- Email: [email protected]
Increased Bacterial Load and Expression of Antimicrobial Peptides in Skin of Barrier-Deficient Mice with Reduced Cancer Susceptibility
Accepted Article Preview: Published ahead of advance online publication www.jidonline.org
Increased Bacterial Load and Expression of Antimicrobial Peptides in Skin of Barrier-Deficient Mice with Reduced Cancer Susceptibility Ken Natsuga, Sara Cipolat, Fiona M Watt
Cite this article as: Ken Natsuga, Sara Cipolat, Fiona M Watt, Increased Bacterial Load and Expression of Antimicrobial Peptides in Skin of Barrier-Deficient Mice with Reduced Cancer Susceptibility, Journal of Investigative Dermatology accepted article preview 30 September 2015; doi: 10.1038/jid.2015.383. This is a PDF file of an unedited peer-reviewed manuscript that has been accepted for publication. NPG are providing this early version of the manuscript as a service to our customers. The manuscript will undergo copyediting, typesetting and a proof review before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers apply.
Received 22 February 2015; revised 24 July 2015; accepted 11 September 2015; Accepted article preview online 30 September 2015
© 2015 The Society for Investigative Dermatology
Increased bacteri al load and expression of antimicrobial peptides i n skin of barrier-deficient mice w ith reduced cancer susceptibility
Ken Natsuga 1 , 2 , Sara Cipolat 1 , 3 , and Fiona M. W att 3 , *
1
Cancer Research UK Cambr idge Research Institut e, Li Ka Shing Centre,
Robinson W ay, Cam bridge CB2 0RE, UK; 2 Department of Der matolog y, Hokkaido Universit y Graduate School of Medicine N15W 7, Sapporo 060 -8638, Japan; 3 Centre f or Stem Cells and Regenerative Medicine, King’s College London, 28t h f loor, Guy’s Tower W ing, London SE1 9RT, UK
*Corresponding author: Fiona. [email protected]. uk
Running title: bacter ial load in barr ier def icient mice
1
© 2015 The Society for Investigative Dermatology
ABSTR ACT Mice lacking three epidermal barr ier pr oteins – envoplakin, periplakin and involucrin - (EPI-/- mice) have a def ective cornif ied layer, reduced epidermal T cells, increased dermal CD4+ T cells and are resistant to developing skin tumours. The tumour-protect ive mechanism involves signalling bet ween Rae-1 expressing keratinocytes and the N a t u r a l K i l l e r G r o u p 2 D ( N K G 2 D ) receptor on immune cells, which also plays a role in host defences a g a i n s t i n f e c t i o n . Given the emerging link bet ween bacter ia and cancer, we investigated whether EPI -/- mice ha ve an altered skin microbiota . The bacter ial phyla wer e similar in wild type and EPI -/- skin. However, bact eria were 3-f old more abundant in EPI-/- skin and penetrated deeper into the epidermis. The major epithelial def ense mechanism against bacteria is product ion of antimicrobial proteins (AMPs). EPI-/- skin exhibited enhanced expression of antimicrobial pept ides. However, r educing the bacter ial load b y antibiotic treatment or breeding mice under specif ic pathogen -f ree condit ions did not reduce AMP expression or alleviate the abnormalit ies in T cell populat ions. W e conclude that the atopic character istics of EPI-/- skin are a consequence of the def ective barr ier rat her than a response to the increase d bacter ial load. It is theref ore unlikel y t hat the increase in skin microbiota contributes dir ectly t o the obser ved cancer resist ance.
2
© 2015 The Society for Investigative Dermatology
INTRO DUCTION In cells of the outer most epidermal layers the plasma membrane is replaced by a cornif ied envelope, f ormed by transglutam inase -1-mediated crosslinking of proteins and lip ids (Nat suga 2014 ). This insoluble structure is a key component of the epidermal barr ier that protects the body f rom envir onment al assaults ( Candi et al. 2005 ). Def ects in the epidermal barr ier are linked to atopic dermat itis in humans and mice (Natsuga 2014 ). Mice t riply def icient in three cornif ied envelope precur sors - envoplakin, periplakin and involucr in (EPI-/- mice) - have epidermal barr ier def ects , a reduction in epidermal T cells and an increase in dermal CD4+ T cells (Sevilla et al. 2007 ).
EPI-/- mice are highly resist ant to developing skin tumours when treated with chemical carcinogens (Cipolat et al. 2014 ). This is due, at least in part, to an exaggerated
atopic
response
to
the
tumour
promoter
1 2 - O - t e t r a d e c a n o y l p h o r b o l - 1 3 - a c e t a t e ( T P A ) , including upregulat ion of the keratinocyte stress-associated antigen Rae -1, which binds NKG2D on immune cells. NKG2D ligands are curr ently being evaluated as therapeut ic targets in a var iet y of cancer types (Spear et al. 2013 ).
Rae-1- NKG2D interactions are not only implicated in cancer ( Strid et al. 2008 ; Cipolat et al. 2014 ), but also mediate host responses to bacter ial and vir al inf ection (Hamerman et al. 2004 ; Lodoen et al. 2004 ; Str id et al. 2008 ; Strid et
3
© 2015 The Society for Investigative Dermatology
al. 2009). Given the strong evidence linking the gut micr obiota, chr onic inf lammation and cancer (Shanahan and O'Toole 2014 ), we speculat ed that the skin microbiome might be alter ed in EPI -/- mice. Human skin is covered by wide var iet y of commensal microbiota , which var y in com position accor ding to body site, age and healt h status (Grice and Segre 2011 ). In addit ion, r ecent studies suggest a regulator y role f or skin commensal bacter ia in skin inf ection and inf lammation (Lai et al. 2009 ; Naik et al. 2012 ). W e theref ore investigated whether the skin microbiome dif f ers between EPI -/- and wild t ype (W T) mice and whether it inf luences keratinocyte expression of Rae -1 and anti-m icrobial proteins (AMPs), nat ural antibiotics that are eff ectors of the innate immune system (Gallo and Hooper 2012 ).
4
© 2015 The Society for Investigative Dermatology
RESULTS Al t ered abundance and distri bution of bacteria in t he ski n of EPI-/- mice To determine whether EPI -/- mice had an alter ed skin m icrobiome, we measured the quantity and diversit y of microbes on the ear skin of six EPI-/and six wild t ype control m ice on the same genetic background. Quant itat ive PCR (qPCR) of 16S r RNA gene revealed that the bacter ial load of EPI -/- mice skin was signif icantly higher than that of control mice ( Figure 1 A). However, pyrosequencing of 16S rRNA gene PCR products showed that the composition of the bacterial f lora of EPI-/- and control skin did not diff er signif icantly (Figure 1B,C).
The microbiota of EPI -/- skin was visua lized by whole -mount f luorescence in situ hybr idization (FISH) of the epiderm is with a universal bacter ial probe . There was punctate labelling on the surface of both EPI -/- and control skin (Figure 2 A). A FISH probe f or Candida dubul inienesis ser ved as a negative control (Biasoli et al. 2010 ) ( Figure 2B). In wild t ype epidermis , bacter ia wer e largely conf ined to the cornif ied layer s, which can be disting uished in vertical sections because the cells lack nuclei (Figure 2C). However, bacteria in EPI-/- epidermis wer e also localized bet ween keratinocyt es in the living layer s, extending in some case into the basal layer (arrowheads , Figure 2C). Quantitat ion of the label conf irm ed that bacter ia in EPI-/- epidermis were located signif icant ly deeper than in control epiderm is (Figure 2D; W T vs
5
© 2015 The Society for Investigative Dermatology
EPI-/- :17.13±0.04 vs 15.89 ±0.02m to median line of epider mis).
At opic features of EPI-/- mouse skin are unaffected b y skin microbiota To determine whether the incr eas ed bacterial load inf luenced t he at opic phenot ype of EPI -/- mice, we used t wo strategies to reduce the bact erial content of the skin. One was to breed EPI-/- mice f or mult iple generat ions under specif ic-pat hogen-f ree (SPF) barr ier condit ions (f lora- def icient EPI -/mice). The other was to treat them syst emically f or two weeks wit h the broad spectrum antibiotic enrof loxacin, a time-frame chosen on th e basis tha t the exaggerated a topic response to TPA tha t we desc ribed p re viou sly (Cipolat e t al., 2014) is induced wi th in 9 da ys. Both treatments reduced the bacterial load of
EPI-/- skin to that of W T control mice (Figure 3 A).
As previously reported (Sevilla et al. 2007 ; Cipolat et al. 2014 ), EPI-/- mice have more dermal CD4 + CD3 + lymphocytes than control mice ( Figure 3B). This diff erence persisted when EPI-/- mice were depleted of bacteria (Figure 3B). The number of e pidermal TCR + CD3 + lymphocytes ( dendritic epidermal T-cells, DETCs) is r educed in EPI -/- mice (Sevilla et al. 2007 ; Cipolat et al. 2014), and again this was not aff ected by reducing the number of skin bacter ia, when evaluated by expression of a pan -TCR marker (Figure 3C) or V3, a specif ic marker of epidermal DETC (Figure 3D).
