Paracetamol reduces influenza-induced immunopathology in a mouse model of infection without compromising virus clearance or the generation of protective immunity

Paracetamol reduces influenza-inducedimmunopathology in a mouse model of infectionwithout compromising virus clearance or thegeneration of protective immunity

Sarah N Lauder,1 Philip R Taylor,1 Stephen R Clark,1 Rhys L Evans,2 James P Hindley,1

Kathryn Smart,1 Heather Leach,1 Emma J Kidd,2 Kenneth J Broadley,2

Simon A Jones,1 Matt P Wise,3 Andrew J Godkin,1 Valerie O’Donnell,1

Awen M Gallimore1

ABSTRACTBackground Seasonal influenza A infection affectsa significant cohort of the global population annually,resulting in considerable morbidity and mortality.Therapeutic strategies are of limited efficacy, and duringa pandemic outbreak would only be available toa minority of the global population. Over-the-countermedicines are routinely taken by individuals sufferingfrom influenza, but few studies have been conducted todetermine their effectiveness in reducing pulmonaryimmunopathology or the influence they exert upon thegeneration of protective immunity.Methods A mouse model of influenza infection wasutilised to assess the efficacy of paracetamol(acetaminophen) in reducing influenza-induced pathologyand to examine whether paracetamol affects generationof protective immunity.Results Administration (intraperitoneal) of paracetamolsignificantly decreased the infiltration of inflammatory cellsinto the airway spaces, reduced pulmonaryimmunopathology associated with acute infection andimproved the overall lung function of mice, withoutadversely affecting the induction of virus-specific adaptiveresponses. Mice treated with paracetamol exhibited anability to resist a second infection with heterologous viruscomparable with that of untreated mice.Conclusions Our results demonstrate that paracetamoldramatically reduces the morbidity associated withinfluenza but does not compromise the development ofadaptive immune responses. Overall, these data supportthe utility of paracetamol for reducing the clinicalsymptoms associated with influenza virus infection.

INTRODUCTIONInfluenza A viruses are responsible for seasonalepidemics accounting for up to 500 000 deaths perannum worldwide; however, the propensity of theseviruses to acquire new genetic material from animalreservoirs generates novel pandemic strains withgreater morbidity and mortality. The global impactof influenza A has been ever-present throughout the20th century with the Spanish (1918), Asian (1957)and Hong Kong (1968) influenza pandemics. The1918e1919 Spanish influenza pandemic was esti-mated to have infected one-third of the world’spopulation and accounted for 50e100 million

deaths.1 In early 2009 attention was drawn to theemergence of a new pandemic H1N1 influenza Astrain that quickly spread to all continents.2 Thera-peutic strategies for prevention and treatment ofinfluenza A viruses remains a complex problem sinceanimal reservoirs and antigenic shift can createpotentially pandemic strains. This makes it difficultand costly to eliminate influenza A or create effec-tive vaccines. Antiviral drugs are the first line ofdefence against new strains, but these are costly,resistance may occur and availability is limited.Many factors contribute to the pathophysiology

of influenza infection, including direct effects ofboth the virus and host factors such as themagnitude and type of immune response.3e6 Over-the-counter medicines that inhibit cyclo-oxygenase(COX) such as aspirin, paracetamol (acetamino-phen) and other non-steroidal anti-inflammatorydrugs (NSAIDs) are often taken to alleviate symp-toms of influenza virus infection. Despite theirubiquitous availability and use, the effects of thisgroup of drugs, including paracetamol, on influenzainfection have been poorly investigated. Althoughwidely used, the pharmacotherapeutic effects ofparacetamol remain controversial, with somestudies suggesting that the drug serves largely as ananalgesic while others report additional anti-inflammatory effects (reviewed in Bertolini et al7).In support of an anti-inflammatory role, Hinz andcolleagues demonstrated that paracetamol inhibitsthe activity of COX-2, an inducible enzyme thatgenerates prostaglandins (PGs), prostacyclinand thromboxane.8 The role of this enzyme ininfluenza infection has been investigated further inCOX-2-deficient mice which exhibit attenuatedpulmonary inflammation and delayed viral clear-ance.9 Administration of paracetamol mighttherefore be expected to limit the severity ofinfluenza-induced lung pathology, but this couldadversely affect patients by impeding antiviralresponses.10 Furthermore, perturbation of theinflammatory response could impair the develop-ment of protective immunity. Although influenza-specific immune responses have not been assessedin COX-2-deficient mice, Prymula et al havereported significantly lower antibody responses tosome vaccines in infants receiving prophylacticparacetamol compared with those that did not.11

