Collagen Cross-linking in Adult Patients with Acute and Chronic Fibrotic Lung Disease Molecular Markers for Fibrotic Collagen 1 - 3

JEROLD A. LAST, TALMADGE E. KING, JR., ANDREAS G. NERLlCH, and KAREN M. REISER

Introduction

Despite tremendous heterogeneity in the pathogenesis and clinical course of pulmonary fibrosis, many forms of the disease seem to progress through similar histopathologic stages in which lung injury is followed by edema, cellular infiltration, and fibrosis. Although the lung damage is apparently reversible at early stages of this process, the long-term sequelae often include irreversible pulmonary fibrosis. Much of our work has focused on the abnormalities in collagen metabolism that precede and constitute this final stage. We investigated aspects of collagen metabolism that are regulated intracellularly by first examining the apparent rate of collagen synthesis in lung minces obtained from rats or monkeys exposed in vivo to fibrotic agents (1-5). In all these studies, we found that lungs from animals exposed to a fibrogenic agent synthesized collagen more rapidly than did lungs from matched control animals. Further, the increase in lung collagen synthesis rate wasaccompanied by an increase in the ratio of type I:type III collagen in newly synthesized collagen (2, 6). This change in the relative amounts of specific collagen types being synthesized by (pre)fibrotic lungs corresponds to similar changes in the collagen types observed in lungs of human patients dying of idiopathic pulmonary fibrosis (7), adult respiratory distress syndrome (8), or infant respiratory distress syndrome (9). In more recent studies, wehave focused on extracellular events, in particular enzymatically mediated collagen cross-linking. In animal models of fibrosis induced by various agents, including bleomycin, ozone, or silica, we consistently observed a dramatic, and selective, increase in the lung collagen content of dihydroxylysino-

SUMMARY Lung tissue from patients with interstitial lung disease (ILO), adult respiratory distress syndrome (AROS), and control subjects with no obvious fibrotic lung disease was analyzed for its content of the collagen cross-links hydroxylysinonorleucine (HLNL), dihydroxylysinonorleucine (OHLNL), and hydroxypyridinium (OHP). We observed significant elevations of the OHLNL:HLNL ratio in patients with AROS, and significant increases In the content of OHP In lungs of patients with ILO. These results are consistent with data from animal models of fibrotic lung disease, suggesting that increases in the OHLNL:HLNL ratio of lung collagen may serve as a marker of an acute fibrotic episode, whereas increased lung collagen OHP content serves as a marker of chronic lung fibrosis. Wesuggest that the underlying mechanism for the changes in OHLNLcontent in (pre)fibrotlc acutely injured lung tissue and in OHP content in long-term fibrosis may be an increase In the activity of Iysyl hydroxylase, a key intracellular enzyme responsible for a specific post-translational modification of collagen. AM REV RESPIR DIS 1990; 141:307-313

norleucine (DHLNL),4 a difunctional collagen cross-link formed by condensation of two hydroxylysine residues, as illustrated in figure 1 (10,11). There was no increase in the corresponding monohydroxylated cross-link hydroxylysinonorleucine (HLNL). This increasein DHLNL occurred as early as 1 wk after the pulmonary insult. Over a period of 6 to 10 wk after exposure, we also observed significant increases in hydroxypyridinium (OHP), a reaction product of DHLNL and hydroxylysine (10-12). Although DHLNL levels eventually return to normal in acute models of fibrosis, we found that OHP content remained elevated, thus serving as an apparently permanent marker of the fibrotic insult (11). Our observations in animal models led us to formulate the following hypotheses concerning human pulmonary fibrosis: (1) that the ratio of DHLNL to HLNL would be increased in acute fibrotic disorders; (2) that an increased ratio of DHLNL to HLNL in chronic fibrotic disease could 4 For the sake of clarity, these abbreviations will refer to both the dehydro and reduced forms of the difunctional cross-links, that is, dehydro-DHLNL and DHLNL and dehydro-HLNL and HLNL will be referred to as DHLNL and HLNL, respectively.

serve as an index of active fibrosis; (3) that OHP content of lung collagen, expressed on a per collagen molecule basis, would be increased in long-term (at least 4 to 6 months at time of assay) fibrotic disease, but would be normal or below normal, as expressed on a per collagen molecule basis, in acute fibrotic disease. In an earlier study using lung tissue from infants dying of respiratory distress syndrome of prematurity (IRDS), we pre-

(Received in original form March 14, 1989 and in revised form June 27, 1989) 1 From the Department of Internal Medicine, School of Medicine and California Primate Research Center, University of California, Davis, California; the National Jewish Center for Immunology and Respiratory Medicine, University of Colorado Health Sciences Center, Denver, Colorado; and the Pathology Institute, University of Munich, West Germany. 2 Supported by Grant No. HL-32690, AG-05324, RR-OOI69, and SCOR Grant No. HL-27353 from the National Institutes of Health. A portion of this work resulted from the U.S.-FRG collaboration in pulmonary research sponsored by the National Heart, Lung, and Blood Institute, NIH. J Correspondence and requests for reprints should be addressed to Jerold A. Last, California Primate Research Center, University of California, Davis, CA 95616.

