.
dMixture of 2E,4E and 2Z,4E isomers.
as estimated by the oxygen uptake data in Fig. 1, the formation of
1 remains relatively constant at around 2% of reaction up to
around 90% oxygen uptake (approximately 50 h), after which time
formation of 1 continues at a declining rate, suggesting the com-
pound may be being released from the polymer.
Autoxidation of crystalline ␤-carotene without solvent
The spontaneous oxidation of pure, finely divided crystalline
␤-carotene in air for 8 weeks was considerably slower than the
reaction in solution, still being only partially complete after
40 days, as determined by FTIR and GPC analyses, compared with
reaction in benzene, which is complete in 3 days. However, the
reaction of the solid was in air, while that in benzene was in
homogeneous solution under 100% oxygen. The product was en-
tirely polymeric with an FTIR spectrum very similar to that of the
polymer of fraction 1 of OxC-beta, as was also found for the corre-
sponding GPC traces (data not shown).
Autoxidation of lycopene and canthaxanthin
Lycopene, partially dissolved in ethyl acetate, underwent full
autoxidation more rapidly than did ␤-carotene under the same
conditions, reaching an endpoint in 5 versus 10–14 days for
␤-carotene. A comparison of the FTIR spectra of OxC-lyc and OxC-
beta shows almost identical absorption profiles (Fig. 10). GPC
analysis indicates much similarity of the OxC-lyc product with
OxC-beta, confirming the predominance of polymeric product
(Fig. 11). The polymer appears to be present in even greater
amount and there is a larger MW spread toward the high MW end
of the fraction than for OxC-beta, with the peak occurring at a
correspondingly higher MW. Also, there was more white precipi-
tate formed during the oxidation, in contrast with ␤-carotene,
where only tiny amounts were sometimes seen. The white powder
was very difficult to dissolve in organic solvents, supporting the
notion that the polymer resembles sporopollenin8 (see Discus-
sion), which is insoluble in essentially all solvents.
Fig. 11. GPC of OxC-lyc. The peak at approximately 5 min
corresponds to polymers with MWs that are similar to or exceed the
column cutoff of approximately 10 000 Da.
The autoxidation of canthaxanthin occurred more slowly in
benzene than for ␤-carotene, taking 8 versus 3 days to reach the
endpoint. Again, there was a large consumption of oxygen, al-
though to a slightly lower extent (approximately 7 molar equiva-
lents versus almost 8 for ␤-carotene). GPC analysis of the oxidation
mixture showed strong similarities with OxC-beta, with a pre-
dominance of polymeric substances (data not shown).
Immunological activity
To gain insight into putative biological activity associated with
products present in fully oxidized ␤-carotene, studies were con-
ducted with OxC-beta in normal human fibroblasts. Evaluation of
cytotoxicity using a standard tetrazole reduction (MTT) assay re-
vealed few adverse effects following OxC-beta treatment over a
72 h time course and up to 15 ␮mol L−1 (expressed in ␤-carotene
equivalents; Fig. 12). Given that immune modulation is a property
that has been associated with carotenoids,1–3 which could actually
be effected through their oxidation products, a high-level evalua-
Published by NRC Research Press

Burton et al.
311
Fig. 12. Evaluation of the in vitro toxicity of OxC-beta. Human
fibroblasts were seeded in 96-well plates (3 × 105 cells per well) and
incubated with OxC-beta at the indicated concentrations for 24, 48,
or 72 h (expressed in ␤-carotene equivalents). Cell viability was
determined by standard tetrazole reduction assay (MTT). Cells
incubated with equivalent concentrations of vehicle (DMSO) alone
served as controls. Values were derived from the mean of
quadruplicate replicates and are expressed as fold changes relative
error compared with vehicle controls. Increases or decreases in MTT
activity exceeding 25% relative to controls are considered to be
indicative of an adverse response.
Table 3. Gene expression in normal human fibroblasts under LPS
challenge.
