Midkine Is Expressed and Differentially Processed during

3 1 4 | L I N G E E T A L . | M O L M E D 1 9 : 3 1 4 - 3 2 3 , 2 0 1 3

INTRODUCTIONThe airways have constitutive host de-

fense mechanisms including mucociliaryclearance, alveolar macrophages sensingpotential pathogens, and antimicrobialhost defense proteins (HDPs) with a po-tency to eradicate bacteria (1). In additionto their antibacterial activity, HDPs mayact as growth factors and regulate leuko-

cyte trafficking (2). Several HDPs are con-stitutively produced by airway epithelialcells to maintain host integrity, althoughmany are induced by inflammation orstored preformed in innate immune cells(for example, LL-37 in neutrophils). Re-cently, we found that the growth factormidkine (MK) has potent bactericidal andfungicidal activities (in vitro) executed

through disruption of microbial mem-branes, similar to the activity exerted byother innate antibiotics such as thehuman β-defensins (hBDs), LL-37 and an-tibacterial chemokines (1,3,4). MK is con-stitutively expressed during healthy con-ditions in several tissues, for example, byepithelial cells in the large airways andby keratinocytes in the skin, where it mayreach bactericidal concentrations in themicroenvironment and thus contribute tohost defense (5,6). Several factors presentat sites of inflammation promote MK ex-pression including retinoic acid (RA), hy-poxia, reactive oxygen species and activa-tion of the transcription factor nuclearfactor (NF)-κB (7–11).

Staphylococcus aureus is an importanthuman pathogen. Although uncommonly

Midkine Is Expressed and Differentially Processed duringChronic Obstructive Pulmonary Disease Exacerbations andVentilator-Associated Pneumonia Associated withStaphylococcus aureus Infection

Helena M Linge,1 Cecilia Andersson,1 Sara L Nordin,1 Anders I Olin,2 Ann-Cathrine Petersson,3

Matthias Mörgelin,2 Amanda Welin,4 Johan Bylund,4 Leif Bjermer,1 Jonas Erjefält,1 and Arne Egesten1*

Sections for 1Respiratory Medicine and Allergology and 2Infection Medicine, Department of Clinical Sciences Lund, Lund University;3Clinical Microbiology, Regional Laboratories of Region Skåne, Lund, Sweden, University Hospital, Lund, Sweden; and 4Departmentof Rheumatology and Inflammation Research, Sahlgrenska Academy at University of Gothenburg, Göteborg, Sweden

Staphylococcus aureus is sometimes isolated from the airways during acute exacerbations of chronic obstructive pulmonarydisease (COPD) but more commonly recognized as a cause of ventilator-associated pneumonia (VAP). Antimicrobial proteins,among them midkine (MK), are an important part of innate immunity in the airways. In this study, the levels and possible process-ing of MK in relation to S. aureus infection of the airways were investigated, comparing COPD and VAP, thus comparing a state ofdisease with preceding chronic inflammation and remodeling (COPD) with acute inflammation (that is, VAP). MK was detectedin the small airways and alveoli of COPD lung tissue but less so in normal lung tissue. MK at below micromolar concentrations killedS. aureus in vitro. Proteolytic processing of MK by the staphylococcal metalloprotease aureolysin (AL), but not cysteine proteasestaphopain A (SA), resulted in impaired bactericidal activity. Degradation was seen foremost in the COOH-terminal portion of themolecule that harbors high bactericidal activity. In addition, MK was detected in sputum from patients suffering from VAP causedby S. aureus but less so in sputum from COPD exacerbations associated with the same bacterium. Recombinant MK was de-graded more rapidly in sputum from the COPD patients than from the VAP patients and a greater proteolytic activity in COPDsputum was confirmed by zymography. Taken together, proteases of both bacteria and the host contribute to degradation of theantibacterial protein MK, resulting in an impaired defense of the airways, in particular, in COPD where the state of chronic in-flammation could be of importance.Online address: http://www.molmed.orgdoi: 10.2119/molmed.2013.00045

Address correspondence to Arne Egesten, BMC B14, Tornavägen 10, SE-221 84 Lund, Swe-

den. Phone: +46-46-222-44-45; Fax: +46-46-15-77-56; E-mail: [email protected].

Submitted May 13, 2013; Accepted for publication September 11. 2013; Epub

(www.molmed.org) ahead of print September 12, 2013.

R E S E A R C H A R T I C L E

M O L M E D 1 9 : 3 1 4 - 3 2 3 , 2 0 1 3 | L I N G E E T A L . | 3 1 5

associated with respiratory tract infec-tions, it is sometimes isolated from spu-tum during acute exacerbations ofchronic obstructive pulmonary disease(COPD), and it is a relatively commoncause of hospital-acquired and, in partic-ular, ventilator-associated pneumonia(VAP). S. aureus is estimated to accountfor 10–20% of all nosocomial pneumonias(12). The reported mortality in VAP patients ranges from 33% to 72%,with the upper range reflecting the in-creased risk for mortality among the elderly, patients with impaired cardio -pulmonary function, immune compro-mised patients and patients who requireprolonged intubation of the airways. S.aureus has a plethora of countermeasuresagainst host defense mechanisms, includ-ing surface proteins and secreted pro-teases (13). The continuing increase in an-tibiotics resistance among S. aureusstrains, and in particular the spread ofmeticillin-resistant S. aureus (MRSA), con-stitute new challenges for health care andhealth care professionals (14).

