An in vitro method for predicting sensitizing properties of inhaled chemicals.

A method for monitoring chemical reactivity in aqueous solutions, at neutral pH and 37 degrees C, was developed. The chemical was allowed to react with a lysine-containing peptide, and the reaction was monitored with high-performance liquid chromatography. Simple acids, bases, and solvents did not react with the peptide, whereas isocyanates, anhydrides, and chloramine-T, substances well known for their sensitizing and asthma inducing properties, did. Thus a positive test strongly suggested that the chemical had the potential to act as a hapten and cause sensitization when inhaled. Prepolymers of diphenylmethane diisocyanate were considerably more reactive than prepolymers of toluene diisocyanate or hexamethylene diisocyanate. Isocyanates blocked with caprolactam, butanone oxime, malonic acid diethylester, or isononyl phenol showed no reactivity. This result suggested a significantly reduced risk of respiratory reactions when such blocked isocyanates are handled at room temperature. One blocked isocyanate showed, however, considerable reactivity.

A substance that is able to bind to a protein or other macromolecule in vivo and act as a hapten can also be anticipated to cause disturbances in the respiratory tract through mechanisms other than immunologic ones. Such substances may, for instance, interfere with receptor proteins or enzymes. This phenomenon has been suggested, on the basis of in vitro studies, for the reactive isocyanates (49). Furthermore, a recent in vitro study suggested that the pharmacological effect of toluene diisocyanate (TDI) may result from an autonomic imbalance between cholinergic and B-adrenergic neural control (50). The known respiratory effects of isocyanates have been reviewed by Musk et al (51).
The hypothesis behind this present work was that there ought to be a correlation between the ability of a substance to react with proteins and its potential to induce respiratory tract dysfunction. Furthermore, we assumed that a well-chosen peptide could replace the protein in an in vitro assay. The use of a peptide was considered advantageous, as we anticipated technical difficulties in demonstrating, and quantifying accurate-Iy, the reaction between a chemical with a low molecular weight and a protein. Thus the objective of this work was to develop a method of monitoring the reaction between a chemical and a suitable peptide.
The choice of the amino acid sequence of the peptide was based on previously published data. Lysine was our first choice, as in vitro studies had shown that, when added to proteins, isocyanates and anhydrides react with this amino acid (52,53). Lysine has an epsilon amino group which can take part in the reaction and has also been shown to be the most likely residue for the covalent binding of chloramine-T to a protein, although tyrosine, methionine, cysteine, and histidine were also modified by the substance (54). Furthermore, lysine has been shown to be involved in the reaction between proteins and ethylene oxide (55).
The second amino acid chosen was tyrosine. This amino acid absorbs ultraviolet light and can thus be detected easily in high-performance liquid chromatography (HPLC). Tyrosine has also been shown to be involved in the reaction between chloramine-T and proteins (54). On the basis of these considerations, three peptides containing lysine-tyrosine were investigated.
After a suitable test procedure had been established, we estimated the reactivity of various chemicals and industrially used isocyanate products to see if the reactivity was related to known respiratory effects of the substances.
Industrially used isocyanates were collected from a Swedish polyurethane factory or were obtained from Bayer AG , Leverkusen, Federal Republic of Germany. Three reactive textile dyes (from the " Remazol" series produced by the Hoechst company) were collected from a Swedish dye house. These dyes were Remazol Black B (color index reactive black 5), Remazol Brilliant Blue R Spec (color index reactive blue 19) and Remazol Brilliant Yellow 4GL (color index reactive yellow 160).
Acetonitrile and tetrahydrofuran, used to dilute the test samples, were dried for 24 h with 10 g of molecular sieve per 100 ml.

High-p erformance liquid chromatography
For reversed phase HPLC, Nova-Pak CI8 Radial-Pak™ cartridges (4 11m particles, column length 10 ern, inner diameter 8 mm) and u-Bondapak CI8 Guard-Pak" precolumn inserts were obtained from Waters Associates , Inc, Milford , Massachusetts, United States. The mobile phase in the HPLC consisted of 58 010 methanol and 42 % aqueous buffer (volume %). The buffer was prepared by dissolving 5 g of I-octane sulfonic acid sodium salt, 1.5 g of l -pentane sulfonic acid sodium salt, and I ml of triethylamine per liter of water. Before the buffer was used, its pH was adjusted to 3.0 by the addition of 10 volume % phosphoric acid.
The samples were chromatographed at a flow rate of 2 mllmin with ultraviolet detection at 280 nm. Ambient temperature and an injected volume of 50111 were used . The reproducibility in HPLC was investigated by 12 consecutive injections of an 0.1 umol /rnl (53 ug /rnl) solution of LTL. Overall variations in the test system were estimated by testing 10 industrially used isocyanate products, repeating the test after 2 d, and studying the performance of duplicate samples according to the method described by Barnett (56).

