LAMP3 Is Critical for Surfactant Homeostasis in Mice

bioRxiv preprint doi: https://doi.org/10.1101/2021.02.05.429758; this version posted February 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 LAMP3 is critical for surfactant homeostasis in mice

2 Lars P. Lunding1,2, Daniel Krause3, Guido Stichtenoth4, Cordula Stamme5, Niklas

3 Lauterbach6, Jan Hegermann7, Matthias Ochs7,8,9, Björn Schuster10, Radislav

4 Sedlacek10, Paul Saftig6, Dominik Schwudke1,3,11, Michael Wegmann1,2,*, Markus

5 Damme6,*

6 1Airway Research Center North, German Center for Lung Research (DZL), Borstel, 7 Germany. 8 2Division of Asthma Exacerbation & Regulation, Research Center Borstel, Leibniz Lung 9 Center, Borstel, Germany. 10 3Bioanalytical Chemistry, Priority Research Area Infections, Research Center Borstel, Leibniz 11 Lung Center, Borstel, Germany. 12 4Department of Pediatrics, University of Lübeck, Lübeck, Germany 13 5Division of Cellular Pneumology, Research Center Borstel, Leibniz Lung Center, Borstel, 14 Germany, and Department of Anesthesiology and Intensive Care, University of Lübeck, 15 Lübeck, Germany. 16 6Institute of Biochemistry, Christian-Albrechts-University Kiel, Kiel, Germany. 17 7Institute of Functional and Applied Anatomy, Research Core Unit Electron Microscopy, 18 Hannover Medical School, Hannover, Germany. 19 8Institute of Functional Anatomy, Charité Medical University of Berlin, Berlin, Germany. 20 9German Center for Lung Research (DZL), Berlin, Germany. 21 10Czech Centre for Phenogenomics, Institute of Molecular Genetics of the Czech Academy of 22 Sciences, Vestec, Czech Republic. 23 11 German Center for Infection Research (DZIF), TTU Tuberculosis, Borstel, Germany. 24 25 ORCID ID's: 26 Lars P. Lunding: 0000-0002-5278-3129 27 Daniel Krause: 0000-0002-7440-6201 28 Cordula Stamme: 0000-0001-8006-2663 29 Matthias Ochs: 0000-0002-0936-6979 30 Paul Saftig: 0000-0003-2637-7052 31 Dominik Schwudke: 0000-0002-1379-9451 32 Michael Wegmann: 0000-0002-1658-1554 33 Markus Damme: 0000-0002-9699-9351

34 *Corresponding authors:

35 Markus Damme, Christian-Albrechts-University Kiel, Institute of Biochemistry, Olshausenstr.

36 40, 24098 Kiel Germany; Tel: +49 431 880 2218; Mail: [email protected]

37 Michael Wegmann, Division of Asthma Exacerbation & Regulation, Priority Area Asthma &

38 Allergy, Research Center Borstel- Leibniz Lung Center, Borstel, Germany. Mail:

39 [email protected]

40 Abstract

41 Lysosome-associated membrane glycoprotein 3 (LAMP3) is a type I transmembrane protein

42 of the LAMP protein family with a cell-type-specific expression in alveolar type II cells in mice

43 and hitherto unknown function. In type II pneumocytes, LAMP3 is localized in lamellar

44 bodies, secretory organelles releasing pulmonary surfactant into the extracellular space to

45 lower surface tension at the air/liquid interface. The physiological function of LAMP3,

46 however, remains enigmatic. We generated Lamp3 knockout mice by CRISPR/Cas9. LAMP3

47 deficient mice are viable with an average life span and display regular lung function under

48 basal conditions. The levels of a major hydrophobic protein component of pulmonary

49 surfactant, SP-C, are strongly increased in the lung of Lamp3 knockout mice, and the lipid

50 composition of the bronchoalveolar lavage shows mild but significant changes, resulting in

51 alterations in surfactant functionality. In ovalbumin-induced experimental allergic asthma, the

52 changes in lipid composition are aggravated, and LAMP3-deficient mice exert an increased

53 airway resistance. Our data suggest a critical role of LAMP3 in the regulation of pulmonary

54 surfactant homeostasis and normal lung function.

55 Abbreviations: Alveolar type II (AT2); Airway hyperresponsiveness (AHR); analysis of

56 variance (ANOVA); bronchoalveolar lavage (BAL); cholesterol ester (CE); ceramide (Cer);

57 cardiolipin (CL); diacylglyceide (DAG); dendritic cells (DC); dipalmitoylphosphatidylcholine

58 (DPPC); hierarchical cluster analysis (HCA); hexosylceramide (HexCer); knockout (KO);

59 lysosome-associated membrane glycoprotein (LAMP); lamellar bodies (LB); lyso-

60 phosphatidylcholine (LPC); ovalbumin (OVA); phosphatidylcholines (PC); principal

61 component analysis (PCA); phosphatidylglycerol (PG); lyso-phosphatidylglycerol (LPG);

62 phosphatidylinositol (PI); lyso-phosphatidylinositol (LPI); phosphatidylethanolamine (PE);

63 lyso-phosphatidylethanolamine (LPE); lyso-cardiolipin (LCL); phosphatidylserine (PS); lyso-

64 phosphatidylserine (LPS); sphingomyelin (SM); surfactant protein (SP); triacylglyceride

65 (TAG); T helper cell (TH2); terminal deoxynucleotidyl transferase dUTP nick end labeling

66 (TUNEL).

67 Introduction

68 The lysosome-associated membrane glycoprotein (LAMP) family consists of five members

69 (LAMP1, LAMP2, LAMP3 / DC-LAMP, CD68 / macrosialin, and LAMP5 (BAD-LAMP)). All

70 family members are heavily N- and O-glycosylated type I transmembrane proteins localized

71 to the limiting membrane of lysosomes and lysosome-related organelles (1). LAMP1 and

72 LAMP2 are ubiquitously expressed (1), while LAMP3 (synonymously called DC-LAMP),

73 CD68, and LAMP5 show cell-type-specific or tissue-specific expression. CD68 expression is

74 restricted to macrophages, monocytes, and microglia (2), whereas LAMP5 is expressed

75 exclusively in the brain (3). The LAMP3-expression pattern between humans and mice

76 differs: LAMP3 is expressed in humans in alveolar type II (AT2) cells and dendritic cells (DC)

77 (eponymously for its alternative name "DC-LAMP" or CD208), respectively. LAMP3 is found

78 exclusively in AT2 cells in mice but not in DCs (4-7). These findings suggest a conserved

79 and critical function of LAMP3 in AT2 cells. On the subcellular levels, LAMP3 is localized in

80 lamellar bodies (LBs) (4). However, the function of LAMP3 in these cells remains enigmatic

81 until now; if and how LAMP3 affects surfactant homeostasis, e.g., mediating surfactant

82 release via LBs, is also poorly understood.

83 The primary function of AT2 cells is mainly attributed to pulmonary surfactant release to

84 lower surface tension at the lung's air/liquid interface. Pulmonary surfactant is a mixture of

85 lipids and proteins (8): The major hydrophobic protein components, surfactant proteins B

86 (SP-B) and C (SP-C), are small proteins deeply embedded into the phospholipids of the

87 surfactant. Notably, deficiency in SP-B or SP-C due to genetic mutations in SFTPB or

88 SFTPC cause hereditary forms of childhood interstitial lung disease (9, 10 ). Sftpb gene-

89 targeted mice die of respiratory failure after birth, associated with irregular and dysfunctional

90 LBs and tubular myelin (11). SP-C deficient mice have a regular life span and only show

91 subtle deficits in lung mechanics and surfactant stability (12).

92 The composition of surfactant lipids has extensively been analyzed (13-17). The most

93 abundant and characteristic lipid class is phosphatidylcholine (PC). PC makes up 3

94 approximately 70% of the total surfactant mass with dipalmitoylphosphatidylcholine (DPPC;

95 PC 16:0/16:0) as the primary lipid species (18). An overall decrease of PC lipids is coupled to

96 higher surface tension and lower surfactant protein inclusion (8). Other hight abundant and

97 functionally essential phospholipid classes in surfactant include phosphatidylglycerol (PG)

98 and phosphatidylinositol (PI), while phosphatidylethanolamine (PE), and sphingomyelin (SM).

99 Are reported as minor components of surfactant and most likely derive from other cell

100 membranes. In addition, lysophosphatidylcholine (LPC) can be found in low abundance in

101 pulmonary surfactant (18, 19). Preassembled surfactant is stored in secretory organelles

102 known as LB that fuse with the cell membranes and release pulmonary surfactant. LBs are

103 multivesicular body / late endosome-derived specialized organelles filled with surfactant

104 characterized by a typical "myelin"-like appearance by electron microscopy. Phospholipids

105 are directly transferred from the endoplasmic reticulum to LB through a vesicle-independent

106 mechanism by the phospholipid transporter "ATP binding cassette subfamily A member 3"

107 (ABCA3) (20, 21). Like SFTPB or SFTPC, loss-of-function mutations in ABCA3 were

108 identified in full-term infants who died from unexplained fatal respiratory distress syndrome

109 (22). ABCA3 mutations represents the most frequent form of congenital surfactant

110 deficiencies (23).

111 A presumably loss-of-function mutation in LAMP3 (p.(E387K)) was recently identified in an

112 Airedale Terrier dog breed, leading to severe symptoms and pathology similar to clinical

113 symptoms of the most severe neonatal forms of human surfactant deficiency (24). The

114 affected puppies' symptoms included lethal hypoxic respiratory distress and occurred within

115 the first days or weeks of life (24). LBs were smaller, contained fewer lamellae, and

116 occasionally revealed a disrupted common limiting membrane (24).