The aim of this study was to investigatethe levels and possible processing of MKin relation to S. aureus infection of the air-ways, comparing COPD and VAP, withthe former being characterized by a stateof chronic inflammation and remodeling,whereas in VAP, the airways respond withan acute inflammation. We found expres-sion and location of MK within COPD air-ways. During infection with S. aureus, MKwas undetectable in COPD sputum, buthigh levels were detected in aspirationsfrom VAP patients. Staphylococcal pro-teases fragmented MK and thereby af-fected its bactericidal activity. COPD spu-tum showed a higher proteolytic activitythan VAP aspirations. We believe this to bean important example of defense impair-ment in the airways during COPD.

MATERIALS AND METHODS

Patient Groups and SamplesIn the present study, lung tissue from 11

COPD (Global Initiative for Chronic Ob-structive Lung Disease [GOLD] stage IV)patients was used (Table 1). In addition,

healthy lung tissue from nine nonsmok-ing controls was obtained during lunglobectomy because of suspected lung can-cer in otherwise healthy nonatopic indi-viduals. Only patients with solid tumorswith visible borders were included in thestudy, and tissues used were obtained asfar from the tumor as possible. In patientswith very severe COPD (GOLD IV),matching lung tissue was collected in as-sociation with lung transplantation. Forall patient groups, care was taken to im-merse the tissue into fixative immediatelyafter surgical excision, and multiple largetissue blocks were prepared for histologi-cal analysis. All subjects gave written in-formed consent to participate in the study,which was approved by the local ethicscommittee in Lund, Sweden (LU412-03).

Sputum from COPD patients and endo-tracheal aspirations from 20 consecutivepatients diagnosed with VAP (bothgroups had positive cultures for S. aureus)were collected at Skåne University Hospi-tal between August 2011 and January2012 and stored at –20°C until analysis.The study design was a cross- sectionalstudy of patients suffering from acute andchronic airway inflammatory responses,respectively, as represented by patientswith S. aureus–associated VAP or COPDexacerbation. Intervention was collectionand culture of sputum or tracheal aspi-rates upon signs of airway infection/ exacerbation. Power analysis assuming a

difference between means of 2 ng/mL and1 standard deviation (SD) indicated thatsix patients in each group were required toachieve a power of 0.8 at α = 0.05. Patientsincluded in other trials (n = 3) or wheresputum sample was too viscous (n = 3)were excluded from all parts of our study.

In patients suffering from VAP, endo-tracheal aspirates were collected byusing a sterile suction catheter via an en-dotracheal tube or a tracheostoma. In pa-tients suffering from COPD, sputum wasmobilized using the huffing technique,without inhalation of nebulized sodiumchloride. Parts of the samples were usedfor culture and parts were frozen at–20°C until use in assays. After thawing,sputum was centrifuged at 5,000g for 5 min and aliquots of the supernatantwere used. Sputum was used withoutaddition of dithiothreitol to make it pos-sible to use antibody-based measurementof MK (that is, enzyme-linked immuno -sorbent assay [ELISA]). The protocol wasapproved by the ethics committee atLund University (#492/2007).

ImmunohistochemistryImmediately after collection, tissues

were placed in 4% buffered formaldehyde,dehydrated and embedded in paraffin,and thin sections (3 μm) were generated.

Single staining of MK. A single stain-ing protocol (EnVision™ Detection sys-tem, K5007; Dako, Glostrup, Denmark)

Table 1. Characteristics of patients with COPD and healthy controls whose tissue was usedfor histological analysis.

Controls COPD GOLD IV

Sex (M/F, n) 2/7, 9 5/6, 11Agea (years) 63 (33–76) 58.5 (53–66)Pack-yearsa (years) 0 40.7 (2560)Exsmokers/current smokers 0 11/0GCS (yes/no/unknown) 0 10/0/1b

β2 Agonist (yes/no/unknown) 0 10/0/1b

Lung functionFEV1

a 2.7 (1.7–5.1) 0.6 (0.4–1.0)FEV1/(F)VCa 85.9 (66–121) 30.2 (17–39)FEV1 % of predicteda 109.8 (82–141) 21.4 (13–27)

FEV1, forced expiratory volume; (F)VC, (forced) vital capacity; GCS, inhaledglucocorticosteroids.aData are given as mean (range).bn = 11; one subject with missing medical history.


M I D K I N E I N C O P D A N D V A P

was used for visualization of MK. Briefly,MK was detected by a rabbit anti-MK an-tibody (Peprotech, London, UK), and sec-ondary antibodies were conjugated withperoxidase polymers. The immunohisto-chemistry protocols were performed byusing an automated immunohistochem-istry robot (Autostainer; Dako). Sectionswere stained with Mayer’s hematoxylinfor visualization of background tissue andwere dehydrated in alcohol and xyleneand mounted in Pertex (Histolab).