6
© 2015 The Society for Investigative Dermatology
DETCs are involved in the lym phoid str ess -sur veillance response, in which tissue damage upr egulates “ stress-associated” genes, including Rae - 1, in keratinocyt es, leading to the act ivat ion of immunor eceptor NKG2D of immune cells ( Strid et al. 2009 ). Rae-1 express ion was increased in EPI -/- mice, as descr ibed previously (Cipolat et al. 2014 ) , and was not reduced by deplet ion of the skin micr obiot a ( Figure 3E). Neutr ophil inf iltrat ion, indicative of chronic inf lammation, is hig hly elevated in T P A t r e a t e d EPI-/- skin (Cipolat et al. 2014).
There was a slight (but not statist ically signif icant) incr ease in t he level of myeloperoxidase ( MPO), a marker of neutrophil act ivat ion, in untreated EPI-/skin (Figure 3F). MPO activit y was not , however, aff ected by deplet ion of the skin microbiot a ( Figure 3F). These findin gs we re supported b y quanti ta ting the number of Ly6 G/Gr -1 positi ve cells (neu trophils) in the de rmis of W T, EPI-/-, flora-deficient EP I-/- and an tibiotic -treated EPI-/- mice (Supplementary Figu re 1).
Hist ologically, untreated adult dorsal skin of EPI -/- mice showed m ild hyperkeratosis (Figure 4 A) and an exag gerated epidermal response to TPA that
includes
hyperkeratosis,
parakeratosis,
epidermal
thickening
and
spongiosis ( Figure 4B, C), as reported pr eviously (Sevilla et al. 2007 ; Cipolat et al. 2014). Deplet ion of the skin microbiot a by antibiotic treatment had no
7
© 2015 The Society for Investigative Dermatology
eff ect on these f eatures (Figure 4 A- C). In addit ion, antibiotics did not reduce spleen size in EPI-/- mice (Figure 4D).
EPI-/- mice have an increased level of epidermal AM Ps, proteases and protease inhi bitors , independent of ski n microbiota Since bacter ial load was not responsible f or the skin inf lammatory phenot ype or keratinocyt e-specif ic stress response of EPI-/- mice, we next invest igated whether the expression of AMPs was inf luenced by skin bact eria . Skin AMPs belong to several dist inct protein f amilies that t arget diff erent types of bacter ia
(Gallo
and
Hooper
2012 ) .
We
examined
expression
of
representatives of three of the f amilies by Q -PCR. EPI-/- epiderm is had elevated
level s
of
Camp –negat ive
gram-positive
and
( -def ensins;
gram -posit ive
(Cathelici din-r elated bacter ia) , and
Def b1,
–negative
ant imicrobial Def b2,
bact eria)
Def b3, and
pept ide ; Def b6 S100A9
(Calprotect ins; Staphylococcus aureus ) compared with control mice ( Figure 5 A-F). However, the expression of those genes was not aff ected by depletion of the skin m icrobiota ( Figure 5 A-F). Immunofluorescence labelling for 3-defensin and Cra mp in sections of mou se back sk in confirmed the se findings. We found that bo th E PI-/- and flora -deficient EPI -/- epidermis ha d more abundan t 3-defensin and Cra mp than W T epidermis (Supplementary F ig ure 2 ).
Antibiotic treatment r educes expression of intestinal AMPs, including Pla2g2a,
8
© 2015 The Society for Investigative Dermatology
Reg3b and Reg3g ( Reikvam et al. 2011 ). Intriguingly, these genes were not upregulated in EPI -/- skin and were not aff ected by t he deplet ion of skin microbiota ( Figure 5G-I). This suggests that the AMPs of the gut and skin respond diff erently t o bacter ial load .
Epidermal proteases are known to process native AMPs to their act ive f orm s (Ovaer e et al. 2009 ) and expression of one of the major skin serine prot ease inhibitor s, Ser pina1b, is increased in EPI -/- skin (Sevilla et al. 2007 ). In contrast to AMPs, Seprina1b levels were aff ected by deplet ion of skin bacter ia in EPI-/- mice, showing a signif icant incr ease in ant ibiotic -treat ed and SPF- housed animals (Figure 5J). Howe ver, when wild-type mice we re trea ted with antibio tics, epide rmal Serpina1b exp re ssion was no t al tered (Supplementary Figure 3).
In contrast to our earlier report (Sevilla et al. 2007 ), total epidermal protease activit y was incr eased in EPI-/- mice. However, total protease activit y was not aff ected by bacter ial depletion ( Figure 5K). Casein gel electrophoresis revealed t wo predominant pr oteins with protease activit y, which, based on electrophoretic mobilit y, likely correspond to elastase 2 and kallikrein 7 (Figure 5L). Both bands were mor e abundant in EPI -/- than W T skin, but were not reduced by bact erial depletion (Figure 5L). These data indicate that , like AMP expression, tot al pr oteases ar e unaff ected by bact erial load in the skin of
9
© 2015 The Society for Investigative Dermatology
EPI-/- mice, not withstanding the increase in Serpina1b levels .
DISCUSSION Our study demonstrates that the epidermal barrier def ect of EPI -/- mice result s in an incr ea sed bact erial load and deeper bacterial penetrat ion of the epidermis. Both phenomena are most likely attributable to the previously reported
f ragilit y
of
the
cornif ied
envelope
and
abnor malities
in
the
intercellular desmosomal juncti ons (Sevilla et al. 2007 ). The increase in bacter ia does no t, however, im pact on t he at opic phenot ype of EPI-/- skin, including inf lammation, upr egulat ion of Rae -1, AMPs and epidermal protease activit y, and does not ref lect any major changes in skin bacterial phyla.
The only exception to these obser vations was Serpina1b, which , as reported previously (Sevilla et al., 2007) , was elevated in EPI -/- skin relat ive to W T. Serpina1b levels were f urther increased by bacter ial deplet ion in EPI-/- but not W T skin. W hile t he under lying mechanism and biological signif icanc e of the Serpina1b f indings remain to be explored, it is worth not ing that elevated expression of Serpins contributes to barr ier dysf unction and inf lammation in allergen-treat ed mouse skin (Sivaprasad et al., 2015). It would be inter esting to exam ine wheth er EPI-/- mice exhibit increased allergic contact dermatit is , since Filaggrin-null mice, which also have a def ective bar rier, display an
10
© 2015 The Society for Investigative Dermatology
enhanced cut aneous response to ovalbumin sensitizat ion ( Kawasaki et al., 2012).
Our obser vations indicate that t he major tumour -protective mechanism identif ied in EPI -/- mice, namely signalling via Rae -1 and NKG2D, is attributable to the inherent structural def ects in the epidermal barrier and not to the incr ease in t he skin microbiota. This is in marked contrast to th e pro-tumorigenic eff ect of f lagellat ed bacteria that colonise skin wounds and signal via Toll like Receptor - 5 (Hoste et al., 2015a, b ).
Our study distinguishes phenot ypes r esulting f rom changes to the epidermal barrier f rom microbiota-induced phenot ypes. Genetic ablation of the serine protease matripase, which pr ocesses Fi laggrin and is required f or normal cornif ied envelope assembly, results in cha nges in the composition of t he skin microbiota and changes in AMP levels, but whether the t wo phenomena ar e related has not been explored
(Scharschmidt et al. 2009 ). The AMP
-def ensin-2 is upregulated in the skin of patients with atopic dermatit is (Ong et al. 2002 ; Asano et al. 2008 ) and S. aur eus inf ections are common in these patients (Ogawa et al. 1994 ). However, the s kin inf lam mation that is a hallmark of atopic dermatitis and other epidermal barrier disorders, such as congenital ichthyosis (Oji et al. 2010 ; Kabashima 2013 ), is not alleviated by
11
© 2015 The Society for Investigative Dermatology
antibiotic treatment . We propose that the altered skin m icrobio me associat ed with atopic dermatit is is a consequence, not a cause, of the disease.
12
© 2015 The Society for Investigative Dermatology
M ATERI AL S AND M ETHO DS Mice Mice def icient f or Ppl/Evpl/Ivl (EPI-/- mice) were generated as described previously on a mixed genetic background (Sevilla et al. 2007 ). Genome scanning performed b y the Jack son Labora tory (Cipolat e t al ., 20 14) established
that their genetic background comprises Sv129 (40.98%, ± 1.52), C57Bl/6 (51.39%, ± 1.62) and BALB/c (4.7%, ± 1.24). On that basis we selected the F2 generation of S v129 and C57BL/6J c rosse s (51.82%, ± 3.35 Sv129; 49.92%, ±
1.98 C57Bl/ 6) as the control f or all exper iments , since their ge netic back ground was 95% identical to that of EPI -/- mice. The F2 control mice were previously
used in studies of the suscept ibilit y of EPI -/- mice to chemical carcinogenesis (Cipolat et al. 2014 ). Matched control and EPI -/- mice were hou sed in the same enviro nments and ha d the same die t.
For some experiments multiple genera ti ons of EPI - /- and co ntrol mice were housed unde r SPF c onditions. A nal ysis was pe rformed on mice tha t had been born in pa thogen free conditions in an SPF barrier f acilit y. Entr y into the f acilit y
required staff to change into scr ubs and protective f oot wear, to take an air shower and use hand disinf ectant . Mice, whet her behind the barrier or in t he conventional f acilit y, were kept in individually vent ilated cages th at were only opened in a lam inar f low cabinet, and were only handled wit h gloved hands.
13
© 2015 The Society for Investigative Dermatology
For antibiotic treatm ent, oral enrof l oxacin (5mg/kg/day) was given to mice kept in the conventional f acilit y f or two weeks .