117 To conclusively address the physiological role of LAMP3 in surfactant homeostasis using a

118 genetically defined animal model, we generated Lamp3 knockout mice (KO, Lamp3-/-) by

119 CRISPR/Cas9. In contrast to dogs bearing a natural mutation in LAMP3, Lamp3-/- mice are

120 vital, display a regular lung function at the basal level, and show no increased prenatal

121 lethality. The lungs of Lamp3-/- animals did neither macroscopically nor microscopically differ

122 from those of wild type littermates. However, biochemical analysis of broncho-alveolar lavage

123 (BAL) fluid clearly revealed increased pro-SP-C levels and altered lipid composition in the

124 Lamp3-/- mice. These differences are further pronounced under diseased conditions as

125 evoked by allergen-induced experimental asthma, and most notably, diseased Lamp3-/- mice

126 revealed significantly increased airway resistance, when stressed during methacholine

127 provocation testing. In conclusion, our data suggest a critical but not vital role of LAMP3 in

128 pulmonary surfactant homeostasis associated with normal lung function.

129

130

131 Material & Methods

132 Reagents and antibodies

133 Analytical grade chemicals were purchased, if not stated otherwise, from Sigma-Aldrich

134 (MO., USA). Antibodies: The following antibodies were used in the study: Rat monoclonal

135 LAMP3, clone 1006F7.05 (Dendritics), SP-C, rabbit polyclonal (Abcam), rabbit polyclonal

136 antibodies against mature SP-B and Pro-SP-B (Seven Hills Bioreagents) were a generous

137 gift from Jeffrey A. Whitsett, polyclonal β-Actin (Santa Cruz Biotechnology, Inc), polyclonal

138 rabbit antiserum against ABCA3 was a kind gift from Michael L. Fitzgerald (20). Fluorophore-

139 conjugated secondary antibodies against the corresponding primary antibody species

140 (AlexaFluor 488) were purchased from Invitrogen / Molecular Probes and were diluted 1:500.

141

142 Generation and housing of Lamp3-/- mice

143 Mice with a null mutation in the Lamp3 gene were generated in a C57BL/6N background

144 using a CRISPR genome‐editing system. For this purpose, Cas9 was combined with two

145 single guide RNAs (sgRNAs) targeting the second exon in the Lamp3 gene with the following

146 spacer sequences: sgRNA_Lamp3 F1: TCATCTACTGACGATACCAT and sgRNA_Lamp3

147 R1: GCTAGACTAGCTCTGGTTGT microinjected into fertilized zygotes. Genome editing was

148 confirmed by PCR amplification in the resulting founder mice with the primers Lamp3F 5´-

149 GATGGGGGAGGGATCTTTTA-3' and Lamp3R 5´-GTTGGCCTCTGATTGGTTGT-3'. A

150 founder harboring a 40 bp deletion within the second exon leading to a frameshift mutation

151 was selected for further breeding. Mice were housed under specific pathogen-free conditions

152 (12 hours' light/dark cycle, constant room temperature, and humidity). Food and water were

153 available ad libitum.

154

155 Ovalbumin model of allergic asthma

156 Female, 6- to 8-week-old Lamp3-/- and wildtype littermates mice (CAU, Kiel, Germany) were

157 used for the allergic asthma model. They received ovalbumin (OVA)-free diet and water ad

158 libitum. All animal studies were approved by the local animal ethics committee. 6

159 Mice were sensitized to OVA by three intraperitoneal (i.p.) injections of 10 µg of OVA (OVA

160 grade VI, Sigma-Aldrich, St. Louis, MO, USA) adsorbed to 150 µg of aluminum hydroxide

161 (imject alum, Thermo Fisher Scientific, Waltham, MA, USA) on days 1, 14 and 21. To induce

162 acute allergic airway inflammation, mice were exposed three times to an OVA (OVA grade V,

163 Sigma-Aldrich) aerosol (1% wt/vol in PBS) on days 26, 27, and 28. Control animals were

164 sham sensitized to PBS and subsequently challenged with OVA aerosol.

165

166 Lung function analysis

167 Steady-state lung parameters midexpiratory flow (EF50), frequency (f), functional residual

168 capacity (Frc), minute volume (MV), peak expiratory flow (PEF), peak inspiratory flow (PIF),

169 expiration time (Te), inspiration time (Ti), and tidal volume (TV) were measured in conscious,

170 spontaneously breathing mice using FinePointe non-invasive airway mechanics double-

171 chamber plethysmography (NAM, Data Science International, St. Paul, MN, USA). Steady-

172 state airway resistance (RI) and dynamic compliance (Cdyn) were measured in anesthetized

173 and ventilated mice using FinePointe RC Units (Data Science International). Airway

174 responsiveness to methacholine (MCh, acetyl‐β‐methylcholine chloride; Sigma-Aldrich)

175 challenge was invasively assessed on day 29 using FinePointe RC Units (Data Science

176 International, St. Paul, MN, USA) by continuous measurement of RI. Therefore, animals were

177 anesthetized with ketamine (90 mg/kg body weight; cp‐pharma) and xylazine (10 mg/kg BW;

178 cp‐pharma) and tracheotomized with a cannula. Mechanical ventilation was previously

179 described (25). Measurements were taken at baseline (PBS) and in response to inhalation of

180 increased concentrations of aerosolized methacholine (3.125; 6.25; 12.5; 25; 50; and

181 100 mg/mL). After assessment of lung function, all animals were sacrificed by cervical

182 dislocation under deep anesthesia.

183

184 Bronchoalveolar lavage

185 For bronchoalveolar lavage, lungs were rinsed with 1 ml of fresh, ice-cold PBS containing

186 protease inhibitor (Complete, Roche, Basel, Switzerland) via a tracheal cannula. Cells in BAL

187 fluid were counted in the Neubauer chamber. Fifty microliter aliquots of lavage fluids were

188 cytospined, and cells were microscopically differentiated according to morphologic criteria.

189

190 Electron microscopy

191 Animals were anesthetized and lungs were fixed by perfusion through the right ventricle with

192 HEPES buffer (pH 7.35) containing 1.5 % glutaraldehyde and 1.5 % formaldehyde, at a fixed

193 positive inflation pressure of 25 cm H2O. After storage of the lungs in the same solution

194 overnight, 3 mm sized pieces of lung tissue were postfixed first in 1 % osmium tetroxide for 2

195 hours and then in 4 % uranyl acetate overnight (both aqueous solutions). Samples were

196 dehydrated in acetone and embedded in Epon. Ultratin sections of 60 nm were poststained

197 with aqueous uranyl acetate and lead citrate (26) and imaged in a Morgagni TEM (FEI,

198 Eindhoven, NL). All animal studies were approved by the local animal ethics committee.

199

200 Histology

201 Lungs were dissected from mice, and the left lung lobe was snap-frozen in liquid nitrogen.

202 The other lobes were fixed ex-situ with 4% (wt/vol) paraformaldehyde via the trachea under

203 constant pressure, removed and stored in 4% paraformaldehyde. Volume of these lobes was

204 determined based on Archimedes principle. In brief, the remaining lung lobes were placed in

205 a water-filled container set on electronic scales. The weight of water displaced by the

206 submerged lung equals the volume of the lung. Then lung tissues were embedded in

207 paraffin. For analysis of lung inflammation, 2 µm sections were stained with periodic acid-

208 Schiff (PAS). Photomicrographs were recorded by a digital camera (DP-25, Olympus, Tokyo,

209 Japan) attached to a microscope (BX-51, Olympus) with a 20-fold magnification objective

210 using Olympus cell^A software. Mucus quantification was performed as previously described

211 (27). Apoptotic cells were stained using ApopTag Peroxidase In Situ Apoptosis Detection Kit

212 (Merck Millipore, MA, USA) according to manufacturers instructions.

213

214 Immunofluorescence staining of lung tissue

215 4% Paraformaldehyde-fixed lungs were incubated in cryoprotectant solution (30% w/v

216 sucrose in 0.1 M phosphate buffer, pH 7.4) over night. 35 µm thin sections were cut with a

217 Leica 9000s sliding microtome (Leica, Wetzlar, Germany) and collected in cold 0.1 M

218 phosphate buffer. Free floating sections were stained by blocking in blocking solution (0.5%

219 Triton-X 100, 4% normal goat serum in 0.1 M PB pH 7.4) for 1 hour at room temperature.

220 Subsequently, sections were incubated in blocking solution containing the primary antibody

221 at 4°C over night. After washing three times with wash solution (0.1 M PB pH 7.4 containing

222 0.25% Triton-X 100), sections were incubated for 2 hours in secondary antibody in solution,

223 washed again three times in wash solution containing 4′,6-Diamidin-2-phenylindol (DAPI) and

224 finally brought on glass slides and embedded in Mowiol/DABCO. Cells were imaged using a

225 Zeiss confocal microscope equipoped with a 63x objective (Zeiss LSM980 AiryScan 2).

226

227 Homogenization of lung tissue

228 Deep frozen lungs were homogenized with mortar and pestle. An aliquot of 30 mg of lung

229 powder was transferred into RLT buffer, and RNA was isolated with RNeasy Mini Kit

230 (Qiagen) according to the manufactures' guidelines. 1 µg of RNA was used for cDNA

231 synthesis (First Strand cDNA Synthesis Kit, Thermo Fisher Scientific, Waltham, MA, USA).

