Lysyl Oxidase Is Required for Vascular and Diaphragmatic ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2003 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 278, No. 16, Issue of April 18, pp. 14387–14393, 2003 Printed in U.S.A.

Lysyl Oxidase Is Required for Vascular and Diaphragmatic Development in Mice* Received for publication, October 3, 2002, and in revised form, December 6, 2002 Published, JBC Papers in Press, December 7, 2002, DOI 10.1074/jbc.M210144200

Ian K. Hornstra‡§¶储, Shonyale Birge‡§¶, Barry Starcher**, Allen J. Bailey‡‡, Robert P. Mecham‡§§¶¶, and Steven D. Shapiro§§储储 From the ‡Department of Medicine, §Division of Dermatology, ¶Barnes-Jewish Hospital, and Departments of §§Pediatrics and ¶¶Cell Biology and Physiology, Washington University School of Medicine, St. Louis, Missouri 63110, the **University of Texas Health Science Center, Tyler, Texas 75708, the ‡‡University of Bristol, Collagen Research Group, Langford B540 5DS, United Kingdom, and the 储储Pulmonary and Critical Care Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts 02115

Lysyl oxidase (LOX) is an enzyme responsible for the cross-linking of collagen and elastin both in vitro and in vivo. The unique functions of the individual members of this multigene family have been difficult to ascertain because of highly conserved catalytic domains and overlapping tissue expression patterns. To address this problem of functional and structural redundancy and to determine the role of LOX in the development of tissue integrity, Lox gene expression was deleted by targeted mutagenesis in mice. Lox-targeted mice (LOXⴚ/ⴚ) died soon after parturition, exhibiting cardiovascular instability with ruptured arterial aneurysms and diaphragmatic rupture. Microscopic analysis of the aorta demonstrated fragmented elastic fiber architecture in homozygous mutant null mice. LOX activity, as assessed by desmosine (elastin cross-link) analysis, was reduced by ⬃60% in the aorta and lungs of homozygous mutant animals compared with wild type mice. Immature collagen cross-links were decreased but to a lesser degree than elastin cross-links in LOXⴚ/ⴚ mice. Thus, lysyl oxidase appears critical during embryogenesis for structural stability of the aorta and diaphragm and connective tissue development.

Lysyl oxidases are a group of enzymes whose members continue to be discovered and their functions defined. Classically, lysyl oxidase (LOX,1 EC 1.4.3.13) is a copper-containing monoamine oxidase extracellular enzyme that catalyzes the conversion of the epsilon amino group of specific lysine residues in collagen and elastin to reactive aldehyde groups (1). This oxidative deamination of lysine produces reactive “allysine” residues (lysine aldehyde residues), which participate in the formation of covalent cross-links. In vitro, LOX can convert various non-peptidyl amine substrates to their corresponding aldehydes. Two lines of evidence support the role of lysyl oxi-

* This work was supported by NIAMS, National Institutes of Health, Grant K08 AR02059; a Barnes-Jewish Hospital Foundation grant (to I. K. H.); NHLRI Grant HL53325 (to R. P. M.); and by NHLBI Grant P01 HL 29594, National Institutes of Health (to S. D. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 储 To whom correspondence should be addressed: Dept. of Medicine, Division of Dermatology, Washington University, 660 South Euclid Ave., Campus Box 8123, St. Louis, MO 63110. Tel.: 314-454-8928; Fax: 314-454-5626; E-mail: [email protected]. 1 The abbreviations used are: LOX, lysyl oxidase; BAPN, ␤-aminopropionitrile; MOPS, 4-morpholinepropanesulfonic acid; DHLNL, dihydroxylysinonorleucine; HLNL, hydroxylysinonorleucine. This paper is available on line at http://www.jbc.org

dase in cross-linking of collagen and elastin. First, lysyl oxidase(s), presumably LOX, has been extracted from bovine aortas using 6 M urea and used as an reagent to demonstrate elastin and collagen cross-linking by tritium release using immature elastin and collagen substrates (2, 3). Second, administration of ␤-aminopropionitrile (BAPN), a naturally occurring inhibitor of lysyl oxidases, results in lathyrism, presumably due to impaired collagen and elastin cross-linking (4 –7). Inhibition by BAPN is dependent upon the carbonyl cofactor in the lysyl oxidases, which in LOX is lysyl-tyrosine quinone (8 –10). Four additional lysyl oxidases have been recently described. All five family members are located on different human chromosomes (11–21) and mouse chromosomes (22–28) (Table I). Messenger RNA expression of all human lysyl oxidase family members appears to be most abundant in adult tissues by Northern blotting (Table I). Under normal conditions, human LOX mRNA expression is low in most tissues with 442 molecules/picogram embryonic skin fibroblast total RNA or about 8% of the mRNA molecules that encode for ␤-actin (29). LOXL1 mRNA expression appears greater than LOX mRNA expression in many adult mouse tissues, while expression of LOXL2, LOXL3, and LOXL4 mRNA appears to be low overall. The cDNA of lysyl oxidase encode proteins with ⬃60 – 85% similarity within the catalytic region to LOX (Table I). Human LOX has the shortest open reading frame of 417 amino acids, while human LOXL2 has the longest, encoding 774 amino acids. All five members of the lysyl oxidase family contain the catalytic lysyl oxidase domain (“LO domain”) comprising ⬃205 amino acids in the carboxyl-terminal half of the protein. The LO domain consists of 4 conserved histidine residues that coordinate copper binding and contains conserved lysine and tyrosine residues that form the cofactor lysyl-tyrosine quinone. The three newest members, LOXL2, LOXL3, and LOXL4, also have four scavenger receptor cysteine-rich domains that mediate ligand binding in many other proteins. Table I summarizes LOX steady state mRNA tissue expression. Thus, while all LOX family members appear to have the structural requirements to be active enzymes, recombinant proteins have not been generated to assess activity. Lysyl oxidase (presumably LOX) has been purified from bovine aorta. Of note, protein expression was also low with mg amounts purified from kilograms tissue (30). Bovine aortic lysyl oxidase has been shown to cross-link both collagen and elastin substrates and other amine substrates in vitro. With a multiplicity of lysyl oxidase-like enzymes, the contribution of any individual gene product to substrate cross-linking and tissue integrity is unknown. To further complicate the situation, mRNA expression patterns of different lysyl oxidases