< Additional methods andfigures are published onlineonly. To view these files pleasevisit the journal online (http://thorax.bmj.com).1Department of Infection,Immunity and Biochemistry,Cardiff University, Heath Park,Cardiff, UK2Division of Pharmacology,Cardiff University, Cathays Park,Cardiff, UK3Adult Critical Care, UniversityHospital of Wales, Cardiff, UK

Correspondence toSarah N Lauder, Department ofInfection, Immunity &Biochemistry, School ofMedicine, Heath Park, CardiffCF14 4XN, UK;[email protected]

VO and AMG contributedequally to this work.

Received 31 August 2010Accepted 9 December 2010Published Online First10 February 2011

This paper is freely availableonline under the BMJ Journalsunlocked scheme, see http://thorax.bmj.com/site/about/unlocked.xhtml

368 Thorax 2011;66:368e374. doi:10.1136/thx.2010.150318

Respiratory infection

In view of the widespread availability of paracetamol andNSAIDs amongst the general population, we sought to examinethe effects of paracetamol and a selective COX-2 inhibitor(celecoxib) in a mouse model of influenza virus infection. Wehypothesised that administration of these drugs followinginfection would diminish the immune response, delay virusclearance and reduce immunopathology in target tissues. Wefurther hypothesised that the adaptive immune response wouldbe adversely affected, thereby compromising the establishmentof protective immunity. In this study, each hypothesis wasaddressed by comparing innate and adaptive immune responses,virus clearance, lung pathology and the ability to resist challengewith heterologous influenza virus in both drug-treated anduntreated mice.

METHODSRefer to the Supplementary information for a detailed descrip-tion regarding the dose selection for celecoxib and paracetamolused during this study, flow cytometric analysis of pulmonarycells and lung intracellular cytokine staining, determination ofserum alanine aminotransferase (ALT), quantification ofpulmonary PGE2, in vivo lung function, in vitro virus titrationassays and the statistical tests used.

Murine influenza infection modelFemale C57BL/6 mice were sourced from Charles River UK andused at 8e10 weeks of age in experimental procedures incompliance with UK Home Office regulations.

Recombinant influenza A virus strains A-PR8-34 (PR8, H1N1)and A/X31 (X31, H3N2) were obtained from the NationalInstitute for Medical Research (NIMR; London, UK). Mice wereinfected intranasally with 50 plaque-forming units (pfu) of PR8in 50 ml of sterile phosphate-buffered saline (PBS), under lightanaesthesia for primary influenza infection. Mice were injecteddaily intraperitioneally with either 10 mg/kg of celecoxib or125 mg/kg paracetamol for 10 days as described in the Supple-mentary material. Memory recall responses to prior influenzainfection were examined by rechallenging previously infectedmice at 7e9 weeks after primary PR8 administration. Mice wereinfected intranasally with 200 pfu of X31 in 50 ml of sterile PBSunder light anaesthesia. Body weight was recorded daily duringthe course of infection until mice were sacrificed or until day 16for memory studies.

Mice were sacrificed at days 4, 7 and 10 during primaryinfection and day 4 after rechallenge, and bronchoalevolarlavage (BAL) was performed. Following BAL, lungs wereisolated and kept on ice. A detailed description regarding BALand lung cell isolation is provided in the Supplementalinformation.

HistopathologyAt day 7 postinfection lungs were perfused with 5 ml of PBS andfixed with 10% neutral buffered formalin solution. Followingfixation, lungs were embedded in paraffin and 5 mm sectionswere cut. Sections were stained with H&E and scored blinded.Lungs were scored using modified scoring criteria from Longhiet al12 and Ashcroft et al.13 The degree of leucocyte infiltration ofthe airway spaces and the formation of lymphoid aggregateswere independently scored on an increasing scale (0e3). Thesewere combined with the extent of perivascular lymphocyticaggregation within each section (0e10), to give a score rangingbetween 0 and 13. The mean score for each clinical parameterwas statistically compared between groups.

Determination of PR8-neutralising antibody titresSerum was isolated from mice at 7e9 weeks postinfection withPR8 and from naive uninfected mice. MadineDarby caninekidney (MDCK) cells at 50% confluence were plated into flat-bottomed 96-well plates and cultured for 48 h in the presence of720 pfu of PR8 and serum at a 1:8 dilution. Cell viability wascalculated using alamar blue (Biosource International, Camarillo,California, USA) according to the manufacturer ’s guidelines.The efficacy of serum from influenza-primed mice to neutralisePR8 was compared with the neutralising efficacy of serum frominfluenza-naive mice.