307

308

LAST, KING, NERLICH, AND REISER

EXTRACELLULAR

INTRACELLULAR

LYS

---+

LYS

1 HYL

----+

sented evidence in support of the first of these hypotheses (13). The present study extends these methods to the study of adult lung tissue samples. Methods Study Population We examined lung tissue from 57 subjects from three institutions. In 33 subjects, the lung tissue was obtained at open lung biopsy, performed for diagnostic purposes, and in 25 subjects, the tissue was obtained at necropsy, as soon as possible after death, from patients who had died of progressive pulmonary dysfunction. In one patient, lung tissue was obtained both at the time of open lung biopsy and 2 months later at necropsy. Informed consent was obtained from each patient, and the protocol was approved by each institution's Institutional Human Review Committee. All samples were immediately frozen until assayed. Samples were coded, and the investigator performing the assay had no knowledge of the diagnosis at the time the assays were conducted. Control lung samples (n = 12) were obtained at necropsy from adults (n = 8) and from children (n = 4) without detectable pathologic pulmonary changes. These subjects' ages ranged from zero to 61 yr (mean, 35 yr). Lung samples were obtained at necropsy from six patients (mean age ± 1 SD = 51.3 ± 10.0 yr; range, 40 to 65 yr) who had died of adult respiratory distress syndrome (ARDS). The duration of illness ranged from 14 days to 2.5 months. Twoof these subjects have been described elsewhere (Patients 1 and 3 in reference 8). The remaining subjects, all from Munich, had the following pertinent clinical characteristics. Patient 3 was a 61-yr~0Id woman with septic shock, 14 days of respirator therapy, and autopsy-confirmed mild fibrosis. Patient 6 was a 43-yr-old man with polytrauma, 25 days of respirator therapy, and autopsy-confirmed severe fibrosis. Patient 12 was a 47-yr-old woman with poly trauma, 22 days of respirator therapy, and autopsy-confirmed fibrosis. Patient 23 was a 4O-yr-old man with septic shock, 26 days of respirator therapy, and autopsy-confirmed fibrosis. Thus, three subjects were male and three were female, with 20.0 ± 5.1 (mean ± 1 SD) days

Fig. 1. Biosynthesis of collagen cross-links. Key intracellular and extracellular steps involved in the formation of HLNL, DHLNL, and OHP are illustrated schematically. Selected lysine residues are hydroxylated intracellularly. After the tropocollagen molecules are secreted into the extracellular matrix, certain lysine and hydroxylysine residues are oxidatively deaminated, a step mediated by the enzyme Iysyl oxidase. The resulting aldehydes may then react with lysine or hydroxylysine residues to form the difunctional cross-links DHLNL and HLNL. DHLNL may then react with hydroxylysine to form the trifunctional cross-link OHP.Only those cross-links discussed in the study are illustrated.

of ventilator support with supplemental oxygen. Lung tissue samples were obtained at open lung biopsy (n = 33) or necropsy (n = 7) in patients with interstitial lung diseases (ILD). Twenty-six had idiopathic pulmonary fibrosis (lPF), four had chronic hypersensitivity pneumonitis (HSP), seven had respiratory bronchiolitis (RB), and one each had sarcoidosis and eosinophilic granuloma of the lung. AU diagnoses were established according to previously described clinical and histologic criteria (14-17). Histopathologic assessment for the presence of fibrosis was performed in all subjects. Additional lung samples obtained at necropsy from subjects in Munich included five IPF and two control samples. Diagnoses were confirmed by pathologic and clinical criteria. In 19 patients with IPF (age, 61 ± 4 yr; 13men and 6 women; 12 current or ex-smokers and 7 never-smokers) who underwent open lung biopsy, a detailed clinical and histopathologic evaluation was performed. The duration of illness, defined as first onset of cough or shortness of breath, was 21 ± 7.7 months before lung biopsy. None of these patients had received corticosteroid or cytotoxic therapy at the time of lung biopsy. The clinical evaluation of these patients included history, physical examination, chest radiograph, pulmonary function studies, arterial blood gas determinations at rest and during exercise, and bronchoalveolar lavage fluid analysis by previously described methods (14,18). The severity of clinical impairment was quantitated by a previously described clinical-radiographic-physiologic (CRP) scoring system (18), which consists of a numerical score for dyspnea, chest radiography, exercise gas exchange, resting AaPo z, thoracic gas volume (Vtg), FVC, and single-breath diffusion capacity for carbon monoxide Dr.co/VA).Scores could potentially range from zero to 100, with higher scores representing more severeimpairment. Age, duration of illness prior to lung biopsy, smoking status, pulmonary function tests, and CRP score were used to examine for correlations with lung collagen content of difunctional collagen cross-links and OHP. In addition, each open lung biopsy was examined by a pathologist for establishment of the diagnosis of IPF and was graded for the presence, extent, and severity of the follow-

ing histopathologic abnormalities frequently present in IPF: alveolar septal inflammation, intra-alveolar round cell accumulation, alveolar septal fibrosis, and honeycombing. Each of these abnormalities was scored by the pathologist as follows: 0 = absence of that abnormality; 1 = mild; 2 = moderate; 3 = severe involvement. The score for each histopathologic abnormality was used to examine for correlations with lung content of the specific collagen cross-links.

Tissue Preparation Lung samples removed at autopsy were dissected free of extraneous tissue, washed, and immediately lyophilized in plastic tubes prior to being mailed to Davis. Biopsy samples were frozen and shipped on dry ice. For analysis, samples were cut into small pieces, washed at 4 0 C for 24 h in phosphate-buffered saline (pH, 7.4), then immersed in incubation buffer for 4 h at room temperature (0.1 M sodium phosphate at pH 7.4) at 1 ml of buffer/1O mg of tissue. Tissues were reduced in a chemical fume hood with sodium borotritide (5 Ci/rnmol; Amersham Corp., Arlington Heights, IL), 10mg/ml in dimethylforrnamide, for 1 h. The amount of sodium borohydride used was one thirtieth of the dry weight of the tissue. Reduction was stopped by the addition of glacial acetic acid (15 M) to a pH of 3. After standing for 15 min, samples were washed in distilled water five times and hydrolyzed in 6N HCI for 18h at 1100 C. The acid was removed by evaporation under nitrogen, and the dried sample was resuspended in 0.2 ml of distilled water. Hydrolysates were filtered through a Rainin filter apparatus (Rainin, Emeryville, CA) lyophilized, and assayed for hydroxyproline by a colorimetric assay (19). The effective reducing capacity of each batch of NaB 3H4 was determined by methods that have been detailed elsewhere (20). Briefly, a solution containing 100 nmol of deltaaminolevulinic acid was reduced with 5 mg of NaB3H4 from each batch used to reduce tissue samples. Oxidized and reduced aminolevulinate were separated by HPLC, and the specific activity of the reduced compound (dpm per nmol) was determined.

