Original Article

Novel insights into osteogenesis and matrix remodelling associatedwith calcific uraemic arteriolopathy

Rafael Kramann1,2*, Vincent M. Brandenburg3,*, Leon J. Schurgers4, Markus Ketteler5,Saskia Westphal2, Isabelle Leisten2, Manfred Bovi2, Willi Jahnen-Dechent6, Ruth Knüchel2,Jürgen Floege1 and Rebekka K. Schneider2

1Division of Nephrology and Clinical Immunology, RWTH Aachen University, Aachen, Germany, 2Institute of Pathology, RWTHAachen University, Aachen, Germany, 3Department of Cardiology, RWTH Aachen University, Aachen, Germany, 4Department ofBiochemistry, Cardiovascular Research Institute Maastricht, Maastricht University, Maastricht, The Netherlands, 5Department ofMedicine III, Division of Nephrology, Klinikum Coburg, Coburg, Germany and 6Institute for Biomedical Engineering,Biointerface Laboratory, Aachen, Germany

Correspondence and offprint requests to: Rafael Kramann; E-mail: rkramann@gmx.net*R.K. and V.M.B. are contributed equally to this work.

AbstractBackground. Calcific uraemic arteriolopathy (CUA) orcalciphylaxis is a rare, life-threatening disease predomi-nantly occurring in patients with end-stage renal disease.Its pathogenesis has been suggested to include ectopicosteogenesis in soft tissue and the vasculature associatedwith extracellular matrix (ECM) remodelling.Methods. To gain further insights into the pathogenesisof CUA, we performed systematic analyses of skin speci-mens obtained from seven CUA patients including histology,immunohistochemistry, electron microscopy, electron disper-sive X-ray analysis (EDX) and quantitative real-time RT-PCR. Skin specimens of (i) seven patients without chronickidney disease and without CUA and (ii) seven dialysispatients without CUA served as controls.Results. In the CUA skin lesions, we observed a significantupregulation of bone morphogenic protein 2 (BMP-2), itstarget gene Runx2 and its indirect antagonist sclerostin.Furthermore, we detected an increased expression of inactiveuncarboxylated matrix Gla protein (Glu-MGP). The upregu-lation of osteogenesis-associated markers was accompaniedby an increased expression of osteopontin, fibronectin,laminin and collagen I indicating an extensive remodellingof the subcutaneous ECM. EDX analysis revealed calcium/phosphate accumulations in the subcutis of all CUA patientswith a molar ratio of 1.68 ± 0.06 matching that of hydroxya-patite mineral. Widespread media calcification in cutaneousarterioles was associated with destruction of the endotheliallayer and partial exfoliation of the endothelial cells (ECs).CD31 immunostaining revealed aggregates of ECs contribut-ing to intraluminal obstruction and consecutive malperfusionresulting in the clinical picture of ulcerative necrosis in allseven patients.Conclusions. Our data indicate that CUA is an active osteo-genic process including the upregulation of BMP-2

signalling, hydroxyapatite deposition and extensive matrixremodelling of the subcutis.

Keywords: bone morphogenic protein 2; calciphylaxis; matrix Glaprotein; matrix remodelling; sclerostin

Introduction

Calcific uraemic arteriolopathy (CUA) or calciphylaxis isa rare, life-threatening disease predominantly affectingpatients with end-stage renal disease (ESRD) or chronickidney disease (CKD). It is characterized by media calci-fication of small- and medium-sized cutaneous vessels,soft-tissue calcification and clinically by progressivepainful skin indurations with ulcerations [1, 2]. The termcalciphylaxis was introduced by Hans Selye in the early1960s [3] based on a rodent model of systemic and localsoft-tissue calcifications. Selye postulated a two-stephypothesis with sensitization factors (i.e. hyperparathyreoid-ism, vitamin D or a diet high in calcium and phosphorus)followed by challenging factors (i.e. local trauma, or theapplication of iron salt, polymycin, egg albumin or gluco-corticoids). Subsequently, a similar syndrome with periph-eral ischaemic tissue necrosis and vascular calcification wasreported in a woman with acute renal failure and this wasalso named calciphylaxis [4]. However, the experimentalmodel of Selye was free of vascular calcification and, there-fore, the human disease is more appropriately described bythe term CUA [5]. The prevalence of CUA is nowadaysestimated to range from less than 1–4% among ESRDpatients [1, 6, 7].The pathogenesis of CUA remains incompletely under-

stood. Multiple clinical risk factors such as female gender,diabetes, obesity, warfarin use, hyperphosphataemia,hyperparathyroidism, the use of calcium-containing

© The Author 2012. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.For Permissions, please e-mail: journals.permissions@oup.com

Nephrol Dial Transplant (2012) 0: 1–12doi: 10.1093/ndt/gfs466

NDT Advance Access published December 6, 2012 at M

edizinische Bibliothek on D

ecember 19, 2012

http://ndt.oxfordjournals.org/D

ownloaded from

phosphate binders and particularly the presence of severeCKD have been identified [8, 9].

Vascular calcification is now considered an active cell-mediated process similar to bone morphogenesis. Never-theless, both origins of the mineralizing cell population andinitiating stimuli remain controversial. One possible stimu-lus is bone morphogenic protein 2 (BMP-2). It belongs tothe transforming growth factor-β superfamily and acts bybinding to its specific type II receptor followed by the acti-vation of the type I receptor [10]. BMP-2 function can beinhibited by matrix Gla protein (MGP) [11], but only bythe active Gla-MGP [12]. MGP activation is vitamin K de-pendent which needs to be activated via gamma-carboxyla-tion [13]. Vitamin K antagonists such as warfarin inhibitthe recycling of vitamin K, thereby limiting the activationof MGP. [14]. Vitamin K antagonist usage appears to be anindependent risk factor for the development of CUA [15].If BMP-2 is not inhibited by MGP, the BMP-2 downstreamsignalling cascade leads to an increased expression of thekey osteogenic transcription factor Runx2 [16]. Runx2 con-trols the expression of matrix proteins of osteoblastic differ-entiation such as osteopontin and collagen I [17].Sclerostin, the glycoprotein product of the SOST gene, isexpressed by mature osteocytes and also antagonizes theeffect of BMP-2 indirectly via inhibition of the Wnt/β-catenin signalling pathway [18, 19].