Double staining of MK and surfac-tant protein A (SP-A) or myeloperoxi-dase (MPO). After antigen retrieval anda blocking step (protein-blocking, ×0909;Dako), sections were incubatedovernight in 4°C with a rabbit anti-MKantibody, and immunoreactivity was vi-sualized after 1-h incubation at roomtemperature (RT) with a goat anti-rabbitAlexa Fluor 555– conjugated secondaryantibody (1:200; Molecular Probes/LifeTechnologies, Carlsbad, CA, USA). SP-Awas detected after 1-h incubation at RTwith a mouse anti-SPA (1:50; Dako) andvisualized after incubation at RT for 1 hwith an Alexa Fluor 488–conjugated goatanti-mouse secondary antibody (1:200;Molecular Probes/Life Technologies).Consecutive sections were stained with adirectly conjugated rabbit anti-MPO(1:12,000; Dako) by using a Rabbit AlexaFluor 488 nm Zenon labeling kit accord-ing to the manufacturer’s instructions(Molecular Probes/Life Technologies).Nuclei were detected by Hoechts 33342,and sections were mounted in TBS/glyc-erin and frozen until quantification. Forall immunohistochemical procedures,markers and tissues, staining was absentin sections using isotype-matched controlantibodies (Dako) that were used insteadof, and at the same concentration as, theprimary antibody.

Bacterial Strains, Determination ofAntibiotics Resistance Profile andScreening of Virulence-AssociatedGenes

The S. aureus strains 5120, Newman,8325-4 and 8325-4 sarA (sarA; deficient ofthe repressor sarA, resulting in overex-

pression of proteases released extracellu-larly [15]) were used for the in vitro stud-ies and routinely cultured in Todd-Hewittmedium (TH; BD, Franklin Lakes, NJ,USA) at 37°C with 5% CO2.

The antibiotics susceptibility of the S. aureus strains isolated from COPD andVAP patients were tested by using thedisc diffusion method according to theSwedish Reference Group for Antibiotics(SRGA) and its subcommittee on meth-odology (SRGA-M; www.srga.org) or by E-test (bioMeriéux, France). Discs withcefoxitin, clindamycin, fusidic acid,trimethoprim-sulfamethoxazole, nor-floxacin, tigecyclin, gentamicin, linezolidand rifampicin were obtained fromOxoid (Basingstoke, UK). The van-comycin susceptibility was tested usingE-tests. All isolates were screened for thepresence of virulence-associated genesencoding Panton Valentine leucocidine(lukS-PV and lukF-PV) and tss encodingthe TSS-1 toxin using polymerase chainreaction and spa typed as previously described (16).

Bactericidal AssayBacteria was cultured to the midexpo-

nential growth phase (OD620 ≈ 0.4),washed in incubation buffer (10 mmol/LTris/HCl, 5 mmol/L glucose; pH 7.5)and adjusted to a concentration of 2 × 106

colony-forming units (CFUs)/mL. Fiftymicroliters bacterial solution (~105 CFUs)was incubated in buffer or with variousconcentrations of the indicated protein ormixtures for 1 h at 37°C. To quantify thebactericidal activity, serial dilutions in in-cubation buffer were plated on TH-agarin duplicates and incubated overnight at37°C. The percent bacterial killing wascalculated by comparison with the num-ber of CFUs when incubated in bufferalone.

Electron MicroscopyBacteria were incubated with MK

(1 μmol/L) in incubation buffer for 1 h at37°C and then subjected to scanning elec-tron microscopy as described (17). Im-munogold labeling and transmissionelectron microscopy (TEM) of ultrathin

sections was performed as previouslydescribed (5). Specimens were examinedin a JEOL (Tokyo, Japan) 1200EX trans-mission electron microscope operated at60 kV accelerating voltage.

Protein Detection by ELISA, SodiumDodecyl Sulfate–Polyacrylamide GelElectrophoresis (SDS-PAGE) andWestern Blotting

The MK ELISA was from Peprotech.For Western blotting, standard tech-niques were used. After separation on aTris-tricine gel, samples were electroblot-ted onto a polyvinylidene fluoride mem-brane. After blocking, membranes wereprobed with antibodies toward MK (Peprotech), followed by a secondarygoat–anti-rabbit antibody (Bio-Rad, Her-cules, CA). Blots were developed byusing the Bio-Rad Chemidoc system(Bio-Rad).

MK Proteolytic Cleavage AssayMK (5 μg) was incubated for the indi-

cated time points with 0.5 μg of eitherstaphylococcal proteases aureolysin (AL)or staphopain A (SA; Biocentrum, Krakow,Poland). Sterile filtered bacterial super-natants (10–15 μL) from overnight cultureswere incubated with recombinant MK pro-tein for indicated time points before eitherSDS-PAGE followed by Western blot anal-ysis or N-terminal sequencing. In a sepa-rate experiment, MK was incubated withCOPD sputa or VAP aspirations (patientIDs: B6, A10, B9 and A2, A6, A7, respec-tively) for 1 and 18 h. Samples were thenanalyzed by SDS-PAGE and standard pro-tein staining methods.

ZymographyZymography was performed by using

precast gels containing collagen (Bio-Rad). The material used in zymographyassays were from COPD patients A1, A9and B5 and VAP patients A3, A4 and B3,respectively.