Quantitation of Cross-links by HPLC Amino acids and difunctional cross-links were analyzed by reversed-phase HPLC using a C18 column, 0.46 x 25 em (12). Briefly, hydrolysates containing 25 to 50 ug of hydroxyproline were injected onto the column via an AItex Model 210 injector (Altex, Berkeley, CA) and were eluted isocratically with 24070 n-propanol in 0.01 M phosphoric acid, 0.3% sodium dodecyl sulfate (pl-l, 2.84) at a flow rate of 0.8 mllmin. Altex Model 100 pumps controlled by an Altex microprocessor (Model 420) were used for elution. Amino acids and difunctional cross-links were detected by postcolumn derivatization with o-phthalaldehyde, which was added to the effluent stream via a three-way valve(Rainin) and an Altex Model l00A pump operated manually. Fluorescent

309

COLLAGEN CROSS-LINKING IN CHRONIC FIBROTIC WNG DISEASE

adducts were quantified with a Gilson fluo- comparing its UV absorbance at 295 nm with functional cross-link content, the ratio rometer with filters designed to detect OPA that of a solution of known concentration of of DHLNL to HLNL, and OHP content derivatives (Gilson, Middleton, WI; excita- N-ethyl-3-pyridinol (Aldrich Chemical Co., (table 1). The content of both DHLNL tion at 360 nm, emission at 455 nm). Elution Milwaukee, WI), a compound with the same and HLNL was significantly elevated in times and areas of peaks were recorded on extinction coefficient as OHP at 295 nm. patients with ILD compared with that in a Hitachi D-2000Chromato-Integrator. Frac- OHP was detected in HPLC column eluates control subjects (p = 0.04 and 0.04, retions were collected at 0.5- or l.O-min inter- with a Hitachi variable wavelength fluorovalswith a fraction collector (Microfractiona- meter set at 295 nm excitation and 395 nm spectively); however, the DHLNL:HLNL emission. The absolute concentration of OHP ratio was similar in control subjects and tor; Gilson). Ready-Safe cocktail (Beckman, Palo Alto, CAl was added as scintillant, and was determined by calibration with N-ethyl- in patients with ILD. OHP content was the entire sample was counted by liquid scin- 3-pyridinol, of known absorbance coefficient significantly higher in patients with ILD tillation counting in a Beckman 8000 count- (23). than in control subjects (p = 0.0002). In We have extensively characterized hydrox- patients with ARDS, the DHLNL:HLNL er using preprogrammed settings recommended by the manufacturer. An internal standard ypyridinium as follows: (1) UV and fluores- ratio was significantly higher than that of DHLNL labeled with 14 C obtained from cence spectrum analysis, showing a charac- in control subjects (p = 0.0022), apparcultured rat chondrosarcoma cells was inject- teristic shift in the excitation maximum when ently because of increased DHLNL coned with representative samples to help locate the pH is raised (20); (2) apparent molecular tent, although the increase in DHLNL cross-link peaks. Results were expressed as weight by gel permeation chromatography was not statistically significant, most likemolecules of cross-link per molecule of col- (21); (3) incorporation of lysine during in vilagen, as described in detail elsewhere (21). vo labeling experiments (22); (4) photolytic ly because of the very small number of Briefly, molecules of cross-linkare determined destruction of the putative OHP peak in lung patients in the ARDS group. As can be seen in table 2, there were no from the amount of radioactivity in that peak, tissue hydrolysates (11); (5) stochiometric relawhereas molecules of collagen are determined tionship between DHLNL disappearance and significant differences in collagen crossOHP formation (22). from the hydroxyproline content. links and OHP content among those paIn order to ensure the correct identificatients with IPF, HSP, RB, sarcoidosis, or tion of the reducible difunctional cross-links Statistical Analysis eosinophilic granuloma of the lung, but in a complex tissue such as lung, both Data for groups of subjects are expressed as subgroup size was very small for some DHLNL and HLNL present in lung hydroly- the mean ± standard error. Comparisons besates werecharacterized in detail as previously tween groups of subjects were carried out of these diseases. However, there were differences between some of these subdescribed (10, 11,20,21) in the following ways: using the Mann-Whitney two-sample U test, (1) apparent molecular weight by gel permeawhich does not assume the data to be nor- groups (lPF, HSP, and RB) and control tion chromatography (10, 20, 22); (2) incor- mally distributed. Possible correlations be- subjects and patients with ARDS in terms poration of lysine in both in vivo and in vitro tween the lung content of collagen cross-links of DHLNL, HLNL, and OHP content. labeling experiments (21,22); (3) reducibility and the discontinuous histopathology scores The DHLNL and HLNL content was with NaB3H4 (20); (4) tissue content and dis- were evaluated by Spearman's nonparamet- lower in control subjects than in patients tribution consistent with literature reports (10, ric rank correlation coefficient. Possible corre- with IPF (p = 0.0244 and p = 0.0456, 22). The identity of purified HLNL has been lations between lung collagen content of cross- respectively). The OHP content was lowfurther confirmed by both radioactive and links and both bronchoalveolar lavage fluid er in control subjects than in patients with "cold" Smith degradations. The latter involved cellular constituents and clinical variables oxidation of putative [lH]HLNL with NaI0 4 were evaluated by linear regression analysis IPF (p = 0.0002), HSP (p = 0.0574), or followed by reduction with NaBH 4 ; HPLC on a microcomputer using statistical analy- RD (p = 0.0032). The DHLNL:HLNL analysis revealed that lysine was the only r- sis software (StatView 512;Brainpower, Cala- ratio in lung collagen for the ILD groups was significantly lower than that for the adiolabeled reaction product (20). In the basas, CAl. radioactive Smith degradation, [lH]HLNL ARDS group (lPF, p = 0.0214; HSP, p = was oxidized with NaI0 4 and reduced with 0.0108; RB, p = 0.0164). Also, the lung Results NaB3H4 , resulting in three radiolabeled peaks: collagen content of OHP was significantMeasurement of Lung Collagen lysine, proline, and hydroxynorvaline (20). ly higher in patients with. IPF (p = The identity of DHLNL was confirmed by Cross-links in Control Subjects 0.0002), HSP(p = 0.0142), or RB (p = a cold Smith degradation in which the only and Patients with Acute and 0.0026) than in patients with ARDS. observed radiolabeled reaction product was Chronic Fibrosis hydroxynorvaline. In addition, in the case of Age We found several differences between DHLNL, we have shown in lung tissue that In a previous animal study (23), we demand fibrotic lungs in terms of dicontrol a stochiometric relationship existsbetween the disappearance of the compound identified as DHLNL and its immediate maturational TABLE 1 product, OHP (22).