CUA can be regarded as a particular subtype of vascu-lar calcification. It has been hypothesized that CUA is,similar to vascular calcification in large arteries, an activeprocess with osteoblast-like cells contributing to calcifica-tion [20]. Thus, in calcified skin lesions of CUA patients,increased expression of osteopontin, suggesting an osteo-genic process, was observed [20]. Besides osteogenesis,an altered extracellular matrix (ECM) with increasedexpression of matrix proteins such as collagen I, osteopontinand fibronectin is important in calcification processes [21,22]. Clinically, hardening of the soft tissue with a dense,plate-like aspect is a frequent early finding in CUA patients,which proceeds later to ulcerations. And similarly, oncesoft-tissue ulcerations have developed, these lesions areoften surrounded by areas of tissue induration typicallypalpable as leather-like rigidifications of skin areas in associ-ation with calcified and necrotic lesions [1].

Our study aimed to gain further insights into mechanisticaspects of CUA development with a special focus on the in-terplay between altered ECM remodelling, osteogenesis,ectopic calcification and vascular obstruction finally leadingto tissue ulcerations in patients with CUA. We performed asystematic, detailed analysis of calcified skin specimens inseven CUA patients by histology, immunohistochemistry,electron microscopy, electron dispersive X-ray analysis (EDX)and quantitative real-time RT-PCR (Qt-RT-PCR). We particu-larly investigated markers involved in osteogenesis such asBMP2, Runx2, sclerostin and key proteins of the ECM suchas collagen I, IV, fibronectin, osteopontin, MGP and laminin.

Materials and methods

CUA Patients

Based on data collected in our calciphylaxis registry initiative (Inter-national Collaborative Calciphylaxis Network), we retrospectively

identified all patients who were treated for calciphylaxis in the UniversityHospital Aachen since 2006 and from whom tissue samples were avail-able (n = 7; two males/five females, age 50 ± 19 years, range 21–77). Weexamined skin specimens obtained from consented clinical autopsies(n = 2) and operative resections (n = 5) of these seven CUA patients. Thediagnosis of CUA of all patients was based on the clinical picture andconfirmed by histological analysis of one or more skin sections. We usedtwo different patient groups as controls: (i) patients without CUA andwithout CKD (non-CKD controls) and (ii) dialysis patients without CUA(dialysis controls). (i) As non-CKD controls, we matched skin specimensof seven non-CKD, non-CUA patients for age, gender, the presence orthe absence of obesity and diabetes (n = 2 men, age 49 ± 18 years, range25–76; diabetes in n = 4, obesity in n = 4). Specimens from non-CKDcontrols were obtained during consented clinical autopsies (n = 2), andby surgical skin resections in n = 5. The cause of death in all autopsypatients was septicaemia. In autopsy cases, the skin samples were alwaysresected within 8 h after death, and significant signs of autolysis wereexcluded. Due to our experience, early sampling of tissues of deceasedsubjects does not reduce the reliability of immunohistochemical or EDXanalysis and is also routinely applied for clinical autopsies. Only twoautopsy cases were included into the PCR analysis (one in the CUAgroup and one in the non-CKD control group. (ii) As dialysis controls,we included skin samples obtained from dialysis patients without CUA(three males/four females, age 54 ± 10, range 45–71, diabetes in n = 4,obesity in n = 3, time on dialysis 4.2 ± 3.4 years). Specimens of dialysiscontrol patients were obtained from two lower leg amputations and fromfive surgical resections of healthy skin after skin cancer (melanoma). Allpatients and controls were of Caucasian ethnicity. Data about themedical history and medications were obtained by retrospective chartreview and obtained from the above-mentioned calciphylaxis registry(www.calciphylaxie.de). The study was approved by the ethical commit-tee of the RWTH Aachen University Hospital and carried out accordingto the principles of the Declaration of Helsinki.

Histomorphological analysis

For histological and immunohistochemical analyses, skin specimenswere fixed in 3.7% formaldehyde for 24 h. Skin specimens were paraf-fin-embedded, cut with a rotating microtome at 3 µm thickness (Leica)and stained according to routine histology protocols. Immunohistochem-ical analysis was performed in an autostainer (DAKO cytomation) usingprimary antibodies specific for fibronectin (mouse monoclonal, 1:200;Sigma, Hamburg, Germany), collagen I (mouse monoclonal, 1:2000;Sigma), collagen IV (mouse monoclonal, 1:250; Sigma), laminin (mousemonoclonal, 1:1000; Sigma), CD68 (mouse monoclonal, 1:400; DAKO,Hamburg, Germany), CD31 (mouse monoclonal, 1:500; DAKO), BMP-2 (rabbit polyclonal, 1:200; Abcam, Cambridge, UK), Glu- and Gla-MGP (mouse monoclonal, 1:200; Vascular Products, Maastricht, theNetherlands) and osteopontin (mouse monoclonal, 1:3000; Santa CruzBiotechnology, Heidelberg, Germany) as described before [23, 24]. Theimmunostaining for sclerostin was performed manually. After blockingthe sections for endogenous avidin/biotin (Avidin-Biotin Blocking Kit,Vector Laboratories Burlingame, CA, USA) and peroxidase activity (3%H2O2), slides were immediately incubated with a goat polyclonal anti-sclerostin antibody (1:200, R&D Systems; Minneapolis, MN, USA).Biotinylated horse monoclonal anti-goat antibody (Vector Laboratories)was used as a secondary antibody. Detection was realized using theVectastin ABC Kit (Vector Laboratories).

Semi-quantitative immunohistochemical analysis

Protein expression and von Kossa staining were quantified by twoblinded investigators (R.K. and R.K.S.) using a semi-quantitative scoringsystem (0, no expression; 1,weak expression; 2 moderate expression; 3,strong expression; 4, very strong expression) in high-power fields (×200magnification) as described before [25, 26]. The number of small arteriesand arterioles with CD31+ endoluminal debris was estimated by countingall arteries and arterioles containing endoluminal CD31+ cells in onesection (magnification ×200) and dividing by all arteries and arterioles inthe corresponding section.

Scanning electron microscopy

Tissue sections were fixed in 3% glutaraldehyde for at least 24 h, rinsedwith sodium phosphate buffer (0.2 M, pH 7.39, MERCK, Darmstadt,Germany) and dehydrated by incubating consecutively in ascending

2 R. Kramann et al.

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

acetone series (30, 50, 70 and 90%) with a final incubation in 100%acetone for 10 min. The tissue sections were critical-point dried in liquidCO2. Samples were analysed using an environmental scanning electronmicroscope (ESEM XL 30 FEG, FEI, PHILIPS, Eindhoven, the Nether-lands) in a high vacuum environment. EDX analysis was performedusing an EDAX Falcon Genesis Spectrum 5.21 energy-dispersive X-rayspectroscopy system with an ultrathin window liquid nitrogen cooled Si(Li) X-ray detector (EDAX Inc., Mahwah, NJ, USA).