Fold
Gene ID
Name
regulation
Pathogen defense
IFNGR1
TLR4
TLR2
Interferon-␥ receptor
Toll-like receptor 4
Toll-like receptor 2
Cluster of differentiation 14
Lymphocyte antigen 96
191
68
63
12
CD14
LY96/MD2
Inflammation
CD55/DAF
TNFRSF1 A
CCL2/MCP-1
IL1B
TNF
IL1RAP
NOS2
7.4
Complement decay accelerating factor
TNF receptor
Monocyte chemotactic protein
Interleukin-1␤
Tumour necrosis factor
IL-1 receptor accessory protein
Nitric oxide synthase (inducible)
Interleukin-6
22
−4.4
−4.7
−6.6
−7.1
−8.1
−8.7
−11
IL-6
CASP1
Caspase-1
−16
Signal transduction
CHUK/IKKa
NF␬BIA/I␬B␣
IRF1
TRAF6
IRAK1
Inhibitor of NF␬B kinase
Inhibitor of NF␬B
27
15
−4.7
−5.4
−5.7
−7.6
Interferon regulatory factor 1
TNF receptor associated factor 6
IL-1 receptor associated kinase 1
Nuclear factor ␬-light-chain enhancer
of activated B-cells
NFKB2
tion of the ability of OxC-beta to influence the expression of genes
relevant to core immune responses was conducted using quanti-
tative, real-time PCR arrays (QRT-PCR). A lipopolysaccharide (LPS)
challenge model, in which cells were initially treated with
5 ␮mol L−1 OxC-beta and then exposed to LPS, was selected to
mimic infection conditions. Several trends emerged from the
gene expression profiles (Table 3). These included the capacity for
OxC-beta to (i) increase the expression of genes associated with
pathogen recognition and host defense, such as toll-like receptors
(TLRs) and their support molecules, (ii) decrease the expression of
inflammatory genes such as cytokines, cytokine receptors, and
effector proteins, and (iii) regulate the expression of signal trans-
duction molecules involved in TLR and inflammatory responses,
most notably inhibitors and mediators within the nuclear factor
␬-light chain enhancer of activated B-cells (NF␬B) pathway. Taken
together, these results suggested that OxC-beta could influence a
diverse array of cellular responses associated with pivotal innate
immune processes.
MYD88
Myeloid differentiation gene 88
−9.3
Surprisingly, the existence of the polymeric product appears to
have received little or no notice, despite the many studies of
␤-carotene oxidation carried out over many years. However, bio-
chemical and geochemical work carried out in the 1960s and 1970s
has led to plausible evidence that oxidative polymerization of
carotenoids is widespread in nature as a fundamental contributor
to the formation of the sporopollenin outer coat (exine) of many
species of pollen and spores17 (see below).
Polymeric carotenoid oxidation products would be predicted to
form on the basis of the known autoxidation behaviour of related
reactive unsaturated olefinic compounds. For example, the com-
prehensive studies carried out in the 1950s with styrene as a
model substrate showed that autoxidation proceeds via two com-
peting pathways involving (a) copolymerization addition of mo-
lecular oxygen to and (b) oxidative cleavage of the carbon double
bond, with a being the dominant pathway above about 25 Torr
partial pressure of oxygen.18 The autoxidation is self-initiated,
with copolymerization yielding styrene polyperoxide, having a
limiting formula of (C8H8O2)n, and with oxidative cleavage yield-
ing benzaldehyde and formaldehyde products in equal amounts.
Carotenoid autoxidation appears to proceed in a very similar
manner via the dual pathways of oxygen addition and oxidative
cleavage, respectively (Scheme 1). For ␤-carotene, the reaction is
self-initiated, as supported by the fact that neither of the chain-
breaking antioxidants ␣-tocopherol and BHA is able to fully in-
hibit oxidation. In fact, in the presence of either antioxidant,
oxidation proceeds at the same slow steady rate until the antiox-
idant is consumed. This, despite the superior antioxidant effi-
ciency of ␣-tocopherol,19 suggesting the antioxidant can readily
trap the peroxyl radicals formed during initiation. Initiation pre-
sumably occurs by direct, partially reversible reaction between
oxygen and ␤-carotene, which is supported by the observation of
trans-to-cis isomerization of ␤-carotene during the reaction.5
Although carotenoids appear to autoxidize in a manner closely
similar to styrene, there are more complex product outcomes
because the 11 double bonds in ␤-carotene versus 1 in styrene
afford multiple points for initiation and numerous subsequent
Discussion
The polymer fraction is the predominant product in the autox-
idation of ␤-carotene. The predominance is demonstrated by the
overall stoichiometry of the ␤-carotene reaction, consuming al-
most 8 equivalents of molecular oxygen (Fig. 1), the more than
30% increase in weight of the product, the GPC trace (Fig. 2), and
the elemental composition of OxC-beta itself (empirical for-
mula C40H60O15 versus C40H56 for ␤-carotene). This is further re-
flected in the composition of the first three fractions isolated from
OxC-beta by precipitation, which contain 31%–34% oxygen (Table 1).
The empirical formulae calculated for fractions 1 and 2 in Table 1
are consistent formally with addition of slightly more than eight
molecules of oxygen per ␤-carotene molecule.
Lycopene and canthaxanthin behave similarly, albeit oxidizing
at different rates (lycopene > ␤-carotene > canthaxanthin). Poly-
mer formation predominates in all three reactions and is illus-
trated in the GPC traces of OxC-lyc (Fig. 11) and OxC-beta (Fig. 2).
Furthermore, the FTIR spectra of OxC-beta and OxC-lyc are very
similar (Fig. 10), implying a corresponding degree of similarity of
the chemical makeup of the polymeric fractions. Together, this
evidence strongly suggests that carotenoids autoxidize predomi-
nantly by polymerization.