RESULTSCelecoxib and paracetamol diminish influenza-associatedpulmonary PGE2 and reduce the extent of influenza-inducedairway inflammationThe anti-inflammatory effects of NSAIDs such as celecoxib areattributed mainly to their ability to inhibit the activity ofCOX-2, an enzyme which catalyses the biosynthesis of PGsfrom arachidonic acid. COX-2 is upregulated during an inflam-matory response, thereby increasing levels of PGs, most notablyPGE2, a metabolite known to modulate immune cell activity. Todetermine whether PGE2 in the lungs of influenza-infected miceis COX-2 derived, lipids from lungs of infected mice and infectedmice administered either celecoxib or paracetamol were analysedusing liquid chromatography/tandem mass spectometry. Ourdata indicate that PGE2 is elevated in the lungs of influenza-infected animals compared to uninfected controls (763 +/�75ng/mg vs 408 +/� 128), and that this is inhibited byadministration of either celecoxib (3766122 ng/mg) or para-cetamol (3686111 ng/mg) (data not shown). These observa-tions indicate that elevated PGE2 following influenza infection isCOX-2 derived. A similar extent of PGE2 inhibition was observedwith celecoxib and paracetamol, in line with reports of theircomparable ability to inhibit PGE2 production in human blood.8

Infection of mice with influenza virus typically results inocclusion of the lung airway spaces by inflammatory cell infil-tration. To gain insight into the inflammatory response in theairways, a quantitative analysis of the cellular composition ofBAL fluid of infected mice was performed by flow cytometry. Asexpected, there was a highly significant increase in the numbersof cells recovered from BAL of infected mice compared withuninfected animals (figure 1AeF). Paracetamol administrationsignificantly reduced cell infiltration at days 4, 7 and 10 post-infection. Detailed analysis showed significantly fewer neutro-phils, inflammatory monocytes, eosinophils and natural killer(NK) cells in the BAL of paracetamol-treated animals comparedwith untreated mice (figure 1CeF). A similar pattern wasobserved in the celecoxib-treated group where a trend for fewercells was observed at each time point, although in this casea statistically significant difference was observed only at day 10.Overall, these data imply a protective effect of paracetamol inreducing the extent of influenza-induced immunopathology.

Airway function is improved in mice receiving celecoxib andparacetamolThe observation that inflammation is reduced by paracetamoladministration is further supported by an analysis of airwayfunction in the same groups of mice. A group of influenza-infected, untreated control animals and animals treated witheither paracetamol or celecoxib were subjected to airway func-tion tests performed by whole-body plethysmography. Thistechnique measures changes in air pressure as a result of

Thorax 2011;66:368e374. doi:10.1136/thx.2010.150318 369

Respiratory infection

** **

***

*

*

*

A

B C D

E F

G

PR8

Celecoxib + PR8

Paracetamol + PR8

Naive

PR8

Celecoxib + PR8

Paracetamol + PR8

Naive

0

100

200

Perc

enta

ge in

crea

se in

Pen

H

Day 4 Day 7 Day 100

500000

1000000

1500000

Cel

l Num

ber

PR8

Celecoxib + PR8

Paracetamol + PR8

Naive

Day 4 Day 7 Day 100

25000

50000

75000

100000

Day 4 Day 7 Day 100

50000

100000

150000

Day 4 Day 7 Day 100

100000

200000

300000Alevolar Macrophages Neutrophils Inflammatory Monocytes

Day 4 Day 7 Day 100

10000

20000

Day 4 Day 7 Day 100

10000

20000

30000

Eosinophils NK Cells

**

* ** ***

*** *

* **Cel

l Num

ber

Cel

l Num

ber

Figure 1 Effect of celecoxib or paracetamol upon influenza-induced airway inflammation and immunopathology. Bronchoalveolar lavage (BAL) fluidwas aspirated at day 4, 7 or 10 postinfection and the airway-infiltrating cells were isolated. The total numbers of cells (A) and the numbers of alveolarmacrophages (B), neutrophils (C), inflammatory monocytes (D), eosinophils (E) and natural killer (NK) cells (F) at the selected time points weredetermined by flow cytometry. (G) At day 7 postinfection, airway hyper-responsiveness to inhaled methacholine (10 mg/ml for 5 min) was determinedby measuring changes in Penh (enhanced pause) 5 min after methacholine challenge, expressed as a percentage of the premethacholine baseline.Changes in Penh were measured from naive mice, used as an uninfected control, and virus-infected mice treated or not with celecoxib or paracetamol.Results are expressed as the mean6SEM for each group, n¼6 mice/group. The significance of differences between drug-treated and untreated micewas assessed using one-way analysis of variance with TukeyeKramers post hoc multicomparison test (AeF) and between infected and uninfectedmice using an unpaired t test (G), *p<0.05; **p<0.01; ***p<0.001.