Real-time reverse trancriptase polymerase chain reaction

Total RNAwas isolated from the formalin-fixed, paraffin-embedded skinspecimens using the RNeasy FFPE Kit (Qiagen, Hilden, Germany) ac-cording to the manufacturer instructions. We achieved an isolation ofRNA feasible for real-time PCR from only four CUA patients (patientno. 1, 3, 5, 6) due to the origin of RNA from formalin-fixed paraffin-embedded sections. We compared gene expression of these four patientswith gene expression of n = 4 non-CUA non-CKD controls and n = 4non-CUA dialysis controls. The RNA concentration was determined bymeasuring absorbance at 260 nm (Nanodrop, Thermo Scientific, Wil-mington, USA). One microgram of RNA was reverse transcribed usingthe high capacity cDNA Reverse Transcriptase Kit (Applied Biosystems,7300 Real-Time PCR System, Foster City, CA, USA). Quantitative PCRreactions were carried out with Power SYBR Green PCR Master Mix(Applied Biosystems). For each sample, 1.2 µL of cDNAwas added as atemplate in PCR reactions. Amplification was monitored with the ABIPrism 7300 (Applied Biosystems). The expression of genes of interestwas normalized against the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase in all samples. The cDNA of the skin sectionsfrom non-CUA non-CKD control patients was used as a relative standardfor the genes of interest and set as one for all genes. Relative geneexpression was analysed with the 2− ΔΔCt method. Primers are listed inSupplementary data, Table S1.

Statistical analysis

Data are presented as mean ± standard deviation (SD). After testing fornormal distribution, one-way analysis of variance with the post hocScheffé procedure (PCR data) or Kruskall–Wallis test with post hocDunńs multiple comparison (histological scoring) was used where appro-priate. Statistical significance was defined as P < 0.05. Analyses wereperformed using PASW Statistic 18.0 (SPSS Inc., Chicago, IL, USA)and GraphPad Prism 5.0c (GraphPad Software Inc., La Jolla, CA, USA).

Results

Clinical results

Clinical data of the patients are presented in Table 1. Allcalciphylaxis patients were admitted to the UniversityHospital Aachen due to complicated CUA, and in allpatients CUAwas associated with chronic or severe acutekidney dysfunction: five patients were chronic haemodia-lysis (HD) patients, one patient suffered acute renal failurebefore CUA was diagnosed and one patient had CKDstage III. Six patients were of Caucasian and one was ofAsian ethnicity. Four patients were receiving long-termanti-coagulation with coumadin, four patients were substi-tuted with oral vitamin D3, four patients were receivingsubcutaneous injections of erythropoietin, one patient wasreceiving intravenous iron, six patients were receiving oralphosphorus binders (four sevelamer, one aluminiumhydroxide, one lanthanum carbonate) and one patient wasreceiving cinacalcet. One haemodialysis patient had un-dergone parathyroidectomy. Laboratory data of the sevenCUA patients are shown in Table 2. Three deaths occurredin HD CUA patients due to refractory septic shock.

The clinical picture of CUAwas deep ulcerative proxi-mal lesions in four patients, while three patients revealeddistal calciphylaxis at the lower extremities.

Distribution of calcified lesions and upregulationof BMP-2, sclerostin and MGP

von Kossa staining revealed calcified areas in the subcutisof all CUA patients. Calcification was observed aroundadipocytes (Figure 1A), in the media of subcutaneous ar-terioles (Figure 1B, inset) and in connective tissue strands(Figure 1B) of the subcutis. Calcification was never ob-served in the dermis. To evaluate the hypothesis thatactive bone formation contributes to the process of subcu-taneous calcification in CUA, we analysed the expressionof BMP-2, its receptor bone morphogenic protein receptor2 (BMPR-2), its downstream target gene Runx2 and itsantagonists sclerostin and MGP in the skin of CUApatients. Immunohistochemistry showed areas positive forBMP-2 and sclerostin adjacent to calcified areas in thesubcutis (Figure 1C–F) in all seven CUA patients. In con-trast, the subcutis of controls (non-CKD and dialysis con-trols) was negative for sclerostin and BMP-2 (Table 3,Supplementary data, Figures S1 and S2). The results ofthe Qt-RT-PCR of four patients confirmed this obser-vation at the mRNA level. BMP-2, sclerostin and Runx2were significantly upregulated in the skin of CUA patientscompared with the skin of both controls (Figure 1G).BMPR-2 was non-significantly upregulated in the skin ofCUA patients compared with controls. Staining for inac-tive (uncarboxylated) Glu-MGP and active (carboxylated)Gla-MGP revealed an expression of both forms in CUAskin lesions (Figure 1H and Supplementary data,Figure S3). However, compared with the controls only theinactive Glu-MGP was significantly increased in the CUAskin (Table 3). We observed no differences in theexpression of both MGP forms between coumadin-treatedand non-treated CUA patients (data not shown).

CUA skin lesions exhibit extensive remodellingof the subcutaneous ECM

As fibrotic procalcific remodelling of the ECM supportscalcification in heart valves and vessels [21, 22, 27], wehypothesized that a similar process is detectable in skinindurations of CUA patients which finally end in cal-cification. Semi-quantitative analysis of the proteinexpression in the subcutis revealed a significantly in-creased subcutaneous expression of collagen I, fibronec-tin, osteopontin and laminin in calcified skin specimensof CUA patients compared with non-CKD controls(Table 3). However, despite a trend towards upregulationof all analysed ECM proteins, semi-quantitative histologi-cal scoring revealed a significantly increased expressiononly for osteopontin and laminin in the CUA skin lesionscompared with the dialysis controls (Figure 2 andTable 3). Osteopontin exhibited a comparable distributionpattern to von Kossa staining in CUA with a strongexpression in the subcutis close to adipocytes and in con-nective tissue strands (Figure 2A–B). In contrast to thecontrol skin (Figure 2D, inset and Supplementary data,

Calciphylaxis: matrix remodelling and osteogenesis 3

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

Figure S4), the fine, reticular fibres surrounding adipo-cytes in CUA patients were broadened and strongly posi-tive for fibronectin (Figure 2D).

In CUA, laminin and collagen I showed increasedexpression in association with adipocytes (Figure 2E andG), whereas collagen IV expression did not differ quanti-tatively from control skin, although the expression patternwas irregular around adipocytes in the subcutis of CUApatients, in terms of adiponecrosis (Figure 2F, Table 3). Wecompared the immunohistochemistry distribution patternfor ECM proteins to the von Kossa staining and observedan upregulation of fibronectin, collagen I and laminin evenin areas that did not show overt calcification. Qt-RT-PCRanalysis supported the immunohistochemial findings re-garding the extensive ECM remodelling process in CUAskin (Figure 2H), as osteopontin, fibronectin and lamininwere significantly upregulated.