For TPA treatment, 6 nmol (3.7 g) of TPA (Sigma -Aldrich, Dorset, UK) in 200 l acetone was applied to mouse dorsal skin three times on alternating days (Cipolat et al. 2014 ). Skin, and in some cases, spleen, was harvested 48 hr af ter the last TPA treatment. Epidermal thickness was assessed on haematoxylin- eosin stained sections of TPA-treat ed skin using ImageJ sof tware. At least 6 fields were analysed per skin sect ion.
All exper iments were perf ormed under the terms of a UK gover nment Home Off ice project licence f ollowing inst itutional ethical review.
Anala ysis of bacterial load Extract ion of skin bacter ial DNA was perf ormed as descr ibed previously (Scharschmidt et al. 2009 ). Brief ly, 5mm diameter skin biopsies were collected f rom the ears of age and sex -matched mice using sterile instruments. DNA was extract ed with the PureLink Genome DNA Mini Kit (Lif e Technologies, NY) f ollowi ng the bead- beat ing step using Precellys 24 (Bertin Technolog ies, France). 16S rRNA gene qPCR was carried out using specif ic prim ers (Bact-8F:
AGAGTTTGATCCTGGCTCAG ,
Bact- 338R:
CTGCTGCCTCCCGTAGGAGT ) with f ast SYBR Green detection (Applied
14
© 2015 The Society for Investigative Dermatology
Biosystems) and run on an ABI Pr ism 7900HT Sequence detection system (Applied Biosystems) . Standard cur ve s were prepared by amplif ying serial dilut ion of known quantit ies of E. coli cells. The V1- V2 region of the 16S r RNA gene was amplif ied using barcoded f usion prim ers and PCR products were sequenced on
a 454GS FLSGX
platf orm.
Pyr osequencing
reads
were
uploaded into QII ME and precessed as described previously ( Caporaso et al. 2010).
Whole-mount fluorescent in sit u hybri dization Pieces of ear skin were collected with a sterilized 3mm biopsy punch. Skin samples were f ixed in 1:1 aceton e:methanol and incubated with PNA FISH probes at 55 0 C. A universal bacteria pr obe (BacUni, AdvanDx, MA) an d pr obe f or C. dubuliniensis (AdvanDx, MA) were used. DAPI was used f or nuclear staining. Images and Z-stacks of whole -mount FISH were obtained using a Leica TCS SP5 Tandem Scanner conf ocal microscope. Optical sectioning and Z stack maximum projection images of whole -mount preparations were produced using LAS AF sof tware (Leica). Evaluat ion of bacteria l locat ion was perf ormed on 3D r econstruct ed -skin using Volocit y sof twar e (PerkinElmer, MA). All the BacUni DAPI signals were detected using the sof tware. The median epidermal line was calculated using the localization of the center of epidermal DAPI signals. The distance of each BacUni signal f rom the median
15
© 2015 The Society for Investigative Dermatology
epidermal line was plotted. Positive values indicate signals above the median line while negative values correspond to signals below the median line.
Immunofluorescence Frozen sections wer e f ixed wit h 4% peraf orma ldehyde/PBS and stained wit h the
f ollowing
ant ibodies
conjugated
with
Alexa
488,
f luorescein
isothiocyanate (FITC) or Alexa 633: ant i- CD3 (BDPharmingen, clone 17. A2), anti- CD4 (eBioscience, clone RM4 - 5), anti-γδ TCR (BDPharmingen, clone GL3), anti- Vγ3 TCR (BDPharmingen, clone 536) or anti-Ly6G /Gr-1 (Beckman Coulter, clone RB6 - 8C5). Formalin-f ixed paraff in sections were treated with citrate buff er f or antigen retrieval and incubated with anti- Camp (Abcam) or anti- 3 def ensin (Alpha Diagnost ic) f ollowed by Alexa 488-conj ugated donk e y anti-rabbit ant ibody (Lif e Technologies). DAPI or PI was used f or nuclear staining. Images wer e acquired using a Leica TCS SP5 Tandem Scanner or an Olympus FV1000 conf ocal micr oscope. W hen quantif ying cells that expr essed a particular marker, at least 6 f i elds wer e analysed per skin section.
M yel operoxidase ELIS A For myeloper oxidase ( MPO) quantif icat ion, MPO ELISA kit (Hycult biotech, Nether lands) was used accor ding to the manuf acturer ’s protocol. 3mm punched ear skin was homogenized in lysis buff er (200m M NaCl, 5m M EDTA, 10m M Tris, 10% glycerin, 1m M PMSF) and centrif uged. The supernatants
16
© 2015 The Society for Investigative Dermatology
were used f or MPO ELISA. The absorbance was measured using Sunrise spectrophotomet er (TECAN, Swit zer land).
Quantitati ve RT-PCR (qRT- PCR) RNA was extracted from murine whole skin or f rom epiderm is that had been scraped f rom the skin f ollowing incubat ion in PBS f or 30 seconds at 56 0 C in PBS. The RNeasy mini kit (Qiagen, UK) was ut ilized according to the manuf acturer ’s instr uctions . RNA concentration
was measured using a
ND-100 NanoDrop spectrophotometer ( NanoDrop Technologies , DE). cDNA was synthesized using a Superscript III First -Strand Synthesis Supermix f or qRT-PCR kit (Invitr ogen) accor ding t o the manuf acturer ’s instructions. qRT-PCR was carr ied out using specif ic primers and f ast SYBR Green , and run on an ABI Prism 7900HT Sequence detect ion system. All samples were compared relative to mouse Gapdh as a house keeping contr ol. The f ollowing primers
were
used.
Rae -1:
CTAGTGCCACCTGGGAATTCA
CATCATTAGCTGATCTCCAGCTCA
(reverse);
(f orwar d), Cam p:
CTTCAACCAG CAGTCCCTAGACA (f orward), TCCAGGTCCAGGAGACGGTA (reverse);
Def b2:
GCCAT GAGACTCTCT GCTC
(f orward),
TGCAACAGGGGTTCTTCTCT (reverse); Def b3: ATTTCTCCTGGTGCTCGTGT (f orward),
GGAACTCCACAACTGCCAAT
CTGTTGCTACAAGAGCCTGG (reverse);
Reg 3b:
(f orwar d),
(reverse);
Pla2g2a:
TTTTCTTGTTCCGGGCGAAA
CCTTAGACCGTGCTTTCTGTG
(f orward),
17
© 2015 The Society for Investigative Dermatology
GTCCATGATGCTCTTCAAGACA
(rever se);
Reg3g:
ACATCAACTGGGAGACGAATC (f orward), TTTGGGATCTTGCTTGTGGCTA (reverse);
Serpina1b:
CATAGGAACGGCTTCAAAGA
TCCAGATCCATATCCCCAGA
(f orward),
(reverse);
AACATCAAATGGGGTGAGGCC
(f orward),
Gapdh:
GTTGTCATGGATGACCTTGGC
(reverse).
Epidermal protease acti vit y Epidermal lysat es were obtained as described previously ( Descargues et al. 2005).
Br ief ly,
Precellys24.
epidermis
Af ter
was
overnight
homogenized
extraction
at
in 4C,
1M
acetic
soluble
acid
using
proteins
were
lyophilized and resuspended in MiliQ water. Af ter acetone pr ecipit ation, the BCA protein assay kit (Pierce, IL) was used to quantif y protein concentration. 2ug of the soluble f raction was monit ored f or protease activit y using the EnzChek
Prot ease
Assay
Kit
(Lif e
Technologies) ,
according
to
the
manuf acturer ’s instr uctions. 5ug of the soluble f raction was loaded onto Novex Zymogram (Casein) Gel s (Lif e Technologies) f or electrophor esis. Gels were stained with Coomassie Br illiant blue.
Statistics Statist ical
analysis
was
perf ormed
using
GraphPad
Prism
(GraphPad
Sof twar e, CA). P values were det ermined with the unpaired st udent ’s t-test or
18
© 2015 The Society for Investigative Dermatology
one way ANO VA f ollowed by Tukey’s test. For analysis of bacterial depth in the skin, a non-par ametric Mann -W hitney test was perf omed. P- values are indicated
with:
*
0.01
**
0.01
***
0. 0001
****p<0.0001.
CONFLICT OF INTEREST The authors stat e no conf lict of interest.
ACKNOWLEDGEM ENTS
KN was the r ecipient of an overseas postdoctoral f ellowship f rom the Japan Societ y f or the Promotion of Science. Sara Cipolat was the recipient of a Feder ation of European Biochemical Societ ies Fellowship. FMW gratef ully acknowledges the f inancial support of the Wellcome Trust and MRC. W e thank Esther Hoste f or her input and the cor e facilities of the Cancer Research UK Cambr idge Research Institute, where some of this work was perf ormed, f or superb technical assistance.
19
© 2015 The Society for Investigative Dermatology
REFERENCES Asano S,
Ichikawa
Y,
Kumagai T,
et
al (2008).
Microanalysis of
an
antim icrobial peptide, beta -def ensin-2, in the stratum corneum f rom patients with at opic dermatit is. Br J Der matol 159: 97-104. Biasoli MS, Tosello ME, Luque AG, et al (2010). Adherence, colonizat ion and dissemination of Candida dubliniensis and other Candida species. Med Mycol 48: 291- 297. Candi E, Schmidt R, Melino G ( 2005). The cornif ied envelope: a model of cell death in the skin. Nat Rev Mol Cell Biol 6: 328-340. Caporaso JG, Kuczynski J, Stombaugh J, et al (2010). QII ME allows analysis of
high-throughput
communit y
sequencing
data.