14387

14388

Lysyl Oxidase Is Required for Development in Mice

TABLE I Lysyl Oxidase family member comparison Chr. indicates chromosome. mRNA and predicted protein sizes are derived from the human genes. AA ⫽ amino acids. SRCR ⫽ scavenger receptor cysteine-rich. Table data are complied from published literature (11–28) and the mouse and human NCBI databases. Family member

Mouse Chr.

mRNA and protein size

5

18

LOXL1

15

9

LOXL2

8

14

LOXL3

2

6

LOXL4

10

19

6.8, 4.8 kb 417 AA 2.4 kb 574 AA 4.0 kb 774 AA 3.3 kb 753 AA 3.5 kb 756 AA

LOX

Human Chr.

Highest mRNA levels, adult tissue distribution

% similarity to LOX LO domain

Protein domains

% similarity to LOXL2 LO domain

Lung, skeletal muscle, kidney, heart Lung, heart, spleen, skeletal muscle, pancreas Lung, thymus, skin, testis, ovary

LO

100

63

LO

85

63

4 SRCR, LO

58

100

Heart, uterus, testis, ovary

4 SRCR, LO

65

78

Skeletal muscle, testis, pancreas

4 SRCR, LO

62

79

overlap in many tissues. To better understand the contribution of LOX to extracellular matrix formation and tissue development, LOX-deficient mice were generated by targeted mutagenesis. In this report, we demonstrate that mice homozygous for the targeted Lox allele (LOX⫺/⫺) die at parturition (or within the first hours of life). LOX⫺/⫺ mice are of equal weight and size to their littermate controls, but die following birth trauma, cardiovascular events, and/or diaphragmatic rupture. These results suggest that LOX is required for extracellular matrix development in specific tissues. EXPERIMENTAL PROCEDURES

Lox Gene Targeting—BAC clones were obtained by hybridizing a 129/SvJ filter library (Genome Systems, St. Louis, MO) using a cDNA probe from the 3⬘-untranslated region of the Lox gene. Positive clones were further screened by an exon 1 oligo probe, and a 5-kb HindIII fragment containing exons 1– 4 was cloned into litmus 29 (New England BioLabs, Beverly, MA). This 5-kb HindIII fragment was reduced in size by digestion with BssHII. This size reduction permitted mutagenesis at the ATG codon where the ATG codon and surrounding bases were converted to a NotI restriction sites by site-directed mutagenesis (QuikChange mutagenesis, Stratagene, La Jolla, CA). The BssHII fragment was then replaced, and again, reduced in size by digestion with NdeI and XhoI (polylinker). The 480-bp outside probe generated by PCR is located between the NdeI site and the 3⬘ HindIII site. The 5⬘ HindIII site was now unique and an additional 2.3-kb 5⬘ HindIII fragment (isolated from the same Lox containing BAC clone) was added to the construct. The PGK-Neo (phosphoglycerate kinase-neomycin phosphotransferase) cassette was cloned into an EcoRV site of a modified pBS246 vector (Invitrogen). In the modified pBS246 vector, the 5⬘ LOXP region was modified to a LOXP⬘ site, and the 3⬘ LOXP site was left unmodified (the LOXP⬘ and LOXP sites were incorporated for future gene replacement experiments). A NotI fragment of the modified pBS246 vector including the PGK-Neo cassette was cloned into the NotI site of the mutagenized ATG initiation codon of the Lox gene construct. This construct deletes the ATG initiation codon and replaces it with the PGK-Neo cassette in the opposite transcriptional orientation. Correctly targeted RW4 embryonic stem cells (129/SvJ), carrying this mutation (frequency of five percent), were injected into C57BL/6 blastocysts and chimeric mice were produced. This mutation was transmitted into the germline and mice homozygous for this mutation (LOX⫺/⫺) were generated in a mixed 129/SvJ ⫻ C57BL/6 background. Southern Blot Analysis—DNA from mice tails was prepared by standard techniques (31, 32). The DNA was digested with HindIII, size-fractionated on an agarose gel, and alkali-transferred to a positively charged nylon membrane. The Lox outside probe was labeled by random primer extension using [␣-32P]dCTP, denatured, hybridized to the DNA-containing membrane, washed, and autoradiograms were perform as described by Hornstra and Yang (31). Northern Blot Analysis—Poly(A)⫹ RNA from whole mice embryos was isolated using the Poly(A)⫹ pure RNA isolation kit (Ambion, Austin, TX). Thirty micrograms of mRNA was size-fractionated on formaldehyde-MOPS-agarose gel, SSC-transferred to a nylon membrane, and UV cross-linked. The blot was sequentially hybridized to cDNA probes of the Lox gene, Loxl gene, and Gapdh gene as described above in Southern blotting.