232 Another 30 mg aliquot of lung powder was transferred into RIPA buffer, and proteins were

233 isolated. Protein concentration was determined with BCA assay (Thermo Fisher Scientific,

234 Waltham, MA, USA).

235 Immunoblot analysis

236 Western analysis was performed on lung tissue homogenates from wildtype and LAMP3-

237 deficient mice. Protein content was measured by the bicinchoninic acid reagent (Thermo

238 Fisher Scientific). Defined amounts of protein were separated on SDS-PAGE and transferred

239 to nitrocellulose membrane. Membranes were incubated with anti-pro-SP-C (rabbit

240 monoclonal, 1:1000, Abcam, Cambridge, UK), anti-pro-SP-B (rabbit polyclonal, 1:2000,

241 Seven Hills Bioreagents, CA, USA), anti-mature SP-B (rabbit polyclonal, 1:1000, Seven Hills

242 Bioreagents), both generous gifts from Jeffrey A. Whitsett, and anti-β-actin (mouse 9

243 monoclonal, 1:200, Santa Cruz Biotechnology, Heidelberg, Germany). Donkey anti-rabbit

244 IgG HRP (1:2000, Santa Cruz) and donkey anti-mouse IgG HRP (1:2000, Cell Signaling)

245 served as secondary Abs. Immunoreactive proteins were visualized using the ECL Western

246 blotting detection system (Bio-Rad Laboratories GmbH, Feldkirchen, Germany), band

247 intensity was quantified with Image J 1.52 (NIH), and data were normalized to β-actin levels.

248

249 qPCR

250 Quantitative RT-PCR was performed on the Roche LightCycler 480 Instrument II system

251 using the LightCycler 480 SYBR Green I Master (Roche Applied Science, Mannheim,

252 Germany). Template cDNA was diluted 1:10. RT-PCR was performed in triplicates in a total

253 volume of 10 µl according to the manufacturers' instruction with a final primer concentration

254 of 0,5 µM. The following primer sequences for housekeeping gene RPL32 were used:

255 forward 5‘-AAAATTAAGCGAAACTGGCG-3', reverse 5‘-ATTGTGGACCAGGAACTTGC-3'.

256 QuantiTect Primer Assays from Qiagen (Venlo, The Netherlands) were used for Sftpb

257 (QT00124908) and Sftpc (QT00109424). Negative controls were included to detect possible

258 contaminations. Amplification specificity was checked using the melting curve and agarose

259 gel electrophoreses. Specific mRNA levels were normalized to the level of the housekeeping

260 gene RPL32 in the same sample.

261 Chemicals and lipid standards for lipidomics

262 All solvents, reagents, and lipid standards were used in the highest available purity. Internal

263 standards were acquired from Avanti Polar Lipids (Avanti Polar Lipids, Alabama, USA).

264 Storage solution was created by combining chloroform, methanol, and water (20/10/1.5;

265 v/v/v. ESI spray mixture was created by combining chloroform, methanol with added 0.1%

266 (wt/v) ammonia acetate and 2-propanol (1:2:4; v/v/v).

267

268 Lipid extraction

269 A customized methyl-tert-butyl ether (MTBE)-based lipid extraction method (28) was applied

270 for BAL and lung tissue homogenates. Details can be found in the Supplementary 10

271 Methods. Lipid extracts were stored in a storage solution at -20 °C until further usage.

272 Cholesterol determination was performed as described earlier (29) (Supplementary

273 Methods).

274

275 Shotgun lipidomics measurements

276 Aliquots of 10 µL of the lipid extract and cholesterol derivatization were diluted in 190 µL ESI

277 spray mixture for each measurement, vigorously mixed, and centrifuged prior to loading into

278 a 96 well plate (Eppendorf, Hamburg, Germany). The well plate was sealed with aluminum

279 sealing foil and kept at 15 °C during the measurement process. All measurements were

280 performed in duplicates. All mass spectrometric acquisitions were performed using a Triversa

281 Nanomate (Advion, Ithaca, USA) as an autosampler and nano-ESI source, applying a spray

282 voltage of 1.1 kV and backpressure of 1.1 psi in ionization modes. For the acquisition of

283 tandem mass spectrometric data, a Q Exactive Plus was used (Thermo Fisher Scientific,

284 Bremen, Germany; methodical details are found in the Supplementary Methods).

285

286 Lipidomics data interpretation

287 For data interpretation, the software pipeline described earlier was utilized (see details

288 Supplementary Methods) (30). Briefly, mass spectra raw files were converted to mzML

289 format with the software module "msconvert" from ProteoWizard version 3.0.18212-

290 6dd99b0f6 (31). Converted files were then imported into LipidXplorer version 1.2.8.1 (32).

291 For each ESI mode, a different set of MFQL (33) files is used. The files generated in

292 LipidXplorer were used in lxPostman for further post-processing, including quality control and

293 quantitation.

294

295 Statistical analysis for lipidomic data

296 Technical replicates are averaged and normalized on protein content. Missing data for

297 ANOVA and PCA were imputed via MetImp 1.2 (34) MCAR/MAR mode with Random Forrest

298 was used as a method with a group-wise missing filter set to 70%. Data were analyzed via

299 Cluster 3.0 version 1.58 and Java Treeview version 1.1.6r4 (35), R version 3.6.1 and

300 RStudio version 1.2.1335 and GraphPad Prism version 8.4.2, and Microsoft Excel for

301 Microsoft 365. Two-way ANOVA was performed with the parameters matching time points for

302 each row, mixed-effects with full fitting, and within each row comparing columns with

303 correcting for multiple comparisons via Tukey. Lipid values are expressed as mean ±

304 standard deviation and significance is designated as *, P < 0.05; **, P < 0.01; ***, P < 0.001.

305 For easier visualization of significant lipid results, lipids from two-way ANOVA were

306 autoscaled. In this procedure, data is centered on the means of groups and then divided by

307 the standard deviation, removing the impact of differences between lipid abundance.

308

309 Bubble Surfactometer

310 Cell free BAL was weighed (=400-800 µL) and ultra-centrifuged (J2-MC, Beckman Coulter,

311 Krefeld, Germany) for 1h at 4°C at 38.730 g. The supernatant was decanted. The pellet was

312 weighed and resuspended using saline containing 1.5 mmol/L CaCL2 equivalent to a

313 concentration of 1:50 (w/w) compared to the basic BAL. Two µL of the concentrated BAL

314 were analyzed for content of choline-containing phospholipids (DPPC, lysolecithin,

315 sphingomyelin; LabAssayTM Phospholipids, FUJIFILM WAKO). Then, samples were

316 normalized to final phospholipids-, i.e. choline containing phospholipids-, concentrations of

317 1.5 mg/ml and analyzed in the Pulsating Bubble Surfactometer (PBS; (36)). A PBS-sample

318 chamber was prefilled using degassed 10% sucrose containing saline plus 1.5 mmol/L

319 CaCL2. Then, 5 µL of each normalized BAL sample were pipetted on the top of the sucrose

320 prefill close to the chimney, where it remains by buoyancy. Subsequently, to mounting of the

321 chamber in the PBS, an air bubble is created by aspiration of ambient air through the

322 chimney and phospholipids might adsorb 30s to the air liquid interface. Then, the bubble is

323 cyclically expanded and compressed at a frequency of 20/min, a compression rate of 50%

324 surface area and at 37° C. Surface tension at maximum (gmax) and minimum (gmin) bubble

325 size is calculated by the PBS and recorded for a 300s period.

326

327 Statistical analyses. If not stated otherwise, a two-tailed unpaired t-test was performed using

328 GraphPad Prism Software Version 5.03. Significant values were considered at P < 0.05.

329 Values are expressed as mean ± standard error of the mean (SEM) and significance is

330 designated as *, P < 0.05; **, P < 0.01; ***, P < 0.001.

331

332 Results

333 Lamp3 knockout mice are viable with normal lung anatomy and pulmonary function

334 LAMP3 is expressed highest in humans and mice in the lung's AT2 cells, where it localizes to

335 LBs (4, 5). Notably, a naturally occurring recessive missense LAMP3 variant has recently

336 been associated with a fatal neonatal interstitial lung disease in Airedale Terrier dogs (24).

337 Therefore, we aimed to analyze the function of LAMP3 in the lung in a genetically more

338 defined animal model and generated Lamp3 knockout mice (Lamp3-/-). CRISPR/Cas9 editing

339 of the Lamp3 allele in mice with guide RNAs targeting exon 2 (Figure 1A, B) yielded a 40 bp

340 deletion resulting in an early Stop-codon in one founder and the mutation was transmitted in

341 the germline (Figure 1C). PCR with primers spanning this deletion was used for genotyping

342 (Figure 1D). LAMP3 was readily detectable by immunohistochemistry on lung sections with

343 a monoclonal antibody in AT2 cells of wildtype mice but was completely absent in Lamp3-/-

344 mice, validating the knockout at the protein level (Figure 1E). We hypothesized that

345 knockout of Lamp3 in mice causes abnormal surfactant homeostasis and postnatal death.

346 However, in contrast to dogs with a LAMP3 mutation, homozygous Lamp3-/- mice were

347 surprisingly born according to the expected Mendelian distribution and survived the weaning

348 phase without increased mortality (Figure 1F). No premature death was observed until 15

349 months-of-age, indicating that LAMP3 is not essential for survival in mice. Macroscopically,

350 Lamp3-/- mice were indistinguishable from wildtype littermates with no obvious phenotype.

351 Histology and analysis of semi-thin sections revealed normal micro-anatomy of the lung with

352 a typical distribution of AT2 cells, type I pneumocytes, alveolar macrophages, and regular

353 capillaries (Figure 1G). Macroscopic signs of multifocal emphysema or edema (as previously

354 described in dogs with a mutation in LAMP3, (24)) were absent in Lamp3-/- mice (Figure 3H).