370 Thorax 2011;66:368e374. doi:10.1136/thx.2010.150318

Respiratory infection

inspiration and expiration. In these experiments, enhanced pause(Penh), a measurement of changes in expiratory effort, was usedas an indicator of airway obstruction. Specifically, control anddrug-treated animals infected 7 days previously with influenzavirus were challenged with aerosolised methacholine (10 mg/ml)and Penh was measured. There was a significant increase in thepeak Penh response in all infected mice compared with unin-fected controls; however, this difference was less pronounced incelecoxib- and particularly paracetamol-treated mice (figure 1G).

To assess if this improvement in lung function correspondedto a reduction in influenza-induced pulmonary immunopa-thology, lungs, recovered at day 7 postinfection, were assessedhistologically. Extensive leukocyte infiltration (figure 2A,C) andperivascular aggregation (figure 2B,D) were observed in allinfluenza-infected animals, when compared with uninfectedcontrol animals. These observations, while not significantlydifferent between drug-treated and untreated mice, whencombined with a notable improvement in lung function inparacetamol-treated mice, imply a marked reduction in influ-enza-associated pathology during the acute stages of infection.

The effects of celecoxib or paracetamol on influenza-inducedliver dysfunctionInfluenza infection of humans can be associated with anincrease in serum ALT levels and a transient and self-limitinghepatitis.14 Studies of mouse models of influenza infectionindicate that liver injury during influenza infection is notattributable to influenza infection of the liver but ratherrepresents collateral damage caused by antigen-independentsequestration of immune cells, most notably influenza-specificCD8+ Tcells in the liver. We sought to determine whether ALTlevels were increased in the serum of influenza-infected mice atday 7 postinfection and whether this was reduced by admin-istration of paracetamol or celecoxib. The results, shown infigure 3A, indicate that both drugs virtually abolished influ-enza-induced increases in ALT levels. Weight loss is alsocommonly used as an indicator of the severity of influenza-induced pathology. However, we observed no change in weightloss following influenza infection as a consequence of celecoxibor paracetamol administration (figure 3B and Supplementaryfigure 1).

200um 200um 200um 200um

CD

0

1

2

3

Deg

ree

of L

euko

cyte

Infil

tratio

n

0

5

10

Deg

ree

of P

eriv

ascu

lar A

ggre

gate***

***

nsns

PR8Celecoxib + PR8Paracetamol + PR8Naive

eviaN8RP Celecoxib + PR8 Paracetamol + PR8A

B

Figure 2 Histological analysis of pulmonary sections from treated and untreated mice postinfection with influenza virus. Lungs were harvested frommice at day 7 postinfection and, following perfusion, were formalin fixed, embedded in paraffin wax, sectioned and stained with H&E. (A and B)Representative sections depicting pulmonary leukocytes from influenza-infected, treated and untreated mice were compared with uninfected mice(original magnification310 (A)) Arrowheads highlight areas of lymphocytic perivascular aggregate (original magnification320 (B)). The meanscore6SEM for each group is given, n¼6 mice/group for the severity of influenza infiltrate (C) and degree of perivascular lymphoid aggregate (D). Thesignificance of differences between influenza-infected mice and uninfected mice was determined using one-way analysis of variance withTukeyeKramers multicomparison test, *p<0.05; **p<0.01.

Thorax 2011;66:368e374. doi:10.1136/thx.2010.150318 371

Respiratory infection

Virus clearance is unimpaired in celecoxib- or paracetamol-treated miceAs paracetamol and, to a lesser extent, celecoxib reduces themagnitude of the inflammatory response to influenza virus,

we hypothesised that viral clearance would be delayed in theseanimals compared with non-treated controls. Viral titres in thelungs of influenza-infected mice declined on days 7 and 10compared with day 4 postinfection. Surprisingly we observed nosignificant differences in viral titres among the groups at any ofthe time points analysed (figure 4).