Microscopy: Dissecting, Light, and Transmission Electron Microscopy—Gross dissection, microscopy, and photography of mutant animals were performed at ⫻5–10 magnifications. Perinatal mice were sacrificed; the thorax and abdomen were gently dissected and were perfused through the left ventricle with phosphate-buffered saline, followed by Histochoice fixative. The mice were fixed in Histochoice for 2–3 days and processed into slides. Hematoxylin and eosin and Hart’s elastic stains were performed on both transverse and sagittal sections. For electron microscopy, perinatal animals were prepared as above except that 2.5% glutaraldehyde in phosphate-buffered saline was used for fixation. The thoracic aorta was removed and incubated overnight in glutaraldehyde at 4 °C. Fixed aortas were processed at the Washington University electron microscopy core facility. Aortas were stained with tannic acid in addition to standard processing, thin-sectioning, and staining. Tissue Desmosine Assay—Tissue samples from perinatal mice were collected from the thoracic aorta and total lung. Tissue samples containing from 0.05 to 1 mg of protein were placed in secure-lock centrifuge tubes containing 500 ␮l of 6 N HCl at 100 °C for 24 h. The hydrolysates were evaporated to dryness in a Savant vacuum concentrator and re-dissolved in 200 ␮l of water. Desmosine was quantified by radioimmunoassay in 10 – 40 ␮l of sample as described previously (33). Hydroxyproline was determined in 20 – 80 ␮l of sample by amino acid analysis on a Beckman 6300 analyzer. Protein was determined in 1–10 ␮l of the hydrolysate using a ninhydrin-based assay (34). Analysis of Tissue Collagen Cross-links—The small tissue samples obtained at birth were suspended in phosphate-buffered saline and reduced with sodium borohydride as described previously (35), prior to hydrolysis in 6 N HCl at 110 °C for 16 h. The hydrolysate was eluted from a CF1 cellulose column (Whatman) to separate the cross-linking amino acids from the standard amino acids. The cross-linking amino acids were then quantified using an amino acid analyzer (Alpha Plus, Amersham Biosciences, Loughborough, UK) using a modified gradient configured for maximum separation of the cross-links (36). The location of the cross-links had previously been confirmed with authentic compounds. Quantification was based on the ninhydrin color and known leucine equivalents. RESULTS

Lox Gene Targeting Results in Perinatal Lethality—A genetargeting vector was designed to eliminate Lox expression by converting the translation initiation codon (ATG) into a translationally inactive NotI restriction site and inserting the phosphoglycerate kinase promoter driving the neomycin antibiotic resistance gene (PGK-Neo) (Fig. 1A). This construct was electroporated into embryonic stem cells (129/SvJ), and clones undergoing homologous recombination were injected into blastocysts. Resulting chimeric mice were bred to C57BL/6J mice yielding germ line transmission of the targeted allele. Breeding of heterozygous null LOX mice resulted in 0 of 400 animals that were homozygous for the targeted allele at 3 weeks of age. Timed matings were set up to determine the fate of LOX⫺/⫺ mice. LOX⫺/⫺ mice were observed by genotype late in gestation and at parturition with proper Mendelian inheritance (Fig. 1B). Grossly, LOX⫺/⫺ mice were of similar size to wild type and heterozygous littermates at late gestation and parturition, ex-


FIG. 1. Lox gene targeting. A, map of Lox genomic locus and genetargeting vector. The upper line diagram shows the 5⬘ end of the mouse Lox locus. Exons are indicated as cross-hatched boxes and are numbered below. The middle line diagram indicates the targeting vector where the ATG initiation codon was mutagenized to a NotI site, and the PGK-NEO cassette was inserted. The lower line diagram demonstrates the targeted genomic allele with homologous recombination of the targeting vector into the genome and the location of the outside probe used for screening. B, genomic Southern blot analysis of mouse tail DNA from a timed mating of LOX⫹/⫺ parents. Newborn tail DNA was digested with HindIII, size-fractionated, transferred to a nylon membrane, and hybridized with the outside probe. ⫹/⫹, ⫹/⫺, ⫺/⫺ represent wild type, heterozygous, and homozygous, respectively. WT indicates wild type band, and RC represents the recombinant gene-targeted band. Molecular weight markers are as indicated on the right. C, Northern blot analysis of poly(A) RNA from newborns with three different genotypes. Poly(A) RNA was size-fractionated using a formaldehyde-containing agarose gel, transferred to a nylon membrane, and sequentially hybridized with murine cDNA probes for the Lox cDNA, LoxL1 (lysyl oxidase-like 1) cDNA, and Gapdh (glyceraldehyde 3-phosphate dehydrogenase) cDNA. Lox ⫽ blot probed with Lox cDNA probe. Loxl1 ⫽ blot probed with lysyl oxidase-like cDNA probe. Gapdh ⫽ blot probed with the glyceraldehyde-3-phosphate dehydrogenase cDNA probe. ⫹/⫹ ⫽ LOX⫹/⫹, ⫹/⫺ ⫽ LOX⫹/⫺, and ⫺/⫺ ⫽ LOX⫺/⫺. Molecular weight markers are indicated on the left side of the panel.