355 The lung volume, normalized to the body weight in 3-months-old animals, was comparable

356 between wildtype and Lamp3-/- mice (Figure 1I).

357 Furthermore, lung function analysis revealed no abnormalities in the breathing pattern of

358 Lamp3-/- mice regarding the frequency, flow, volume, and time parameters as non-invasively

359 assessed in spontaneously breathing animals as well as for airway resistance and

360 compliance as recorded invasively in relaxed, ventilated animals (Table 1). Analysis of the

361 BAL cellular composition revealed a tendency towards a higher number of lymphocytes and

362 neutrophils in Lamp3-/- mice, which, however, did not reach statistical significance. The

363 number of macrophages was moderately increased in Lamp3-/- mice (Figure 1J).

364

365 LAMP3 deficiency in mice causes changes in surfactant protein levels

366 We reasoned that a lack of LAMP3 might affect the levels of the surfactant proteins SP-B

367 and SP-C, major components of the LB-stored- and secreted surfactant. Indeed, quantitative

368 PCR analysis revealed reduced levels of Sftpb transcript and a similar trend (though not

369 statistically significant) for Sftpc in total lung RNA (Figure 2A) of Lamp3 knockout mice. In

370 contrast, immunoblot analysis of lung tissue lysates from wildtype and Lamp3-/- mice for SP-B

371 and SP-C showed normal levels of pro-SP-B and a striking increase in the levels of (pro-)

372 SP-C, indicating a critical role of LAMP3 in the biosynthesis or turnover of SP-C (Figure 2B,

373 C). The levels of mature SP-B were also significantly increased, though only to a moderate

374 level compared the increase in SP-C.

375 Immunofluorescence analysis on lung sections from wildtype and Lamp3-/- mice revealed no

376 major differences in the staining pattern of the essential LB proteins ABCA3 or mature SP-B.

377 Both proteins localized in wildtype and Lamp3-/- mice to ring-like structures of AT2 cells,

378 representing LBs as determined by high-resolution fluorescence microscopy (Figure 2D).

379

380 The lipid composition of the bronchoalveolar lavage is altered in Lamp3-/- mice

381 Given the apparent differences in SP-C levels and mature SP-B in protein extracts of lung

382 tissue lysates, we next analyzed the lipid composition of the BAL and total lung homogenates

383 by a semi-targeted shotgun lipidomics approach. After shotgun lipidomics, a total of 208

384 different lipid species in the BAL were quantified. We first determined if the two groups

385 (wildtype and Lamp3-/- BAL lipids) could be separated by analyzing the lipidomics dataset by

386 unsupervised principal component analysis (PCA) (Figure 3A). The apparent separation was

387 characterized by increased PGs and LPG levels in the wildtype while SMs are increased in

388 the Lamp3-/- mice. To gain better insight into the perturbations on the lipid species level, we

389 further applied hierarchical clustering (Figure 3B). In conjunction with the PCA, wildtype and

390 Lamp3-/- mouse BAL lipid profiles were clearly distinguished by the underlying Spearman

391 Rank correlation. According to the genotype, the abundance of lipid species' correlated for

392 PG, LPG, and DAG, showing a relative increase in wildtype mice, which was also confirmed

393 by statistical analysis for the selected indicated lipids species (Figure 3C). In contrast, PI and

394 LPI are decreased in Lamp3-/- mice. Notably, PG is an essential component of the pulmonary

395 surfactant, and as multiple lipids of the same class are affected, a disruption of the normal

396 BAL physical properties is likely. In contrast to BAL's lipid profiles, no statistical differences

397 were identified between wild type and Lamp3-/- mice by lipidomic analysis of the perfused

398 total lung tissue (Supplemental Figure 1A-C). These findings, in summary, indicate that

399 LAMP3 deficiency leads to subtle but statistically significant changes in the lipidome and

400 particularly surfactant-relevant lipids of the BAL.

401

402 Lamp3 knockout leads to changes in the ultrastructure of lamellar bodies and altered

403 biophysical properties of pulmonary surfactant

404 Defects in surfactant homeostasis are typically accompanied by a change of the

405 ultrastructure of LBs (11, 12, 24). This finding prompted the analysis of the Lamp3-/- lungs by

406 transmission electron microscopy, focusing on LBs in the AT2 cells. Lamellar bodies from

407 wildtype mice had the typical lamellar appearance (Figure 4A). In Lamp3-/- mice,

408 conspicuous LBs could occasionally be observed within AT2 cells (Figure 4A). They

409 consisted of several clews of lipid lamellae and seemingly homogenous areas that were

410 surrounded by a common limiting membrane. At higher magnification, the homogenous

411 areas also revealed lamellae that were not immediately obvious at lower magnification

412 (Figure 4A). Accordingly, we tested the surfactant's biophysical properties and functionality

413 obtained from BAL of wildtype and Lamp3-/- mice to address any functional relevance of such

414 morphological alterations. Surfactant samples were investigated during dynamic

415 compression of an air bubble in a pulsating bubble (36). The surface tension of wildtype and

416 Lamp3-/- mice is summarized at minimum (gmin) (Figure 4B) and maximum (gmax) bubble

417 size (Figure 4C). Surfactant standardized at a phospholipid concentration of 1.5 mg/ml of

418 both wildtype and Lamp3-/- mice reach a low physiological gmin close to 0 mN/m during the

419 observational period of 300s, respectively (Figure 4B). During the compression period, gmin

420 (0.95 vs 3.6 mN/m, median, p<0.01) and gmax (29.2 vs. 29.9 mN/m, median, p<0.0001) in

421 Lamp3-/- mice were found slightly but significantly lower compared to wild type surfactant.

422 Thus initially, gmin of samples from Lamp3-/- mice decreases faster, reaching <10 mN/m

423 already after 40s compared to wildtype surfactant after 80s (mean) (Figure 4C), indicating a

424 slightly increased adsorption property of surfactant from Lamp3-/- mice.

425

426 Lamp3-/- mice show increased airway resistance and aggravated BAL lipid changes under

427 pathological conditions

428 The overall phenotype of Lamp3-/- mice was mild and did not affect lung function to the extent

429 that functional lung tests could reveal alterations in healthy, spontaneously breathing animals

430 (Table 1). We, therefore, wondered if LAMP3 is more critical under challenged, i.e.,

431 pathological conditions. We applied a well-described model of experimental allergic asthma

432 (25, 37) (Figure 5A) that is characterized by allergic inflammation of the airways, mucus

433 hyperproduction, and the development of airway hyperresponsiveness (AHR), thus,

434 resembles the pathophysiologic hallmarks underlying the symptoms of human asthma.

435 Analysis of the different immune-cell populations (macrophages, lymphocytes, neutrophils,

436 eosinophils) infiltrating the bronchoalveolar lumen in diseased animals did not unveil any

437 differences between wildtype and Lamp3-/- mice. The number of mucus-producing goblet

438 cells and the amount of mucus stored in the airway mucosa did also not differ

439 (Supplemental Figure 3). The degree of apoptotic cell death as revealed by terminal

440 deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining between wildtype and

441 Lamp3-/- mice remained unchanged (Supplemental Figure 4). Importantly, when animals

442 were subjected to MCh-provocation testing to determine the degree of AHR, Lamp3-/- mice

443 displayed significantly increased airway resistance in response to MCh-inhalation (Figure

444 5B).

445 To further investigate the impact of experimental asthma in mice, the lipid composition of the

446 BAL and total lung tissue was determined (Figure 5D-G). The response to experimentally

447 induced asthma led to further divergence in the BAL lipid composition between Lamp3-/- mice

448 and wildtype. Further increase in PG, LPG, and DAG abundance was observed in wildtype

449 mice while levels of SM and PI increased in in Lamp3-/-. In the lipidomes of the perfused total

450 lung tissue, no further changes were observed (Supplemental Figure 2). Hierarchical

451 cluster analysis further revealed that the opposite response in the BAL lipidome of the

452 wildtype and Lamp3-/- mice affected a wider range of lipid species, compared to the healthy

453 animals (Figure 5E, Figure 3B-D). Spearman Rank correlation between the two distinct

454 groups (wildtype and Lamp3-/-) further confirmed this lipid phenotype by high correlation

455 factor for the two major branches comprising lipid compositions of 8 animals in each group

456 (Figure 5F). Lipid species of mainly PG, LPG, and DAG classes show a relative increase in

457 Lamp3+/+ mice, which were statistically significant for indicated lipids species. In contrast, PI,

458 LPE, Cer, and SM showed increased abundance in Lamp3-/- mice (Figure 5G). In summary,

459 the lipid homeostasis for surfactant production in BAL of Lamp3-/- mice are more disturbed

460 under the pathophysiologic conditions of experimental asthma when compared to the

461 unchallenged animals.

