HYPOTHESIS

Hypothesis

COX-1, COX-2, and COX-3 and the future treatment of chronic

inflammatory disease

Derek A Willoughby, Adrian R Moore, Paul R Colville-Nash

A new generation of non-steroidal anti-inflammatory drugs has been described that selectively targets the
inducible isoform of cyclo-oxygenase, cyclo-oxygenase 2 (COX-2). This isoform is expressed at sites of
inflammation, which has led to the speculation that its inhibition could provide all the benefits of current non-
steroidal anti-inflammatory drugs, but without their major side-effects on the gastrointestinal system (which are
due to inhibition of COX-1). We have shown that COX-2 (identified by use of specific antibodies) is induced during
the resolution of an inflammatory response, inhibition of COX-2 resulting in persistence of the inflammation due to
the prevention of the synthesis of a range of anti-inflammatory prostanoids. We propose that there is a third
isoform of this enzyme family, COX-3, a proposal that will have implication for the prescription of both existing and
new generation anti-inflammatory drugs, and might represent a new therapeutic target.

Vane1 described the mode of action of aspirin-like drugs
to be due to the inhibition of cyclo-oxygenase (COX).
The enzyme that was originally used was COX-1,
a constitutive isoform involved in a range of
physiological functions. Inhibition of gastric
constitutively expressed COX-1 resulted in most of
the unwanted side-effects seen in patients.
Non-steroidal inflammatory drugs (NSAIDs) treated
the symptoms of inflammatory disease without
affecting the underlying disease; indeed many
NSAIDs accelerated cartilage breakdown in
rheumatoid arthritis.2 Nevertheless, for many patients
this family of drugs improved quality of life, if the
patient escaped the unpleasant side-effects on the
gastrointestinal tract. With the more recent discovery of
a second isoform of COX, COX-2, the inducible
isoform of the prostaglandin-forming enzyme that is
expressed at sites of inflammation, we have been
eagerly awaiting the new class of NSAIDs. 3–7 Studies
that use in vitro methods show inhibitory activity
of putative anti-inflammatory drugs on either
COX-1 or COX-2.8 On the basis of these tests,
many of the original NSAIDs were found to have
inhibitory activity against both isoforms—ie, were
dual inhibitors of COX-1 and COX2. The more
selective COX-2 inhibitors had fewer gastrointestinal
side-effects and thus, after 100 years of NSAIDs,
we see the first major step forward—a new class of
NSAIDs giving pain relief and reduction of the
other classic signs of inflammation (heat, redness,
swelling). This new generation of anti-inflammatory
drugs, which are selective COX-2 inhibitors, are
predicted to be, and largely are, free of gastrointestinal
side-effects.
Lancet 2000; 355: 646–48
Department of Experimental Pathology, William Harvey Research
Institute, St Bartholomew’s and Royal London Hospital Schools of
Medicine & Dentistry, Charterhouse Square, London EC1M 6BQ,
UK (Prof D A Willoughby BSc, A R Moore PhD, P R Colville-Nash PhD)
Correspondence to: Prof D A Willoughby
(e-mail: d.a.willoughby@mds.qmw.ac.uk)
Is the discovery of COX-2 the end of this story?
Will these drugs stay with us like aspirin for another
century, or are there hopes of even more dramatic
therapies arising from this group of COX enzymes?
Many pharmaceutical companies use simple tests to
identify anti-inflammatory activity, and this point is
important to recognise. A classic example of a test,
which has been used for decades, is the carrageenan rat
paw oedema model. Putative anti-inflammatory drugs
are generally tested for their effect on paw swelling
at a single time point, 3 h after injection of the irritant. 9
This dependence on a brief window in time to predict
anti-inflammatory activity and mode of action could be
misleading. Thus, if concentrations of COX-2 are
raised at this time point, leading to the production
of proinflammatory prostanoids such as PGE2, but
before or after this time point there are other
enzyme systems and mediators active, these results
cannot reflect an overall mechanism or process.
Indeed, different types of experimental inflammation
are recognised as being brought about by a sequential
release of pharmacologically-active mediators such as
histamine, 5-hydroxytryptamine, kinins, and
prostaglandins.10 When the window of assessment is
at a very early stage, anti-inflammatory activity with
antihistamines is detected, this being the initial
mediator of the acute inflammatory response. 10 Such
a basis for the development of new therapeutics is,
thus, less than satisfactory. A further important factor
during the evolution of these models of inflammation is
the changing cellular migration pattern, initially
polymorphonuclear-cell dominated then switching to a
mononuclear dominated reaction. This switch in cell
populations is brought about by various chemotactic
factors and adhesion-molecule expression, plus the
apoptosis of the polymorphonuclear leucocytes leading
to the mononuclear cell dominance. Thus, to examine
inflamed tissue or exudates for enzyme activity at
different time points is almost akin to examining a
different tissue.
Willoughby and colleagues11 described the effect of
some selective COX-2 inhibitors and some dual