Nat
Methods
7:
335-336. Cipolat S, Hoste E, Natsuga K, et al (2014). Epidermal bar rier def ects link atopic dermat itis with altered skin cancer suscept ibilit y. eLife 3: e01888. Descargues P, Deraison C, Bonnart C, et al (2005). Spink5-def icient mice mimic Net herton syndrome through degradation of desmoglein 1 b y epidermal protease hyperactivit y. Nat Genet 37: 56- 65. Gallo RL, Hooper LV (2012). Epit helial antimicrobial def ence of the skin and intest ine. Nat Rev I mmunol 12: 503- 516. Grice EA, Segre JA ( 2011). The skin microbiome. Nat Rev Microbiol 9: 244-253.
20
© 2015 The Society for Investigative Dermatology
Hamerman JA, Ogasawar a K, Lanier LL ( 2004). Cutt ing edge: Toll - like receptor signa ling in macrophages induces ligands f or the NKG2D receptor. J Immunol 172: 2001-2005. Hoste E, Ar wert EN, Lal R et al (2015a) . I nnate sensing of microbial product s promotes wound -induced skin cancer. Nat Commun 6 : 5 9 3 2 . Hoste E, Cipolat S, Watt FM (2015 b). Understanding allerg y and cancer risk: what are the barr iers? Nat Rev Cancer d o i : 1 0 . 1 0 3 8 / n r c 3 9 0 9 Kabashima K ( 2013) . New concept of the pathogenesis of atopic dermat itis: interplay among the barrier, allerg y, and pruritus as a trinit y. J Dermatol Sci 70: 3-11. Kawasaki H, Nagao K, Kubo A et al (2012). Altered stratum corneum barr ier and enhanced percut aneous immune responses in f ilaggrin -null mice. J Allerg y Clin Immunol . 129: 1538-1546. Lai Y, Di Nardo A, Nakatsuji T et al (2009). Commensal bacteria regulate Toll- like receptor 3 - dependent inf lammation af ter skin injur y. Nat Med 15: 1377-1382. Lodoen MB, Abenes G, Umamoto S et al (2004). The cytomegalovirus m155 gene
product
subverts
natur al
killer
cell
ant iviral
prot ection
by
disruption of H60 - NKG2D int eract ions. J Exp Med 200: 1075- 1081. Naik S, Bouladoux N, W ilhelm C et al ( 2012). Compartmentalized control of skin immunit y by resident commensals. Science 337: 1115- 1119. Natsuga K ( 2014). Epidermal barrier s. Cold Spring Harb Perspect Med 4:
21
© 2015 The Society for Investigative Dermatology
a018218. Ogawa T, Katsuoka K, Kawano K et al ( 1994). Comparative study of staphylococcal f lora on the skin surf ace of atopic dermat itis patients and healthy subjects. J Dermatol 21: 453 -460. Oji V, Tadini G, Akiyam a M et al ( 2010). Revised nom enclature and classif icat ion of inherited ichthyoses: results of the First Ichthyosis Consensus Conf erence in Sor eze 2009. J Am Acad Der matol 63: 607-641. Ong PY, Ohtake T, Brandt C et al (2002). Endogenous anti m icrobial peptides and skin inf ections in atopic dermat it is. N Engl J Med 347: 1151-1160. Ovaere P, Lippens S, Vandenabeele P et al (2009). The emerging roles of serine pr otease cascades in the epider mis. Trends Biochem Sci 34: 453-463. Reikvam DH, Erof eev A, Sandvik A et al (2011). Depletion of murine int est i na l microbiota: eff ects on gut mucosa and epithelial gene expression. PLoS One 6: e17996. Scharschmidt TC, List K, Grice EA et al (2009). Matriptase -def icient mice exhibit ichthyot ic skin with a select ive shif t in skin micr obiota. J Invest Dermatol 129: 2435- 2442. Sevilla LM, Nachat R, Groot KR et al ( 2007). Mice def icient in involucr in, envoplakin, and periplakin have a def ective epidermal barrier. J Cell Biol 179: 1599-1612.
22
© 2015 The Society for Investigative Dermatology
Shanahan F, O'Toole PW (2014). Host-microbe inter actions and spat ial variat ion of cancer in the gut. Nat Rev Cancer 14: 511-512. Sivaprasad U, Kinker KG, Ericksen MB et al (2015). SERPINB3/B4 contributes to early inflammation and barrier dysfunction in an experimental murine model of atopic dermatitis. J Invest Dermatol.
135:160-169.
Spear P, W u MR, Sentman ML et al (2013). NKG2D ligands as therapeut ic targets. Cancer Immun 13: 8. Strid J, Roberts SJ, Filler RB et al (2008). Acute upregulat ion of an NKG2D ligand promotes rapid reorganizat ion of a local immune compartment with pleiotropic eff ects on carcinogenesis. Nat Immunol 9: 146-154. Strid J, Tigelaar RE, Hayday AC (2009). Skin immune surveillance by T cells--a new order ? Semin Immunol 21: 110-120.
23
© 2015 The Society for Investigative Dermatology
FIGURE LEGENDS Figure 1. Characterization of skin microbiota in EPI -/- and WT mice ( A) qPCR of 16sRNA gene in EPI -/- and W T ear skin. Data ar e means ± SEM f rom 6 mice per group. ** p<0.01. (B, C) Micr obial diversit y of EPI-/- and W T ear skin obtained by 16sRNA gene pyroseq uencing f rom 6 mice per group . (B) Data are presented as means of relative abundance ± SEM. ( C) Each colum n shows relat ive abundance of the bact eria l phyla in a single mouse . There were no sta tis tically significant difference s be tween W T and EPI -/- sk in microbiota phyla .
Figure 2. Epidermal penetration by bacteria ( A, B) W hole-mount FISH of ear skin using BacUni ( A) or C. dubuliniensis ( B) pr obes (green), with DAPI counterst ain (blue) . The images are Z-stack maximum projection s of the epidermis. Scale bars: 20 m. (C) Optical sectioning of 3D reconstruct ed whole-mount FISH (BacUni). Arrowheads show bacter ial penetration of the basal epidermal layer. Dotted line indicates posit ion of basement membrane . Scale bars: 10 m. ( D) Quantit ation of location of individual BacUni signals relat ive to the median thickness (dotted line) of the epider mis. The boxes extend f rom the 25 t h to 75 t h per cent iles. W hiskers indicate minimum to maximum. **** p<0.0001 .
Figure 3. Microbiota independent -inflammatory and stress response
24
© 2015 The Society for Investigative Dermatology
phenot ypes of EPI -/- skin ( A) qPCR of 16sRNA gene in W T, EPI -/-, antibiotic-treated EPI -/- and f lora- def icient EPI -/- ear skin samples. ( B) Number of CD4+CD3+ cells per mm 2 der mis. (C) Number of TCR+ cells per mm epidermis. ( D) Number of V3+ cells per mm epiderm is. ( E) qRT-PCR of Rae1 in W T, EPI-/-, antibiotic-treated EPI -/- and f lora-def icient EPI -/epidermis.
( F)
Quantif icat ion
of
myeloperoxidase
in
W T,
EPI-/-
and
f lora-def icient skin lysates. Dat a are means ± SEM f rom at least 4 (A) or 3 (B-F) mice per group. * p<0.05, *** p <0.001; unless indicated no signif icant diff erences wer e f ound ( one-wa y ANOVA followed b y Tuk ey’s te st).
Figure 4. Effect of antibiotic treatment on skin histology. ( A) H&E stained skin sect ions of W T, EPI -/- and antibiotics -treated EPI -/- at steady stat e. Arrowheads: mild hyperkeratosis. ( B) H&E stained sect ions of W T, EPI-/- and antibiotic-treated
EPI -/-
skin
af ter
three
applicat ion s
of
TPA.
Black
arrowheads: mild hyperkeratosis; w hite arrowhead s: severe hyperkeratosis; asterisk s: parakeratosis; black arrows: spongiosis . Scale bar s: 100 um. (C, D) Epidermal
thicknes s
(C)
and
spleen
mass
( D)
of
W T,
EPI -/-
and
antibiotic-treated EPI -/- mice. TPA was applied thr ee times on alternat ing days on all the mice (n = 3 or 4 per condition) . ** p<0.01, *** p<0.001.
Figure 5. AM P expressi on and prot ease acti vit y in EPI-/- epidermis follow ing microbiota modulation . ( A-K) qRT-PCR of Camp ( A) , Def b1 (B),
25
© 2015 The Society for Investigative Dermatology
Def b2 (C), Def b3 ( D), Def b6 (E), S100A9 ( F), Pla2g2a ( G), Reg3b ( H), Reg3g (I) and Serpina1b ( J) in W T, EPI-/-, antibiotic-treat ed EPI -/- and f lora-def icient EPI-/- epidermis. Data are means ± SEM f rom at least 4 m ice per group. * p<0.05, ** p<0.01 . (K) Total e pidermal protease activit ies of W T, EPI -/-, antibiotic-treated EPI -/- and f lora-def ient EPI -/- skin. Data ar e means ± SEM f rom 3-4 mice per group. (L) Casein gel electrophoresis of epidermal lysates. Each lane is lysate f rom a diff erent mouse. Posit ions of m olecular mass markers (31kDa, 17kDa) are shown. The t wo major bands of p roteolyt ic activit y are indicated by arrowheads .