Published by NRC Research Press

312
Can. J. Chem. Vol. 92, 2014
Scheme 1. Carotenoid autoxidation yields two classes of products: carotenoid oxygen copolymers and cleavage compounds. Oxygen
copolymerization is the dominant pathway. Norisoprenoid products are shown as being formed by direct cleavage or indirectly by scission of
the polyperoxyl radical. A, ␤-carotene; B, canthaxanthin; C, lycopene.
R
A: R=
R
+O2
Initiation
-O2
B: R=
Carotenoid Epoxides and
Long-Chain Norisoprenoids
Carotenoid Peroxyl Diradical
O
O2
Propagation
(Addition)
O2
Cleavage
O2
C: R=
Polyperoxyl Radical
Short-Chain Norisoprenoids
Scission
Secondary Product
Termination
Oxygen Copolymers
(Polyperoxide)
Primary Product
branch points in the chain copolymerization reaction. Polymer
formation occurs early in the reaction (Fig. 8), rather than later,
as might occur by recombinations of ␤-carotene oxidation frag-
ments that would otherwise be potential precursors of cleavage
compounds.
from the polyperoxyl radical rather than by a direct oxidative
cleavage of the double bond.18
The polymer itself may be susceptible to oxidative decomposi-
tion. This would also alter the size of the polymer and is consistent
with the behaviour of sporopollenin, which is highly resistant to
all chemical and biochemical reagents except strong oxidizing
agents (see below).8 The very highly oxidized nature of some of the
norisoprenoid products (e.g., geronic acid, a product obtained
by ozonization of ␤-carotene9) implies that strong oxidizing con-
ditions exist during ␤-carotene autoxidation.
Many norisoprenoid products are present in minor amounts in
OxC-beta. The identified compounds are all short-chain noriso-
prenoids smaller in size than vitamin A (Fig. 9). A majority of the
compounds are well-known oxidation products.
Geronic acid (1), the most abundant norisoprenoid at 2.4%
(Table 2), together with the other four compounds, 4, 5, 12, and 14,
represent almost 6% (w/w) of the OxC-beta product, or about 40% of
the monomer fraction, which itself is 15% of OxC-beta. Given the
large number of identified and unidentified cleavage compounds
that make up the remaining 60% of cleavage products (i.e., 9% of
OxC-beta), we estimate each to be present at less than, or most
likely much less than, 1% levels.
The amount of ␤-carotene consumed in the formation of com-
pounds 1, 4, 5, 12, and 14 can be estimated. Compounds 1, 5, 12,
and 14 each result from oxidative attack on one of the two cyclo-
hexyl rings of ␤-carotene. The mole percent contribution relative
to the initial ␤-carotene of each compound is shown in Table 2.
Together, the five compounds account for approximately 11% of
the ␤-carotene consumed.
The longer chain compounds (i.e., vitamin A and larger com-
pounds) that have been reported to form early in the oxidation5,20
are notably absent in OxC-beta. These compounds themselves are
susceptible to further oxidation and once formed can feed back
into the continuing oxidation, undergoing further cleavage to
smaller norisoprenoids as well as participating in the copolymer-
ization reaction (Scheme 1).
Although we have focused on the nonvolatile (or relatively low
volatility) oxidation products in our investigation, we recognize
there are volatile products that can escape by gaseous evolution
during the reaction or by evaporation during workup. In this
regard, it is to be noted that there is a discrepancy in the weight
gain from oxygen incorporation observed in the formation of
OxC-beta (approximately 30%) and the gain that would be calcu-
lated from the OxC-beta empirical formula (approximately 46%)
(Table 1). Evolution of 1–2 molar equivalents of carbon dioxide
We surmise, by analogy with styrene,18 that initial oxygen ad-
dition to ␤-carotene leads to carotene polyperoxide structures,
starting with formation of a peroxyl diradical5 that in the ensuing
chain reaction undergoes further addition of oxygen, with con-
comitant cross-linking with other carotenoid molecules via per-
oxyl radical additions5 (Scheme 1). However, the growing, still
unsaturated, polyperoxyl radical faces competing steps of (a) fur-
ther addition of oxygen, (b) scission leading to simultaneous for-
mation of a cleavage fragment (norisoprenoid precursor) and an
associated shortened polyperoxyl radical, and (c) termination. The
preponderance of polymer product indicates that reaction a is
favoured over reaction b before c becomes the ultimate fate of the
polyperoxyl radical. However, because the most abundant poly-
mer species in the GPC trace of OxC-beta in Fig. 2 corresponds to
a MW of roughly 800 Da, the chain length for the most part must
be relatively short, producing a largely oligomeric product. The
scission reactions limit the “average” polymer size, which, while
mostly oligomeric for ␤-carotene (approximately 800 Da), is larger
for lycopene (cf. Figs. 2 and 11).