Calcification of the subcutaneous adipose tissue involvescalcium hydroxyapatite formation

Scanning electron microscopy (SEM) confirmed calcifiedareas in the skin sections of all CUA patients (Figure 3).The most extensive calcification was detected in themedia of subcutaneous arterioles (Figure 3B) and withinthe subcutaneous connective tissue (Figure 3F). Compar-able with von Kossa staining, we detected calcificationsaround adipocytes (Figure 3E). Energy-dispersive X-rayspectroscopy (EDX) showed carbon, oxygen, phosphorusand calcium in all calcified regions (Figure 3D), with amolar ratio of calcium/phosphate of 1.68 ± 0.06 matchingthat of calcium hydroxyapatite mineral. Sodium was

detected in five, iron in three, magnesium in two and sili-cium in one skin section of CUA patients (data notshown). Control spectra from skin sections of non-CKDcontrols showed carbon and oxygen (Figure 3C), butnever calcium or phosphate.

Clogging of subcutaneous vessels by endothelial cells

In CUA lesions, lumina of small cutaneous vessels wereoften obstructed by cell debris (Figures 1B and 3B, inset).This observation had previously led to a ‘2-hit model’where vascular calcification is followed by endoluminalobstruction, thrombus formation and tissue necrosis [1,28]. Anti-CD31 staining revealed CD31+ endothelial cells(ECs) within the endoluminal debris in about 30% of thecalcified cutaneous arteries and arterioles of calcifiedCUA skin sections (Figure 4A and B), while the normalCD31+ ECs layer of the intima was often destroyed inthose vessels. In contrast, anti-CD31 staining of skin sec-tions from non-CKD controls showed a regular CD31+

layer of intimal EC and never intraluminal cellular debrisaccumulation (Figure 4C). Furthermore, we observed his-tological signs of thrombosis in very small cutaneousvessels (diameter 15 μm) with remnants of CD31+ cellswithin the thrombus (Figure 4D–E).

Haemosiderophages in the subcutis

Iron has been discussed as a potential contributor to CUAdevelopment [29, 30] and, as noted above, we detectediron by EDX analysis in skin lesions of three of the sevenCUA patients. In line with our EDX analysis, Prussian

Table 1. Clinical characteristics of seven CUA patients

Patient no. Age (year) Sex Obesity DM Wound distribution Cause of CKD/ESRD RRT mode Time on HD (years) Outcomea

1 44 F Yes Yes Abdominal Diabetic HD 5 Expired2 43 M Yes Yes Distal legs Hypertensive HD 6 Alive3 77 F No No Distal leg Hypertensive HD 3 Alive4 67 F No No Abdominal Hypertensive HD 5 Expired5 21 M Yes No Abdominal ARFb No RRT – Alive6 39 F Yes Yes Abdominal Diabetic HD 3 Expired7 56 F Yes Yes Distal leg CKD III diabetic No RRT – Alive

RRT, renal replacement therapy; ESRD, end-stage renal disease; DM, diabetes mellitus; HD, haemodialysis; CKD, chronic kidney disease; F, female;M, male. Obesity was defined as body mass index >30.aAt discharge of hospital.bIncrease of creatinine up to 4.2 mg/dL in medical history.

Table 2. Laboratory values of seven CUA patients upon admission

Patient no. Albumin (g/L) Calcium (mmoL/L) Phosphorus (mmoL/L) AP (U/L) PTH (ng/L) Creatinine (mg/dL) CRPa (mg/L)

1 17.3 2.4 0.8 196 60 4.9 2302 33 2.1 2.7 115 896 12.2 383 36.8 2.7 1.4 108 787 3.1 174 32.5 2.2 1.2 96 276 3.3 935 32.8 1.4 1.0 157 4 1 626 10.7 2.5 0.8 276 118 4.9 847 41.5 2.4 1.2 105 80 0.7 5

AP, alkaline phosphatase; CRP, C-reactive protein.aNormal value <5 mg/L.

4 R. Kramann et al.

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

blue staining was positive in only these three skin sec-tions. In two sections, only small Prussian blue-positiveareas were observed in association with small subcu-taneous vessels (Figure 5A, inset), while in the skin

section of one CUA patient, we observed large Prussianblue-positive areas in the subcutaneous connective tissue(Figure 5A and B). SEM and EDX analyses revealed thatthe iron deposits were not closely correlated with the

Fig. 1. Up-regulation of osteogenic markers in skin lesions of CUA patients. (A and B) von Kossa positive subcutaneous adipose tissue (A),connective tissue (B) and arteriole (B, inset). (C and D) BMP-2 staining and (E and F) sclerostin staining indicated an osteogenic process in the skinof CUA patients. We observed BMP2-positive areas in the dermis (C arrows) and the subcutaneous adipose tissue (D arrows) only in the calcifiedskin and subcutaneous areas of the CUA patients. Similarly, sclerostin positive areas were observed only in the subcutaneous adipose tissue ofcalcified skin in CUA patients (E and F arrows). The subcutis of control subjects was completely negative for these stainings (Supplementary data,Figures S1 and S2). (G) mRNA expression of BMP-2, sclerostin (SOST) and Runx2 was significantly enhanced in the skin lesions of CUA patientscompared with the skin of both control groups (non-CKD controls = expression set as 1 for all genes). (H) Uncarboxylated inactive Glu-MGPexpression was increased in the subcutaneous arterioles (arrow) and in the connective tissue surrounding adipocytes of the subcutis (H, insets,arrowheads) *P < 0.05 versus non-CKD controls, #P < 0.05 versus dialysis controls, values are expressed as mean ± SD. All scale bars 200 µm, insets100 µm.

Calciphylaxis: matrix remodelling and osteogenesis 5

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

calcified areas, as we observed carbonate, oxygen,sodium, phosphorus and sulphate associated with the ironspectra, but not calcium (Figure 5C and D). Rather, anti-CD68 staining confirmed CD68+ macrophages within theiron deposits (Figure 5E and F), identifying them ashaemosiderophages.

We observed no significant differences in any per-formed analyses between the proximal and distal form ofCUA. This might be due to the small patient cohort.

Discussion

Once thought to be a passive process, vascular calcifica-tion has emerged as an active and tightly regulated cell-mediated process including matrix remodelling and depo-sition [16]. CUA can be regarded as a high-speed tem-plate for ectopic, vascular calcification processes with arapid onset. Our current study points towards a number ofyet unidentified issues in the pathogenesis of CUA at thetissue level: (i) we describe the type and distribution ofsoft-tissue (not just microvascular) calcification in CUAspecimens; (ii) we present data on extensive remodellingof the subcutaneous tissue, (iii) our data present evidencefor an interplay of increased BMP-2, Glu-MGP and scler-ostin expression to calciphylaxis and (iv) demonstrate evi-dence for destruction of the endothelial cell layer insubcutaneous vessel leading to vascular obstruction.