462

463 Discussion

464 The two best-characterized family members of the LAMP family, LAMP1 and LAMP2, are

465 ubiquitously expressed and quantitatively account for a substantial fraction of the lysosomal

466 membrane proteome (21, 38). Though the function of LAMP1 and LAMP2 are still not fully

467 understood, they are supposed to play pivotal roles in the fusion of membranes, autophagy,

468 microtubule-dependent positioning of lysosomes, and especially as structural components of

469 the lysosomal membrane (39). In contrast, the expression of the structurally related LAMP3

470 protein is limited to specific cell types, implicating a more defined and tissue/cell-type-specific

471 function. Together with the fact that LAMP3 localizes to LBs rather than lysosomes, a critical

472 function of LAMP3 in LB function and surfactant homeostasis is suggested. In fact, the recent

473 finding that a natural mutation (p.(E387K) in LAMP3 leads to clinical symptoms and

474 pathology similar to the most severe neonatal forms of surfactant deficiency in an Airedale

475 Terrier dog bred suggested a critical role in surfactant biology and lung physiology for

476 LAMP3.

477 Therefore, we aimed to investigate the role of LAMP3 in lung physiology by using a genuine

478 knockout model. To our surprise, our findings obtained from Lamp3-/- mice revealed striking

479 differences to those found in dogs with the recessive missense LAMP3 variant: While lethal

480 hypoxic respiratory distress and failure lead to the premature death of dog puppies within the

481 first days or week of living, Lamp3--/- mice developed normally and did not display signs of

482 lung pathology and dysfunction after birth. We can only speculate about the apparent

483 difference between the different animal models: On the one hand, in contrast to laboratory

484 mouse strains, dog breeds like the Airedale Terrier were generated by inbreeding that aimed

485 to pronounce or produce distinct anatomical features. Thus, it could not be excluded that

486 other genetic factors, possibly homozygous in the inbred dogs, impact this phenotype. On

487 the other hand, the point mutation (p.(E387K)) found in the dogs resulted in the expression of

488 a LAMP3 variant that could still be detected with a LAMP3-specific antibody in lung tissue,

489 suggesting that the overall protein structure of the LAMP3 variant is at least in part

490 untouched by the mutation. This finding leaves the possibility of an impaired function of the

491 mutated variant, which is not the case when LAMP3 is completely knocked-out by

492 CRISPR/Cas9. These findings, however, should be considered when human patients

493 suffering from surfactant deficiencies are analyzed: Our data suggest that LAMP3 mutations

494 might also present with a mild clinical phenotype, maybe not immediately presented with the

495 typical signs of a full surfactant deficiency in human patients suffering from childhood

496 interstitial lung disease. Since neither snap-frozen tissue samples nor BAL from the affected

497 Airedale Terrier puppies could be obtained, characterization of the phenotype caused by a

498 mutation of the LAMP3 gene in dogs is quite limited in regard to the lipidome profile of BAL

499 and the function of the surfactant.

500 The lipidomics analyses performed in this study are in good agreement with earlier reported

501 quantities of main lipid species and classes by Prueitt et al. (16). We find PC at 67.1 mol% in

502 wildtype mice (16 mol% neutral lipids excluded as in study) compared to the previously

503 reported 75.2 mol% (16). The main contributing PC class lipid species is DPPC (PC

504 16:0/16:0) with 43.9% compared to the 30-60% reported in the literature (14, 17, 40, 41). PG

505 was found at 9 mol% in agreement with the values reported by others (16). Our analyses

506 show a clear separation of the genotypes according to different lipid compositions and upon

507 induction of experimental asthma. In the lipidomes of the perfused lung tissues, only minor

508 Lamp3 genotype-dependent changes were observed, which underlines the importance of

509 LAMP3 to regulate the lipid secretion in the surfactant. It is further noteworthy that the BAL

510 lipid composition was mostly affected by the reduction in the abundance of PGs, LPG, and

511 DAG in the Lamp3-/- mice.

512 Importantly, experimentally induced asthma further reduced the levels of essential surfactant

513 lipids and in parallel, led to an increase in lipids like SM, PI, and PS. Interestingly lipids

514 containing arachidonic acid (AA, FA 20:4) were elevated in Lamp3-/- mice. Further affects by

515 signaling was not investigated in this study, where AA plays a crusial role as educt of

516 numerous lipid mediators. These lipids might further indicate a higher degree of cell lysis in

517 conjunction with increased number of immune cells in BAL after OVA treatment and an

518 excessive immune reaction. However, it should be noted that we did not observe major

519 differences in the number of apoptotic cells in lung sections.

520 While our data support a role of LAMP3 in surfactant homeostasis, we can only speculate

521 about the mechanisms of how LAMP3 might regulate surfactant levels secreted by AT2 cells.

522 Different scenarios are feasible: On one hand, it has been speculated previously that LAMP3

523 plays a role in surfactant recycling by retrieving remnants of secreted LB and re-endocytosis

524 (4). On the other hand, other LAMP family members are acting as accessory subunits of

525 multi-transmembrane spanning proteins like LAMP1/LAMP2 for the lysosomal polypeptide

526 transporter TAPL (42). Likewise, UNC-46, the C. elegans orthologue of LAMP5, acts as a

527 trafficking chaperone, essential for the correct targeting of the nematode vesicular GABA-

528 transporter UNC-47 (43). LAMP3 could take over a similar function, e.g., by modulating or

529 regulating the trafficking or stability of one of the integral LB proteins like SP-C (which is

530 synthesized as a type II transmembrane protein). However, such a mechanism might not be

531 essential and become relevant only under conditions, where surfactant levels need to be

532 regulated more tightly. It is interesting to note that an aggravation of the lipid changes under

533 the pathological conditions of experimental allergic asthma may explain the lung function

534 differences observed after MCh-provocation, when lung function was pushed to the limit. In

535 sensitized animals, inhalation of allergen aerosols results in an acute inflammatory reaction

536 with an infiltration of T helper 2 (TH2) cells and eosinophils into the airways and therelease of

537 a plethora of cytokines, chemokines, and cytotoxic metabolites. Consequently, this causes

538 activation of submucosal glands and differentiation of goblet cells, ultimately resulting in

539 increased mucus production. TH2-type cytokines, together with eosinophil products, trigger

540 the development of AHR that leads to an increased airway resistance in response to non-

541 allergic stimuli and the formation of the typical asthma attack. Therefore, in mice with

542 experimental allergic asthma, the degree of allergic airway inflammation is closely correlated

543 with the degree of mucus production, and AHR. Remarkably, this is not the case in Lamp3-/-

544 mice: While numbers of inflammatory cells such as eosinophils and lymphocytes as well as 21

545 the number of goblet cells and amount of mucus stored in the airway mucosa are not

546 different between wildtype and Lamp3-/- mice, Lamp3-/- mice display a markedly increased

547 airway resistance in response to MCh inhalation. Since this response appears to be

548 independent of both the degree of airway inflammation and mucus production, it is possible

549 that the increase in airway resistance is not due to an increased AHR but could originate in

550 an altered composition and thus a function of surfactant lining the alveoli and airways of

551 Lamp3-/- mice. Hence, in mice, not only the alveoli but also the ductūs alveolares are affected

552 by the tension-reducing effects of surfactant. LAMP3 deficiency leads to an altered lipid

553 composition of surfactant and consequently to a reduced effect on surface tension, which

554 could in turn physiologically lead to increased surface tension in the distal parts of the airway

555 system and, thus, to an increased total resistance under the fierce conditions of a MCh-

556 provocation test.

557 Taken together, whereas the lack of LAMP3 seems to be compensable under physiological

558 conditions, it impacts basal surfactant homeostasis and lung lipid composition and airway

559 resistance in experimental allergic asthma.

560

561 Acknowledgment

562 We thank Gabriele Walter, Maike Langer, Linda Lang, Juliane Artelt, Franziska Beyersdorf,

563 and Dominika Biedziak for excellent technical assistance. We thank Dr. Nils Hoffmann (ISAS

564 Dortmund) for support in data upload and data base management of the LIFS webportal. The

565 authors declare no financial and non-financial competing interests. Radislav Sedlacek's work

566 was supported by funding of the Ministry of Education, Youth and Sports to the Czech Centre

567 for Phenogenomics (LM2018126) and by the Institute of Molecular Genetics of the Czech

568 Academy of Sciences (RVO 68378050). Research in Dominik Schwudke's lab was supported

569 by Lipidomics Informatics for Life Science (LIFS) of the German Network for Bioinformatics

570 Infrastructure (de.NBI, BMBF).

571

572 Author contribution

573 LPL, DK, DT, CS, NL, JH, MO, BS, RS, PS, DS, MW, and MD contributed to data

574 acquisition, analysis, and interpretation. LPL, MW, and MD concepted and designed the

575 study. MD and MW draft the initial version of the manuscript. DK and DS designed and

576 performed all lipidomics experiments and all related data analysis. All authors were involved

577 in drafting the final manuscript, have provided final approval of this manuscript for

578 submission, and agreed to be accountable for their specific contribution of this study.

579 References

580 1. Furuta K, Yang XL, Chen JS, Hamilton SR, August JT. Differential expression of the

581 lysosome-associated membrane proteins in normal human tissues. Arch Biochem Biophys.

582 1999;365(1):75-82.

583 2. Holness CL, Simmons DL. Molecular cloning of CD68, a human macrophage marker

584 related to lysosomal glycoproteins. Blood. 1993;81(6):1607-13.

585 3. Tiveron MC, Beurrier C, Ceni C, Andriambao N, Combes A, Koehl M, et al. LAMP5

586 Fine-Tunes GABAergic Synaptic Transmission in Defined Circuits of the Mouse Brain. PLoS

587 One. 2016;11(6):e0157052.

588 4. Salaun B, de Saint-Vis B, Pacheco N, Pacheco Y, Riesler A, Isaac S, et al.

589 CD208/dendritic cell-lysosomal associated membrane protein is a marker of normal and

590 transformed type II pneumocytes. Am J Pathol. 2004;164(3):861-71.