Analysis of OHP Hydrolysates containing approximately 5 Ilg of hydroxyproline were analyzed for their OHP content by isocratic HPLC on a C18 reversed-phasecolumn using 16% acetonitrile in 0.01 M hexafluorobutyric acid as eluant at a flow rate of 1 mllmin as described previously (23). An OHP standard was prepared from bovine achilles tendon by hydrolysis in 6 N HCI for 24 h, followed by gel filtration on a Biogel P2 column with 0.1 M pyridine acetate as eluant. The concentration of the partially purified standard was determined by

COLLAGEN CROSS-LINKS IN LUNG TISSUE FROM CONTROL SUBJECTS ANO PATIENTS WITH AROS OR ILO* Cross-link Content (mol/mol of collagen) Group

OHLNL

HLNL

OHLNUHLNL

OHP (mmol/mol of collagen)

Control

0.066 ± 0.026 (n = 8)

0.015 ± 0.004 (n = 8)

3.247 ± 0.349 (n = 14)

16.707 ± 2.704 (n = 14)

AROS

0.120 ± 0.048 (n = 4)

0.020 ± 0.007 (n = 4)

6.852 ± 0.872 (n = 6)

14.717 ± 0.514 (n = 6)

Interstitial lung disease

0.095 ± 0.01 (n = 31)

0.027 ± 0.003 (n = 31)

3.814 ± 0.366 (n = 40)

32.9 ± 1.946 (n = 44)

• Values are mean ::t: 1 SEM, with n, the number of subjects, given in parentheses. Differences between mean values were analyzed for significance by the Mann-Whitney two-tailed, unpaired U test (see text).

310

LAST, KING, NERlICH, AND REISER

Relationship between Lung Collagen Cross-link Content and Clinical Parameters, Bronchoalveolar Lavage Cellular Content, and Histopathologic Changes in Patients with IPF Complete clinical and histopathologic evaluation was obtained in 19 patients with IPF (mean age, 61 ± 3.5 yr; 13men, 6 women). As a group, the patients with IPF demonstrated the following functional abnormalities: FVC = 63 ± 4070 predicted, FEV 1 = 69 ± 5070 predicted, FEV l/FVC = 61 ± 3010 predicted, Vtg = 82 ± 6070 predicted, TLC = 76 ± 5070 predicted, OLeo/VA = 79 ± 6% predicted, coefficient of retraction = 11.2 ± 1.6 em H 20/L, resting AaPo 2 = 21.6 ± 2.6 mm Hg, and AaPo 2 at maximal steadystate exercise = 37.3 ± 4.1 mm Hg. For the patients with IPF, there was no correlation between the duration of illness (mean of the group, 23 ± 8 months) prior to open lung biopsy and lung collagen content of cross-links. The degree of clinical impairment quantitated by the CRP score (mean, 48 ± 5; range 21 to 84) did not correlate with collagen cross-link content. However, the OHP

TABLE 2 COLLAGEN CROSS-LINKS IN LUNG TISSUE FROM SUBGROUPS OF PATIENTS WITH ILO* Cross-link Content (mol/mol of collagen) Group IPF Respiratory bronchiolitis Hypersensitivity pneumonitis Eosinophilic granuloma Sarcoidosis

HLNL

OHLNL

0.106 ± 0.012 (n == 21)

0.052

0.065 ± 0.009 (n == 4)

0.02

0.07 ± 0.028 (n == 4)

0.032

0.028

:I:

(n == 21)

0.006

:I:

(n

= 4) 0.01

:I:

OHP (mmol/mol of collagen)

OHLNUHLNL

(n == 4)

4.102 ± 0.493 (n == 28)

33.377 ± 2.445 (n == 31)

3.587 ± 0.654 (n == 6)

31.471 ± 3.128 (n = 7)

2.39 ± 0.325 (n == 4)

34.175 ± 7.582 (n == 4)

0.06

0.02

3.77

13.4

(n == 1)

(n == 1)

(n == 1)

(n == 1)

0.13

0.05

2.87

42.5

(n == 1)

(n == 1)

(n == 1)

(n == 1)