N-Terminal SequencingThe identification of MK fragments

generated by S. aureus protease cleav-age was performed by N-terminal se-



quencing by Edman degradation ofeight sequential amino acids (HEJAnalyser; Karolinska Institutet, Stock-holm, Sweden).

Molecular ModelingMolecular modeling of MK was per-

formed as described (4).

RESULTS

Expression of MK in Lung Tissue fromCOPD

Immunohistochemistry followed bymorphometric image analysis showed thatCOPD airways, including the airway wall(Figure 1A), epithelial cells (Figure 1B) and

the small airway parenchyma (Figure 1C)have a significantly higher expression ofMK than the airways of controls. MK pro-tein, as detected by immunohistochem-istry in small airways (Figure 1D), as seenin a higher magnification (Figure 1E), ispredominantly expressed in bronchial ep-ithelial cells. In alveoli, weaker presence

Figure 1. MK as detected by immunofluorescence in lung tissue from COPD. The immunoreactivity of MK expression, expressed as positivepixels per mm2, in the small airway wall (A), small airway epithelium (B) and alveolar parenchyma (C). Data expressed as scatter dot plotsand line indicate the median value. Asterisks show statistical difference when compared with controls: *p < 0.05, **p < 0.01. Immunohisto-chemical staining of MK in small airways (D), small airway epithelium at higher magnification (E) and alveolar parenchyma (F). Triple fluo-rescent immunohistochemical staining of nuclei (DAPI, blue) and MK (Alexa Fluor 555, red) together with MPO of neutrophils show weakcolocalization (G) and surfactant protein A (SP-A) expressed by type 2 pneumocytes of the alveoli and is shown in green (H and I: AlexaFluor 488). In (I), alveolar macrophages containing SP-A (that is, likely to have been phagocytosed) also show presence of MK. Scalebars: D, F = 200 μm; E, H, I = 50 μm; and G = 100 μm.



of MK was seen (Figure 1F). By immuno-fluorescent double staining, MK (red) wasvisualized in bronchial epithelium ofsmall airways, showing weak colocaliza-tion with myeloperoxidase (green) of neutrophils. Neutrophils mainly showedsubmucosal positioning (Figure 1G). MKwas also detected in the alveolar paren -chyma, showing to some degree colocal-ization with SP-A of type 2 pneumocytes(Figure 1H). In addition, alveolar macro -phages showed presence of both MK andSP-A (Figure 1I), capable of phagocytosingthe latter (18).

To localize MK on an ultrastructurallevel, immunogold labeling followed byTEM analysis was performed (Figure 2).This process revealed the presence of MK

on the apical surface and associationwith cilia of bronchial epithelial cells inCOPD airways (Figures 2A, B); for closerdetail see Figures 2C, D.

Figure 2. Ultrastructural detection of MK byimmunoelectron microscopy of lung tissuein COPD. (A) Immunogold-labeling of ultra-thin sections was performed on small airwaylung tissue from a COPD patient. Colloidalgold particles indicate location of boundantibodies against MK. MK is present on thesurface of cilia (B) and on the bronchial ep-ithelial cell surface (C and D). Scale bars: A = 10 μm, B = 1 μm and C and D = 100 nm.

Figure 3. Bactericidal activity of MK against S. aureus is impaired by AL but not SA. (A) MKwas compared with the classic HDP LL-37 in the viable counts assay using the S. aureusstrain 5120 as target. Bacteria were incubated with increasing MK concentrations for 1 hand then plated, and CFUs were enumerated after an overnight incubation at 37°C. MKwas the more potent antibacterial agent at lower concentrations, killing approximately90% of the bacteria at 0.1 μmol/L. (B) MK antibacterial activity has potent bactericidalactivity against the S. aureus strains Newman and 8325-4. (C) Scanning electron mi-croscopy showing intact S. aureus after incubation in buffer alone for 1 h at 37°C (i) anddisrupted bacteria with leakage of intracellular contents after incubation with MK (1 μmol/L) (ii). (D) MK was incubated with AL and SA for 3 and 18 h. Samples were sepa-rated by SDS-PAGE and the fragmentation pattern is visualized by Coomassie staining. ALgenerated bands of approximately 5 and 6 kDa, respectively, whereas SA generatedtwo bands of similar sizes. (E) AL and SA degradation products of MK show differences inbactericidal activity. After incubation of MK with AL, only 25% killing was retained at a MKconcentration corresponding to 1 μmol/L of the holoprotein. SA-generated fragments re-tained bactericidal activity in the order of the MK holoprotein. (F) Incubation of MK withculture supernatant from S. aureus 8325-4 wt and sarA (a strain overexpressing and re-leasing proteases) resulted in degradation of MK. A faint band is seen with supernatantfrom wt, and a distinct band is seen with sarA (arrow). In addition, a smaller band is ob-served in the latter case (*). The bands correspond to the molecular sizes generated withrecombinant AL.