591 5. Akasaki K, Nakamura N, Tsukui N, Yokota S, Murata S, Katoh R, et al. Human

592 dendritic cell lysosome-associated membrane protein expressed in lung type II pneumocytes.

593 Arch Biochem Biophys. 2004;425(2):147-57.

594 6. de Saint-Vis B, Vincent J, Vandenabeele S, Vanbervliet B, Pin JJ, Ait-Yahia S, et al. A

595 novel lysosome-associated membrane glycoprotein, DC-LAMP, induced upon DC

596 maturation, is transiently expressed in MHC class II compartment. Immunity. 1998;9(3):325-

597 36.

598 7. Salaun B, de Saint-Vis B, Clair-Moninot V, Pin JJ, Barthelemy-Dubois C,

599 Kissenpfennig A, et al. Cloning and characterization of the mouse homologue of the human

600 dendritic cell maturation marker CD208/DC-LAMP. Eur J Immunol. 2003;33(9):2619-29.

601 8. Veldhuizen EJ, Haagsman HP. Role of pulmonary surfactant components in surface

602 film formation and dynamics. Biochim Biophys Acta. 2000;1467(2):255-70.

603 9. Nogee LM, de Mello DE, Dehner LP, Colten HR. Brief report: deficiency of pulmonary

604 surfactant protein B in congenital alveolar proteinosis. N Engl J Med. 1993;328(6):406-10.

605 10. Nogee LM, Dunbar AE, 3rd, Wert SE, Askin F, Hamvas A, Whitsett JA. A mutation in

606 the surfactant protein C gene associated with familial interstitial lung disease. N Engl J Med.

607 2001;344(8):573-9.

608 11. Clark JC, Wert SE, Bachurski CJ, Stahlman MT, Stripp BR, Weaver TE, et al.

609 Targeted disruption of the surfactant protein B gene disrupts surfactant homeostasis, causing

610 respiratory failure in newborn mice. Proc Natl Acad Sci U S A. 1995;92(17):7794-8.

611 12. Glasser SW, Burhans MS, Korfhagen TR, Na CL, Sly PD, Ross GF, et al. Altered

612 stability of pulmonary surfactant in SP-C-deficient mice. Proc Natl Acad Sci U S A.

613 2001;98(11):6366-71.

614 13. Yu S, Harding PGR, Smith N, Possmayer F. Bovine Pulmonary Surfactant -

615 Chemical-Composition and Physical-Properties. Lipids. 1983;18(8):522-9.

616 14. Kahn MC, Anderson GJ, Anyan WR, Hall SB. Phosphatidylcholine Molecular-Species

617 of Calf Lung Surfactant. Am J Physiol-Lung C. 1995;269(5):L567-L73.

618 15. Agudelo CW, Kumley BK, Area-Gomez E, Xu YM, Dabo AJ, Geraghty P, et al.

619 Decreased surfactant lipids correlate with lung function in chronic obstructive pulmonary

620 disease (COPD). PLoS One. 2020;15(2).

621 16. Prueitt JL, Chi EY, Lagunoff D. Pulmonary Surface-Active Materials in Chediak-

622 Higashi-Syndrome. J Lipid Res. 1978;19(4):410-5.

623 17. Postle AD, Heeley EL, Wilton DC. A comparison of the molecular species

624 compositions of mammalian lung surfactant phospholipids. Comp Biochem Phys A.

625 2001;129(1):65-73.

626 18. Lopez-Rodriguez E, Perez-Gil J. Structure-function relationships in pulmonary

627 surfactant membranes: from biophysics to therapy. Biochim Biophys Acta.

628 2014;1838(6):1568-85.

629 19. Arbibe L, Koumanov K, Vial D, Rougeot C, Faure G, Havet N, et al. Generation of

630 lyso-phospholipids from surfactant in acute lung injury is mediated by type-II phospholipase

631 A2 and inhibited by a direct surfactant protein A-phospholipase A2 protein interaction.

632 Journal of Clinical Investigation. 1998;102(6):1152-60.

633 20. Fitzgerald ML, Xavier R, Haley KJ, Welti R, Goss JL, Brown CE, et al. ABCA3

634 inactivation in mice causes respiratory failure, loss of pulmonary surfactant, and depletion of

635 lung phosphatidylglycerol. J Lipid Res. 2007;48(3):621-32.

636 21. Chen JW, Pan W, D'Souza MP, August JT. Lysosome-associated membrane

637 proteins: characterization of LAMP-1 of macrophage P388 and mouse embryo 3T3 cultured

638 cells. Arch Biochem Biophys. 1985;239(2):574-86.

639 22. Shulenin S, Nogee LM, Annilo T, Wert SE, Whitsett JA, Dean M. ABCA3 gene

640 mutations in newborns with fatal surfactant deficiency. N Engl J Med. 2004;350(13):1296-

641 303.

642 23. Somaschini M, Nogee LM, Sassi I, Danhaive O, Presi S, Boldrini R, et al.

643 Unexplained neonatal respiratory distress due to congenital surfactant deficiency. J Pediatr.

644 2007;150(6):649-53, 53 e1.

645 24. Dillard KJ, Ochs M, Niskanen JE, Arumilli M, Donner J, Kyostila K, et al. Recessive

646 missense LAMP3 variant associated with defect in lamellar body biogenesis and fatal

647 neonatal interstitial lung disease in dogs. PLoS Genet. 2020;16(3):e1008651.

648 25. Webering S, Lunding LP, Vock C, Schroder A, Gaede KI, Herzmann C, et al. The

649 alpha-melanocyte-stimulating hormone acts as a local immune homeostasis factor in

650 experimental allergic asthma. Clin Exp Allergy. 2019;49(7):1026-39.

651 26. Reynolds ES. The use of lead citrate at high pH as an electron-opaque stain in

652 electron microscopy. J Cell Biol. 1963;17:208-12.

653 27. Lunding LP, Webering S, Vock C, Behrends J, Wagner C, Holscher C, et al.

654 Poly(inosinic-cytidylic) acid-triggered exacerbation of experimental asthma depends on IL-

655 17A produced by NK cells. J Immunol. 2015;194(12):5615-25.

656 28. Matyash V, Liebisch G, Kurzchalia TV, Shevchenko A, Schwudke D. Lipid extraction

657 by methyl-tert-butyl ether for high-throughput lipidomics. J Lipid Res. 2008;49(5):1137-46.

658 29. Liebisch G, Binder M, Schifferer R, Langmann T, Schulz B, Schmitz G. High

659 throughput quantification of cholesterol and cholesteryl ester by electrospray ionization

660 tandem mass spectrometry (ESI-MS/MS). Bba-Mol Cell Biol L. 2006;1761(1):121-8.

661 30. Schwudke D, Shevchenko A, Hoffmann N, Ahrends R. Lipidomics informatics for life-

662 science. J Biotechnol. 2017;261:131-6.

663 31. Chambers MC, Maclean B, Burke R, Amodei D, Ruderman DL, Neumann S, et al. A

664 cross-platform toolkit for mass spectrometry and proteomics. Nat Biotechnol.

665 2012;30(10):918-20.

666 32. Herzog R, Schuhmann K, Schwudke D, Sampaio JL, Bornstein SR, Schroeder M, et

667 al. LipidXplorer: A Software for Consensual Cross-Platform Lipidomics. PLoS One.

668 2012;7(1).

669 33. Herzog R, Schwudke D, Schuhmann K, Sampaio JL, Bornstein SR, Schroeder M, et

670 al. A novel informatics concept for high-throughput shotgun lipidomics based on the

671 molecular fragmentation query language. Genome Biol. 2011;12(1).

672 34. Wei RM, Wang JY, Su MM, Jia E, Chen SQ, Chen TL, et al. Missing Value Imputation

673 Approach for Mass Spectrometry-based Metabolomics Data. Sci Rep. 2018;8.

674 35. Saldanha AJ. Java Treeview-extensible visualization of microarray data.

675 Bioinformatics. 2004;20(17):3246-8.

676 36. Enhorning G. Pulsating bubble technique for evaluating pulmonary surfactant. J Appl

677 Physiol Respir Environ Exerc Physiol. 1977;43(2):198-203.

678 37. Lunding L, Webering S, Vock C, Schroder A, Raedler D, Schaub B, et al. IL-37

679 requires IL-18Ralpha and SIGIRR/IL-1R8 to diminish allergic airway inflammation in mice.

680 Allergy. 2015;70(4):366-73.

681 38. Wilke S, Krausze J, Bussow K. Crystal structure of the conserved domain of the DC

682 lysosomal associated membrane protein: implications for the lysosomal glycocalyx. BMC

683 Biol. 2012;10:62.

684 39. Eskelinen EL, Tanaka Y, Saftig P. At the acidic edge: emerging functions for

685 lysosomal membrane proteins. Trends Cell Biol. 2003;13(3):137-45.

686 40. King RJ. Pulmonary Surfactant. J Appl Physiol (1985). 1982;53(1):1-8.

687 41. Holm BA, Wang ZD, Egan EA, Notter RH. Content of dipalmitoyl phosphatidylcholine

688 in lung surfactant: Ramifications for surface activity. Pediatr Res. 1996;39(5):805-11.

689 42. Demirel O, Jan I, Wolters D, Blanz J, Saftig P, Tampe R, et al. The lysosomal

690 polypeptide transporter TAPL is stabilized by interaction with LAMP-1 and LAMP-2. J Cell

691 Sci. 2012;125(Pt 18):4230-40.

692 43. Schuske K, Palfreyman MT, Watanabe S, Jorgensen EM. UNC-46 is required for

693 trafficking of the vesicular GABA transporter. Nat Neurosci. 2007;10(7):846-53.