* Values are means ± 1 SEM, with n, the number of subjects, given in parentheses. Differences between mean values were analyzed for significance by the Mann-Whitney two-tailed, unpaired U test.

onstrated that there were age-associated changes in collagen cross-linking in the lungs of monkeys and rats. Therefore, because our patient and control populations spanned a wide age range, we sought to determine whether the differences noted above could be a function of age. We found that there was no correlation between age and DHLNL or HLNL content and the DHLNL:HLNL ratio in lung collagen from control subjects. On the other hand, as shown in figure 2, the lung collagencontent of OHP wasstrongly correlated with age in our control human subjects (r = 0.81, p < 0.001). The OHP content as a function of age for control subjects and patients with IPF is shown in figure 2, and the corresponding data for control subjects and patients with ARDS are shown in figure 3. There were no apparent age-associated effects in the patients with fibrotic lung disease. It is of interest that in all of the patients with IPF younger than 50 yr of age (and in many of them older than 50 yr of age), OHP values were increased above the 95 0J0 confidence limits for age-matched control subjects.

Smoking The smoking history was known in 32 subjects with ILD; 12werenever-smokers and 20 werecurrent or ex-smokers. There were no differences in the collagen crosslinks between patients with ILD who smoked and those who never smoked. Although there was no correlation between age and pack-years of smoking, there was a strong correlation between lung collagen OHP content and packyearsof smoking (r = 0.704, p = 0.00(5), as shown in figure 4. The comparison of lung collagen cross-links in control sub-

jects and never-smokers or ever-smokers with ILD yielded similar results, as described above, i.e., the OHP content was significantly lower in control subjects than in either of these groups (control subjects, 16.707 ± 2.704; never-smokers, 35.167 ± 3.523, p = 0.0022; smokers, 35.805 ± 2.976, p = 0.0002).

70 (J)

60 ~

50

'"-I ~

40

Z

8

o

0

Fig. 2. Scattergram illustrating correlation between lung collagen OHP content and age in controls (squares) and patients with IPF (circles). OHP content increased with age in control subjects (r == 0.81, P < 0.001); there was no significant correlation between age and OHP content in patients with IPF; 95% confidence intervals are shown for control subjects.

Z

Q

U

C.

:: Q

30 20 10

10

20

30

40

50

60

70

80

AGE

30

!Z

25

~

20

Q

15

Z

Fig. 3. Scattergram Illustrating correlation between OHP content and age in patients with ARDS (squares); values for control subjects, identical to those shown in figure 2, are shown for comparison.

U

C.

:: Q

10

. 5+--~F--,.--.---..--,.--.,---r_--....--.----,-~---r---'--+ ·10

10

20

30

40

50

60

70

AGE

65

o

60 ~

Z

0

55

'"-I

!Z Q

U

C.

:: Q

50 45

Fig. 4. Scattergram showing correlation between lung collagen OHP content and pack0.82, P < 0,01). years for patients with IPF (r

40 35

=

30 25 20 ·1

PK/YRS

311

COLLAGEN CROSS-LINKING IN CHRONIC FIBROTIC WNG DISEASE

content tended to decrease as the FEV 1/ FVC ratio decreased (r = 0.53, p = 0.0237)and tended to increase as the TLC increased (r = 0.416, p = 0.0859). Also, the OHP content tended to decrease as the coefficient of retraction increased (r = 0.508, p = 0.053). The percentage of instilled fluid recovered from the patients with IPF (mean, 46 ± 4%) was lower than that in healthy volunteers studied in our laboratory (mean, 75 ± 20/0) (14). The concentration of cells per milliliter of recovered fluid was greater in patients with IPF (35.9 x 104 ± 3.35 compared with 15.4 x 104 ± 2.2 in healthy volunteers). The content of alveolar macrophages per milliliter of BAL fluid was higher in patients with IPF (22.6 ± 3.4 x 104 ) than in healthy volunteers (14.3 ± 2.1 x 104 ) . The other BAL fluid cellular contents were higher in patients with IPF than in healthy volunteers: neutrophils, 5.7 ± 1.4 x 104 versus 0.1 ± 0.04 x 104; lymphocytes, 6 ± 1.2 x 104 versus 1.0 ± 0.2 x 104 ; eosinophils, 1.6 ± 0.44 x 104 versus 0.02 ± 0.01 x 104. In order to determine if BAL fluid cellular constituents were markers of any particular collagen crosslink content in the lung, we evaluated the correlations between BAL cells and the content of collagen cross-links. The only significant correlations among BAL cellular constituents were between the eosinophil content (expressed as cells per milliliter) and the DHLNL content (r = 0.673, p = 0.0233) and the DHLNL:HLNL ratio (r = 0.548, p = 0.0227). The biopsy specimens from all patients demonstrated features fulfilling the criteria for IPF, but at various stages of fibrosis. No significant correlation between the extent and severity of specific histopathologic abnormalities or overallscores for cellularity or fibrosis and the levels of collagen cross-links were apparent, except for a significant correlation (p < 0.045) between lung collagen DHLNL content and severity of alveolar epithelial desquamation observed histologically.

CD

10

~

I

9 8

7 Fig. 5. Comparison of DHLNL:HLNL ratios in lung collagen obtained from control subjects and from patients with IPF or ARDS. Samples from IPF lungs included tissue obtained by biopsy (open circles) and at autopsy (closed circles). Control lung samples and ARDS samples were obtained at autopsy. Horizontal bars denote group means.