cept the LOX⫺/⫺ mice died shortly after delivery. Examination of total embryo Poly(A)⫹ RNA by Northern analysis demonstrated essentially absent levels of the LOX mRNA (Fig. 1C) in LOX⫺/⫺ mice. LOXL1 mRNA expression was detectable in total embryos, without compensatory changes in the absence of LOX. Gross Aortic and Diaphragmatic Abnormalities in Lox Null Mutant Mice—Wild type and LOX⫺/⫺ mice were of similar size, limb length, musculature, and overall gross morphology. Major differences were found in the thorax and abdomen in lox⫺/⫺ versus wild type mice (Fig. 2, A and B). The diaphragm was often ruptured in LOX⫺/⫺ mice, allowing herniation of abdominal contents into the thorax. In Fig. 2B, the liver and stomach, no longer constrained by the diaphragm, displaced the heart from the mid-chest and compressed the lungs. In contrast, the diaphragm in wild type mice distinctly separated the thorax and abdominal contents. Following removal of the sternum and anterior distal ribs (Fig. 2, C and D), the wild type mice had an intact diaphragm, while in LOX⫺/⫺ mice we observed the muscular rim of the diaphragm connected to the body wall, but only fragments of the central tendon were visible. This suggests a critical weakness in the tensile strength of the collagen-rich central tendon of the diaphragm. Diaphragmatic rupture was found to be either unilateral (left or right) or bilateral in different mice. Diaphragmatic rupture occasionally occurred in

14389

FIG. 2. Perinatal gross morphology demonstrates tortuous aorta and diaphragmatic rupture in LOXⴚ/ⴚ mice. A and B, anterior surface of LOX⫹/⫹ (A) and LOX⫺/⫺ (B) mice. Necropsy was performed by dissection of the skin, ribs, and peritoneum and photomicrographs (⫻10) of the anterior surface are displayed. Heart ⫽ H, the lungs ⫽ Lu, the liver ⫽ L, diaphragm ⫽ Dia (arrows pointing to diaphragm), and S ⫽ stomach. Note the diaphragm rupture with abdominal contents filling the thorax in LOX⫺/⫺. C and D, posterior thorax of the LOX⫹/⫹ (C) and LOX⫺/⫺ (D) mice. Following necropsy, photomicrographs (⫻10) of the anterior surface of the posterior chest wall are displayed. The thoracic aorta is indicated with arrows marking the right and left sides. Note the tortuousity of the LOX⫺/⫺ aorta with maximal displacement to the right and left indicated by arrows.

LOX⫺/⫺ mice in the last days of gestation but most frequently occurred at birth when the animals started breathing. Most lox⫺/⫺ mice had hemothorax and/or hemoperitoneum. The descending aorta of LOX⫺/⫺ mice was always markedly tortuous (Fig. 2D). This contrasted with the uniformly straight path of the aorta in heterozygous and wild type mice (Fig. 2C). The aortic curvature in the LOX⫺/⫺ mice represents a lengthening of the aorta in vivo in the absence of LOX. While LOX is critical for aorta and diaphragmatic development, other organs, including the limbs and skeletal system of LOX⫺/⫺ mice, appeared grossly normal. Microscopic Arterial Abnormalities in LOX⫺/⫺ Mice—Microscopic analyses of transverse sections from neonatal mice stained with trichrome demonstrated similar morphology between the wild type and LOX⫺/⫺ mice at low power (⫻40) in the skin, musculature, vertebrae, and lungs (Fig. 3, upper panels). Higher power view (Fig. 3, lower panels, ⫻200) of the lungs following Hart’s elastic stain revealed less prominent staining of elastic fibers around blood vessels (labeled V) and conducting

14390


FIG. 3. Light microscopy of thorax and lungs in wild type (ⴙ/ⴙ) and homozygous LOX mutant (ⴚ/ⴚ) mice. The upper panels represent neonate thorax stained with trichrome at ⫻40. The lower panels shows neonatal lung sections (⫻200) stained with Hart’s elastin stain. Conducting airways are indicated by the letter C and arrows. Pulmonary vessels are marked by the letter V and arrows.

airways (labeled C) in the LOX⫺/⫺ mice as compared with wild type mice. Light microscopy (⫻400) of the descending thoracic aorta demonstrated significant abnormalities at term in the absence of LOX (Fig. 4). Hematoxylin and eosin staining shows areas of disrupted smooth muscle cell contact in the LOX⫺/⫺ mice (clear spaces, arrows), not observed in the wild type (arrow). trichrome staining reveals discontinuity of collagen (arrow) in the aortic lamellae as well as impaired continuity of smooth muscle cells in the LOX null aorta, while the wild type aorta had concentric blue lamellae and cohesive cell contact (arrow). Hart’s elastic stain demonstrates a well developed internal elastic lamina and concentric lamellae in wild type (arrow). LOX⫺/⫺ mice, however, had a fragmented and discontinuous internal elastic lamina with similarly fragmented and irregular lamellae (arrow). Aorta wall thickness was more variable and often significantly thinner in LOX⫺/⫺ than in wild type mice. Histologically, elastin fiber abnormalities in LOX⫺/⫺ mice appear as early as E14.5 and progress in parallel with the development of elastic fibers (data not shown). Aortic rupture is a late event in development occurring most frequently in the perinatal period. Defects in the aortas of full-term LOX⫺/⫺ mice were even more pronounced when examined by transmission electron microscopy stained with tannic acid to accentuate the extracellular matrix (Fig. 5). In wild type mice, the internal elastic lamina just below the luminal endothelial cell layer is well developed with a nearly continuous thick band of elastin (upper arrows). Proceeding away from the lumen, nearly linear and continuous concentric lamellae are visualized between cell layers (lower arrows). In contrast, the internal elastic lamina is poorly developed, fragmented, and discontinuous in LOX⫺/⫺ mice (upper arrows). The concentric lamellae are similarly fragmented and discontinuous with increased space between smooth muscle cells in the LOX⫺/⫺ mice (lower arrows). Only irregular collections of matrix were seen between the LOX⫺/⫺ cells, and some of the cells appeared less polarized when compared with wild type. Collagen fibers were of similar diameters