694

695

696

697 Tables

698 Table 1. Steady state lung parameters in wildtype and Lamp3-/- mice. Airway resistance

699 (RI) and dynamic compliance were measured in anesthetized and ventilated mice using

700 FinePointe RC Units. Midexpiratory flow (EF50), frequency (f), functional residual capacity

701 (Frc), minute volume (MV), peak expiratory flow (PEF), peak inspiratory flow (PIF), expiration

702 time (Te), inspiration time (Ti), and tidal volume (TV) were measured in conscious,

703 spontaneously breathing mice using FinePointe non-invasive airway mechanics (NAM)

704 double chamber plethysmography. Values displayed as mean ± SEM, n= 7 per genotype.

parameter (abbr.) unit mean ± SEM mean ± SEM p-value (WT) (KO) airway resistance (RI) cmH2O*secs/mL 1.423 ± 0.19 1.282 ± 0.16 0.5869 dynamic compliance (Cdyn) mL/cm H2O 0.041 ± 0.01 0.051 ± 0.01 0.5126 midexpiratory flow (EF50) ml/sec 5.028 ± 0.32 4.958 ± 0.31 0.8772 frequency (f) BPM 235.6 ± 8.1 234.0 ± 9.2 0.8968 functional residal capacity (Frc) ml 0.7667 ± 0.14 0.7275 ± 0.05 0.7946 minute volume (MV) ml/min 123.9 ± 7.9 121.2 ± 6.7 0.7985 peak expiratory flow (PEF) ml/sec 5.995 ± 0.38 5.884 ± 0.30 0.8215 peak inspiratory flow (PIF) ml/sec 5.571 ± 0.38 5.314 ± 0.31 0.6088 expiration time (Te) sec 0.1308 ± 0.005 0.1308 ± 0.005 0.9982 inspiration time (Ti) sec 0.1281 ± 0.004 0.1300 ± 0.004 0.7522 tidal volume (TV) ml 0.5234 ± 0.018 0.5190 ± 0.014 0.8480 705

706

707

708 Figure legends

709 Figure 1. CRISPR/Cas9-mediated knockout of Lamp3 in mice. (A) CRISPR guide-RNA

710 sequences used for targeted editing of the murine Lamp3 allele. The Protospacer Adjacent

711 Motif (PAM) is underlined. (B) Schematic drawing of CRISPR/Cas9-mediated targeting of

712 exon 2 of the Lamp3 locus. (C) Sanger sequencing chromatogram of a PCR-product

713 covering the targeted segment with genomic tail-DNA of one wild type and one homozygous

714 Lamp3-/- mouse as a template. CRISPR target site one is underlined. The frameshift resulting

715 from endogenous repair mechanisms results in an early stop codon. (D) 2% Agarose gel with

716 PCR-products of a PCR covering parts of exon 2 of the Lamp3 locus used for genotyping.

717 (E) Immunohistochemistry staining of lung sections of wildtype and Lamp3-/- mice with an

718 antibody specific for LAMP3. (F) Genotype distribution of the litters from heterozygote

719 Lamp3+/--breeding pairs after weaning (3-4 weeks after birth). N = 208 individuals. (G)

720 Hematoxylin / Eosin staining of lung sections from adult wildtype and Lamp3-/- mice. (H)

721 Photos of the inflated lungs of adult wildtype and Lamp3-/- mice. (I) Lung volume/body weight

722 ratio of wild type and Lamp3-/- mice. n = 8 (per genotype), ns = not significant. (J) The

723 number of inflammatory cells (lymphocytes, neutrophils, eosinophils, and macrophages) in

724 the BAL of wildtype and Lamp3-/- mice. n = 8 (per genotype), ns = not significant; 0.05; * = p

725 ≤ 0.05.

726 Figure 2. LAMP3-deficiency results in altered (pro-)surfactant C levels. (A) Quantitative

727 real-time PCR of total RNA from wildtype and Lamp3-/- mice for SP-B and SP-C. The

728 expression is normalized to the housekeeping gene RPL32. n = 8 (per genotype), ns = not

729 significant; ** = p ≤ 0.01. (B) Representative immunoblots of lung tissue lysates from wildtype

730 and Lamp3-/- mice for pro-SP-C, pro-SP-B, and mature SP-B. Actin is depicted as a loading

731 control. (C) Quantification of immunoblots for pro-SP-C, pro-SP-B, and mature SP-B

732 normalized to Actin as a loading control. n = 8 (per genotype); ns = not significant; *** = p ≤

733 0.001. (D) Immunofluorescence staining of lung sections from adult wildtype and Lamp3-/-

734 mice for ABCA3 and mature SP-B (green). Nuclei are stained with DAPI (blue).

735 Figure 3. Deficiency of LAMP3 results in minor changes in the BAL lipidome under

736 basal conditions. (A) Three-dimensional principal component analysis (PCA) loadings plot

737 of the BAL lipidome of wild type (black) and Lamp3-/- (green) mice. Each point represents one

738 individual animal. Arrows show the 10 lipid species with the strongest effect across all

739 principal components. Ellipses show the 95% confidence interval of the group. PC1 to PC3

740 explain 65% of the variability in the data set. (B) Hierarchical clustering of 140 lipid species

741 (of 208 after application of a 90% occupation threshold) identified in eight wild type and

742 seven Lamp3-/- mice BAL fluid samples. Each row represents a lipid species and each

743 column a sample. Red boxes indicate enlarged selections (C2) and (C3). (C1) The

744 hierarchical clustered tree of sample groups correlates both groups into two major branches.

745 (C2) The Upper selected segment of the HCA with 31 lipids shown with a relative decrease

746 to wildtype mice. (C3) The lower selected segment of the HCA was shown with a relative

747 increase of 7 lipids to the control. (D) Differences in the BAL lipidome of wild type and

748 Lamp3-/- mice mutant using autoscaled ranges. Samples are color-coded according to their

749 group, black for Lamp3+/+ and green for Lamp3-/- mice. Boxplots show the median with the

750 interquartile range. Whiskers show maximum and minimum with outliers in the respective

751 colors. Data is on the basis of mole percentage. Only lipid species are shown with a

752 significant difference computed with a two-way ANOVA.

753 Figure 4. Deficiency of LAMP3 leads to altered LB morphology and reduced surfactant

754 functionality, (A) Transmission electron microscopy of LB in an alveolar AT2 cells of a

755 wildtype and a Lamp3-/- mouse. Wildtype: Individual Lamellar Bodies are clearly discernible

756 and separated (left), containing continuous lipid lamellae (right). Lamp3-/-: Left: Multiple

757 clews of lipid lamellae (white asterisks) are located within shared limiting membranes (white

758 arrowheads). The space between the clews (black asterisks) is entirely filled with light grey

759 material. Right: enlargement of the boxed area indicated left. Higher magnification reveals

760 lamellae also inside the light grey material, which in some cases obviously are continuations

761 of the darker stained lamellae inside the clew (top) or the stack of lamellae bottom left. (B, C)

762 Surface tension at minimum (gmin; C)) and maximum (gmax; B) bubble size, derived from a 31

763 pulsating bubble surfactometer (36) during a compression period of 300s. Eight pellets of

764 BAL from Lamp3-/- and wild type mice were prepared by ultracentrifugation and resuspended

765 in saline CaCl2 to standardize phospholipids at 1.5 mg/ml each. Phospholipid concentration

766 was determined using a commercial kit. Data are plotted as mean +/-SD. Statistical analysis

767 was performed using an unpaired t-test.

768 Figure 5. The stress-induced allergic asthma model aggravates phenotypes in Lamp3-/-

769 mice and causes a genotype-dependent increase in airway resistance. (A) Experimental

770 setup of the stress-challenged model of allergic asthma and the experimental groups. (B)

771 Airway resistance of the four experimental groups (wildtype, Lamp3-/-, wildtype + OVA,

772 Lamp3-/- +OVA). N = 8 (per genotype), significances calculated for methacholine provocation

773 test after 100 mg/ml methacholine exposure, ns = not significant; 0.05; **** = p ≤ 0.0001 (C)

774 Number of inflammatory cells in the BAL of wildtype, Lamp3-/-, wildtype + OVA, Lamp3-/-

775 +OVA mice. N = 8 (per genotype), ns = not significant; 0.05; * = p ≤ 0.05. (D) Two-

776 dimensional PCA loadings plot of the BAL lipidome of wild type (black), Lamp3-/- (green), wild

777 type +OVA (blue), and Lamp3-/- +OVA (red) mice. The datasets for wild type (black), Lamp3-/-

778 (green) are identical to (Figure 3A). Each point represents one individual. Arrows show the

779 10 lipid species with the strongest effect on all principal components. Ellipses indicate the

780 95% confidence interval of the experimental groups, where PC1 and PC2 describe 48.9% of

781 the variability in the data set. The separation of the groups is in PC1, and PC2 is not

782 complete for all groups. A clear separation between the Lamp3+/+ +PBS and Lamp3+/+ +OVA

783 can be seen. The separation between untreated Lamp3-/- and treated Lamp3-/- is not

784 complete. (E) Hierarchical clustering of 140 lipid species (of 208 after application of a 90%

785 occupation threshold) identified in 8 wild type +OVA and 8 Lamp3-/- +OVA mice BAL fluid

786 samples. Each row represents a lipid species and each column a sample. Red boxes

787 indicate selections shown in (F). The two major branches of the tree represent the genotypes

788 without false assignments. Selected segments of the hierarchical clustered tree where a

789 relative decrease for 21 lipids for Lamp3-/-, mainly DAG and PG species, compared to the

790 Lamp3+/+ mice are observed. The lower selection shows, in contrast, a cluster of 26 lipids 32

791 with a relative increase in quantity in the mutant compared to Lamp3+/+. (G) Differences in the

792 BAL lipidome of wild type +OVA and Lamp3-/- +OVA applying autoscaling. Samples are

793 color-coded according to their group, red for Lamp3+/+ +OVA and blue for Lamp3-/- +OVA.