....J

Z

....J

I

6

:::J

5

:E o

4

Z

found increased hydroxylation of lysine, increased ratios of DHLNL:HLNL, and increased amounts of OHP in lung collagen. In limited studies of human tissue from neonates dying of IRDS, we observed significantly increased ratios of DHLNL:HLNL as compared with ratios in the lungs of age-matched infants with no known lung disease (13).Such experiments led us to predict that DHLNL: HLNL ratios should be elevated in lungs of patients undergoing acute fibrotic episodes with rapid deposition of fibrotic collagen, and that OHP content of lung collagen should be increased in patients with chronic pulmonary fibrosis (13). The experiments reported here were specifically designed to test these hypotheses. It should be emphasized that all of the cross-links shown in figure 1 (DHLNL, HLNL, and OHP) occur uniquely in collagen and not in other proteins. Our data (figure 5) suggest that our first hypothesis, that DHLNL:HLNL raDiscussion tios are increased in acute fibrotic disOn the basis of experiments performed orders such as ARDS, is valid in human in several animal models of fibrosis, we fibrotic disease and follows the pattern have suggested that fibrotic lung colla- observed in animal models. The data obgen contains relatively more hydroxyly- tained in patients with IPF are harder to sine and hydroxylysine-derived cross- interpret. The highest DHLNL:HLNL links than does lung collagen from age- values were observed in autopsy specimatched control animals. For example, mens. In contrast, most of the DHLNL: in rats with lung fibrosis induced by in- HLNL values in biopsy-obtained specitratracheal administration of silica (10) mens were in the normal range, even in or bleomycin (11), and in monkeys ex- patients with clinically "active" disease. posed to ozone by inhalation (12), we A potential explanation is that patients

3 2

•

I

....• •

•

o

••

o ,,-

an, '.'. '

~. 8 -:3;: ..

.•..

:;:.:

IPF

ARDS

with IPF frequently are hospitalized and treated with high flow oxygen for days or weeks prior to death. Consequently, this therapy and the underlying illness may have resulted in new lung injury causing acute fibrosis in the presence of the existing chronic fibrosis. Intralung heterogeneity may provide additional difficulties in interpretation. For example, in one case, marked variability was observed in samples obtained from four separate lung lobes from the same patient: DHLNL content of tissue from the left upper lobe and the left lower lobe were in the normal range, whereas DHLNL content from the right upper lobe and the right lower lobe were very high. Two lung lobes were available from one other patient; DHLNL:HLNL values werenormal in both lobes. We clearly have insufficient data to determine whether such heterogeneity of DHLNL content is a consistent feature of IPF or not. It is well described that there is considerable histopathologic heterogeneity in the lungs of patients with IPF, such that there are areas of varying degrees of interstitial fibrosis with foci of honeycombing and mononuclear cell infiltration throughout the involved lung (24). Therefore, human lung fibrosis differs from experimental fibrosis in this respect. We have found that in animal models the entire lung tends to respond homogeneously to a fibrogenic stimulus. We have not observed large intralobe or interlobe vari-

312

LAST. KING, NERUCH . AND REISER

ability in collagen synthesis rate or crosslink content when we have systematically analyzed multiple samples from rat lungs (I). . Our second hypothesis suggested that we would find increased lung OHP content in long-term fibrosis, but not in acute fibrosis. We did indeed find that patients with IPF had increased lung OHP content, whereas patients with ARDS, who had generally been sick a month or less, had lower than normal amounts of OHP in their lung collagen (figures 2 and 3). These data in human tissue are consistent with our earlier data in animal models, and they suggest that fibrotic collagen evolves in the lung over several months, as illustrated schematically in figure 6. During the early stages of disease (that is, within the first 2 months), new fibrotic collagen has extremely low levels of OHP since the biosynthesis of this crosslink is relatively slow, requiring 4 to 6 wk in vivo (10, 22). Thus, OHP content, expressed as molecules of cross-link per molecule of collagen, will actually be lower than normal. (Of course, if OHP were expressed on a per lung basis , content would be higher than normal since total lung collagen is increased.) After several months, some of the excess DHLNL matures into OHP to form mature fibrotic collagen in which the number of molecules of OHP per molecule of collagen is higher than normal. Our data suggest that in humans as well as in animal models, OHP serves as a marker for the presence of structurally abnormal collagen synthesized in response to a pulmonary insult, whether exogenously or endogenously inflicted. When possible, we attempted to de-

termine if correlations existed between clinical parameters and changes in lung collagen cross-link content in patients with ILD. Even though we started with a large series of patients (64 in all, 33 of whom had open lung biopsies and 32 of whom had necropsies performed), values of n for various subgroups defined by disease (table 2) or by age (figure 2) were small. Thus, although it is likely that in some cases a lack of statistical significance may have resulted from small group size rather than from lack of biologic significance, we believe that the archival value of these data justify the attempt to perform such analyses. Difunctional cross-link and OHP contents, as determined by the analysis shown in tables 1 and 2, were normalized to the collagen content of the lung samples on the basis of chemical analysis of lung tissue for its 4-hydroxyproline content. Thus, it is relevant to ask whether collagen (4-hydroxYPfoline) content of the whole lung was likely to have increased, decreased, or remained the same in the samples analyzed. If the DHLNL, HLNL, and OHP analyzed were to occur in the same population of collagen, this issue would be of no consequence; however, if the various cross-links measured were in different collagens of the lung, then we might be concerned that apparent increases in, for example , molecules of DHLNL could arise from either increased content of DHLNL or decreased content of total lung collagen, with preferential degradation of collagen molecules not containing this cross-link (that is, data expressed as DHLNL/4-hydroxyproline could increase because of an increase in the numerator or a decrease in the de-

NORM AL LUNG

t~ E W

FIBROTIC COLLAGEN

EAR L Y FI BROSIS
2 MONTH S

Fig. 6. Schematic illustration of the evolution of fibrosis. During the early stages , excess newly synthes ized collagen is relatively deficient in OHP ; thus , OHP content expressed as mol cross-link/mol collagen will be lower in early fibrotic lungs than in control lungs . During later stages, some of the excess DHLNL matures into OHP to form mature fibrotic collagen . in which mol cross-l ink/mol collagen is higher Ihan normal. In some cases, areas of "active fibrosis" may also be present, in which collagen relatively poor in OHP is also present.