646 THE LANCET • Vol 355 • February 19, 2000


Arbitrary inflammatory response
Peak of
inflammation
Onset
Cox-1 Cox-1
+ +
Cox-2
Time
FIgure 1: Hypothetical time course showing the expression of
COX enzymes during the onset and resolution of an
inflammatory response.
(COX-1 and COX-2) inhibitors on carrageenan pleurisy
in the rat over a time course ranging from 0–48 h after
injection of the irritant. This investigation gave
surprising results in that the COX-2 inhibitors showed
anti-inflammatory activity early in the onset of the
inflammatory response, coincident with the expression
of COX-2 protein, as did the older dual inhibitors
such as indometacin. However, by 6 h the COX-2
inhibitors were without effect although the dual
inhibitors still showed efficacy. At this point the COX-2
protein, as shown by western blotting, was no
longer present.
Even more surprising, at 48 h—the time of near
complete resolution of inflammation in this model—
there was a second substantial increase in COX-2-
protein expression detected by western blotting.
This COX-2 protein did not produce the
proinflammatory prostaglandin PGE 2 in an ex vivo
biochemical assay with exogenously supplied
arachidonic acid, nor could this be detected at this time
in vivo. By contrast, anti-inflammatory prostaglandins
(PGD2 and PGF2␣) plus a member of the
cyclopentenone family (15deoxy∆12–14PGJ2) were
produced in vivo at this time.
When the same group12 suppressed this newly
formed COX by treatment either with COX-2 selective
inhibitor or the COX-2 dual inhibitor 24–48 h after
application of the irritant, these agents now showed
a different effect in that the inflammatory response
was not inhibited but continued and did not resolve, as
seen in the untreated animals.12 The recognition of
the enzyme protein was by western blot, by means
of an antibody that was raised to a small epitope on the
COX-2 protein. We postulate that this large expression
of protein is not COX-2 but may represent a third
isoform of COX, COX-3, which unlike COX-1 and
COX-2 does not produce proinflammatory prostanoids
but produces anti-inflammatory members of that
family (figure).
The expression of a catalytic variant of COX-2 or a
third COX enzyme has been shown in vitro in
transformed macrophages treated with high doses of
NSAIDs (apart from aspirin and paracetamol) for
48 h.13 This enzyme (again identified by an antibody
against COX-2) is sensitive to inhibition by paracetamol
but less sensitive to other NSAIDs compared with
COX-2 induced by lipopolysaccharide. However, this
scenario is clearly different from that which exists in the
resolution phases of inflammation in an untreated
THE LANCET • Vol 355 • February 19, 2000
HYPOTHESIS
animal. Furthermore, the COX-2 like protein in the
transformed macrophages produced PGE 2 in response
to exogenous arachidonic acid, unlike the COX-2-like
protein in samples from the 48 h pleurisy in which PGE 2
Resolution synthesis was not seen in response to exogenous
arachidonic acid nor could PGE2 be detected in the
pleural exudates at this time. Perhaps, therefore,
there exists yet another isoform of COX that might
be expressed under certain circumstances. One
? Cox-3 potential mechanism for the induction of this COX-2-
Heme-oxygenase 1 like protein in the transformed cells is via the activation
of a group of orphan nuclear transcription factors,
the peroxisome proliferator-activated receptors
(PPARs).14 This group of receptors might be activated
by various chemicals known as peroxisome proliferators,
including drugs such as the fibrate-based compound
clofibrate and NSAIDs14 an activation that induces
expression of a COX-2 like protein. Again, unlike
COX-2 induced by other stimuli, this COX-2 does not
produce cellular PGE2 although, again, exogenous
arachidonic acid will result in PGE2 synthesis.
Notably, one of the proposed anti-inflammatory
prostanoids proposed to be important in the
resolution of the carrageenan pleurisy is
15deoxy∆12–14PGJ. Inhibition of the production
of 15deoxy∆12–14PGJ2 results in an exacerbation of
the model, a process than can be reversed by its
exogenous replacement.12 This agonist is a
naturally occurring PPAR␥ agonist, and activation of
this system in macrophages is associated with anti-
inflammatory effects in macrophages. 15–17
If this hypothesis is true, expression of this third
inducible isoform of COX could result in the typical
periods of remission seen in many clinical cases of
chronic inflammatory disease such as rheumatoid
arthritis. If this hypothesis is further proved in man, an
urgent need for markers of disease activity would be