794 Boxplots show the median with the interquartile range, and whiskers show maximum and

795 minimum with outliers in the respective colors. Data is based on mole percentage. Selected

796 lipid species showed a significant difference by two-way ANOVA. (H) Bar plots of absolute

797 lipid values normalized on protein content. Error bars signify standard deviations.

33 bioRxiv preprint doi: https://doi.org/10.1101/2021.02.05.429758; this version posted February 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 1

A C TTC A T C T A C T G A C G A T A CCA TTGGC TC TGGC TCA

target site 1: TCATCTACTGACGATACCAT-TGG +/+ target site 2: CCT-ACAACCAGAGCTAGTCTAGC Lamp3 B S S T D D T I G S G S Cas9 TTC A T C T A C T G A C G A T A C A CCA G A G C T AG T C T AG

Lamp3-/- guide RNA genomic DNA S S T D D T P E L V STOP double strand break Lamp3 Exon2

+/+ Lamp3 Exon2 /- Lamp3 +/+ + -/- D F Lamp3 +/- repair O -/- H 2 Lamp3 Lamp3Lamp3 23 % Lamp3 40 bp deletion 33 % -559 bp -519 bp Lamp3 Exon2 43 % n = 208 E G Lamp3+/+ Lamp3+/+

LAMP3 DAPI merge Lamp3-/- Lamp3-/-

LAMP3 DAPI merge 10 μm

50 μm H I J +/+ Lamp3 lymphocytes eosinophils neutrophils macrophages 0.03 ns 0.4 ns 4 * ) 4 0.3 3 0.02

0.2 ns 2 Lamp3-/- 0.01 ns 0.1 1 Inflammatory cells cells/ml BAL (x10 n=8 n=8 n=8 n=8 n=8 n=8 n=8 Lung volume / bodyweight n=8 0.00 0.0 n=8 0 /+ /- - - - - + - /+ / /+ / /+ / /+ / + - + - + - + - 3 3 3 3 p p mp3 mp mp3 m Lamp3 a a a amp3 amp a Lam L L L Lamp3 L L L Lamp3 bioRxiv preprint doi: https://doi.org/10.1101/2021.02.05.429758; this version posted February 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 2

A B 30 ** 8 +/+ -/- ns Lamp3 Lamp3 kDa 6 CT] CT] 28- 20 pro   SP-C 4 55- expression rel. 10 expression rel. Actin 2 36- to RPL32 [ to RPL32 [ Sftpc Sftpb n=8 n=8 0 n=8 n=8 0 pro - - + -/ /+ +/ + -/ 25- 3 3 3 3 SP-B p m a amp amp amp 55- C L L L L Actin 20000 6000 ns 15000 36- *** *** 15000 mat 4000 10000 15- SP-B 10000 pro SP-C pro SP-B 2000 5000

5000 mature SP-B n=8 -Actin normalized) [a.u.] -Actin normalized) [a.u.] -Actin normalized) [a.u.] n=8 n=8   n=8  n=8  n=8 ( ( 0 ( 0 0 + - - / -/ + - /+ -/ + +/ -/ + 3 3 3 3 3 p 3 amp m amp L amp a amp amp L L L L L D Lamp3+/+ Lamp3+/+ Lamp3+/+ Lamp3+/+

ABCA3 ABCA3 mature SP-B mature SP-B Lamp3-/- Lamp3-/- Lamp3-/- Lamp3-/-

ABCA3 100 μm ABCA3 10 μm mature SP-B 100 μm mature SP-B 10 μm bioRxiv preprint doi: https://doi.org/10.1101/2021.02.05.429758; this version posted February 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 3

A B Lamp3+/+ Lamp3-/- C

Corr.=0.0566

Corr.=0.3000

Corr.=0.6047

Lamp3+/+ D Lamp3-/-

DAG 14:0_18:2 DAG 16:0_16:0 DAG DAG 16:0_18:1 DAG 16:0_18:2 LPG 16:0 LPG Corr.=0.7524 LPG 20:4

LPI LPI 16:0 LPI 20:4 PC O- PC O− 14:0/16:0 PE 16:1_20:4 PE PE 18:2_18:1 PE O- PE O− 16:0/20:4 PG 14:0_16:0 PG 14:0_18:2 PG 15:0_18:2 PG 16:0_16:0 PG 16:0_22:4 PG PG 16:1_16:0 PG 17:0_18:2 PG 18:0_20:4 PG 18:2_20:3 PG 18:3_18:2 PI 16:0_18:1 PI 16:0_20:4 PI PI 18:0_20:4 PI 18:1_20:4 SM 34:1;0 SM SM 42:1;0 SM 42:3;0 -2 -1 0 1 2 Autoscale bioRxiv preprint doi: https://doi.org/10.1101/2021.02.05.429758; this version posted February 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 4

Lamp3+/+

Lamp3-/- BC Gmin Gmax 35 25 <0.0001**** 20 0.0066** 30 15 [mN/m] [mN/m] 10 min max   25 Lamp3+/+ 5 20 Lamp3-/- 10 0 0 0 100 200 300 0 100 200 300 time [sec] time [sec] bioRxiv preprint doi: https://doi.org/10.1101/2021.02.05.429758; this version posted February 6, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 5 AB150 +/+ **** Sensitization Challenge Lamp3 Lamp3-/- OVA/Alum i.p./ OVA/Aerosol +/+ 100 Lamp3 OVA PBS -/- † Lamp3 OVA

Day 1 14 21 25 26 27 28 29 50 ns Groups: Lamp3+/+ PBS Lamp3-/- PBS

+/+ rel. to baseline (%) Lamp3 OVA 0 -/-

Lamp3 OVA Increase Resistance Airway 0 3.125 6.25 12.5 25 50 100 MCh concentration [mg/ml] C macrophages neutrophils D 6 ns lymphocytes 80 eosinophils 20 ns ) ns 4 ns ns 60 PG 18:2_18:1 4 10 PE 18:0_20:4 LPG 20:4 PG 18:3_18:2 LPE 22:6 40 LPG 18:2 SM 34:1;0 PG 16:0_18:2 PI 18:0_20:4 0 SM 42:1;0 2 20 inflammatory cells

cells/ml BAL (x10 ns

ns PC2 (20.1%) ns -10 n=8 n=8 n=8 n=8 n=8 n=8 0 n=8 n=8 n=8 n=8 0 Lamp3+/+ PBS S S A A S S A A S S A A S S A A B B V B B V B B V B B V -/- /- - /- - - - Lamp3 PBS + P /- P + OV/- O /+ P - P + O -/ OV /+ P - P + OV-/ O + P -/ P + OV-/ O -20 +/ - +/ - + +/ + +/ +/ +/ 3 3 3 3 3 3 3 3 +/+ 3 p 3 3 p 3 3 p 3 3 3 p Lamp3 OVA p p m p p p m m am a m amp m amp a am a ampL amp amp ampL am amp a L a am -/- L L L L L L L L L L L L L Lamp3 OVA −30 -20 -10 0 10 20 +/+ -/- EFGE Lamp3 Lamp3 PC1 (28.8%)

Cer 16:1/16:0;0 Cer Cer 18:0/14:1;0 DAG 14:0_16:0 DAG 14:0_18:2 Corr.=0.316 Corr.=0.295 DAG 16:0_16:0 DAG DAG 16:0_16:1 DAG 16:0_18:1 DAG 16:0_18:2 DAG 16:0_20:4 LPC LPC 18:2 LPG 16:0 LPG 18:1 LPG LPG 18:2 LPG 20:4 LPI 16:0 LPI LPI 18:0 LPI 20:4 PC 14:0_16:1 PC PC 14:0_18:2 PC 16:0_18:2 PC 16:1_20:4 PC O− 14:0/16:0 PC O- PC O− 16:1/16:0 Corr.=0.567 PE 16:0_18:0 PE 16:1_20:4 PE PE 16:1_22:5 PE 18:2_18:1 PE 18:3_18:0 PE O- PE O− 16:0/20:4 PG 14:0_16:0 PG 14:0_18:2 PG 15:0_18:2 PG 16:0_16:0 PG 16:0_18:1 PG 16:0_18:2 PG 16:0_20:3 PG 16:0_22:4 PG PG 16:1_16:0 PG 16:1_20:4 PG 17:0_18:1 PG 17:0_18:2 PG 18:0_20:4 PG 18:1_18:1 PG 18:2_18:0 PG 18:2_18:1 PG 18:2_20:3 PG 18:2_22:4 PG 18:3_18:2 PI 16:0_18:1 Corr.=0.525Coo PI 16:0_20:4 PI PI 18:0_20:4 PI 18:1_20:4 PS 18:0_20:4 PS PS 18:0_22:4 PS 18:1_18:1 SM 34:1;0 SM 42:1;0 SM SM 42:2;0 SM 42:3;0 -3 -2 -1 0 1 2 3 Autoscale Lamp3+/+ OVA Lamp3-/- OVA