The high oxygen content of fractions 1 and 2 (Table 1; C40
H57–59O17.5) implies commitment to oxidation of most, but not all,
of the 11 available double bonds in ␤-carotene. Although a basic
polyperoxide structure, by analogy with styrene oxygen copoly-
merization,18 is consistent with a structure having a high oxygen
content, the substantial presence of carbonyl and hydroxyl
groups in the FTIR spectra of OxC-beta and OxC-lyc (Fig. 10) indi-
cates that the structure is more complex. The complexity of the
structure relative to a simple polyperoxide can be attributed to
thecompetingscissionreactionspartially“unzipping”thepolyper-
oxyl radical (Scheme 2).5 This is supported by the amount of poly-
mer reaching a maximum before the reaction endpoint, then
slowly declining, and the continued steady release of geronic acid
and presumably other monomers (Fig. 8). This process limits the
growth and therefore size of the radical: the scission reaction
ejects norisoprenoid moieties and other groups (e.g., carbon diox-
ide) while simultaneously transforming the radical structure,
leading ultimately to formation of carbonyl and hydroxyl groups
in various forms (e.g., keto, aldehyde, carboxyl, and peracid
groups).5 In support, it has been reported in the much less com-
plex autoxidation of styrene that cleavage products originate
Published by NRC Research Press

Burton et al.
313
Scheme 2. Illustration of a possible route for C=C bond cleavage to yield keto and aldehyde norisoprenoid products via scission (“unzipping”)
of the growing polyperoxide radical. The hatched lines in the polyperoxide radical indicate the bond scissions required for cleavage of the
carbon backbone. R represents any unsaturated moiety, including ␤-carotene, capable of participating in peroxyl radical addition.
O OR
O OR
O2
ROO
(peroxyl
radical)
O
O
β-Carotene
Propagation
R, O2
O
O OR
+
+
RO
O
aldehyde
ketone
O
O
Scission
+
Shortened polyperoxide
O
OR
Unzipping
during the reaction could account at least partly for the discrep-
ancy and would further require a corresponding additional up-
take of 1–2 molar equivalents of oxygen (i.e., 9–10 total) to help
maintain mass balance. Indeed, it was reported earlier that car-
bon dioxide is produced during autoxidation as well as when
OxC-beta is heated in vacuo.5
In a direct chemical test, reaction of ␤-carotene in dichlorometh-
ane with oxygen in the presence of boron trifluoride led to the
formation of a polymeric substance strongly resembling sporopol-
lenin.8 Later, it was established that the reaction also required UV
light,24 consistent with a free radical reaction. Synthetic, oxida-
tively polymerized carotenoids prepared in this manner show
much similarity with natural sporopollenins, as established by a
number of chemical criteria, including elemental analysis and IR
spectra8 and, more recently, solid-state 13C NMR spectroscopy.25
The elemental compositions of sporopollenins obtained from
numerous sources are distinguished by a high content of oxygen.8
Expressed in terms of the arbitrary C90 basis first used by Zetz-
sche,7 the range is encompassed by the composite formula
C90H115–158O10–44 with the number of oxygens heavily weighted to
the high end.8 In a direct comparison of native and synthetic
sporopollenins, the empirical formulae for Muco mucedo sporopol-
lenin, synthetic ␤-carotene polymer, Lilium henryi sporopollenin,
and the synthetic polymer of carotenoids and carotenoid esters
Other volatile products that are purged by the flow of oxygen
during the reaction or possibly evaporate during the workup also
will contribute to the oxygen discrepancy. Examples of products
are provided by studies into the generation of aromas and off-
flavours by autoxidation in ␤-carotene-containing foods.21 Of par-
ticular note in the present work, we have observed the presence of
a volatile yellow compound in the condensate of the ethyl acetate
reaction solvent recovered during rotary evaporation of the reac-
tion mixture. The colour faded within a day. Although no attempt
was made to identify the yellow compound, it may have been
diacetyl, which has been tentatively identified as a volatile prod-
uct of ␤-carotene autoxidation.21 Further, when the oxygen bub-
bled through the solution of ␤-carotene in ethyl acetate was fed to
another flask containing a ␤-carotene solution (i.e., two reactors
connected serially), the reaction in the second reactor proceeded
noticeably faster, as if volatile reactive species carried from the
first reactor accelerated the overall reaction in the second flask.
obtained by extraction of Lilium henryi were found to be C90H130O33
,
C90H130–132O27–32, C90H142O36, and C90H148O38, respectively.23 For
comparison, the formulae for OxC-beta and its main isolated poly-
mer fraction (fraction 1, Table 1) are C90H135O35 and C90H132O40
respectively.