Bactericidal Activity of MK AgainstDifferent Strains of S. aureus

A previous study identified MK as anantibacterial against S. aureus (4). In thecurrent study, we demonstrate bacterialkilling of S. aureus strains 5120, New-man and 8325-4 (Figures 3A, B). MKwas the more potent antimicrobialagent, killing 5120 at lower concentra-tions than the classic HDP LL-37. Bacte-ria exposed to buffer alone remained in-tact, as imaged by standard error of themean (Figure 3C, top panel i), whereasexposure to MK resulted in leakage ofintracellular material from the bacteria,suggesting membrane disruption as alikely mode of action (Figure 3C, lowerpanel ii).

MK Fragments Generated byStaphylococcal Proteases

In the host–microbe encounter, bacte-rial proteases are likely to have accessto constitutively expressed HDPs suchas MK. Incubation of recombinant MK with the S. aureus proteases AL and SAfor 3 h generated fragments, which re-mained present after 18 h of incubation(Figure 3D) as detected by proteinstaining. AL generated fragments of

MK of the approximate size of 5 and6 kDa, as estimated from the stainedgel. N-terminal sequencing combinedwith mass spectrometry identified theband of higher molecular weight to cor-respond to the true NH terminus of theholoprotein, whereas the other bandconstituted a mixture of peptides lo-cated in the COOH terminus of theholoprotein. SA generated an MK frag-ment of the apparent size of 4 kDa, andN-terminal sequencing of this bandshowed a mixture of two equally abun-dant peptides—one corresponding tothe true NH terminus of the holopro-tein and one from the COOH terminus.These results are displayed in Figure 4.

To investigate the effect of AL and SAcleavage on MK bactericidal activity, MKwas preincubated with respective pro-tease and then used in the viable countassay. The results show that AL reducedMK killing to 25% (Figure 3E), whereaskilling remained close to 100% after incu-bation with SA. MK was then incubatedwith supernatants from cultures of S. au-reus wild-type (wt) strain 8325-4 or itsprotease overexpressing derivative sarA(Figure 3F). Immunoblotting revealedthat MK remained intact after incubation

with wt supernatant, whereas it wascleaved by the sarA supernatant. Takentogether, these data demonstrate that MKis bactericidal toward S. aureus and thatAL, but not SA, by proteolytic cleavageacts as a bacterial countermeasure tolimit MK-mediated eradication.

COPD Sputa and EndotrachealAspirations from VAP PatientsAssociated with S. aureus InfectionShow Differences in MK Levels andProteolytic Activity

Patient characteristics and the results ofthe virulence-associated genes screeningare shown in Table 2. The average age ofthe patient groups was 73 years forCOPD and 41 years for VAP. The VAPgroup spent an average of 3.3 d on a ven-tilator before the introduction of antibi-otics or death occurred. None of theCOPD patients required intubation. Theresults of the virulence gene screening re-vealed that only one bacterial isolate (A4)was positive for the pvl gene, and three(A4, B5, B9) were positive for the super-antigen tsst-1. The spa-types t002 and t056were found in two isolates each (A6, B7and A7, A9, respectively). All testedstrains were sensitive to methicillin.

Figure 4. Molecular model of MK depicting the cleavage sites AL and SA, respectively. Fragments generated after cleavage with AL andSA, respectively, were characterized by N-terminal protein sequencing combined with mass spectrometry. The cleavage sites for AL areindicated in red and SA in blue. Both enzymes cleaved MK at A22.



Significantly less MK was detected insputum from COPD patients (mean: 0.15 ng/mL, range 0–0.49; n = 5, 2 of 5undetectable) than in endotracheal aspi-rations from VAP patients (n = 8, mean:2.62 ng/mL, range 0–8.41; 1 of 8 unde-tectable) (p = 0.030 as determined by theMann-Whitney U test; Figure 5A). Fig-ure 5B shows degradation patterns of re-combinant MK exposed for 1 or 18 h toeither COPD sputa or VAP aspirations,respectively. In COPD sputum, MK deg-radation kinetics was more rapid than inaspirations from VAP patients. To eluci-date why MK was degraded more rap-idly in COPD sputa, zymography wasperformed. The substrate clearance areawas quantified and shows that COPDsputa had a significantly higher prote-olytic activity than VAP aspirations (p =0.01, Figure 5C).

DISCUSSIONAccording to the World Health Organ-

ization, 65 million people are affected byCOPD, and, based on data from 2005, thedisease accounts for 5% of all deaths glob-ally (http://www.who.int/respiratory/copd/burden/en). COPD exacerbationsare linked to recurrent bacterial or viral

infections that influence the outcome ofthe disease negatively. The most com-monly isolated bacterial species areStreptococcus pneumoniae, Haemophilus in-fluenzae and Moraxella catarrhalis, but S.aureus species are also found (19,20), par-ticularly in elderly patients (21).