26
© 2015 The Society for Investigative Dermatology
© 2015 The Society
© 2015 The Society
© 2015 The Society
© 2015 The Society
© 2015 The Society
Increased Bacterial Load and Expression of Antimicrobial Peptides in Skin of Barrier-Deficient Mice with Reduced Cancer Susceptibility Ken Natsuga, Sara Cipolat, Fiona M Watt
Cite this article as: Ken Natsuga, Sara Cipolat, Fiona M Watt, Increased Bacterial Load and Expression of Antimicrobial Peptides in Skin of Barrier-Deficient Mice with Reduced Cancer Susceptibility, Journal of Investigative Dermatology accepted article preview 30 September 2015; doi: 10.1038/jid.2015.383. This is a PDF file of an unedited peer-reviewed manuscript that has been accepted for publication. NPG are providing this early version of the manuscript as a service to our customers. The manuscript will undergo copyediting, typesetting and a proof review before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers apply.
Received 22 February 2015; revised 24 July 2015; accepted 11 September 2015; Accepted article preview online 30 September 2015
© 2015 The Society for Investigative Dermatology
Increased bacteri al load and expression of antimicrobial peptides i n skin of barrier-deficient mice w ith reduced cancer susceptibility
Ken Natsuga 1 , 2 , Sara Cipolat 1 , 3 , and Fiona M. W att 3 , *
1
Cancer Research UK Cambr idge Research Institut e, Li Ka Shing Centre,
Robinson W ay, Cam bridge CB2 0RE, UK; 2 Department of Der matolog y, Hokkaido Universit y Graduate School of Medicine N15W 7, Sapporo 060 -8638, Japan; 3 Centre f or Stem Cells and Regenerative Medicine, King’s College London, 28t h f loor, Guy’s Tower W ing, London SE1 9RT, UK
*Corresponding author: Fiona. [email protected]. uk
Running title: bacter ial load in barr ier def icient mice
1
© 2015 The Society for Investigative Dermatology
ABSTR ACT Mice lacking three epidermal barr ier pr oteins – envoplakin, periplakin and involucrin - (EPI-/- mice) have a def ective cornif ied layer, reduced epidermal T cells, increased dermal CD4+ T cells and are resistant to developing skin tumours. The tumour-protect ive mechanism involves signalling bet ween Rae-1 expressing keratinocytes and the N a t u r a l K i l l e r G r o u p 2 D ( N K G 2 D ) receptor on immune cells, which also plays a role in host defences a g a i n s t i n f e c t i o n . Given the emerging link bet ween bacter ia and cancer, we investigated whether EPI -/- mice ha ve an altered skin microbiota . The bacter ial phyla wer e similar in wild type and EPI -/- skin. However, bact eria were 3-f old more abundant in EPI-/- skin and penetrated deeper into the epidermis. The major epithelial def ense mechanism against bacteria is product ion of antimicrobial proteins (AMPs). EPI-/- skin exhibited enhanced expression of antimicrobial pept ides. However, r educing the bacter ial load b y antibiotic treatment or breeding mice under specif ic pathogen -f ree condit ions did not reduce AMP expression or alleviate the abnormalit ies in T cell populat ions. W e conclude that the atopic character istics of EPI-/- skin are a consequence of the def ective barr ier rat her than a response to the increase d bacter ial load. It is theref ore unlikel y t hat the increase in skin microbiota contributes dir ectly t o the obser ved cancer resist ance.
2
© 2015 The Society for Investigative Dermatology
INTRO DUCTION In cells of the outer most epidermal layers the plasma membrane is replaced by a cornif ied envelope, f ormed by transglutam inase -1-mediated crosslinking of proteins and lip ids (Nat suga 2014 ). This insoluble structure is a key component of the epidermal barr ier that protects the body f rom envir onment al assaults ( Candi et al. 2005 ). Def ects in the epidermal barr ier are linked to atopic dermat itis in humans and mice (Natsuga 2014 ). Mice t riply def icient in three cornif ied envelope precur sors - envoplakin, periplakin and involucr in (EPI-/- mice) - have epidermal barr ier def ects , a reduction in epidermal T cells and an increase in dermal CD4+ T cells (Sevilla et al. 2007 ).
EPI-/- mice are highly resist ant to developing skin tumours when treated with chemical carcinogens (Cipolat et al. 2014 ). This is due, at least in part, to an exaggerated
atopic
response
to
the
tumour
promoter
1 2 - O - t e t r a d e c a n o y l p h o r b o l - 1 3 - a c e t a t e ( T P A ) , including upregulat ion of the keratinocyte stress-associated antigen Rae -1, which binds NKG2D on immune cells. NKG2D ligands are curr ently being evaluated as therapeut ic targets in a var iet y of cancer types (Spear et al. 2013 ).
Rae-1- NKG2D interactions are not only implicated in cancer ( Strid et al. 2008 ; Cipolat et al. 2014 ), but also mediate host responses to bacter ial and vir al inf ection (Hamerman et al. 2004 ; Lodoen et al. 2004 ; Str id et al. 2008 ; Strid et
3
© 2015 The Society for Investigative Dermatology
al. 2009). Given the strong evidence linking the gut micr obiota, chr onic inf lammation and cancer (Shanahan and O'Toole 2014 ), we speculat ed that the skin microbiome might be alter ed in EPI -/- mice. Human skin is covered by wide var iet y of commensal microbiota , which var y in com position accor ding to body site, age and healt h status (Grice and Segre 2011 ). In addit ion, r ecent studies suggest a regulator y role f or skin commensal bacter ia in skin inf ection and inf lammation (Lai et al. 2009 ; Naik et al. 2012 ). W e theref ore investigated whether the skin microbiome dif f ers between EPI -/- and wild t ype (W T) mice and whether it inf luences keratinocyte expression of Rae -1 and anti-m icrobial proteins (AMPs), nat ural antibiotics that are eff ectors of the innate immune system (Gallo and Hooper 2012 ).
4
© 2015 The Society for Investigative Dermatology
RESULTS Al t ered abundance and distri bution of bacteria in t he ski n of EPI-/- mice To determine whether EPI -/- mice had an alter ed skin m icrobiome, we measured the quantity and diversit y of microbes on the ear skin of six EPI-/and six wild t ype control m ice on the same genetic background. Quant itat ive PCR (qPCR) of 16S r RNA gene revealed that the bacter ial load of EPI -/- mice skin was signif icantly higher than that of control mice ( Figure 1 A). However, pyrosequencing of 16S rRNA gene PCR products showed that the composition of the bacterial f lora of EPI-/- and control skin did not diff er signif icantly (Figure 1B,C).
The microbiota of EPI -/- skin was visua lized by whole -mount f luorescence in situ hybr idization (FISH) of the epiderm is with a universal bacter ial probe . There was punctate labelling on the surface of both EPI -/- and control skin (Figure 2 A). A FISH probe f or Candida dubul inienesis ser ved as a negative control (Biasoli et al. 2010 ) ( Figure 2B). In wild t ype epidermis , bacter ia wer e largely conf ined to the cornif ied layer s, which can be disting uished in vertical sections because the cells lack nuclei (Figure 2C). However, bacteria in EPI-/- epidermis wer e also localized bet ween keratinocyt es in the living layer s, extending in some case into the basal layer (arrowheads , Figure 2C). Quantitat ion of the label conf irm ed that bacter ia in EPI-/- epidermis were located signif icant ly deeper than in control epiderm is (Figure 2D; W T vs
5
© 2015 The Society for Investigative Dermatology
EPI-/- :17.13±0.04 vs 15.89 ±0.02m to median line of epider mis).
At opic features of EPI-/- mouse skin are unaffected b y skin microbiota To determine whether the incr eas ed bacterial load inf luenced t he at opic phenot ype of EPI -/- mice, we used t wo strategies to reduce the bact erial content of the skin. One was to breed EPI-/- mice f or mult iple generat ions under specif ic-pat hogen-f ree (SPF) barr ier condit ions (f lora- def icient EPI -/mice). The other was to treat them syst emically f or two weeks wit h the broad spectrum antibiotic enrof loxacin, a time-frame chosen on th e basis tha t the exaggerated a topic response to TPA tha t we desc ribed p re viou sly (Cipolat e t al., 2014) is induced wi th in 9 da ys. Both treatments reduced the bacterial load of
EPI-/- skin to that of W T control mice (Figure 3 A).
As previously reported (Sevilla et al. 2007 ; Cipolat et al. 2014 ), EPI-/- mice have more dermal CD4 + CD3 + lymphocytes than control mice ( Figure 3B). This diff erence persisted when EPI-/- mice were depleted of bacteria (Figure 3B). The number of e pidermal TCR + CD3 + lymphocytes ( dendritic epidermal T-cells, DETCs) is r educed in EPI -/- mice (Sevilla et al. 2007 ; Cipolat et al. 2014), and again this was not aff ected by reducing the number of skin bacter ia, when evaluated by expression of a pan -TCR marker (Figure 3C) or V3, a specif ic marker of epidermal DETC (Figure 3D).
6
© 2015 The Society for Investigative Dermatology
DETCs are involved in the lym phoid str ess -sur veillance response, in which tissue damage upr egulates “ stress-associated” genes, including Rae - 1, in keratinocyt es, leading to the act ivat ion of immunor eceptor NKG2D of immune cells ( Strid et al. 2009 ). Rae-1 express ion was increased in EPI -/- mice, as descr ibed previously (Cipolat et al. 2014 ) , and was not reduced by deplet ion of the skin micr obiot a ( Figure 3E). Neutr ophil inf iltrat ion, indicative of chronic inf lammation, is hig hly elevated in T P A t r e a t e d EPI-/- skin (Cipolat et al. 2014).