Our first major finding was that BMP-2 was upregu-lated in skin lesions of CUA patients associated with asignificant upregulation of the downstream key osteogenictranscription factor ‘Runt-related transcription factor 2’(Runx2, also called core-binding factor alpha 1, Cbfa1).BMP-2 induction in the skin may be related to inflam-mation or reactive oxygen species (ROS) [9, 31–33]. Onepotential source of BMP-2 expression might be ECs as itis known that BMP can be upregulated in ECs by oscil-latory shear stress, inflammatory cytokines and ROS [34].In one case report of CUA, expression of BMP-4 in theperiarteriolar adventitia was described [35]. Both BMP-2and BMP-4 have also been described in calcified areasof atherosclerotic lesions [36, 37]. BMP-4 is thought

to promote endothelial proliferation and angiogenesis,whereas BMP-2 has a direct link to calcification as itpromotes mineralization [34]. The increased BMP-2expression is accompanied by an increased expression ofMGP in its inactive, uncarboxylated form (Glu-MGP),which might result in a decreased capability of MGP toinhibit BMP-2 signalling. Our finding of increasedexpression of inactive Glu-MGP in CUA skin lesionsmight reflect local vitamin K deficiency. However, we didnot observe any differences between coumadin-treatedand non-treated CUA patients in our small cohort.Runx2 induction is an essential early step responsible

for pre-osteoblast formation and plays an important rolein enchondral ossification [38]. Further, Runx2 upregu-lates the expression of ECM proteins critical for osteogen-esis such as osteopontin and collagen I [39]. This theoryof a cascade in CUA development was supported by oursecond major finding, namely a significant upregulationof several ECM proteins critical for mineralization (osteo-pontin, fibronectin, collagen I and laminin) in skin lesionsof CUA patients. ECM remodelling is important forosteogenesis and calcification in bone [38], heart valves[27] and vessels [21, 40]. Osteopontin expression in skinlesions of CUAwas first described by Ahmed et al. [20].These authors predominantly observed osteopontin-posi-tive areas in the media of calcified cutaneous vessels. Weadditionally detected osteopontin in the calcified connec-tive tissue surrounding adipocytes. As osteopontin is animportant regulator of mineralization and blocks apatitecrystal growth [41, 42], its increased expression in CUAmight be a feedback mechanism to control uncontrolledgrowth of crystals in the CUA lesions. The most pro-nounced alteration of the ECM in the subcutis of CUApatients compared with control subjects was the strongexpression of fibronectin even in non-calcified areas. Fi-bronectin might be a critical ECM protein for the calcifi-cation observed in the CUA lesions as it promotes thecalcification of vascular smooth muscle cells in vitro [22].Collagen I is another ECM protein in the CUA lesionsand the major protein of bone that is essential for osteo-genesis and also plays an important role in atherosclerosisand vascular calcification [21]. Collagen I expression was

Table 3. Semiquantitative evaluation of protein expression and von Kossa staining in the subcutis of calcified CUA skin and control skin (from non-CKD patients and dialysis patients)

Non-CKDcontrols (n = 7)

Dialysiscontrols (n = 7)

Non-CKD controls versusdialysis controls (P-valuea)

CUA(n = 7)

CUAversus non-CKDcontrols (P-valuea)

CUAversus dialysiscontrols (P-valuea)

von Kossa 0 0.07 ± 0.19 ns 3.57 ± 0.35 <0.001 <0.01BMP-2 0 0.21 ± 0.27 ns 1.57 ± 0.19 <0.001 <0.01Sclerostin 0 0.14 ± 0.38 ns 1.79 ± 0.64 <0.001 <0.01Glu-MGP 0.17 ± 0.29b 0.43 ± 0.53 ns 2.5 ± 0.71 <0.05 <0.01Gla-MGP 0.67 ± 0.29b 1.07 ± 0.67 ns 1.57 ± 0.45 ns nsOsteopontin 0 0.21 ± 0.27 ns 3.64 ± 0.48 <0.001 <0.01Collagen I 0.5 ± 0.5 1.14 ± 0.24 ns 1.86 ± 0.48 <0.001 nsCollagen IV 1.79 ± 0.27 1.57 ± 0.45 ns 1.64 ± 0.38 ns nsFibronectin 0.86 ± 0.24 1.57 ± 0.45 ns 3.43 ± 0.45 <0.001 nsLaminin 0.57 ± 0.53 0.92 ± 0.19 ns 1.71 ± 0.39 <0.01 <0.05

aKruskall–Wallis test with post hoc Dunn’s multiple comparison test.bMGP staining in healthy controls was only performed in n = 3 skin samples.ns, non significant.

6 R. Kramann et al.

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

apparently upregulated in CUA skin specimen. Finally,we detected a significant upregulation of laminin in CUAskin, i.e. a key regulator of cell migration and adhesion

[43]. This is in line with our previous finding of increasedexpression of laminin in calcified vessels of dialysispatients [44]. In summary, we were able to detect a

Fig. 2. ECM remodelling in the subcutis of CUA skin lesions. Immunohistochemistry demonstrated expression of various ECM proteins in CUAskin lesions in comparison with skin sections of dialysis controls (insets). (A and B) Osteopontin staining revealed positive areas with a similardistribution pattern as von Kossa staining (in Figure 1) around adipocytes (A and B arrows) and in the connective tissue strands of the subcutis(asterisk in A). (C and D) Fibronectin staining demonstrated strongly positive subcutaneous connective tissue with thickening of periadipocytic(arrows) and connective tissue (asterisk) fibres. (E) Laminin expression was slightly enhanced in CUA skin (F). Staining for collagen IV showed amore irregular distribution pattern in CUA subcutis (arrows) compared with skin of dialysis controls (inset). (G) Collagen I expression was increasedin the subcutis of CUA patients (arrows). (H) mRNA expression of osteopontin (OPN) and fibronectin (FN) was significantly increased in the skin ofCUA patients compared with both non-CKD and dialysis controls, whereas laminin (Lam) expression was only significantly increased in CUAversusnon-CKD controls, collagen I (Col I) expression was non-significantly increased and collagen IV (Col IV) expression remained unchanged comparedwith skin of controls. Non-CKD controls = expression set as 1 for all genes; *P < 0.05 versus non-CKD controls, #P < 0.05 versus dialysis controls;values are expressed as mean ± SD; all scale bars 200 µm, except (F) and (F, inset) 100 µm.