In healthy individuals, MK is constitu-tively expressed in bronchial epitheliumof large airways and at least partially de-pends on retinoic acid (5). In the currentstudy, we found COPD lung tissue to ex-press MK in small airways, in particularwithin the airway wall, epithelium andparenchyma. This result is in contrast tohealthy lungs, where MK is expressed inlarge but not small airways (5). Addition-ally, we found the presence of MK intype 2 pneumocytes and alveolarmacrophages in the present study. Oneimportant question is whether MK con-centrations reach bactericidal concentra-tions. In a previous study, using an airliquid system, we calculated the MK con-centration of the airway surface liquid toan amount of 0.7 μmol/L (that is, a bac-tericidal concentration) (5). In addition,MK was detectable in induced sputum ofhealthy individuals (5). Several factorsare likely to contribute to increased MK

expression during S. aureus infection ofthe airways. The MK gene has a retinoicacid (RA)-responsive element in its pro-moter region, and several factors presentduring inflammation increase the genera-tion of RA from vitamin A, for example,activation of TLR2 by peptidoglycan ofgram-positive bacteria (e.g., S. aureus)(22,23). In addition, MK expression is en-hanced by the proinflammatory tran-scription factor NF-κB, reactive oxygenspecies, tumor necrosis factor-α, and in-terleukin-1β (10,11). In severe infection,hypoxia of tissues may occur and, inter-estingly, hypoxia enhances MK expres-sion via the transcription factor hypoxia-inducible factor 1-α (HIF-1α) (9).

Others found significantly elevatedMK in sera from hypoxemic patientscompared with healthy controls (24), butthis study did not include examination oflung tissue or sputa. The difference inMK detection in lung tissue versus sputaquestions the reliability of sputum as asampling method. In addition, process-ing of sputa, for example by addition ofthe reducing agent dithiothreitol to lowerviscosity, may result in dramatic differ-ences with respect to detection limits ofvarious readouts (25). The sputum and

Table 2. Characteristics of COPD and VAP patients, bacterial methicillin resistance and virulence-associated gene profile.

Virulence-associated

Methicillin resistance: genes

ID number Gender Age (years) Underlying disease/injury Days on ventilator cefoxitin spa pvl tsst-1

A1 M 79 COPD 0 S t091 N NA10 F 70 COPD 0 S t9929 N NB1 M 42 COPD 0 S t449 N NB5 F 79 COPD 0 S t166 N POSB6 F 80 COPD 0 S t094 N NB9 M 86 COPD 0 S NT N POSA2 F 67 Trauma 4 S t005 N NA3 M 50 Cerebral hemorrhage 1 S t2094 N NA4 M 25 Trauma 3 S t127 POS POSA6 M 27 Trauma 3 S t002 N NA7 M 46 Acute myocardial infarction 3 S t056 N NA9 F 48 Subarachnoid hemorrhage 4 S t056 N NB3 M 47 Cerebral infarction 3 S t1743 N NB7 F 59 Neurosurgery 6 S t002 N NB10 M 36 Trauma 3 S t1931 N NC1 F 1 Sepsis unknown S t065 N N

The number of days on a ventilator was calculated from day of admittance until day of antibiotics administration or until death. S, sensitive;N, negative; POS, positive; NT, nontypeable.



aspiration samples used in our studywere without additions, increasing thereliability of the results.

Electron microscopy analysis locatedMK to the airway lumen and on the ciliaof the epithelia of COPD airways (Figure

2). MK is believed to exert its antibacter-ial activity through bacterial membranedisruption and the extracellular locationidentified by electron microscopy, whereit can encounter invading or colonizingpathogens, supporting the role of MK asan antibacterial factor (3). Earlier studieshave localized the highest antibacterialactivity of MK to the COOH-terminalhalf and identified its ability to kill gram-positive and gram-negative bacteria aswell as fungi (3–5). S. aureus is namedamong the isolated pathogens found inacute exacerbations of COPD but, be-sides skin infections, it remains a morerecognized cause of necrotizing pneumo-nia, HAP, VAP and upper airway infec-tions. For these reasons, we used S. au-reus as an experimental target for MKfunction and compared airway-expelledmaterial from two different patientgroups infected with this pathogen. Asexpected, MK readily killed all testedstrains of S. aureus in vitro. AL but not SAdestructed the antibacterial activity ofMK. Both NH- and COOH-terminal re-gions of MK contain antibacterial activ-ity, but only the activity of the COOHdomain including the tail region equatesto that of the holoprotein. Our data sug-gest that AL degrades the COOH- terminal half where its most potent anti-bacterial activity is found, althoughfurther experiments are needed to con-firm this. Future studies will addresswhat role, if any, the protease-generatedfragments of MK have.

Not only do the proteases used in thisstudy act on host proteins (26) such asMK, they also regulate the profile of theextracellular proteins on the bacterialsurface (15). Thereby, these proteaseshave the ability to substantially influ-ence bacterial virulence and host inter-play. The theme is known from otherhost–pathogen encounters (27), and re-cent work has shown that proteasesfrom Streptococcus pyogenes and Fine-goldia magna act proteolytically to medi-ate bacterial evasion of host defensefunctions in skin infections (6). Bacterialcolonization has over the last decade hasbecome more recognized as a driving