Protective immunity is not impaired in mice treated withparacetamol or celecoxib during primary infection withinfluenza virusA reduction in influenza-induced innate immune responses asa consequence of celecoxib or paracetamol treatment could alterthe magnitude of the virus-specific adaptive immunity. Toinvestigate this, interferon g (IFNg)- and tumour necrosis factora (TNFa)-producing CD4+ and CD8+ T cells in the lungs weremeasured at day 10 postinfection. The number of cytokine-producing cells was, as expected, elevated as a result of theinfection, but there was no statistically significant differencebetween influenza-infected groups treated or not with drugs

A

B

0

250

500

AL

T U

/m

l

*** *** ***

PR8Celecoxib + PR8Paracetamol + PR8

Naive

0 4 8 12 1670

80

90

100

110

Days Post Infection

Pe

rc

en

ta

ge

o

f S

ta

rtin

g W

eig

ht PR8

Celecoxib + PR8

Paracetamol + PR8

Naive

Figure 3 Effect of cyclo-oxygenase 2 (COX-2) inhibition upon clinicalparameters of primary influenza infection. (A) Serum collected from miceat day 7 postinfection was used to determine circulating alanineaminotransferase (ALT) levels during acute influenza infection. Resultsare expressed as the mean6SEM, n¼6e16 mice/group. (B) Post-PR8infection mice were weighed daily for 16 days, and changes in bodyweight were expressed as a percentage of their starting weight. Eachdata point represents the mean6SEM, n¼6 mice/group. Thesignificance of differences between drug-treated and untreated micewas assessed using one-way analysis of variance with post hocTukeyeKramer multiple mean comparison,*p<0.05; **p<0.01;***p<0.001.

1 4 7 10 130

25000

50000

75000PR8

Celecoxib + PR8

Paracetamol + PR8

Naive

Days Post Infection

PR8

pfu/

lung

Figure 4 Pulmonary viral load during acute influenza infection. At days4, 7 and 10 postinfection mice were sacrificed and the lungs wereextracted and homogenised. Pulmonary viral titres were determinedusing the previously described MDCK (MadineDarby canine kidney)virus quantitation assay; results are given as the mean for eachgroup6SEM, n¼4e9 mice/group.

Figure 5 Adaptive immune responsesfollowing acute influenza infection andinhibition of cyclo-oxygenase 2 (COX-2).(AeC) Lungs were harvested at day 10postinfection and a single cellularsuspension was achieved followingcollagenase digestion. Lung cells werestimulated for 5 h in vitro with PMA(phorbol 12-myristate 13-acetate) andionomycin; the numbers of CD4+ (A)and CD8+ T cells (B) secreting tumournecrosis factor a (TNFa) and interferong (IFNg) were determined by flowcytometry. (C) Alternatively, lung cellswere not stimulated and the percentageof CD8+ influenza-specific tetramer+

cells was determined by flowcytometry. (D) Serum from both treatedand untreated mice was isolated at7e9 weeks postinfection with50 plaque-forming units of PR8. Therelative neutralising antibody activity ofeach group of mice was determinedusing the neutralising antibody titreassay described in the Methods section.Each data point given is themean6SEM, n¼3e6 mice/group.

A

Cns

D

B

ns

ns

ns

ns

TNFα IFNγ0

25000

50000

Cel

l Num

ber

TNFα IFNγ0

10000

20000

Cel

l Num

ber

0

2

4

6

8

Perc

enta

ge o

f NP-

64 T

etra

mer

+

CD

8+ T

cells

0

30

60

90

120

Perc

enta

ge C

ell V

iabi

lity

ns

CD4+ CD8+

PR8Celecoxib + PR8

Paracetamol + PR8Naive

372 Thorax 2011;66:368e374. doi:10.1136/thx.2010.150318

Respiratory infection

(figures 5A,B and Supplementary figure 2). As an additionalmeasure of the influenza-specific CD8+ T cell response, nucleo-protein-64 (NP-64) tetramers were used. The results shown infigure 5C indicate that the administration of either celecoxibor paracetamol did not significantly alter the induction ofvirus-specific CD8+ T cell responses.

Finally, the virus-neutralising capacity of serum recoveredfrom treated and untreated mice was assessed and, again, nosignificant difference was observed between mice in the differenttreatment groups (figure 5D). As expected, the neutralisingability of serum from all influenza-infected memory mice wassignificantly greater than that from naive uninfected mice.Overall these data indicate that while administration of para-cetamol or celecoxib reduces the inflammatory response inducedby influenza virus infection, there is no effect on the adaptiveimmune response.