,
In percentage terms, the ranges of carbon, hydrogen, and
oxygen values for sporopollenins from many species have been
reported as 56%–63%, 7%–9%, and 29%–37%, respectively.26 From
Table 1, it can be seen that the carbon, hydrogen, and oxygen
percentages for fractions 1–3 and OxC-beta all fall within these
ranges.
The IR spectra of sporopollenins are very similar and character-
ized by strong hydroxyl and carbonyl stretching frequencies at
3400–3500 and 1650–1750 cm−1, respectively.26,27 The IR spectra of
Lilium henryi pollen exines and the synthetic polymer obtained
by oxidative copolymerization of its extracted carotenoids and
carotenoid esters are very similar.24 Similarly, the IR spectra of
sporopollenin isolated from Lycopodium clavatum and the synthetic
oxidative polymer from ␤-carotene show strong similarity.24 The
latter comparison is reproduced in Fig. 10, which also shows the IR
spectra for OxC-beta and OxC-lyc. It is evident that the spectra are
all remarkably similar.
The natural connection
Carotenoid-derived norisoprenoids are widely distributed in na-
ture, for example, as ubiquitous constituents in plant-derived aro-
mas and flavours.22 Polymeric autoxidative carotenoid addition
products also would appear to be ubiquitous and yet have re-
ceived relatively scant attention. The natural existence of such
carotenoid products was proposed in 1968 by Brooks and Shaw,
who postulated that copolymerization of carotenoids and their
esters with oxygen is the underlying basis for the formation of
sporopollenin.17 This extremely robust substance is an integral,
structural polymeric component of the outer walls (exines) of
pollen, spores, and related species (including some microbial or-
ganisms), being recognized as critical for their protection and
longevity.
Shaw and co-workers observed that pollen grain development
in Lilium henryi is accompanied by formation of carotenoids (in-
cluding carotenoid esters) in major amounts in plant anthers.8
There also have been reports of increases in carotene content
during sexual reproduction in members of the Mucorales fungi
order.23
There is one significant difference in the polymeric product
from OxC-beta and that obtained by Shaw and co-workers using
ionic catalysts (e.g., BF3). Our results demonstrate that ␤-carotene
and other carotenoids spontaneously polymerize with ambient
Published by NRC Research Press

314
Can. J. Chem. Vol. 92, 2014
oxygen, and for ␤-carotene, this occurs whether it is present in
solution or as a solid. The OxC-beta polymer is much less polym-
erized, being largely oligomeric, which also makes it much more
tractable (e.g., solubility; however, note that the product of autox-
idation of lycopene was more heavily polymerized, producing
some amount of white precipitate that was insoluble in tetrahy-
drofuran and ethyl acetate and only slightly soluble in DMSO with
heating). However, if the reaction were to be conducted under the
emulsion polymerization conditions that Brooks and Shaw pro-
posed in their detailed scenario of sporopollenin formation dur-
ing sporogenesis,8,27 the probability of chain termination would
be decreased, leading to longer chain lengths and therefore a
more polymerized product. (The OxC-beta oligomer/polymer
likely corresponds to the pre-sporopollenin proposed by Brooks
and Shaw to be contained in orbicules.8) A peroxidase or
lipoxygenase could well accelerate the polymerization in situ by
increasing the number of peroxyl radicals. In the synthetic reac-
tion of Brooks et al. a catalyst such as BF3 could promote cross-
linking and thereby a much higher degree of polymerization.
Despite the plausibility of the Brooks and Shaw oxidative carot-
enoid polymer proposal, there remains considerable controversy
surrounding both the structure of sporopollenin and the nature
of its constituent monomer(s). To this day, these details are essen-
tially unresolved except at the broadest level. It appears that the
current view is that sporopollenin is not derived by oxidative
carotenoid polymerization. However, a significant difficulty that
we note with the alternative ideas is that they appear not to ade-
quately address the high oxygen content established for the ma-
jority of sporopollenin species analyzed. It is also possible that
there is more than one biochemical pathway to structures that
function similarly to sporopollenin.28 A useful summary of the
various ideas and the experimental support behind them has
been provided by Barrier.26
sporopollenin formation, as occurs, for example, in flowering
plants. The immunological activity of fully autoxidized ␤-carotene
confirms the idea that non-vitamin A oxidation products are ca-
pable of an activity that in the past has been attributed to the
parent carotenoid.
Experimental
Materials and general methods are described in the supplemen-
tary material (see “Supplementary material” section).