Figure 5. Detection and stability of MK in airway secretions from COPD and VAP, respec-tively. (A) Airway secretions from patients suffering from COPD exacerbations or VAP associ-ated with S. aureus infection were analyzed for MK by ELISA. The results show significantlyless MK in sputum from COPD patients compared with higher levels in VAP patients (p =0.030, as determined by the Mann-Whitney U test). (B) To investigate the stability of MK inairway secretions, recombinant MK was incubated in sputa from three COPD patients orthree VAP patients for 1 or 18 h and the integrity of the molecule was subsequently ana-lyzed by Western blot. The kinetics of degradation was more rapid in COPD sputa, showinglittle or no MK remaining after 1 h in 2 of 3 patients. (C) To investigate and compare pro-tease activity in airway secretions, sputa obtained from three COPD and three VAP patientswere analyzed for protease activity by zymography. COPD sputa generated significantlylarger clearing zones in the collagen matrix, indicating the presence of higher proteolyticactivity (p < 0.01). The sizes of the clearing zones were measured, and the data representmean ± SD. p value was calculated using the Student t test. n.d., Not detectable.



force behind exacerbations in COPD.The role of bacterial proteases frompathogens more commonly associatedwith COPD exacerbations remains anopen area of research.

S. aureus has many ways of evadingthe immune system, with expression ofsurface protein A, an adherence- promoting surface protein encoded bythe spa gene, being one of them. The clin-ical isolates in our study showed geneticheterogeneity in terms of spa type; onlytwo types were found in more than onepatient. This result suggests a low risk ofpatient-to-patient or hospital- mediatedspread. One of the strains in our studycarried t127, which is also found inclones of US100. Methicillin resistancehas historically been a larger problem inthe U.S. than in Europe (14). Althoughstaphylococcal virulence does not de-pend on methicillin resistance alone butrelies on a combination of factors, it cre-ates major problems for hospitals andhealth care professionals. Frequency ofMRSA isolation remains low in Sweden(http://www.smittskyddsinstitutet.se[continuously updated information]),and fortunately, we found all strains inthis study sensitive to methicillin.

That we detected MK in COPD air-ways by histology analysis but not insputa led us to investigate whether COPDsputa contained more proteolytic activ-ity than VAP aspirations. Degranulationof extravasated neutrophils in the air-ways together with a protease–proteaseinhibitor imbalance is a hallmark of ex-acerbations in COPD as described in thearticle by Celli and Barnes (28). ThatCOPD samples showed larger clearingzones in the zymography assay is likelydue to higher concentrations of pro-teases originating from the increasednumber of neutrophils, which is a hall-mark of the condition. The serine pro-teases not only degrade MK, as shownin this study, but also contribute to ep-ithelium injury, increased mucus pro-duction and chemokine production (29).Because both patient groups were in-fected with S. aureus, it appears that thecontribution of host proteases to MK

degradation in COPD is greater than thecontribution by bacterial proteases. Thepatient groups differ by age and ventila-tion therapy and the sample size issmall. Also, the VAP group may not suf-fer from the protease inhibitor imbalanceor other unknown COPD-related innateimmunity deficits. The composition ofsputum (COPD) compared with trachealaspirates (VAP) could differ. One exam-ple is the levels of anionic molecules (forexample, mucins, free DNA and osteo-pontin) that could interact with MK andthus affect measurements. In addition,we cannot rule out that contaminatingsaliva could have a diluting effect on theMK concentrations observed in COPD.However, in a previous study, sputumwas induced through inhalation of nebu-lized sodium chloride in healthy indi-viduals. In these samples, MK was de-tectable using the same ELISA as used inthe current study (5).

CONCLUSIONOur data demonstrate the possibility

that S. aureus modulates and corrupts hostairway defense lines such as MK by frag-mentation in both immuno-competentand -suppressed patient groups.

ACKNOWLEDGMENTSWe thank Pia Andersson and Maria

Baumgarten for excellent technical assis-tance and Staffan Arvidson for providingthe S. aureus strains 8325-4 and sarA.

This work was supported by theSwedish Research Council (projectsA0615601 HML and 2010-4224AE); theSwedish Heart and Lung Foundation(20100164); the Medical Faculty of LundUniversity; Swedish Government Fundsfor Clinical Research (ALF); and thefoundations of Bergh, Greta and JohanKock, and Alfred Österlund.

DISCLOSUREThe authors declare that they have no

competing interests as defined by Molec-ular Medicine, or other interests thatmight be perceived to influence the re-sults and discussion reported in thispaper.

REFERENCES1. Bartlett JA, Fischer AJ, McCray PB Jr. (2008) In-

nate immune functions of the airway epithelium.Contrib. Microbiol. 15:147–63.

2. Gallo RL. (2008) Sounding the alarm: multiplefunctions of host defense peptides. J. Invest. Der-matol. 128:5–6.

3. Svensson SL, et al. (2010) Midkine and pleiotrophinhave bactericidal properties: preserved antibacterialactivity in a family of heparin-binding growth fac-tors during evolution. J. Biol. Chem. 285:16105–15.

4. Nordin SL, Sonesson A, Malmsten M, MorgelinM, Egesten A. (2012) The epithelium-producedgrowth factor midkine has fungicidal properties.J. Antimicrob. Chemother. 67:1927–36.

5. Nordin SL, et al. (2013) Midkine is part of the an-tibacterial activity released at the surface of dif-ferentiated bronchial epithelial cells. J. InnateImmun. 5:519–30.

6. Frick IM, et al. (2011) Constitutive and inflammation-dependent antimicrobial peptidesproduced by epithelium are differentiallyprocessed and inactivated by the commensalFinegoldia magna and the pathogen Streptococ-cus pyogenes. J. Immunol. 187:4300–9.