As the adaptive immune response was unimpaired in micetreated with COX-2 inhibitors, we considered it likely thatthere would be no loss of protective immunity in these mice.To determine this, mice infected with the PR8 virus at least8 weeks previously were rechallenged with the heterologousinfluenza virus, X-31, enabling assessment of hetero-subtypicimmunity. In this case, T cells would mediate protectiveimmunity given that neutralising antibodies raised followingPR8 infection do not cross-neutralise the X-31 virus. Asexpected, prior infection with the PR8 virus did confer someprotection against infection with X-31 as these animalsdisplayed significantly less weight loss than animals that hadnot experienced a prior exposure to influenza (figure 6A). Inagreement with this, viral titres were significantly reduced inPR8-immune mice, including those treated with either para-cetamol or celecoxib during the primary infection, comparedwith mice exposed to influenza virus for the first time (figure6B). These findings indicate that administration of paracetamolor celecoxib does not compromise the development of protectiveT cell-mediated immunity.

DISCUSSIONThe current study set out to examine the impact of paracetamolon the course of influenza virus infection, focusing on viralclearance, the generation of innate and adaptive immuneresponses and the establishment of protective immunity. Thedata clearly demonstrate that in the context of the influenza-infected lung, paracetamol reduces pulmonary PGE2 andinflammation without compromising virus clearance or thegeneration of protective immunity.

During the early stages of virus infection (days 4e10) wefound that paracetamol reduced the magnitude of infiltration ofinnate cells into the lungs. A similar trend, although to a lesserextent, was observed in those treated with the selective COX-2inhibitor celecoxib. These results are similar to findings usingCOX-2-deficient mice.9 The effect of influenza infection was lesssevere in COX-2-deficient animals, as demonstrated by decreasedmortality and a blunted inflammatory response compared withwild-type animals. However, elevated virus titres were observedin the animals, indicating, in contrast to our own findings, thatthe abrogated inflammatory response was associated withpoorer control of virus. The more pronounced effects observed inCOX-2-deficient animals compared with those treated withCOX-2 inhibitors may reflect developmental differences inthe genetically modified animals affecting the function of theimmune system and/or the difference between a complete(COX-2�/� animals) and transient (drug administration) blockof COX activity.

Influenza infection caused airway hyper-responsiveness to theinhaled spasmogen, methacholine, in agreement with previousreports,15 which was reduced following treatment with para-cetamol or celecoxib. Thus, paracetamol appears to playa protective role, reducing lung pathology and helping topreserve lung function. Patients infected with seasonal influenzaA, who subsequently develop severe respiratory failure, arecharacterised by a profound early inflammatory infiltrate andcytokine release. The same was true for the 2009 novel H1N1influenza A pandemic where severe lung injury occurred inyoung adults in whom it was frequently difficult to maintainoxygenation. Autopsies on patients succumbing to this newinfluenza strain showed pulmonary pathology similar toprevious pandemics, with profound cellular infiltrates anddiffuse alveolar damage.16e18 Thus paracetamol may, throughinhibition of COX-2, afford protection for patients infected withinfluenza and complicated by respiratory failure through atten-uation of the proinflammatory response. The importance of theCOX-2 pathway in severe respiratory failure is furthersubstantiated by experiments using avian H5N1 influenza Avirus.19 Avian H5N1 infection has a case death approaching 30%but is not transmissible between humans; it is characterised bysevere respiratory failure with inflammatory infiltrates withinthe lung and high levels of inflammatory cytokines such asTNFa and IFNg.20 21 Compared with seasonal H1N1 influenzaA virus, avian H5N1 differentially upregulates COX-2 in primaryhuman macrophages in vitro and epithelial cells in vivo, theincreased expression resulting from autocrine or paracrine