Reaction studies of ␤-carotene autoxidation
Oxygen uptake kinetics were measured directly using a closed
system with ␤-carotene fully dissolved in an oxygen-saturated ben-
zene solution. In a typical reaction, a flask containing a solution of
␤-carotene in benzene (50–100 mL, 20 mmol L−1) saturated with
pure oxygen maintained at 1 atm was connected to a simple,
custom-built apparatus and shaken in the dark in a water bath at
30 °C (further details of the procedure are provided in the supple-
mentary material). Oxygen uptake leveled off at approximately
72 h. The fully autoxidized ␤-carotene product OxC-beta was re-
covered by rotary evaporation of the solvent under reduced pres-
sure to give a resin-like yellow residue.
The effects upon the kinetics of the initial concentration of
␤-carotene (0.2, 2, and 20 mmol L−1) and of the chain-breaking
antioxidants ␣-tocopherol and BHA were each studied under the
same conditions.
Preparation of fully autoxidized ␤-carotene (OxC-beta)
Preparation was carried out on a larger scale more conveniently
using an open system by bubbling oxygen into a solution of
␤-carotene partially dissolved in ethyl acetate at room tempera-
ture (approximately 12.5 g L−1), taking no precautions to exclude
light. Reaction was considerably slower than in benzene, taking
10–14 days to reach the endpoint, which was determined by mea-
suring the absorbance at 380 nm on a sample of the reaction
solution using a quartz cell with 0.1 cm path length and the reac-
tion solvent as a reference. Details are provided in the supplemen-
tary material.
Immunological activity
The present work provides the first evidence for the existence of
non-vitamin A immunological activity within the OxC-beta prod-
uct mixture. The snapshot of gene expression associated with
OxC-beta treatment presented herein provides insight into poten-
tial events and mechanisms, but the pattern of activity suggests
that it is consistent with two key impacts. In the first, OxC-beta
upregulates the expression of genes encoding products that func-
tion in pathogen sensing and the detection of pathogen-
associated molecule patterns (PAMPs), including TLRs and other
proteins that act as cofactors for PAMP detection, such as CD14
(cluster of differentiation 14) and lymphocyte antigen 92 (LY96/
MD2). In the second, OxC-beta also appears to down-regulate the
expression of genes associated with the initiation and propaga-
tion of inflammatory responses, suggesting anti-inflammatory po-
tential. This effect was observed for inflammatory cytokines such
as tumour necrosis factor (TNF) and interleukin-1␤, but also for
cytokine receptors and other molecules that promote an inflam-
matory reaction. Of particular interest was the finding that OxC-
beta inhibited key signaling molecules and regulators of the NF␬B
pathway. This pathway plays pivotal roles in signaling events ini-
tiated by both the TLR system and inflammatory mediators such
as TNF, suggesting a potential common mechanism behind both
patterns of gene expression. Further evaluation of OxC-beta and
its role in modulating immune response is warranted, especially
the extension of these studies into immune cells. It also remains
to be determined if the activity originates within the polymer or
monomer fractions of OxC-beta.
Autoxidation of solid ␤-carotene
Finely divided, crystalline ␤-carotene was allowed to stand in air
over a period of up to 8 weeks in an open, clear glass vessel with no
attempt to exclude light.
Fully autoxidized lycopene (OxC-lyc)
A suspension of lycopene (100 mg) in ethyl acetate (10 mL, nom-
inally 19 mmol L−1) was fully autoxidized at room temperature
under an atmosphere of oxygen in essentially the same manner as
for the preparative-scale synthesis of OxC-beta. The reaction was
complete after 5 days, giving a yellow solution and some white
precipitate. The mixture was centrifuged and the liquid decanted.
The solvent from the liquid fraction was evaporated and the resi-
due dried under vacuum to give a yellow solid (approximately
100 mg). The white precipitate was dried under vacuum to provide
a white powder (approximately 40 mg) that was water insoluble
and had very little or no solubility in organic solvents.
Fully autoxidized canthaxanthin (OxC-can)
Canthaxanthin was oxidized in benzene under conditions iden-
tical to those initially used for ␤-carotene. A solution of canthax-
anthin in benzene (20 mmol L−1) saturated with oxygen was
incubated in a shaker bath, in the dark, at 30 °C under pure
oxygen at atmospheric pressure. After 8 days, approximately
7 molar equivalents of oxygen had been consumed. The solvent
was evaporated under reduced pressure to give a resin-like yellow
residue.
Conclusions
In the presence of oxygen, ␤-carotene and other carotenoids
spontaneously oxidize to form predominantly carotenoid−oxygen
copolymers and minor amounts of numerous norisoprenoids. The
findings support the oxidative carotenoid polymer hypothesis of
Published by NRC Research Press

Burton et al.