needed, thus making it possible to stop prescribing these
and similar drugs (dual or COX-2 inhibitors) at periods
when the patients should naturally be moving into
remission. Indeed, prescribing these drugs could retard
normal periods of remission.
Furthermore, if this third isoform, COX-3, is found to
have these anti-inflammatory effects in human beings,
could we promote the production of this enzyme and
use the naturally occurring factors it produced to
modulate the inflammatory response? The production of
cyclopentenone prostaglandins by COX-3 at 48 h in the
carrageenan pleurisy12 provide a link to the induction of
other endogenous anti-inflammatory mechanisms such
as the stress protein heme-oxygenase 1. 15 We have
previously proposed that modulation of this naturally
produced heat-shock protein could represent a unique
therapeutic mechanism. Experimentally, an increase in
concentration of this inducible enzyme has anti-
inflammatory activity that is equivalent to or greater
than that seen with steroids in inflammatory models. 18
This pathway has, as yet, also to be specifically targeted
therapeutically in human beings.
If these hypotheses are proven we hope that some of
the pharmaceutical companies would consider these
alternative therapeutic targets rather than following
what will inevitably be a rush to produce COX-2
inhibitors, thus producing a series of “me too”
therapeutic agents.
647
HYPOTHESIS
References
1 Vane JR. Towards a better aspirin. Nature 1994; 367:
215–16.
2 Rashad S, Revell P, Hemingway A, Low F, Rainsford K, Walker F.
Effect of non-steroidal anti-inflammatory drugs on the course of
osteoarthritis. Lancet 1989; ii: 519–22.
3 Masferrer JL. Zweifel BS, Seibert K, Needleman P. Selective
regulation of cellular cyclooxygenase by dexamethasone and
endotoxin in mice. J Clin Invest 1990; 86: 1375–79.
4 Xie WL, Chipman JG, Robertson DL, Erikson RL, Simmons DL.
Expression of a mitogen-responsive gene encoding prostaglandin
synthase is regulated by mRNA splicing. Proc Natl Acad Sci USA
1991; 88: 2692–96.
5 Hla T, Neilson K. Human cyclooxygenase-2 cDNA. Proc Natl
Acad Sci USA 1992; 89: 7384–88.
6 Masferrer JL, Seibert K, Zweifel B, Needleman P. Endogenous
glucocorticoids regulate an inducible cyclooxygenase enzyme. Proc
Natl Acad Sci USA 1992; 89: 3917–21.
7 Vane JR, Mitchell JA, Appleton I, et al. Inducible isoforms of
cyclooxygenase and nitric-oxide synthase in inflammation. Proc Natl
Acad Sci USA 1994; 91: 2046–50.
8 Warner TD, Giuliano F, Vojnovic I, Bukasa JA, Mitchell JA, Vane
JR. Nonsteroid drug selectivities for cyclo-oxygenase-1 rather than
cyclo-oxygenase-2 are associated with human gastrointestinal
toxicity: a full in vitro analysis. Proc Natl Acad Sci USA 1999; 96:
7563–68.
9 Winter CA, Risley EA, Nuss GW. Carrageenan-induced edema in
hind paw of the rat as an assay for anti-inflammatory drugs. Proc Soc
Exp Biol Med 1962; 111: 547.
10 Di Rosa M, Giroud JP, Willoughby DA. Studies on the mediators
of the acute inflammatory response induced in rats in
648
different sites by carrageenan and turpentine. J Pathol 1971;
104: 15–29.
11 Gilroy DW, Tomlinson A, Willoughby DA. Differential effects of
inhibitors of cyclooxygenase (cyclooxygenase 1 and cyclooxygenase
2) in acute inflammation. Eur J Pharmacol 1998; 355:
211–17.
12 Gilroy DW, Colville-Nash PR, Willis D, Chivers J, Paul-Clark MJ,
Willoughby DA. Inducible cyclooxygenase may have
anti-inflammatory properties. Nat Med 1999; 5: 698–701.
13 Simmons DL, Botting RM, Robertson PM, Madsen ML,
Vane JR. Induction of an acetaminophen-sensitive cyclooxygenase
with reduced sensitivity to nonsteroid anti-inflammatory drugs.
Proc Natl Acad Sci USA 1999; 96: 3275–80.
14 Lehmann JM, Lenhard JM, Oliver BB, Ringold GM,
Kliewer SA. Peroxisome proliferator-activated receptors alpha
and gamma are activated by indomethacin and other non-steroidal
anti-inflammatory drugs. J Biol Chem 1997; 272: 3406–10.
15 Colville-Nash PR, Qureshi S, Willis D, Willoughby DA. Inhibition
of inducible nitric oxide synthase by peroxisome proliferator-
activated receptor agonists: correlation with induction of heme
oxygenase 1. J Immunol 1998; 161: 978–84.
16 Ricote M, Li AC, Willson TM, Kelly CJ, Glass CK.
The peroxisome proliferator-activated receptor-gamma is
a negative regulator of macrophage activation. Nature 1998; 391:
79–82.
17 Jiang C, Ting AT, Seed B. PPAR-gamma agonists inhibit
production of monocyte inflammatory cytokines. Nature 1998;
391: 82–86.
18 Willis D, Moore AR, Frederick R, Willoughby DA. Heme
oxygenase: a novel target for the modulation of the inflammatory
response. Nat Med 1996; 2: 87–90.
THE LANCET • Vol 355 • February 19, 2000