7. Kadomatsu K, Tomomura M, Muramatsu T.(1988) cDNA cloning and sequencing of a newgene intensely expressed in early differentiationstages of embryonal carcinoma cells and in mid-gestation period of mouse embryogenesis.Biochem. Biophys. Res. Comm. 151:1312–8.

8. Tomomura M, Kadomatsu K, Matsubara S, Muramatsu T. (1990) A retinoic acid-responsivegene, MK, found in the teratocarcinoma system:heterogeneity of the transcript and the nature of the translation product. J. Biol. Chem.265:10765–70.

9. Reynolds PR, Mucenski ML, Le Cras TD, NicholsWC, Whitsett JA. (2004) Midkine is regulated byhypoxia and causes pulmonary vascular remod-eling. J. Biol. Chem. 279:37124–32.

10. Hobo A, et al. (2009) The growth factor midkineregulates the renin-angiotensin system in mice.J. Clin. Invest. 119:1616–25.

11. You Z, et al. (2008) Midkine is a NF-kappaB- inducible gene that supports prostate cancer cellsurvival. BMC Med. Genomics. 1:6.

12. Lode H, Raffenberg M, Erbes R, Geerdes-FengeH, Mauch H. (2000) Nosocomial pneumonia: epi-demiology, pathogenesis, diagnosis, treatmentand prevention. Curr. Opin. Infect. Dis. 13:377–84.

13. Veldkamp KE, van Strijp JA. (2009) Innate im-mune evasion by staphylococci. Adv. Exp. Med.Biol. 666:19–31.

14. Defres S, Marwick C, Nathwani D. (2009) MRSAas a cause of lung infection including airway in-fection, community-acquired pneumonia andhospital-acquired pneumonia. Eur. Resp. J.34:1470–6.

15. Karlsson A, Saravia-Otten P, Tegmark K, Mor-feldt E, Arvidson S. (2001) Decreased amounts ofcell wall-associated protein A and fibronectin-binding proteins in Staphylococcus aureus sarA



mutants due to up-regulation of extracellularproteases. Infect. Immun. 69:4742–8.

16. Petersson AC, Olsson-Liljequist B, Miorner H,Haeggman S. (2010) Evaluating the usefulness ofspa typing, in comparison with pulsed-field gelelectrophoresis, for epidemiological typing ofmethicillin-resistant Staphylococcus aureus in alow-prevalence region in Sweden 2000–2004.Clin. Microbiol. Infect. 16:456–62.

17. Oehmcke S, Morgelin M, Herwald H. (2009) Acti-vation of the human contact system on neutrophilextracellular traps. J. Innate. Immun. 1:225–30.

18. Kingma PS, Whitsett JA. (2006) In defense of thelung: surfactant protein A and surfactant proteinD. Curr. Opin. Pharmacol. 6:277–83.

19. Sethi S. (2010) Infection as a comorbidity ofCOPD. Eur. Resp. J. 35:1209–15.

20. Sethi S, Evans N, Grant BJ, Murphy TF. (2002)New strains of bacteria and exacerbations ofchronic obstructive pulmonary disease. N. Engl.J. Med. 347:465–71.

21. Albertson TE, Louie S, Chan AL. (2010) The diag-nosis and treatment of elderly patients withacute exacerbation of chronic obstructive pul-monary disease and chronic bronchitis. J. Am.Geriatr. Soc. 58:570–9.

22. Muramatsu T. (2002) Midkine and pleiotrophin:two related proteins involved in development, sur-vival, inflammation and tumorigenesis. J. Biochem.132:359–71.

23. Manicassamy S, et al. (2009) Toll-like receptor 2-dependent induction of vitamin A-metabolizingenzymes in dendritic cells promotes T regulatoryresponses and inhibits autoimmunity. Nat. Med.15:401–9.

24. Krzystek-Korpacka M, et al. (2008) Respiratoryinsufficiency related to COPD accelerates sys-temic inflammation, under-nutrition, and angio-genesis in esophageal malignancies. Exp. Oncol.30:75–80.

25. Wang F, He B. (2009) The effect of dithiothreitolon chemotactic factors in induced sputum ofchronic obstructive pulmonary disease patients.Respiration. 78:217–22.

26. Sieprawska-Lupa M, et al. (2004) Degradation ofhuman antimicrobial peptide LL-37 by Staphylo-coccus aureus-derived proteinases. Antimicrob.Agents Chemother. 48:4673–9.

27. Egesten A, et al. (2009) SpeB of Streptococcuspyogenes differentially modulates antibacterialand receptor activating properties of humanchemokines. PLoS One. 4:e4769.

28. Celli BR, Barnes PJ. (2007) Exacerbations ofchronic obstructive pulmonary disease. Eur. Resp.J. 29:1224–38.

29. Hiemstra PS, van Wetering S, Stolk J. (1998) Neu-trophil serine proteinases and defensins inchronic obstructive pulmonary disease: effects onpulmonary epithelium. Eur. Resp. J. 12:1200–8.

Documents

Midkine Is Expressed and Differentially Processed during