Calciphylaxis: matrix remodelling and osteogenesis 7

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

sophisticated concert of proteins in CUA lesions pointingtowards a pro-calcifying ECM environment. This remo-delling of the subcutaneous ECM in CUA may also con-tribute to the clinical finding of skin induration of CUApatients, which is a frequent early finding in areas thatlater progress to ulcerative lesions.

ECM remodelling in the so-called matrix maturationphase is associated with the expression of osteopontinwhich is thought to promote the deposition of mineral byregulating the amount and size of hydroxyapatite crystalsformed [45]. This is consistent with our third majorfinding, namely the detection of hydroxyapatite within thecalcified lesions of CUA patients. Calcium and phosphatewere described within the calcified lesions of CUApatients before [5, 20], but we are the first to identify a

molar ratio of calcium to phosphate of 1.6 matching thatof hydroxapatite mineral. This finding supports ourhypothesis of active osteogenesis in CUA lesions. Hydro-xyapatite is characterized by the incorporation of numer-ous foreign ions, which is consistent with our detection ofsmall amounts of magnesium, sodium and the rare earthelement silicium in the calcified CUA sections [46, 47].However, our study cohort is too small to exclude the pres-ence of any other calcium biominerals in CUA. Interest-ingly, besides hydroxyapatite, the presence of whitlockitewas also reported in uraemic vascular calcification [48, 49].Our fourth major novel finding was a significant upre-

gulation of the osteocyte marker sclerostin in skin lesionsof CUA patients. Sclerostin, first identified in 2001 [50],is an inhibitor of canonical Wnt signalling via binding to

Fig. 3. Hydroxyapatite accumulation in the subcutis of CUA skin lesions. SEM demonstrated calcified areas in the subcutis (B, E and F) of CUApatients, whereas no calcification was observed in the skin of non-CKD controls (A). Adipose tissue (A) and subcutaneous vessels of non-CKDcontrol (A, inset) did not show any calcification. In contrast, in CUA subcutis multiple calcified areas (B arrowhead, E and F) with strongly calcifiedsubcutaneous arteries (B, inset) that partly contain endoluminal thrombosis (asterisk) were found. The calcification pattern surrounded the adipocytes(E arrows). EDX analysis only showed carbon and oxygen of the biological matrix in non-CKD skin (C), whereas calcium and phosphate weredetected in the skin of CUA patients with a molar ratio of 1.6 matching that of hydroxyapatite mineral (D). The calcified areas in the subcutis ofCUA patients were partially finely granular around adipocytes (E) and partially massive in the connective tissue (F). Scale bars 100 µm, except(F, inset) 20 µm.

8 R. Kramann et al.

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

the frizzled-LRP5/6 membrane receptor complex leadingto increased bone formation [51]. The regulation of scler-ostin expression in osteocytes is not yet fully understood[52]. Calcitriol, glucocorticoids, TNF-α, BMP2, 4 and 6are able to increase the sclerostin expression [52]. In-creased serum values of sclerostin in ESRD patients com-pared with healthy control subjects have been described[53]. Recently, an upregulated sclerostin expressionduring vascular smooth muscle cell calcification in vitroand in the calcified aorta of ectonucleotide pyropho-sphate/phosphodiesterase 1-null (Enpp1−/−) mice wereobserved [54]. Didangelos et al. [55] detected sclerostinexpression in intact human aortas obtained during aorticvalve replacement. Overall, the role of sclerostin inuraemic bone disease and uraemic vascular calcification isincompletely understood. However, our immunohisto-chemistry and PCR data point towards an important roleof sclerostin also in CUA. At this time-point, we areunable to establish any causality. However, regarding itsrole in the bone, the upregulation of sclerostin might bean antiregulatory process, as a last effort to prevent soft-tissue calcification.

The devastating clinical picture of the very severe formof calciphylaxis—the proximal ulcerative form [1]—ismainly caused by large ulcerative areas that often triggerinfectious complications (septicaemia). It is likely that inCUA the primarily fibrotic and calcifying processes ulti-mately result in a fatal malperfusion syndrome. Indeed,

we detected detached endoluminal CD31+ ECs and a de-stroyed endothelial cell layer of small- and medium-sizedsubcutaneous arteries in calcified CUA skin. Others pre-viously also noted that the endothelium is ‘lifted off’ incalcified subcutaneous vessels of CUA skin sections [20].Furthermore, we detected signs of total vascular occlusionin subcutaneous terminal vessels with thrombosis andremnants of CD31+ cells. Thus, a sequence may exist,whereby vascular calcification triggers endothelialdamage finally leading to off-sheared ECs, which may bethe nidus for thrombogenesis resulting in malperfusionand necrosis of the skin. The subcutaneous reduced bloodflow might also promote the calcification of the subcu-taneous soft tissue. In the context of thrombus formation,it is also interesting that BMPs (BMP-2/4) induce a proin-flammatory endothelial phenotype and cause an inductionof endothelial adhesion molecules in ECs with increasedmonocyte adhesion capability [31, 33, 56].Prussian blue-positive areas have previously been re-

ported in close spatial relation with vessels in CUA skinbiopsies [29]. Iron deposition is considered a ‘challengingfactor’ for CUA in analogy to what Hans Selye postulatedin his animal models. Indeed, we detected iron via Prus-sian blue staining in the subcutis of three CUA patients,and EDX analysis confirmed the presence of iron by threetypical iron peaks. However, we do not consider iron asbeing a trigger factor for CUA. Given our observation thatCD68 staining revealed an accumulation of macrophages

Fig. 4. Obstruction of subcutaneous vessels by ECs in CUA. CD31 staining demonstrated a completely destroyed endothelial cell layer withclogging of small and medium subcutaneous arterioles (A and B arrows) by detached ECs in about 30% of the subcutanoues vessels in CUA patientsin comparison with (C) regular endothelium in the subcutis of a non-CKD control (arrows CD31). We observed total endoluminal occlusion withresidual CD31+ areas (arrow D and E) within thrombotic formations in smaller subcutaneous vessels (D and E). Scale bars A–C 50 µm, D and E 10 µm.

Calciphylaxis: matrix remodelling and osteogenesis 9

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

exactly in these ‘iron-positive’ areas, the iron accumulationin the subcutis of CUA patients might be due to haemosi-derophages. Interestingly, the one patient in our study whoreceived i.v. iron therapy did not exhibit iron accumulationin the skin, which is in line with our hypothesis of second-ary iron deposition due to haemorrhages.