in all genotypes (data not shown). These histological abnormalities in the LOX⫺/⫺ mice likely predispose to the aortic rupture with hemothorax and hemoperitoneum that was observed. Biochemical Analysis of Elastin and Collagen Cross-links and Hydroxyproline—LOX activity was assessed by quantifying elastin cross-linking using a desmosine radioimmunoassay (33) in tissue samples from litters of full-term mice. Compared with wild type mice, LOX⫹/⫺ mice had 86 and 103% of desmosine in aorta and lung tissues respectively. LOX⫺/⫺ mice had only 39% of the desmosine cross-links in aortic hydrolysates and 45% of the desmosine in lung hydrolysates when compared with wild type (Table II). Thus, desmosine analysis agrees with the ultrastructural studies in documenting decreased elastin content. Hydroxyproline was measured by amino acid analysis from litters of full-term mice to quantify collagen content in aortic and lung hydrolysates. LOX⫺/⫺ mice aortas had 74%, and lungs had 68% hydroxyproline content as compared with wild type mice. Heterozygous LOX (LOX⫹/⫺) aortas and lungs had 97 and 93% of hydroxyproline content as wild type mice (Table III). Immature collagen cross-links were analyzed in both the whole body and lungs of mice. As shown in Table IV, there are statistically significant decreases in DHLNL (dihydroxylysinonorleucine), and HLNL (hydroxylysinonorleucine) of 43 and 39%, respectively, in LOX⫺/⫺, as compared with wild type total body collagen cross-links. Collagen cross-link data are normalized to moles of collagen as determined by hydroxyproline content. In LOX⫹/⫺ mice, DHLNL is 100% of wild type, while HLNL is only 64% of wild type content. This could represent a greater role for LOX in HLNL cross-links, with a decrement observed with quantitative reduction in LOX (50%) in heterozygous mice. Collagen cross-link data from the lungs was also determined. The analysis of E20.5 lungs was at the limit of assay detection and showed only a trend toward decreased HLNL in heterozygote and homozygous null mice (14 and 32%, respectively). Lung DHLNL was not different between the genotypes. These data demonstrate a quantitative reduction of elastin-derived cross-links, collagen content, and immature col-


14391

FIG. 4. Light microscopy of thoracic aortas in wild type (ⴙ/ⴙ) and homozygous LOX mutant (ⴚ/ⴚ) mice. Serial sections of thoracic aortas (⫻400) are displayed. The upper two panels represent hematoxylin and eosin (H&E)stained sections. Note the disrupted smooth muscle cells in LOX⫺/⫺, but not wild-type, mice (arrows). The middle panels stained with trichrome demonstrate discontinuity of collagen in LOX⫺/⫺, but not wild-type, mice (arrows). Lower panel, photomicrographs of tissue sections stained with Hart’s elastin stain. The arrows highlight a well developed internal elastic lamina and concentric lamellae in wild type mice, whereas it is fragmented and discontinuous in LOX⫺/⫺ mice (arrows).

lagen cross-links in LOX⫺/⫺ mice, all of which support and help explain the phenotypic abnormalities observed. Yet, incomplete loss of collagen and elastin cross-linking likely explains the fact that not all organs were grossly abnormal. DISCUSSION

LOX gene targeting results in decreases in both collagen and elastin cross-links in vivo, confirming the catalytic capacity of LOX demonstrated in vitro. Despite the presence of five lysyl oxidase enzymes, LOX deficiency is lethal, suggesting that other lysyl oxidases are unable to effectively compensate during development for the physiological stresses associated with parturition and life ex utero. LOXL1 mRNA does not increase to compensate for LOX deficiency in the total embryo. The inability of the other lysyl oxidases to compensate may relate to either distinct substrate specificities of individual enzymes or temporally and spatially unique expression patterns. The consequences of decreased lysyl oxidase activity appear most significant in the development of the aorta and diaphragm, two organs where tensile strength requirements increase significantly at birth (37). LOX⫺/⫺ mice survive in utero, probably due to feto-maternal circulation, low systemic blood pressure, and low intrathoracic pressure. At parturition, the fetus is subjected to many physiologic stresses, including physical trauma from passage through the birth canal, increased

systemic blood pressure (37), and conversion of liquid filled to air filled lungs with the onset of breathing requiring diaphragmatic contraction. It is precisely at this time shortly after birth when the majority of LOX⫺/⫺ mice die, likely due to the inability to handle increased stress placed on the vascular bed and diaphragm. Although elastin cross-links were significantly decreased in the aorta and lungs of LOX⫺/⫺ mice, they were not completely eliminated. This suggests that other enzyme(s) participate in cross-linking elastin to produce the ⬃40% of desmosine that remain in these organs. Whether LOX catalyzes a specific subset of lysine residues within tropoelastin or whether it contributes 60% of total activity on random crosslinks is not known. LOX⫺/⫺ mice also have 40% reduced immature collagen cross-links (DHLNL and HLNL) in the whole mouse. Although the exact relationship between collagen cross-links and tensile strength is unknown, 40% overall reduction in cross-linking could certainly impair mechanical strength, particularly if it were greater in individual tissues. The combined reduction in elastin and collagen cross-linking resulted in loss of structural integrity of the aorta. Impaired collagen cross-linking most likely explains the diaphragmatic abnormalities, since this is the major matrix component of this tissue. Due to the small size of these organs at birth,