nominator). For IPF, it is known that total lung collagen (hydroxyproline) content is increased in patients autopsied after a chronic course of this disease (25). It has also been shown that patients surviving for approximately 2 wk or longer with ARDS and ventilatory support with supplemental oxygen administration have increased total lung collagen content (8, 26). Thus, if there is any influence of the denominator term (4-hydroxyproline content of lungs) used for data normalization in this study, the effect would be to systematically underestimate the changes in collagen cross-linking occurring in the diseased lungs. We have examined the issue of whether degradation of mature lung collagen at the biochemical level is an important component of the pathogenesis of pulmonary fibrosis in an established animal model of ARDS: intratracheally instilled bleomycin in the rat. These studies indicate that there is no detectable turnover of cross-linked collagen of lungs of rats administered bleomycin 6 wk after lung collagen was labeled by systemic administration of ['Hllysine to neonatal rat pups (27). Thus, the changes in difunctional collagen cross-links observed in the patients with ARDS studied, and in hydroxypyridinium content of lung collagen in the patients with lPF studied, are unlikely to have occurred as a result of collagen degradation in these diseased lungs. If we assume that the pathogenic mechanisms of lung fibrosis are similar in the bleomycin model in the rat and in human fibrotic lung disease, then the changes in lung collagen cross-linking reported in this study reflect alterations in collagen biosynthesis in pulmonary fibrosis. The significance of such alterations in collagen biosynthesis in fibrotic lungs for optimal therapeutic strategy for treatment of these diseases remains to be determined. What might be the mechanisms underlying the observed changes in collagen cross-linking in pulmonary fibrosis? The observed selective increase in the dihydroxylated cross-link DHLNL and its maturational product, 0 HP, in both human fibrosis and animal models suggest that alterations in levels of lysine hydroxylation may playa role. Recent studies have provided more direct evidence implicating the enzyme lysyl hydroxylase in early events in fibrosis (28). In these studies, lysyl hydroxylase activity was assayed in lungs from control rats and from rats that had received 1.0 unit of bleomycin 1 wk previously, a dose sufficient to

313

COLLAGEN CROSS·LINKING IN CHRONIC FIBROTIC WNG DISEASE

cause increased lung collagen content of both DHLNL and hydroxylysine. Lysyl hydroxylase activity was significantly elevated in the bleomycin-exposed rat lungs. Given the similarities in cross-link abnormalities between experimental fibrosis and human pulmonary fibrosis, one might speculate that they may share underlying mechanisms. It should be noted, however, that although there is a general consensus concerning abnormalities in collagen observed in animal models of fibrosis, conflicting observations have been reported in human fibrotic disease. For example, Nerlich and coworkers (29) have reported decreases in type I:type III collagen ratios in ARDS, as well as decreases in lysine hydroxylation in extracted and isolated collagens; similarly, Bateman (30), Kirk (31), and Bateman and coworkers (32) have also suggested that there may be early decreases in type I:type III ratios in IPF and in IRDS. In contrast, others have reported increased type I:type III collagen ratios in both ARDS and IRDS (8, 9). As discussed by Nerlich and coworkers (29), there are many possible reasons for these apparent discrepancies. For example, there may be important differences in the study population and/or the time of sampling; unlike animals with experimentally induced fibrosis, patients with ARDS make up an extremely heterogeneous population, with myriad differences in age, race, etiology, concurrent diseases, past environmental exposures, and ongoing therapeutic intervention. Perhaps the most. important factor contributing to apparent discrepancies may be the choice of analytical technique, as there are potential pitfalls in both biochemical and immunohistochemical approaches to determining collagen type ratios, as discussed by us in detail elsewhere (33). Despite these conflicting reports regarding some aspects of collagen metabolism in human fibrotic disease, we feel that the data on cross-linking abnormalities presented herein do suggest that alterations in lysine hydroxylation may playa role in the abnormalities observed in lung collagen in human fibrosis as wellas in experimental fibrosis in animals. Acknowledgment The writers thank S. Arlene Nicolli, Alma Kervitsky, Martin Wallace, Suzanne Hennessey,

and Trudy McDermott for their contributions to the studies that form the basis of this report; James Waldron Jr., M.D., Ph.D., for his review of many of the lung biopsies, and our patients and their referring physicians for allowing us to participate in their care.