A

B

******

***

PR8 Only + X31Celecoxib + X31Paracetamol + X31

Uninfected + X31

0 1 2 3 4 575

85

95

105

Days Post Infection

Perc

enta

ge o

f Sta

rting

Wei

ght

*** ***

10 2

10 3

10 4

10 5

X31

pfu/

lung

hom

ogen

ate

PR8

Celecoxib + PR8

Paracetamol + PR8

Naive

Figure 6 Protective immune responses to secondary influenzainfection. (A and B) At 7e9 weeks postinfection mice were rechallengedwith 200 plaque-forning units (pfu) of X31. (A) Mice were weighed priorto secondary infection to provide a baseline 100% starting weight and for4 days thereafter. Each data point represents the mean6SEM for eachgroup, n¼3e6 mice/group. (B) At day 4 after secondary infection, micewere sacrificed and the lungs were harvested following perfusion withphosphate-buffered saline. Lungs were homogenised and the viral loadwas determined using the previously described MDCK (MadineDarbycanine kidney) plaque assay. Plots represent the mean of eachgroup6SEM, n¼6 mice/group, 3 mice/group for PR8-naive mice. Thesignificance of differences between previously PR8-infected mice andPR8-naive mice was evaluated using one-way analysis of variance withTukeyeKramer multiple comparison post hoc test, ***p<0.001.

Thorax 2011;66:368e374. doi:10.1136/thx.2010.150318 373

Respiratory infection

mediators rather than direct effects of the virus.19 Expression ofinflammatory cytokines is driven by COX-2 upregulation andcan be effectively inhibited with selective COX-2 inhibitors.19 Inanother murine model of avian H5N1 infection, addition of theselective COX-2 inhibitor celecoxib to the neuraminidaseinhibitor zanamivir reduced lung inflammation and virus titresand increased survival from 13.3% to 53.3% and mean survivaltime from 8.4 to 13.3 days compared with zanamivir alone.22 Itis clear from this study and our own that attenuating theinflammatory response to influenza does not compromise virusclearance. Indeed, overall we observed a trend for better viruscontrol in the treated animals.

Overall these data support a protective role for paracetamolduring the acute stage of infection with influenza virus. Weconsidered it important that the impact of paracetamoladministration at this time on the generation of influenza-specific responses was also assessed. As the innate immuneresponse influences the development of the adaptive response,attenuation of this response through the effects of paracetamolcould suppress the virus-specific immune response. In addition,PGE2 has been shown in vivo to promote T helper 1 (Th1) celldifferentiation, thus inhibition of PGE2 could directly down-modulate development of influenza-specific T cell and antibodyresponses.23 In support of this, a recent report detailing antibodyresponses in 226 infants receiving vaccines demonstratedsignificantly lower antibody titres in those receiving prophy-lactic paracetamol.11 Our data clearly show, however, nosignificant differences in the influenza-specific immune responsein drug-treated and untreated animals. Importantly, the abilityof drug-treated mice to resist rechallenge with a heterotypicinfluenza virus was unimpaired.

The current study demonstrates that both paracetamol andcelecoxib, which inhibit COX-2, attenuate pulmonary inflamma-tion without impeding virus clearance or the development ofspecific immunity in mice. While the doses of paracetamol used aredifferent for mice and humans, as described in the Supplementarymaterial, the implications of the study are clear. Over-the-countermedicines such as paracetamol, which inhibit COX-2, are cheap,widely available and target the host response rather than the virus,and therefore offer an attractive therapeutic option.24 However,with the information currently available on the effects of para-cetamol upon influenza infection, it is possible that its impact maydiffer in mice and humans. It is important that these observations,generated using a mouse model, are validated with respect to theaction of paracetamol and its impact on the immune system inclinical studies of human influenza virus infection.

Acknowledgements The authors would like to thank Dr Emma Jones for herexpertise with the histological images in figure 2, Dr Sivasankar Baalasubramanianfor assistance in performing the experiments in figure 1, Dr Mike Badminton forfacilitating the measurements of alanine transaminase, and Mr Lloyd Morgan forproviding assistance in the generation of preliminary data for the study.

Funding This work was supported by a Wellcome Trust project grant (080340), anMRC Senior Non-Clinical Fellowship awarded to AMG (G117/488) and PRT(G0601617), and a European Union Marie Curie Fellowship awarded to SRC.

Competing interests None.

Provenance and peer review Not commissioned; externally peer reviewed.

REFERENCES1. Murray CJ, Lopez AD, Chin B, et al. Estimation of potential global pandemic

influenza mortality on the basis of vital registry data from the 1918e20 pandemic:a quantitative analysis. Lancet 2006;368:2211e18.

2. Garten RJ, Davis CT, Russell CA, et al. Antigenic and genetic characteristics ofswine-origin 2009 A(H1N1) influenza viruses circulating in humans. Science2009;325:197e201.

3. Conn CA, McClellan JL, Maassab HF, et al. Cytokines and the acute phase responseto influenza virus in mice. Am J Physiol 1995;268:R78e84.