14392

Lysyl Oxidase Is Required for Development in Mice TABLE IV Immature collagen-derived cross-links DHLNL represents dihydroxylysinonorleucine in mol/mol collagen and HLNL represents hydroxylysinonorleucine in mol/mol collagen. Averages ⫾ S.D. are shown. Parentheses indicate percentage of wild type. DHLNL and HLNLa Whole body

Lung

LOX⫹/⫹, n⫽3

2.01 ⫾ 0.19 (100) 0.58 ⫾ 0.10 (100)

1.62 ⫾ 0.26 (100) 2.31 ⫾ 1.14 (100)

LOX⫹/⫺, n⫽3

2.02 ⫾ 0.81 (100) 0.37 ⫾ 0.06 (64)b

1.82 ⫾ 0.75 (112) 1.98 ⫾ 1 (86)

LOX⫺/⫺, n⫽3

1.15 ⫾ 0.37 (57)b 0.35 ⫾ 0.06 (61)b

1.64 ⫾ 0.36 (102) 1.58 ⫾ 0.18 (68)

a First entry of each line refers to DHLNL, and the second entry refers to HLNL. b p ⬍ 0.05 as compared with wild type using a t test.

FIG. 5. Transmission electron microscopy of wild type (ⴙ/ⴙ) and homozygous LOX mutant (ⴚ/ⴚ) thoracic aortas. Disordered and fragmented elastic fibers in LOX⫺/⫺ thoracic aortas are seen by transmission electron microscopy (⫻3000 and ⫻7000 magnification as shown on the left). The arrows point to elastic lamellae. L represents the lumen of the aorta. TABLE II Desmosine, elastin cross-linking in aorta and lung Average desmosine in pmol/mg of protein is shown ⫾ S.D. Parentheses indicate percentage of wild type.

⫹/⫹

LOX n⫽5 LOX⫹/⫺ n⫽7 LOX⫺/⫺ n⫽6 a

Aorta

Lung

846 ⫾ 164 (100)

103 ⫾ 26 (100)

729 ⫾ 129 (86)

106 ⫾ 14 (103)

327 ⫾ 77 (39)a

46 ⫾ 7 (45)a

p ⬍ 0.05 as compared with wild type using t test.

TABLE III Hydroxyproline, collagen content in aorta and lung Average hydroxyproline in nmol/mg of protein is shown ⫾ S.D. Parentheses indicate percentage of wild type.

⫹/⫹

LOX n⫽5 LOX⫹/⫺ n⫽6 LOX⫺/⫺ n⫽5 a

Aorta

Lung

150 ⫾ 29 (100)

30 ⫾ 8 (100)

146 ⫾ 39 (97)

28 ⫾ 4 (93)

111 ⫾ 20 (74)

21 ⫾ 3 (68)a

p ⬍ 0.05 as compared with wild type using a t test.

detection of collagen cross-links in these individual tissues was below the sensitivity of the assay. Morphologically, the elastin defects were more apparent which is likely because 36 of 40 lysine residues in tropoelastin are cross-linked in mature elastin, while there are only one to two collagen cross-links per triple helical collagen unit (38). In addition, the inability to cross-link tropoelastin renders this precursor molecule sensitive to trypsin-like proteases (38). LOX⫺/⫺ mice clearly show a decreased quantity of elastin by electron microscopy, and the decreased desmosine cross-links appear to correlate with decreased elastin deposition and presumably increased tropoelastin turnover. In contrast, morpho-

logical changes in collagen are less apparent, since collagen cross-links are not required for collagen fibril assembly. LOX⫺/⫺ mice had a 30% reduction in hydroxyproline content, which may be influenced by decreased hydroxyproline from both collagen and elastin. Collagen cross-links are reduced by 40% in LOX⫺/⫺ mice, which certainly appears to have influenced the tensile strength of some connective tissues. Insight into the physiologic relevance of impaired elastin and collagen in LOX⫺/⫺ mice can be elucidated from other loss of function mice. Elastin null mutant mice die by P4.5 likely due to an obliterative aortic occlusive disease (39). Mice heterozygous or haploinsufficient for the elastin gene have 2.5 times more aortic lamellar units compared with control mice (40, 41). This increase in lamellar units is also observed in humans with supravalvular aortic stenosis characterized by haploinsuffiency of the elastin gene. Thus, elastin appears to regulate smooth muscle proliferation and the quantity of aortic lamellar units (40), but primary deficiency of elastin does not correlate with arterial rupture. Also, we do not observe smooth muscle proliferation and increased lamellar units in LOX⫺/⫺ mice. Ehlers-Danlos syndrome Type IV is characterized by a primary collagen defect (42). Type IV or the vascular type of Ehlers-Danlos syndrome in many patients is a consequence of mutations in the COL3A1 gene. The disease phenotype includes arterial rupture, thin skin, bruising, uterine rupture, and small joint hyperextensibility. Most Col3a1 gene-targeted mice die perinatally due to blood vessel rupture, and those that make into adulthood have one-fifth of the normal life span also dying from aortic rupture (43). Aortic aneurysm formation occurs between the media and the adventitia (43), where collagen is relatively more prominent than elastin at this interface. Therefore, the lethal phenotypic changes in LOX-deficient mice are probably more related to decreased collagen cross-links and the resultant change in tensile strength in the aorta and diaphragm than to changes in elastin content or maturation. In LOX⫺/⫺ mice, while aortic and diaphragmatic changes are the most obvious phenotypic abnormalities, other organs and tissues are certainly also affected. Upon dissection, LOX⫺/⫺ mouse skin tears easily, and the ribs are softer and cut with minimal force. LOX⫺/⫺ paraffin-infiltrated embryos deform more readily upon transverse or sagittal cutting. Furthermore, the elastin in the conducting airways and pulmonary vessels is altered in less developed in LOX null animals. Similarly, the heart is probably affected with elastin defects in the heart valves, but since the animals die at parturition, these physiological defects are difficult to examine in detail. Thus, LOX-induced cross-linking is not required for organogenesis but is required for the development of tensile