In summary, although the present cross-sectional studydoes not allow a clear causality to be established, our datapoint towards an active, cell-mediated, BMP-2-driven os-teogenic process with extensive ECM remodelling anddeposition of hydroxyapatite mineral in the subcutis ofCUA patients. Although yet unproven, the present dataconfirm our clinical impression of a cascade starting withmatrix remodelling, followed by calcification, endothelialdamage and thrombus formation, and finally leading toluminal obstruction with tissue necrosis. Future researchshould focus on early causative factors for matrix remo-delling and on the identification of driving forces for theprogress of the above-postulated cascade in CUA.

Supplementary data

Supplementary data are available online at http://ndt.oxfordjournals.org.

Funding. This study was funded by intramural funding to R.K. andR.S. (START grants) from RWTH Aachen University. V.B. and M.K.received a project grant from AMGEN for the German CalciphylaxisRegistry conduct. The results presented in this paper have not beenpublished previously in whole or part.

Acknowledgements. We thank Norina Labude (Institute of Pathology)and Esther Stuettgen (Department of Nephrology) for excellent technicalassistance in the lab.

Conflict of interest statement. None declared.

References

1. Brandenburg VM, Cozzolino M, Ketteler M. Calciphylaxis: a stillunmet challenge. J Nephrol 2011; 24: 142–148

Fig. 5. Haemosiderophages in the subcutis of CUA lesions. (A and B) Positive Prussian blue staining in the subcutaneous connective tissue strandsof CUA lesions was observed in close association to vessels (asterisks). (C and D) SEM and EDX analyses revealed iron with small amounts ofsulphate and phosphorus in the connective subcutaneous tissue (arrows in C). (E and F) CD68 staining demonstrated an accumulation ofmacrophages (arrows) in all ‘iron-positive’ subcutaneous areas. Comparable with iron deposition in (A and B), macrophages aggregate in theconnective tissue strands of the subcutis (E) and around the subcutaneous vessels (asterisk in F). All scale bars 200 µm, except (C, inset) 20 µm.

10 R. Kramann et al.

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

2. Rogers NM, Coates PT. Calcific uraemic arteriolopathy: an update.Curr Opin Nephrol Hypertens 2008; 17: 629–634

3. Selye H, Grasso S, Dieudonne JM. On the role of adjuvants in calci-phylaxis. Rev Allergy 1961; 15: 461–465

4. Anderson DC, Stewart WK, Piercy DM. Calcifying panniculitiswith fat and skin necrosis in a case of uraemia with autonomoushyperparathyroidism. Lancet 1968; 2: 323–325

5. Coates T, Kirkland GS, Dymock RB et al. Cutaneous necrosis fromcalcific uremic arteriolopathy. Am J Kidney Dis 1998; 32: 384–391

6. Hayden MR, Goldsmith D, Sowers JR, Khanna R. Calciphylaxis:calcific uremic arteriolopathy and the emerging role of sodium thio-sulfate. Int Urol Nephrol 2008; 40: 443–451

7. Don BR, Chin AI. A strategy for the treatment of calcific uremicarteriolopathy (calciphylaxis) employing a combination of therapies.Clin Nephrol 2003; 59: 463–470

8. Singh RP, Derendorf H, Ross EA. Simulation-based sodium thiosul-fate dosing strategies for the treatment of calciphylaxis. Clin J AmSoc Nephrol 2011; 6: 1155–1159

9. Sowers KM, Hayden MR. Calcific uremic arteriolopathy: pathophy-siology, reactive oxygen species and therapeutic approaches. OxidMed Cell Longev 2010; 3: 109–121

10. Hruska KA, Mathew S, Saab G. Bone morphogenetic proteins invascular calcification. Circ Res 2005; 97: 105–114

11. Zebboudj AF, Imura M, Bostrom K. Matrix GLA protein, a regulatoryprotein for bone morphogenetic protein-2. J Biol Chem 2002; 277:4388–4394

12. Wallin R, Schurgers L, Wajih N. Effects of the blood coagulationvitamin K as an inhibitor of arterial calcification. Thromb Res 2008;122: 411–417

13. Schurgers LJ, Cranenburg EC, Vermeer C. Matrix Gla-protein: thecalcification inhibitor in need of vitamin K. Thromb Haemost 2008;100: 593–603

14. Chatrou ML, Winckers K, Hackeng TM, Reutelingsperger CP,Schurgers LJ. Vascular calcification: the price to pay for anticoagulationtherapy with vitamin K-antagonists. Blood Rev 2012; 26: 155–166

15. Hayashi M, Takamatsu I, Kanno Y et al. A case-control study ofcalciphylaxis in Japanese end-stage renal disease patients. NephrolDial Transplant 2012; 27: 1580–1584

16. Johnson RC, Leopold JA, Loscalzo J. Vascular calcification: patho-biological mechanisms and clinical implications. Circ Res 2006; 99:1044–1059

17. Xiao G, Jiang D, Ge C et al. Cooperative interactions between activat-ing transcription factor 4 and Runx2/Cbfa1 stimulate osteoblast-specificosteocalcin gene expression. J Biol Chem 2005; 280: 30689–30696

18. van Bezooijen RL, Papapoulos SE, Lowik CW. Bone morphogeneticproteins and their antagonists: the sclerostin paradigm. J EndocrinolInvest 2005; 28: 15–17

19. Rachner TD, Khosla S, Hofbauer LC. Osteoporosis: now and thefuture. Lancet 2011; 377: 1276–1287

20. Ahmed S, O’Neill KD, Hood AF, Evan AP, Moe SM. Calciphylaxisis associated with hyperphosphatemia and increased osteopontinexpression by vascular smooth muscle cells. Am J Kidney Dis 2001;37: 1267–1276

21. Adiguzel E, Ahmad PJ, Franco C, Bendeck MP. Collagens in theprogression and complications of atherosclerosis. Vasc Med 2009;14: 73–89

22. Watson KE, Parhami F, Shin V, Demer LL. Fibronectin and collagenI matrixes promote calcification of vascular cells in vitro, whereascollagen IV matrix is inhibitory. Arterioscler Thromb Vasc Biol1998; 18: 1964–1971

23. Kramann R, Couson SK, Neuss S et al. Exposure to uremic seruminduces a procalcific phenotype in human mesenchymal stem cells.Arterioscler Thromb Vasc Biol 2011; 31: e45–54

24. Schurgers LJ, Teunissen KJ, Knapen MH et al. Novel conformation-specific antibodies against matrix gamma-carboxyglutamic acid(Gla) protein: undercarboxylated matrix Gla protein as marker for vascu-lar calcification. Arterioscler Thromb Vasc Biol 2005; 25: 1629–1633