315
Note of caution
the expression of 84 genes involved in inflammation and the host
response to pathogen infection. Total RNA was extracted from
pooled cells using an RNeasy minikit (Qiagen) and cDNA gener-
ated using an RT2 First Strand synthesis kit (SABiosciences). The
array analysis was carried out on a MyiQ iCycler (Biorad) using a
RT2 qPCR Master Mix (SABiosciences) according to the manufac-
turer’s instructions. The arrays for each test condition were re-
peated once to evaluate the reproducibility of results. For analysis,
the average normalized difference in threshold cycle (⌬Ct) was
calculated relative to the selected housekeeping gene (glycerade-
hyde 3-phosphate dehydrogenase) and relative gene expression
(2−⌬Ct) determined for each gene in both OxC-beta-treated (test)
and vehicle control arrays. Fold up- or down-regulation was deter-
mined from the ratio of gene expression in test samples versus
control samples. Each array contained a positive PCR control and
controls for reverse transcriptase efficiency and the presence
of contaminating genomic DNA.
The oxidation reactions involve exposure of flammable solvents
to pure oxygen. Precautions must be taken to avoid any potential
sources of ignition. Additionally, OxC-beta contains peroxides and
may be prone to decomposition under certain conditions that
may generate localized heating. We have prepared more than a
hundred samples of OxC-beta with only one incident. On one
occasion, when the freshly prepared brittle product was rapidly
ground and compacted, the powder started to warm up and the
evolving heat resulted in a meltdown and decomposition of the
sample with evolution of gases. For this reason, only small sam-
ples of dry OxC-beta should be prepared and grinding should be
avoided.
Composition of OxC-beta: polymeric and monomeric
fractions
The relative proportions of the polymeric and monomeric frac-
tions were determined first by a gravimetric method, taking ad-
vantage of the low solubility of the polymer in hexane to isolate it
by precipitation. A second independent and more convenient
method used relative polymer and monomer UV absorbance peak
areas determined from an OxC-beta GPC trace. Details of both
methods are provided in the supplementary material.
Supplementary material
Supplementary material is available with the article through
the journal Web site at http://nrcresearchpress.com/doi/suppl/
10.1139/cjc-2013-0494.
Time course of polymer formation
Acknowledgements
Polymer formation during the oxidation reaction was followed
using the direct GPC−UV absorbance method for measuring poly-
mer and monomer proportions in OxC-beta. The formation of
geronic acid (compound 1, Fig. 9), the most abundant cleavage
compound in OxC-beta, also was measured as an index of
monomer formation. Details are provided in the supplemen-
tary material.
We thank the Atlantic Canada Opportunities Agency and the
National Research Council of Canada Industrial Research Assis-
tance Program for their financial support of this work.
References
(1) Bendich, A.; Olson, J. A. FASEB J. 1989, 3, 1927.
(2) Bendich, A. J. Nutr. 1989, 119, 112.
(3) Chew, B.; Park, J. In Carotenoids: Nutrition and Health; Britton, G.; Liaaen-
Jensen, S.; Pfander, H., Eds. Birkhäuser Verlag: Basel, 2009; Vol. 5,
pp. 363–382.
(4) Burton, G. W.; Ingold, K. U. Science 1984, 224, 569. doi:10.1126/science.
6710156.
(5) Mordi, R. C.; Walton, J. C.; Burton, G. W.; Hughes, L.; Ingold, K. U.;
Lindsay, D. A. Tetrahedron Lett. 1991, 32, 4203. doi:10.1016/S0040-4039(00)
79905-9.
Identification of main monomeric products and synthesis
of reference compounds
The supplementary material provides details on the identifica-
tion of cleavage compounds and the synthesis of reference stan-
dards. Compound numbers refer to chemical structures depicted
in Fig. 9.
(6) Mordi, R. C.; Walton, J. C.; Burton, G. W.; Hughes, L.; Ingold, K. U.;
Lindsay, D. A.; Moffatt, D. J. Tetrahedron 1993, 49, 911. doi:10.1016/S0040-
4020(01)80333-1.
(7) Zetzsche, F.; Kalt, P.; Liechti, J.; Ziegler, E. J. Prakt. Chem. 1937, 148, 267.
doi:10.1002/prac.19371480903.
(8) Shaw, G. In Sporopollenin; Brooks, J.; Grant, P. R.; Muir, M.; Gijzel, P. V.; Shaw,
G., Eds.; Academic Press: London and New York, 1971; pp. 305–348.
(9) Strain, H. H. J. Biol. Chem. 1933, 102, 137.
(10) Isoe, S.; Hyeon, S. B.; Katsumura, S.; Sakan, T. Tetrahedron Lett. 1972, 25, 2517.
doi:10.1016/S0040-4039(01)84863-2.
Synthesis of deuterium-labelled standards and quantitation
of selected norisoprenoids
Quantities of several of the norisoprenoids were measured by
GC−MS-based methods using deuterated analogs as internal
standards and nondeuterated reference standards. Deuterium-
labelled analogs of 1, 4a, 5, 12, and 14 were prepared as internal
standards. Full synthesis, characterization, and analytical details
are available in the supplementary material.