Lysyl Oxidase Is Required for Development in Mice strength in collagen and elastogenesis needed for the organism to survive ex utero. A human disease of primary LOX deficiency is not known. If the phenotype were identical to what we see in LOX-deficient mice, one would expect to find stillborns or newborns unable to meet cardiopulmonary physiological stresses ex utero. Since this has not been described, it is likely that either a LOX deficiency either manifests itself as a more severe deficiency impairing embryonic viability earlier in gestation, or compensation by other LOXLs may lead to a milder phenotype that is unappreciated but could predispose to aortic aneurysms/ dissection, diaphragmatic hernias, or other connective tissue phenotypes. Although primary LOX deficiency has not been described in humans, secondary LOX deficiency resulting from low enzyme activity is associated with at least three human conditions: lathyrism, copper deficiency, and Menke’s syndrome. Lathyrism is a condition of decreased lysyl oxidase activity secondary to ingestion of BAPN, a naturally occurring and specific inhibitor of lysyl oxidase. The severity of the phenotype in lathyrism is proportional to the quantity and timing of BAPN ingestion. In animals, BAPN exposure early in life leads to aortic aneurysms and death, while later in life it manifests in bone and tendon abnormalities (44). Primary human copper deficiency manifesting as anemia, neutropenia, and bone fractures is rarely seen in developed countries except in some malabsorption syndromes and very low birth weight infants (45). In developing countries, it is observed in malnourished children with low copper diets (45). Menke’s syndrome is characterized by a loss of function mutation in the X-linked ATP7A gene that encodes a copper transporter necessary for copper absorption to make active cuproenzymes, which includes lysyl oxidases (46). This results in decreased total lysyl oxidase activity, likely affecting all active LOX family members. Menke’s patients exhibit developmental delay, loose skin, fragile bones, and aortic abnormalities. Aortic defects manifest as tortuousity, fragmented discontinuous elastic lamellae, and consequent aneurysm formation (47, 48). Thus, humans with Menke’s syndrome have connective tissue defects that overlap with the phenotype of LOX⫺/⫺ mice. If gene disruption of the Lox gene results in lethality, what phenotypes could be expected from the other four-lysyl oxidaselike genes? The answer to this question requires knowing what lysine substrates the other four lysyl oxidase-like genes catalyze in vivo, the spatial and compartmental location of these enzymes, and the temporal patterns of gene expression. These enzymes have been notoriously difficult to purify. It has also been difficult to express large quantities of recombinant protein. Thus, gene targeting provides a method to define the roles of each lysyl oxidase in a biological context. Of the five lysyl oxidases, LOX and LOXL1 are most homologous, while LOXL2, LOXL3, and LOXL4 represent another homologous group. Whether either group has a greater affinity toward collagen versus elastin is unknown. These lysyl oxidases may have other substrates other than the extracellular matrix proteins, collagen and elastin. Future research in to the roles of lysyl oxidases in biology will be revealing. REFERENCES 1. Kagan, H. M., and Trackman, P. C. (1991) Am. J. Respir. Cell Mol. Biol. 5, 206 –210 2. Jordan, R. E., Milbury, P., Sullivan, K. A., Trackman, P. C., and Kagan, H. M.