4. Korteweg C, Gu J. Pathology, molecular biology, and pathogenesis of avianinfluenza A (H5N1) infection in humans. Am J Pathol 2008;172:1155e70.

5. Kuiken T, Taubenberger JK. Pathology of human influenza revisited. Vaccine2008;26(Suppl 4):D59e66.

6. Perrone LA, Plowden JK, Garcia-Sastre A, et al. H5N1 and 1918 pandemic influenzavirus infection results in early and excessive infiltration of macrophages andneutrophils in the lungs of mice. PLoS Pathog 2008;4:e1000115.

7. Bertolini A, Ferrari A, Ottani A, et al. Paracetamol: new vistas of an old drug. CNSDrug Rev 2006;12:250e75.

8. Hinz B, Cheremina O, Brune K. Acetaminophen (paracetamol) is a selectivecyclooxygenase-2 inhibitor in man. FASEB J 2008;22:383e90.

9. Carey MA, Bradbury JA, Seubert JM, et al. Contrasting effects of cyclooxygenase-1(COX-1) and COX-2 deficiency on the host response to influenza A viral infection.J Immunol 2005;175:6878e84.

10. Graham NM, Burrell CJ, Douglas RM, et al. Adverse effects of aspirin, paracetamol,and ibuprofen on immune function, viral shedding, and clinical status inrhinovirus-infected volunteers. J Infect Dis 1990;162:1277e82.

11. Prymula R, Siegrist CE, Chlibek R, et al. Effect of prophylactic paracetamoladministration at time of vaccination on febrile reactions and antibody responsesin children: two open-label, randomised controlled trials. Lancet2009;374:1339e50.

12. Longhi MP, Williams A, Wise M, et al. CD59a deficiency exacerbates influenza-induced lung inflammation through complement-dependent and -independentmechanisms. Eur J Immunol 2007;37:1266e74.

13. Ashcroft T, Simpson JM, Timbrell V. Simple method of estimating severity ofpulmonary fibrosis on a numerical scale. J Clin Pathol 1988;41:467e70.

14. Polakos NK, Cornejo JC, Murray DA, et al. Kupffer cell-dependent hepatitisoccurs during influenza infection. Am J Pathol 2006;168:1169e78;quiz 1404e5.

15. Bozanich EM, Gualano RC, Zosky GR, et al. Acute Influenza A infectioninducesbronchial hyper-responsiveness in mice. Respir Physiol Neurobiol2008;162:190e6. 17.

16. Bautista E, Chotpitayasunondh T, Gao Z, et al. Clinical aspects of pandemic 2009influenza A (H1N1) virus infection. N Engl J Med 2009;362:1708e19.

17. Gill JR, Sheng ZM, Ely SF, et al. Pulmonary pathologic findings of fatal 2009pandemic influenza A/H1N1 viral infections. Arch Pathol Lab Med2009;134:235e43.

18. Mauad T, Hajjar LA, Callegari GD, et al. Lung pathology in fatal novel humaninfluenza A (H1N1) infection. Am J Respir Crit Care Med 2009;181:72e9.

19. Lee SM, Cheung CY, Nicholls JM, et al. Hyperinduction of cyclooxygenase-2-mediated proinflammatory cascade: a mechanism for the pathogenesis of avianinfluenza H5N1 infection. J Infect Dis 2008;198:525e35.

20. Cheung CY, Poon LL, Lau AS, et al. Induction of proinflammatory cytokines in humanmacrophages by influenza A (H5N1) viruses: a mechanism for the unusual severity ofhuman disease? Lancet 2002;360:1831e7.

21. Lee SM, Gardy JL, Cheung CY, et al. Systems-level comparison of host-responseselicited by avian H5N1 and seasonal H1N1 influenza viruses in primary humanmacrophages. PLoS One 2009;4:e8072.

22. Zheng BJ, Chan KW, Lin YP, et al. Delayed antiviral plus immunomodulatortreatment still reduces mortality in mice infected by high inoculum of influenzaA/H5N1 virus. Proc Natl Acad Sci USA 2008;105:8091e6.

23. Bloom ET, Babbitt JT. Prostaglandin E2, monocyte adherence and interleukin-1 inthe regulation of human natural killer cell activity by monocytes. Nat Immun CellGrowth Regul 1990;9:36e48.

24. Fedson DS. Confronting an influenza pandemic with inexpensive generic agents: canit be done? Lancet Infect Dis 2008;8:571e6.

374 Thorax 2011;66:368e374. doi:10.1136/thx.2010.150318

Respiratory infection