References 1. Greenberg DB, Reiser KM, Last JA. Correlation of biochemical and morphological manifestations of acute pulmonary fibrosis in rats administered paraquat. Chest 1978; 74:421-5. 2. Reiser KM, Last JA. Pulmonary fibrosis in experimental acute respiratory disease. Am Rev Respir Dis 1981; 123:58-63. 3. Hesterberg TW, Gerriets JE, Reiser KM, et a/. Bleomycin-induced pulmonary fibrosis: characterization of the rat model. Toxicology 1981;60:360-7. 4. Last JA, Hesterberg TW, Reiser KM, et a/. Ozone-induced alterations in collagen metabolism of monkey lungs: use of biopsy obtained material. Toxicol Appl Pharmacol 1981; 60:579-85. 5. Reiser KM, Hesterberg TW, Haschek WM, Last JA. Experimental silicosis. I. Acute effects of intratracheally instilled quartz on collagen metabolism and morphology of rat lungs. Am J Pathol 1982; 107:176-85. 6. Haschek WM, Reiser KM, Klein-Szanto AJ, et a/. Potentiation of butylated hydroxytolueneinduced acute lung damage by oxygen. Am Rev Respir Dis 1983; 127:28. 7. Seyer JM, Hutcheson ET, Kang AH. Collagen polymorphism in idiopathic chronic pulmonary fibrosis. J Clinical Invest 1976; 57:1498-1507. 8. Last JA, Siefkin A, Reiser KM. Type I collagen is increased in lungs of patients with adult respiratory distress syndrome. Thorax 1983; 38:364-8. 9. Shoemaker CT, Reiser KM, Goetzman BW, Last JA. Elevated ratios of type I/Ill collagen in the lungs of chronically ventilated neonates with respiratory distress. Pediatr Res 1984; 18:1176-80. 10. Reiser KM, Last JA. Collagen crosslinking in lungs of rats with experimental silicosis. Coli Relat Res 1986; 6:313-24. 11. Reiser KM, Tryka F, Lindenschmidt RC, Last JA, Witschi HR. Changes in collagen cross-linking in bleomycin-induced pulmonary fibrosis. J Biochem Toxicol 1986; 1:83-91. 12. Reiser KM, Tyler WS, Hennessy SM, Dominguez 11, Last JA. Long-term consequences of exposure to ozone: II. Structural alterations in lung collagen of monkeys. Toxicol Appl Pharmacol1987; 89:314-22. 13. Reiser KM, Last JA. A molecular marker for fibrotic collagen in lungs of infants with respiratory distress syndrome. Biochem Med 1987; 37:16-21. 14. Watters LC, Schwarz MI, Cherniack RM, et a/. Idiopathic pulmonary fibrosis. Pretreatment bronchoalveolar lavage constituents and their relationships with lung histopathology and clinical response to therapy. Am Rev Respir Dis 1987; 135: 696-704. 15. Kawanami D, Basset F, Barrios R, Lacronique JG, Ferrans VJ, Crystal RG. Hypersensitivity pneumonitis in man: light and electron microscopic studies in 161ung biopsies. Am J Patho11983; 110: 275-89. 16. Myers JL, VealCF, Shin MS, Katzenstein ALA.

~espiratory.b~onchiolitis causing interstitial lung disease. A clinicopathologic study of six cases. Am Rev Respir Dis 1987; 135:880-4. 17. ~ing TE, Schwartz MI, Dreisin RB, Pratt DS, !heofllopoulos AN. Circulating immune complexes III pulmonary eosinophilic granuloma. Ann Intern Med 1979; 91:397-9. 18. Watters LC, King TE, Schwarz MI, Waldron JA, Stanford RE, Cherniack RM. A clinical radiographic, and physiologic scoring system for the !on~itud~nal assessment of patients with idiopathIC fibrosis. Am Rev Respir Dis 1986; 133:97-103. 19.. Wo.ess~er JE The determination of hydroxyproline III tissue and protein samples containing small proportions of this imino acid. Arch Biochem Biophys 1961; 93:440-7. 20. Reiser KM, Last JA. Analysis of collagen composition and biosynthesis by HPLe. Liquid Chromatogr 1983; 1:498-502. 21. Reiser KM, Last JA. Biosynthesis of collagen crosslinks: in vivo labelling of neonatal skin, tendon and bone in rats. Connect Tissue Res 1986; 14:293-306. 22. Last JA, Summers P, Reiser KM. Biosynthesi.s .of collagen crosslinks. II. In vivo labelling, stability and turnover of collagen in the developing rat lung. Biochim Biophys Acta 1989; 990: 182-9. 23. ReiserKM, Hennessy SM, Last JA. Analysis of age-associated changes in collagen crosslinking in the skin and lung in monkeys and rats. Biochim Biophys Acta 1987; 926:339-48. 24. Katzenstein ALA, Myers JL, Prophet WD, Corley LS, Shin MS. Bronchiolitis obliterans and usual interstitial pneumonia. A comparative clinicopathologic study. Am J Surg Patho11986; 10:373-81. 25. Selman M, Montano M, Ramos C, Chapela R. Concentration, biosynthesis, and degradation of collagen in idiopathic pulmonary fibrosis. Thorax 1986; 41:355-9. 26. Zapol WM, Trelsted RL, Coffey JW, Tsai I, Salvador RA. Pulmonary fibrosis in severe acute respiratory failure. Am Rev Respir Dis 1979; 119: 547-54. 27. Last JA, Reiser KM. Biosynthesis of collagen crosslinks. II I. in vivo labelling and stability of lung collagen in rats with bleomycin-induced pulmonary fibrosis. Am J Respir Cell Mol Bioi 1989; 1:111-7. 28. Gerriets JE, Last JA, Gelzlechter T, Reiser KM. In vitro lysine hydroxylation. Fed Proc 1989; 245:1456. 29. Nerlich AG, Nerlich ML, Muller PK. Pattern of collagen types and molecular structure of collagen in acute post-traumatic pulmonary fibrosis. Thorax 1987; 42:863-9. 30. Bateman ED. Cryptogenic fibrosing alveolitis: prediction of fibrogenic activity from immunohistochemical studies of collagen types in lung biopsy specimens, Thorax 1983; 38:93-101. 31. Kirk JME. Quantitation of types I and III collagen in biopsy lung samples from patients with fibrosing alveolitis. Coli Relat Res 1984; 4:169-82. 32. Bateman E, Turner-Warwick M, AdelmannGrill Be. Immunohistochemical study of collagen types in human foetal lung and fibrotic lung disease. Thorax 1981; 36:645-53. 33. Reiser KM, Last JA. Assessment offibrogenic potential of particulates and gases. In: Saxena J, ed. Hazard assessment of chemicals. Vol. 5. Washington, D.C.: Hemisphere Publishing Corp., 1987; 133-70.