14393

(1977) Adv. Exp. Med. Biol. 79, 531–542 3. Melet, J., Vianden, G. D., and Bachra, B. N. (1977) Anal. Biochem. 77, 141–146 4. Siegel, R. C. (1976) J. Biol. Chem. 251, 5786 –5792 5. Layman, D. L., Narayanan, A. S., and Martin, G. R. (1972) Arch Biochem. Biophys 149, 97–101 6. Narayanan, A. S., Siegel, R. C., and Martin, G. R. (1972) Biochem. Biophys. Res. Commun. 46, 745–751 7. Tang, S. S., Trackman, P. C., and Kagan, H. M. (1983) J. Biol. Chem. 258, 4331– 4338 8. Wang, S. X., Mure, M., Medzihradszky, K. F., Burlingame, A. L., Brown, D. E., Dooley, D. M., Smith, A. J., Kagan, H. M., and Klinman, J. P. (1996) Science 273, 1078 –1084 9. Wang, S. X., Nakamura, N., Mure, M., Klinman, J. P., and Sanders-Loehr, J. (1997) J. Biol. Chem. 272, 28841–28844 10. Akagawa, M., and Suyama, K. (2001) Biochem. Biophys. Res. Commun. 281, 193–199 11. Hamalainen, E. R., Jones, T. A., Sheer, D., Taskinen, K., Pihlajaniemi, T., and Kivirikko, K. I. (1991) Genomics 11, 508 –516 12. Mariani, T. J., Trackman, P. C., Kagan, H. M., Eddy, R. L., Shows, T. B., Boyd, C. D., and Deak, S. B. (1992) Matrix 12, 242–248 13. Kenyon, K., Modi, W. S., Contente, S., and Friedman, R. M. (1993) J. Biol. Chem. 268, 18435–18437 14. Kim, Y., Boyd, C. D., and Csiszar, K. (1995) J. Biol. Chem. 270, 7176 –7182 15. Jourdan-Le Saux, C., Le Saux, O., Donlon, T., Boyd, C. D., and Csiszar, K. (1998) Genomics 51, 305–307 16. Jourdan-Le Saux, C., Tronecker, H., Bogic, L., Bryant-Greenwood, G. D., Boyd, C. D., and Csiszar, K. (1999) J. Biol. Chem. 274, 12939 –12944 17. Asuncion, L., Fogelgren, B., Fong, K. S., Fong, S. F., Kim, Y., and Csiszar, K. (2001) Matrix Biol 20, 487– 491 18. Ito, H., Akiyama, H., Iguchi, H., Iyama, K., Miyamoto, M., Ohsawa, K., and Nakamura, T. (2001) J. Biol. Chem. 276, 24023–24029 19. Jourdan-Le Saux, C., Tomsche, A., Ujfalusi, A., Jia, L., and Csiszar, K. (2001) Genomics 74, 211–218 20. Maki, J. M., and Kivirikko, K. I. (2001) Biochem. J. 355, 381–387 21. Maki, J. M., Tikkanen, H., and Kivirikko, K. I. (2001) Matrix Biol 20, 493– 496 22. Kenyon, K., Contente, S., Trackman, P. C., Tang, J., Kagan, H. M., and Friedman, R. M. (1991) Science 253, 802 23. Mock, B. A., Contente, S., Kenyon, K., Friedman, R. M., and Kozak, C. A. (1992) Genomics 14, 822– 823 24. Chang, Y. S., Svinarich, D. M., Yang, T. P., and Krawetz, S. A. (1993) Cytogenet. Cell Genet. 63, 47– 49 25. Lossie, A. C., Buckwalter, M. S., and Camper, S. A. (1993) Mamm. Genome 4, 177–178 26. Tchernev, V. T., Yang, T. P., and Kingsmore, S. F. (1997) Mamm. Genome 8, 621– 622 27. Wydner, K. S., Kim, Y., Csiszar, K., Boyd, C. D., and Passmore, H. C. (1997) Genomics 40, 342–345 28. Jourdan-Le Saux, C., Le Saux, O., Gleyzal, C., Sommer, P., and Csiszar, K. (2000) Matrix Biol. 19, 179 –183 29. Hamalainen, E. R., Kemppainen, R., Kuivaniemi, H., Tromp, G., Vaheri, A., Pihlajaniemi, T., and Kivirikko, K. I. (1995) J. Biol. Chem. 270, 21590 –21593 30. Palamakumbura, A. H., and Trackman, P. C. (2002) Anal. Biochem. 300, 245–251 31. Hornstra, I. K., and Yang, T. P. (1992) Mol. Cell. Biol. 12, 5345–5354 32. Sambrook, J., and Russell, D. (2001) Molecular Cloning: A Laboratory Manual, 3rd Ed., pp. 6.1– 6.62, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY 33. Starcher, B., Green, M., and Scott, M. (1995) Respiration 62, 252–257 34. Starcher, B. (2001) Anal. Biochem. 292, 125–129 35. Light, N. D., and Bailey, A. J. (1982) Methods Enzymol. 82, 360 –372 36. Sims, T. J., and Bailey, A. J. (1992) J. Chromatogr. 582, 49 –55 37. Nichols, D. G., Cameron, D. E., Greeley, W. J., Lappe, D. G., Ungerleider, R. M., and Wetzel, R. C. (eds) (1995) Critical Heart Disease in Infants and Children, pp. 18 –22, Mosby-Yearbook, Inc., St. Louis, MO 38. Brown-Augsburger, P., Tisdale, C., Broekelmann, T., Sloan, C., and Mecham, R. P. (1995) J. Biol. Chem. 270, 17778 –17783 39. Li, D. Y., Brooke, B., Davis, E. C., Mecham, R. P., Sorensen, L. K., Boak, B. B., Eichwald, E., and Keating, M. T. (1998) Nature 393, 276 –280 40. Dietz, H. C., and Mecham, R. P. (2000) Matrix Biol. 19, 481– 488 41. Li, D. Y., Faury, G., Taylor, D. G., Davis, E. C., Boyle, W. A., Mecham, R. P., Stenzel, P., Boak, B., and Keating, M. T. (1998) J. Clin. Invest. 102, 1783–1787 42. Mao, J. R., and Bristow, J. (2001) J. Clin. Invest. 107, 1063–1069 43. Liu, X., Wu, H., Byrne, M., Krane, S., and Jaenisch, R. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 1852–1856 44. Weaver, A. L. (1967) Arthritis Rheum. 10, 470 – 478 45. Uauy, R., Olivares, M., and Gonzalez, M. (1998) Am. J. Clin. Nutr. 67, 952S–959S 46. Mercer, J. F. (2001) Trends Mol. Med. 7, 64 – 69 47. Pasquali-Ronchetti, I., Baccarani-Contri, M., Young, R. D., Vogel, A., Steinmann, B., and Royce, P. M. (1994) Exp. Mol. Pathol. 61, 36 –57 48. Oakes, B. W., Danks, D. M., and Campbell, P. E. (1976) Exp. Mol. Pathol. 25, 82–98