There was a slight (but not statist ically signif icant) incr ease in t he level of myeloperoxidase ( MPO), a marker of neutrophil act ivat ion, in untreated EPI-/skin (Figure 3F). MPO activit y was not , however, aff ected by deplet ion of the skin microbiot a ( Figure 3F). These findin gs we re supported b y quanti ta ting the number of Ly6 G/Gr -1 positi ve cells (neu trophils) in the de rmis of W T, EPI-/-, flora-deficient EP I-/- and an tibiotic -treated EPI-/- mice (Supplementary Figu re 1).
Hist ologically, untreated adult dorsal skin of EPI -/- mice showed m ild hyperkeratosis (Figure 4 A) and an exag gerated epidermal response to TPA that
includes
hyperkeratosis,
parakeratosis,
epidermal
thickening
and
spongiosis ( Figure 4B, C), as reported pr eviously (Sevilla et al. 2007 ; Cipolat et al. 2014). Deplet ion of the skin microbiot a by antibiotic treatment had no
7
© 2015 The Society for Investigative Dermatology
eff ect on these f eatures (Figure 4 A- C). In addit ion, antibiotics did not reduce spleen size in EPI-/- mice (Figure 4D).
EPI-/- mice have an increased level of epidermal AM Ps, proteases and protease inhi bitors , independent of ski n microbiota Since bacter ial load was not responsible f or the skin inf lammatory phenot ype or keratinocyt e-specif ic stress response of EPI-/- mice, we next invest igated whether the expression of AMPs was inf luenced by skin bact eria . Skin AMPs belong to several dist inct protein f amilies that t arget diff erent types of bacter ia
(Gallo
and
Hooper
2012 ) .
We
examined
expression
of
representatives of three of the f amilies by Q -PCR. EPI-/- epiderm is had elevated
level s
of
Camp –negat ive
gram-positive
and
( -def ensins;
gram -posit ive
(Cathelici din-r elated bacter ia) , and
Def b1,
–negative
ant imicrobial Def b2,
bact eria)
Def b3, and
pept ide ; Def b6 S100A9
(Calprotect ins; Staphylococcus aureus ) compared with control mice ( Figure 5 A-F). However, the expression of those genes was not aff ected by depletion of the skin m icrobiota ( Figure 5 A-F). Immunofluorescence labelling for 3-defensin and Cra mp in sections of mou se back sk in confirmed the se findings. We found that bo th E PI-/- and flora -deficient EPI -/- epidermis ha d more abundan t 3-defensin and Cra mp than W T epidermis (Supplementary F ig ure 2 ).
Antibiotic treatment r educes expression of intestinal AMPs, including Pla2g2a,
8
© 2015 The Society for Investigative Dermatology
Reg3b and Reg3g ( Reikvam et al. 2011 ). Intriguingly, these genes were not upregulated in EPI -/- skin and were not aff ected by t he deplet ion of skin microbiota ( Figure 5G-I). This suggests that the AMPs of the gut and skin respond diff erently t o bacter ial load .
Epidermal proteases are known to process native AMPs to their act ive f orm s (Ovaer e et al. 2009 ) and expression of one of the major skin serine prot ease inhibitor s, Ser pina1b, is increased in EPI -/- skin (Sevilla et al. 2007 ). In contrast to AMPs, Seprina1b levels were aff ected by deplet ion of skin bacter ia in EPI-/- mice, showing a signif icant incr ease in ant ibiotic -treat ed and SPF- housed animals (Figure 5J). Howe ver, when wild-type mice we re trea ted with antibio tics, epide rmal Serpina1b exp re ssion was no t al tered (Supplementary Figure 3).
In contrast to our earlier report (Sevilla et al. 2007 ), total epidermal protease activit y was incr eased in EPI-/- mice. However, total protease activit y was not aff ected by bacter ial depletion ( Figure 5K). Casein gel electrophoresis revealed t wo predominant pr oteins with protease activit y, which, based on electrophoretic mobilit y, likely correspond to elastase 2 and kallikrein 7 (Figure 5L). Both bands were mor e abundant in EPI -/- than W T skin, but were not reduced by bact erial depletion (Figure 5L). These data indicate that , like AMP expression, tot al pr oteases ar e unaff ected by bact erial load in the skin of
9
© 2015 The Society for Investigative Dermatology
EPI-/- mice, not withstanding the increase in Serpina1b levels .
DISCUSSION Our study demonstrates that the epidermal barrier def ect of EPI -/- mice result s in an incr ea sed bact erial load and deeper bacterial penetrat ion of the epidermis. Both phenomena are most likely attributable to the previously reported
f ragilit y
of
the
cornif ied
envelope
and
abnor malities
in
the
intercellular desmosomal juncti ons (Sevilla et al. 2007 ). The increase in bacter ia does no t, however, im pact on t he at opic phenot ype of EPI-/- skin, including inf lammation, upr egulat ion of Rae -1, AMPs and epidermal protease activit y, and does not ref lect any major changes in skin bacterial phyla.
The only exception to these obser vations was Serpina1b, which , as reported previously (Sevilla et al., 2007) , was elevated in EPI -/- skin relat ive to W T. Serpina1b levels were f urther increased by bacter ial deplet ion in EPI-/- but not W T skin. W hile t he under lying mechanism and biological signif icanc e of the Serpina1b f indings remain to be explored, it is worth not ing that elevated expression of Serpins contributes to barr ier dysf unction and inf lammation in allergen-treat ed mouse skin (Sivaprasad et al., 2015). It would be inter esting to exam ine wheth er EPI-/- mice exhibit increased allergic contact dermatit is , since Filaggrin-null mice, which also have a def ective bar rier, display an
10
© 2015 The Society for Investigative Dermatology
enhanced cut aneous response to ovalbumin sensitizat ion ( Kawasaki et al., 2012).
Our obser vations indicate that t he major tumour -protective mechanism identif ied in EPI -/- mice, namely signalling via Rae -1 and NKG2D, is attributable to the inherent structural def ects in the epidermal barrier and not to the incr ease in t he skin microbiota. This is in marked contrast to th e pro-tumorigenic eff ect of f lagellat ed bacteria that colonise skin wounds and signal via Toll like Receptor - 5 (Hoste et al., 2015a, b ).
Our study distinguishes phenot ypes r esulting f rom changes to the epidermal barrier f rom microbiota-induced phenot ypes. Genetic ablation of the serine protease matripase, which pr ocesses Fi laggrin and is required f or normal cornif ied envelope assembly, results in cha nges in the composition of t he skin microbiota and changes in AMP levels, but whether the t wo phenomena ar e related has not been explored
(Scharschmidt et al. 2009 ). The AMP
-def ensin-2 is upregulated in the skin of patients with atopic dermatit is (Ong et al. 2002 ; Asano et al. 2008 ) and S. aur eus inf ections are common in these patients (Ogawa et al. 1994 ). However, the s kin inf lam mation that is a hallmark of atopic dermatitis and other epidermal barrier disorders, such as congenital ichthyosis (Oji et al. 2010 ; Kabashima 2013 ), is not alleviated by
11
© 2015 The Society for Investigative Dermatology
antibiotic treatment . We propose that the altered skin m icrobio me associat ed with atopic dermatit is is a consequence, not a cause, of the disease.
12
© 2015 The Society for Investigative Dermatology
M ATERI AL S AND M ETHO DS Mice Mice def icient f or Ppl/Evpl/Ivl (EPI-/- mice) were generated as described previously on a mixed genetic background (Sevilla et al. 2007 ). Genome scanning performed b y the Jack son Labora tory (Cipolat e t al ., 20 14) established
that their genetic background comprises Sv129 (40.98%, ± 1.52), C57Bl/6 (51.39%, ± 1.62) and BALB/c (4.7%, ± 1.24). On that basis we selected the F2 generation of S v129 and C57BL/6J c rosse s (51.82%, ± 3.35 Sv129; 49.92%, ±
1.98 C57Bl/ 6) as the control f or all exper iments , since their ge netic back ground was 95% identical to that of EPI -/- mice. The F2 control mice were previously
used in studies of the suscept ibilit y of EPI -/- mice to chemical carcinogenesis (Cipolat et al. 2014 ). Matched control and EPI -/- mice were hou sed in the same enviro nments and ha d the same die t.
For some experiments multiple genera ti ons of EPI - /- and co ntrol mice were housed unde r SPF c onditions. A nal ysis was pe rformed on mice tha t had been born in pa thogen free conditions in an SPF barrier f acilit y. Entr y into the f acilit y
required staff to change into scr ubs and protective f oot wear, to take an air shower and use hand disinf ectant . Mice, whet her behind the barrier or in t he conventional f acilit y, were kept in individually vent ilated cages th at were only opened in a lam inar f low cabinet, and were only handled wit h gloved hands.
13
© 2015 The Society for Investigative Dermatology
For antibiotic treatm ent, oral enrof l oxacin (5mg/kg/day) was given to mice kept in the conventional f acilit y f or two weeks .
For TPA treatment, 6 nmol (3.7 g) of TPA (Sigma -Aldrich, Dorset, UK) in 200 l acetone was applied to mouse dorsal skin three times on alternating days (Cipolat et al. 2014 ). Skin, and in some cases, spleen, was harvested 48 hr af ter the last TPA treatment. Epidermal thickness was assessed on haematoxylin- eosin stained sections of TPA-treat ed skin using ImageJ sof tware. At least 6 fields were analysed per skin sect ion.