(11) Oyler, A. R.; Motto, M. G.; Naldi, R. E.; Facchine, K. L.; Hamburg, P. F.;
Burinsky, D. J.; Dunphy, R.; Cotter, M. L. Tetrahedron 1989, 45, 7679. doi:10.
1016/S0040-4020(01)85785-9.
(12) Clark, K. B.; Howard, J. A.; Oyler, A. R. J. Am. Chem. Soc. 1997, 119, 9560.
doi:10.1021/ja970774o.
(13) Bougioukou, D. J.; Smonou, I. Tetrahedron Lett. 2002, 43, 4511. doi:10.1016/
S0040-4039(02)00819-5.
(14) Brouwer, M. S.; Hulkenberg, A.; Kok, J. G. J.; van Moorselaar, R.;
Overbeek, W. R. M.; Wesselman, P. G. J. Recl. Trav. Chim. Pays-Bas 1979, 98,
316. doi:10.1002/recl.19790980513.
(15) Wu, Z.; Robinson, D. S.; Hughes, R. K.; Casey, R.; Hardy, D.; West, S. I. J. Agric.
Food Chem. 1999, 47, 4899. doi:10.1021/jf9901690.
(16) Caris-Veyrat, C.; Amiot, M.-J.; Ramasseul, R.; Marchon, J.-C. New J. Chem.
2001, 25, 203. doi:10.1039/b006975m.
(17) Brooks, J.; Shaw, G. Nature 1968, 219, 532. doi:10.1038/219532a0.
(18) Miller, A. A.; Mayo, F. R. J. Am. Chem. Soc. 1956, 78, 1017. doi:10.1021/
ja01586a042.
(19) Burton, G. W.; Ingold, K. U. J. Am. Chem. Soc. 1981, 103, 6472. doi:10.1021/
ja00411a035.
Cells and test conditions
Normal human fibroblasts derived from neonatal foreskin
(ATCC No. CRL-2097) used in this study were cultured in Dulbec-
co’s modified essential medium (DMEM) supplemented with
10% fetal bovine serum and antibiotics. OxC-beta was initially pre-
pared as a 50 mmol L−1 carotene-equivalent stock solution
(26.85 mg OxC-beta mL−1 DMSO) with subsequent working solu-
tions prepared by dilution in DMEM and filter sterilization. Vehi-
cle concentration never exceeded 0.25% in these studies. Cell
viability was assessed at 24, 48, and 72 h post-treatment with
OxC-beta using a commercial MTT assay (Roche) according to the
manufacturer’s instructions. For challenge studies, triplicate
wells of fibroblasts were pre-incubated with 5 ␮mol L−1 OxC-beta
or vehicle alone for 24 h prior to challenge with 100 ng mL−1 LPS
for an additional 24 h. Following LPS treatment, cells from each
condition were collected and pooled for analysis.
(20) Handelman, G. J.; van Kuijk, F. J. G. M.; Chatterjee, A.; Krinsky, N. I. Free
Radic. Biol. Med. 1991, 10, 427. doi:10.1016/0891-5849(91)90051-4.
(21) Chiba, N. M.Sc. thesis, Oregon State University, 1967. Available from
http://ir.library.oregonstate.edu/xmlui/bitstream/handle/1957/26796/
CHIBANAOKI1967.pdf?sequence=1.
Gene expression analysis
QRT-PCR analysis was conducted using an RT2 Profiler Innate &
Adaptive Immune Response array (SABiosciences) that profiled
Published by NRC Research Press

316
Can. J. Chem. Vol. 92, 2014
(22) Winterhalter, P.; Rouseff, R. L., Eds. Carotenoid-derived Aroma Compounds;
American Chemical Society: Washington, DC, 2001.
(23) Gooday, G.; Fawcett, P.; Green, D.; Shaw, G. Microbiology 1973, 74, 233. doi:
10.1099/00221287-74-2-233.
(24) Geisert, M.; Rose, T.; Zahn, R.; Shaw, G.; Brooks, J. Journal f. prakt. Chemie
1988, 330, 476. doi:10.1002/prac.19883300321.
(26) Barrier, S. Ph.D. thesis, University of Hull, 2008. Available from https://
hydra.hull.ac.uk/catalog/hull:6412.
(27) Brooks, J.; Shaw, G. In Pollen Development and Physiology; Heslop-Harrison, J.,
Ed.; Butterworths: London, 1971; pp. 99–114.
(28) de Azevedo Souza, C.; Kim, S. S.; Koch, S.; Kienow, L.; Schneider, K.;
McKim, S. M.; Haughn, G. W.; Kombrink, E.; Douglas, C. J. Plant Cell 2009, 21,
507. doi:10.1105/tpc.108.062513.
(25) Shaw, G.; Apperley, D. C. Grana 1996, 35, 125. doi:10.1080/00173
139609429483.
Published by NRC Research Press