25. Amann K, Kronenberg G, Gehlen F et al. Cardiac remodelling inexperimental renal failure–an immunohistochemical study. NephrolDial Transplant 1998; 13: 1958–1966

26. Koleganova N, Piecha G, Ritz E et al. Arterial calcification inpatients with chronic kidney disease. Nephrol Dial Transplant 2009;24: 2488–2496

27. Akat K, Borggrefe M, Kaden JJ. Aortic valve calcification: basicscience to clinical practice. Heart 2009; 95: 616–623

28. Janigan DT, Hirsch DJ, Klassen GA, MacDonald AS. Calcified sub-cutaneous arterioles with infarcts of the subcutis and skin (‘calciphy-laxis’) in chronic renal failure. Am J Kidney Dis 2000; 35: 588–597

29. Farah M, Crawford RI, Levin A, Chan Yan C. Calciphylaxis in thecurrent era: emerging ‘ironic’ features? Nephrol Dial Transplant2011; 26: 191–195

30. Amuluru L, High W, Hiatt KM et al. Metal deposition in calcificuremic arteriolopathy. J Am Acad Dermatol 2009; 61: 73–79

31. Csiszar A, Smith KE, Koller A et al. Regulation of bone morpho-genetic protein-2 expression in endothelial cells: role of nuclearfactor-kappaB activation by tumor necrosis factor-alpha, H2O2, andhigh intravascular pressure. Circulation 2005; 111: 2364–2372

32. Fukui N, Ikeda Y, Ohnuki T et al. Pro-inflammatory cytokine tumornecrosis factor-alpha induces bone morphogenetic protein-2 in chon-drocytes via mRNA stabilization and transcriptional up-regulation.J Biol Chem 2006; 281: 27229–27241

33. Csiszar A, Ahmad M, Smith KE et al. Bone morphogenetic protein-2 induces proinflammatory endothelial phenotype. Am J Pathol2006; 168: 629–638

34. Bostrom KI, Rajamannan NM, Towler DA. The regulation of valvu-lar and vascular sclerosis by osteogenic morphogens. Circ Res 2011;109: 564–577

35. Griethe W, Schmitt R, Jurgensen JS et al. Bone morphogenicprotein-4 expression in vascular lesions of calciphylaxis. J Nephrol2003; 16: 728–732

36. Bostrom K, Watson KE, Horn S et al. Bone morphogenetic proteinexpression in human atherosclerotic lesions. J Clin Invest 1993;91: 1800–1809

37. Dhore CR, Cleutjens JP, Lutgens E et al. Differential expression ofbone matrix regulatory proteins in human atherosclerotic plaques.Arterioscler Thromb Vasc Biol 2001; 21: 1998–2003

38. Cohen MM, Jr. The new bone biology: pathologic, molecular, andclinical correlates. Am J Med Genet A 2006; 140: 2646–2706

39. Ducy P, Zhang R, Geoffroy V, Ridall AL, Karsenty G. Osf2/Cbfa1:a transcriptional activator of osteoblast differentiation. Cell 1997; 89:747–754

40. Pai AS, Giachelli CM. Matrix remodeling in vascular calcificationassociated with chronic kidney disease. J Am Soc Nephrol 2010; 21:1637–1640

41. Morikawa S, Mabuchi Y, Kubota Y et al. Prospective identification,isolation, and systemic transplantation of multipotent mesenchymalstem cells in murine bone marrow. J Exp Med 2009; 206: 2483–2496

42. Bianco P, Robey PG, Simmons PJ. Mesenchymal stem cells: revisit-ing history, concepts, and assays. Cell Stem Cell 2008; 2: 313–319

43. Hamill KJ, Kligys K, Hopkinson SB, Jones JC. Laminin depositionin the extracellular matrix: a complex picture emerges. J Cell Sci2009; 122: 4409–4417

44. US Renal Data System: USRDS 2008 Annual Data Report: Atlas ofChronic Kidney Disease and End-Stage Renal Disease in the UnitedStates. Bethesda, MD: National Institutes of Health, National Insti-tute of Diabetes and Digestive and Kidney Disease, 2008.

45. Clarke B. Normal bone anatomy and physiology. Clin J Am SocNephrol 2008; 3(Suppl 3): S131–S139

46. Bigi A, Cojazzi G, Panzavolta S et al. Chemical and structuralcharacterization of the mineral phase from cortical and trabecularbone. J Inorg Biochem 1997; 68: 45–51

47. Joschek S, Nies B, Krotz R, Goferich A. Chemical and physico-chemical characterization of porous hydroxyapatite ceramics madeof natural bone. Biomaterials 2000; 21: 1645–1658

48. Schlieper G, Aretz A, Verberckmoes SC et al. Ultrastructural analy-sis of vascular calcifications in uremia. J Am Soc Nephrol 2010; 21:689–696

49. Contiguglia SR, Alfrey AC, Miller NL, Runnells DE, Le Geros RZ.Nature of soft tissue calcification in uremia. Kidney Int 1973; 4:229–235

Calciphylaxis: matrix remodelling and osteogenesis 11

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from

50. Moester MJ, Papapoulos SE, Lowik CW, van Bezooijen RL. Scler-ostin: current knowledge and future perspectives. Calcif Tissue Int2010; 87: 99–107

51. Baron R, Rawadi G. Targeting the Wnt/beta-catenin pathway toregulate bone formation in the adult skeleton. Endocrinology 2007;148: 2635–2643

52. Drueke TB, Lafage-Proust MH. Sclerostin: just one more player inrenal bone disease? Clin J Am Soc Nephrol 2011; 6: 700–703

53. Cejka D, Herberth J, Branscum AJ et al. Sclerostin and Dickkopf-1in renal osteodystrophy. Clin J Am Soc Nephrol 2011; 6: 877–882

54. Zhu D, Mackenzie NC, Millan JL, Farquharson C, MacRae VE. Theappearance and modulation of osteocyte marker expression during

calcification of vascular smooth muscle cells. PLoS One 2011; 6:e19595.

55. Didangelos A, Yin X, Mandal K et al. Proteomics characterizationof extracellular space components in the human aorta. Mol Cell Pro-teomics 2010; 9: 2048–2062

56. Sorescu GP, Song H, Tressel SL et al. Bone morphogenic protein 4produced in endothelial cells by oscillatory shear stress inducesmonocyte adhesion by stimulating reactive oxygen species pro-duction from a nox1-based NADPH oxidase. Circ Res 2004; 95:773–779

Received for publication: 11.4.12; Accepted in revised form: 22.8.12

12 R. Kramann et al.

at Medizinische B

ibliothek on Decem

ber 19, 2012http://ndt.oxfordjournals.org/

Dow

nloaded from