All exper iments were perf ormed under the terms of a UK gover nment Home Off ice project licence f ollowing inst itutional ethical review.
Anala ysis of bacterial load Extract ion of skin bacter ial DNA was perf ormed as descr ibed previously (Scharschmidt et al. 2009 ). Brief ly, 5mm diameter skin biopsies were collected f rom the ears of age and sex -matched mice using sterile instruments. DNA was extract ed with the PureLink Genome DNA Mini Kit (Lif e Technologies, NY) f ollowi ng the bead- beat ing step using Precellys 24 (Bertin Technolog ies, France). 16S rRNA gene qPCR was carried out using specif ic prim ers (Bact-8F:
AGAGTTTGATCCTGGCTCAG ,
Bact- 338R:
CTGCTGCCTCCCGTAGGAGT ) with f ast SYBR Green detection (Applied
14
© 2015 The Society for Investigative Dermatology
Biosystems) and run on an ABI Pr ism 7900HT Sequence detection system (Applied Biosystems) . Standard cur ve s were prepared by amplif ying serial dilut ion of known quantit ies of E. coli cells. The V1- V2 region of the 16S r RNA gene was amplif ied using barcoded f usion prim ers and PCR products were sequenced on
a 454GS FLSGX
platf orm.
Pyr osequencing
reads
were
uploaded into QII ME and precessed as described previously ( Caporaso et al. 2010).
Whole-mount fluorescent in sit u hybri dization Pieces of ear skin were collected with a sterilized 3mm biopsy punch. Skin samples were f ixed in 1:1 aceton e:methanol and incubated with PNA FISH probes at 55 0 C. A universal bacteria pr obe (BacUni, AdvanDx, MA) an d pr obe f or C. dubuliniensis (AdvanDx, MA) were used. DAPI was used f or nuclear staining. Images and Z-stacks of whole -mount FISH were obtained using a Leica TCS SP5 Tandem Scanner conf ocal microscope. Optical sectioning and Z stack maximum projection images of whole -mount preparations were produced using LAS AF sof tware (Leica). Evaluat ion of bacteria l locat ion was perf ormed on 3D r econstruct ed -skin using Volocit y sof twar e (PerkinElmer, MA). All the BacUni DAPI signals were detected using the sof tware. The median epidermal line was calculated using the localization of the center of epidermal DAPI signals. The distance of each BacUni signal f rom the median
15
© 2015 The Society for Investigative Dermatology
epidermal line was plotted. Positive values indicate signals above the median line while negative values correspond to signals below the median line.
Immunofluorescence Frozen sections wer e f ixed wit h 4% peraf orma ldehyde/PBS and stained wit h the
f ollowing
ant ibodies
conjugated
with
Alexa
488,
f luorescein
isothiocyanate (FITC) or Alexa 633: ant i- CD3 (BDPharmingen, clone 17. A2), anti- CD4 (eBioscience, clone RM4 - 5), anti-γδ TCR (BDPharmingen, clone GL3), anti- Vγ3 TCR (BDPharmingen, clone 536) or anti-Ly6G /Gr-1 (Beckman Coulter, clone RB6 - 8C5). Formalin-f ixed paraff in sections were treated with citrate buff er f or antigen retrieval and incubated with anti- Camp (Abcam) or anti- 3 def ensin (Alpha Diagnost ic) f ollowed by Alexa 488-conj ugated donk e y anti-rabbit ant ibody (Lif e Technologies). DAPI or PI was used f or nuclear staining. Images wer e acquired using a Leica TCS SP5 Tandem Scanner or an Olympus FV1000 conf ocal micr oscope. W hen quantif ying cells that expr essed a particular marker, at least 6 f i elds wer e analysed per skin section.
M yel operoxidase ELIS A For myeloper oxidase ( MPO) quantif icat ion, MPO ELISA kit (Hycult biotech, Nether lands) was used accor ding to the manuf acturer ’s protocol. 3mm punched ear skin was homogenized in lysis buff er (200m M NaCl, 5m M EDTA, 10m M Tris, 10% glycerin, 1m M PMSF) and centrif uged. The supernatants
16
© 2015 The Society for Investigative Dermatology
were used f or MPO ELISA. The absorbance was measured using Sunrise spectrophotomet er (TECAN, Swit zer land).
Quantitati ve RT-PCR (qRT- PCR) RNA was extracted from murine whole skin or f rom epiderm is that had been scraped f rom the skin f ollowing incubat ion in PBS f or 30 seconds at 56 0 C in PBS. The RNeasy mini kit (Qiagen, UK) was ut ilized according to the manuf acturer ’s instr uctions . RNA concentration
was measured using a
ND-100 NanoDrop spectrophotometer ( NanoDrop Technologies , DE). cDNA was synthesized using a Superscript III First -Strand Synthesis Supermix f or qRT-PCR kit (Invitr ogen) accor ding t o the manuf acturer ’s instructions. qRT-PCR was carr ied out using specif ic primers and f ast SYBR Green , and run on an ABI Prism 7900HT Sequence detect ion system. All samples were compared relative to mouse Gapdh as a house keeping contr ol. The f ollowing primers
were
used.
Rae -1:
CTAGTGCCACCTGGGAATTCA
CATCATTAGCTGATCTCCAGCTCA
(reverse);
(f orwar d), Cam p:
CTTCAACCAG CAGTCCCTAGACA (f orward), TCCAGGTCCAGGAGACGGTA (reverse);
Def b2:
GCCAT GAGACTCTCT GCTC
(f orward),
TGCAACAGGGGTTCTTCTCT (reverse); Def b3: ATTTCTCCTGGTGCTCGTGT (f orward),
GGAACTCCACAACTGCCAAT
CTGTTGCTACAAGAGCCTGG (reverse);
Reg 3b:
(f orwar d),
(reverse);
Pla2g2a:
TTTTCTTGTTCCGGGCGAAA
CCTTAGACCGTGCTTTCTGTG
(f orward),
17
© 2015 The Society for Investigative Dermatology
GTCCATGATGCTCTTCAAGACA
(rever se);
Reg3g:
ACATCAACTGGGAGACGAATC (f orward), TTTGGGATCTTGCTTGTGGCTA (reverse);
Serpina1b:
CATAGGAACGGCTTCAAAGA
TCCAGATCCATATCCCCAGA
(f orward),
(reverse);
AACATCAAATGGGGTGAGGCC
(f orward),
Gapdh:
GTTGTCATGGATGACCTTGGC
(reverse).
Epidermal protease acti vit y Epidermal lysat es were obtained as described previously ( Descargues et al. 2005).
Br ief ly,
Precellys24.
epidermis
Af ter
was
overnight
homogenized
extraction
at
in 4C,
1M
acetic
soluble
acid
using
proteins
were
lyophilized and resuspended in MiliQ water. Af ter acetone pr ecipit ation, the BCA protein assay kit (Pierce, IL) was used to quantif y protein concentration. 2ug of the soluble f raction was monit ored f or protease activit y using the EnzChek
Prot ease
Assay
Kit
(Lif e
Technologies) ,
according
to
the
manuf acturer ’s instr uctions. 5ug of the soluble f raction was loaded onto Novex Zymogram (Casein) Gel s (Lif e Technologies) f or electrophor esis. Gels were stained with Coomassie Br illiant blue.
Statistics Statist ical
analysis
was
perf ormed
using
GraphPad
Prism
(GraphPad
Sof twar e, CA). P values were det ermined with the unpaired st udent ’s t-test or
18
© 2015 The Society for Investigative Dermatology
one way ANO VA f ollowed by Tukey’s test. For analysis of bacterial depth in the skin, a non-par ametric Mann -W hitney test was perf omed. P- values are indicated
with:
*
0.01
**
0.01
***
0. 0001
****p<0.0001.
CONFLICT OF INTEREST The authors stat e no conf lict of interest.
ACKNOWLEDGEM ENTS
KN was the r ecipient of an overseas postdoctoral f ellowship f rom the Japan Societ y f or the Promotion of Science. Sara Cipolat was the recipient of a Feder ation of European Biochemical Societ ies Fellowship. FMW gratef ully acknowledges the f inancial support of the Wellcome Trust and MRC. W e thank Esther Hoste f or her input and the cor e facilities of the Cancer Research UK Cambr idge Research Institute, where some of this work was perf ormed, f or superb technical assistance.
19
© 2015 The Society for Investigative Dermatology
REFERENCES Asano S,
Ichikawa
Y,
Kumagai T,
et
al (2008).
Microanalysis of
an
antim icrobial peptide, beta -def ensin-2, in the stratum corneum f rom patients with at opic dermatit is. Br J Der matol 159: 97-104. Biasoli MS, Tosello ME, Luque AG, et al (2010). Adherence, colonizat ion and dissemination of Candida dubliniensis and other Candida species. Med Mycol 48: 291- 297. Candi E, Schmidt R, Melino G ( 2005). The cornif ied envelope: a model of cell death in the skin. Nat Rev Mol Cell Biol 6: 328-340. Caporaso JG, Kuczynski J, Stombaugh J, et al (2010). QII ME allows analysis of
high-throughput
communit y
sequencing
data.
Nat
Methods
7:
335-336. Cipolat S, Hoste E, Natsuga K, et al (2014). Epidermal bar rier def ects link atopic dermat itis with altered skin cancer suscept ibilit y. eLife 3: e01888. Descargues P, Deraison C, Bonnart C, et al (2005). Spink5-def icient mice mimic Net herton syndrome through degradation of desmoglein 1 b y epidermal protease hyperactivit y. Nat Genet 37: 56- 65. Gallo RL, Hooper LV (2012). Epit helial antimicrobial def ence of the skin and intest ine. Nat Rev I mmunol 12: 503- 516. Grice EA, Segre JA ( 2011). The skin microbiome. Nat Rev Microbiol 9: 244-253.
20
© 2015 The Society for Investigative Dermatology
Hamerman JA, Ogasawar a K, Lanier LL ( 2004). Cutt ing edge: Toll - like receptor signa ling in macrophages induces ligands f or the NKG2D receptor. J Immunol 172: 2001-2005. Hoste E, Ar wert EN, Lal R et al (2015a) . I nnate sensing of microbial product s promotes wound -induced skin cancer. Nat Commun 6 : 5 9 3 2 . Hoste E, Cipolat S, Watt FM (2015 b). Understanding allerg y and cancer risk: what are the barr iers? Nat Rev Cancer d o i : 1 0 . 1 0 3 8 / n r c 3 9 0 9 Kabashima K ( 2013) . New concept of the pathogenesis of atopic dermat itis: interplay among the barrier, allerg y, and pruritus as a trinit y. J Dermatol Sci 70: 3-11. Kawasaki H, Nagao K, Kubo A et al (2012). Altered stratum corneum barr ier and enhanced percut aneous immune responses in f ilaggrin -null mice. J Allerg y Clin Immunol . 129: 1538-1546. Lai Y, Di Nardo A, Nakatsuji T et al (2009). Commensal bacteria regulate Toll- like receptor 3 - dependent inf lammation af ter skin injur y. Nat Med 15: 1377-1382. Lodoen MB, Abenes G, Umamoto S et al (2004). The cytomegalovirus m155 gene
product
subverts
natur al
killer
cell
ant iviral
prot ection
by
disruption of H60 - NKG2D int eract ions. J Exp Med 200: 1075- 1081. Naik S, Bouladoux N, W ilhelm C et al ( 2012). Compartmentalized control of skin immunit y by resident commensals. Science 337: 1115- 1119. Natsuga K ( 2014). Epidermal barrier s. Cold Spring Harb Perspect Med 4:
21
© 2015 The Society for Investigative Dermatology
a018218. Ogawa T, Katsuoka K, Kawano K et al ( 1994). Comparative study of staphylococcal f lora on the skin surf ace of atopic dermat itis patients and healthy subjects. J Dermatol 21: 453 -460. Oji V, Tadini G, Akiyam a M et al ( 2010). Revised nom enclature and classif icat ion of inherited ichthyoses: results of the First Ichthyosis Consensus Conf erence in Sor eze 2009. J Am Acad Der matol 63: 607-641. Ong PY, Ohtake T, Brandt C et al (2002). Endogenous anti m icrobial peptides and skin inf ections in atopic dermat it is. N Engl J Med 347: 1151-1160. Ovaere P, Lippens S, Vandenabeele P et al (2009). The emerging roles of serine pr otease cascades in the epider mis. Trends Biochem Sci 34: 453-463. Reikvam DH, Erof eev A, Sandvik A et al (2011). Depletion of murine int est i na l microbiota: eff ects on gut mucosa and epithelial gene expression. PLoS One 6: e17996. Scharschmidt TC, List K, Grice EA et al (2009). Matriptase -def icient mice exhibit ichthyot ic skin with a select ive shif t in skin micr obiota. J Invest Dermatol 129: 2435- 2442. Sevilla LM, Nachat R, Groot KR et al ( 2007). Mice def icient in involucr in, envoplakin, and periplakin have a def ective epidermal barrier. J Cell Biol 179: 1599-1612.
22
© 2015 The Society for Investigative Dermatology
Shanahan F, O'Toole PW (2014). Host-microbe inter actions and spat ial variat ion of cancer in the gut. Nat Rev Cancer 14: 511-512. Sivaprasad U, Kinker KG, Ericksen MB et al (2015). SERPINB3/B4 contributes to early inflammation and barrier dysfunction in an experimental murine model of atopic dermatitis. J Invest Dermatol.
135:160-169.
Spear P, W u MR, Sentman ML et al (2013). NKG2D ligands as therapeut ic targets. Cancer Immun 13: 8. Strid J, Roberts SJ, Filler RB et al (2008). Acute upregulat ion of an NKG2D ligand promotes rapid reorganizat ion of a local immune compartment with pleiotropic eff ects on carcinogenesis. Nat Immunol 9: 146-154. Strid J, Tigelaar RE, Hayday AC (2009). Skin immune surveillance by T cells--a new order ? Semin Immunol 21: 110-120.
23
© 2015 The Society for Investigative Dermatology
FIGURE LEGENDS Figure 1. Characterization of skin microbiota in EPI -/- and WT mice ( A) qPCR of 16sRNA gene in EPI -/- and W T ear skin. Data ar e means ± SEM f rom 6 mice per group. ** p<0.01. (B, C) Micr obial diversit y of EPI-/- and W T ear skin obtained by 16sRNA gene pyroseq uencing f rom 6 mice per group . (B) Data are presented as means of relative abundance ± SEM. ( C) Each colum n shows relat ive abundance of the bact eria l phyla in a single mouse . There were no sta tis tically significant difference s be tween W T and EPI -/- sk in microbiota phyla .
Figure 2. Epidermal penetration by bacteria ( A, B) W hole-mount FISH of ear skin using BacUni ( A) or C. dubuliniensis ( B) pr obes (green), with DAPI counterst ain (blue) . The images are Z-stack maximum projection s of the epidermis. Scale bars: 20 m. (C) Optical sectioning of 3D reconstruct ed whole-mount FISH (BacUni). Arrowheads show bacter ial penetration of the basal epidermal layer. Dotted line indicates posit ion of basement membrane . Scale bars: 10 m. ( D) Quantit ation of location of individual BacUni signals relat ive to the median thickness (dotted line) of the epider mis. The boxes extend f rom the 25 t h to 75 t h per cent iles. W hiskers indicate minimum to maximum. **** p<0.0001 .
Figure 3. Microbiota independent -inflammatory and stress response
24
© 2015 The Society for Investigative Dermatology
phenot ypes of EPI -/- skin ( A) qPCR of 16sRNA gene in W T, EPI -/-, antibiotic-treated EPI -/- and f lora- def icient EPI -/- ear skin samples. ( B) Number of CD4+CD3+ cells per mm 2 der mis. (C) Number of TCR+ cells per mm epidermis. ( D) Number of V3+ cells per mm epiderm is. ( E) qRT-PCR of Rae1 in W T, EPI-/-, antibiotic-treated EPI -/- and f lora-def icient EPI -/epidermis.
( F)
Quantif icat ion
of
myeloperoxidase
in
W T,
EPI-/-
and
f lora-def icient skin lysates. Dat a are means ± SEM f rom at least 4 (A) or 3 (B-F) mice per group. * p<0.05, *** p <0.001; unless indicated no signif icant diff erences wer e f ound ( one-wa y ANOVA followed b y Tuk ey’s te st).
Figure 4. Effect of antibiotic treatment on skin histology. ( A) H&E stained skin sect ions of W T, EPI -/- and antibiotics -treated EPI -/- at steady stat e. Arrowheads: mild hyperkeratosis. ( B) H&E stained sect ions of W T, EPI-/- and antibiotic-treated
EPI -/-
skin
af ter
three
applicat ion s
of
TPA.
Black
arrowheads: mild hyperkeratosis; w hite arrowhead s: severe hyperkeratosis; asterisk s: parakeratosis; black arrows: spongiosis . Scale bar s: 100 um. (C, D) Epidermal
thicknes s
(C)
and
spleen
mass
( D)
of
W T,
EPI -/-
and
antibiotic-treated EPI -/- mice. TPA was applied thr ee times on alternat ing days on all the mice (n = 3 or 4 per condition) . ** p<0.01, *** p<0.001.
Figure 5. AM P expressi on and prot ease acti vit y in EPI-/- epidermis follow ing microbiota modulation . ( A-K) qRT-PCR of Camp ( A) , Def b1 (B),
25
© 2015 The Society for Investigative Dermatology
Def b2 (C), Def b3 ( D), Def b6 (E), S100A9 ( F), Pla2g2a ( G), Reg3b ( H), Reg3g (I) and Serpina1b ( J) in W T, EPI-/-, antibiotic-treat ed EPI -/- and f lora-def icient EPI-/- epidermis. Data are means ± SEM f rom at least 4 m ice per group. * p<0.05, ** p<0.01 . (K) Total e pidermal protease activit ies of W T, EPI -/-, antibiotic-treated EPI -/- and f lora-def ient EPI -/- skin. Data ar e means ± SEM f rom 3-4 mice per group. (L) Casein gel electrophoresis of epidermal lysates. Each lane is lysate f rom a diff erent mouse. Posit ions of m olecular mass markers (31kDa, 17kDa) are shown. The t wo major bands of p roteolyt ic activit y are indicated by arrowheads .
26
© 2015 The Society for Investigative Dermatology
© 2015 The Society
© 2015 The Society
© 2015 The Society
© 2015 The Society
© 2015 The Society