Pilot experiments
The reaction between isocyanates and pep tides was studied as a model for all reactive chemicals. Various combinations of isocyanate and peptide concentrations were used, and several mobile phases were tested. The reaction between ethylene oxide, at a high concentration , and LTL was also studied.

Test procedure
LTL was dissolved in an 0.02 molll sodium phosphate buffer (a mixture of 0.02 molll sodium dihydrogen phosphate and 0.02 molll sodium hydrogen phosphate pH 7.0) to a final concentration of 0.1 umol /rnl. Ali-.quots of the solution were used the same day or stored frozen at -20°C.
The test sample was dissolved in water to a final concentration of 4.0 urnol /rnl. Samples that were not soluble, or stable, in water were dissolved in acetonitrile instead.
To test a sample, 50 111 of the sample solution (4.0 umol/ml) was added to 500 III of the LTL solution (0.1 umol /rnl) in a test-tube at ambient temperature. The capped test-tube was then immediately placed in a 37°C water bath for 10 min and , before injection, in a 20°C (ambient temperature) water bath for I min.
During the HPLC, 50 IIIof the mixture was injected, and the area for the LTL peak was measured. This peak area was compared with the area for the unreacted LTL peak, ie, the peak obtained when 50 III of water (or acetonitrile) was mixed with 500 III of the LTL solution. The percentage of LTL that had reacted with the test sample was calculated, divided by 10, and rounded off to the nearest integer. The resulting figure was named the "peptide reactivity index," and it ranged from 0, when no detectable reaction could be noted, to 10, when all LTL had reacted.
As the molecular weight of many industrially used isocyanates was not known (or could not be defined) and as mixtures were often used, a slight modification of the procedure that has already been presented was adopted for such samples. A hypothetical molecular weight of 500 was assumed, ie, 4 umol/rnl corresponded to 2 mg/m\. Furthermore, some isocyanate products were difficult to dissolve directly in acetonitrile. These products were dissolved in tetrahydrofuran (20 mg/ml) as a first step and then diluted with acetonitrile to 2 mg /ml. Thus 50 III of the isocyanate solution (2 mg /ml in acetonitrile) was added to 500 III of the LTL solution (0.1 umol/rnl) in a test-tube and treated as has already been described.

Pilot experiments
Acetonitrile cou ld be used as a solvent as it did not affect the peptides. All three peptides (LTL, LTG and LT) had the ability to react with th e isocyanates MOI , TOI (a mixture of the 2,4-and 2,6-isomers) and HOI. LTL was , howe ver, found to be the peptide with the lon gest retention time. Generally, LTL was the peptide that showed the best separa tion from the other components in the chro matograms. For this reason, LTL was chosen for the exper iment s that followed . The detection limit of LTL was approximately 0.001 Ilmo l/ml.
The relative standard deviat ion [coefficient of variation (CV)] of the HPLC, investigated by 12 consecutive injections of an 0.1 umol /rnl (53 ug/rnl) solution of LTL, was 2 070. The standard deviation for the camplete test procedure, estimated from tests of isocyanate products on two different occasions and the determination of the differences between duplicate samples, corresponded to approximately 4 % of the original % pept ide re acted 100 LTL peak area . The retention time for LT L decreased linearly as the column temperature increa sed. The retent ion time at 50°C was approximatel y half that obtained at 20°C. Ambient temperature was chosen, but the exper iment showed that temperature was a variable that could ha ve been chan ged if the LTL peak had not been fully resolved in the chromatogram. MOI , TOI , and HOI were allowed to react with LTL (0.1 umol/rnl) at various molar ratios (figure 1). The percent age of the peptide th at reacted increased as the isocyanate:peptide ratio increased, and the differences between the three isoc yanates were noted . The potential to react with LTL was highe st for TOI , followed by MOI and HOI in that order. From the results illustrated in figure I, a molar ratio of 4 between the test sample and the LTL peptide was chosen . At this ratio , TOI reached the 100 % level of reacted peptide. Furthermore, differences between the individual isocyanates were clearly noted, whereas a considerably higher ratio would eliminate this difference. A typical chromatogram from a test of HOI is shown in figure 2. As can be seen, HOI itself did not produce peaks that interfered with LTL. It will also be seen that onl y about 30 % of the original LTL peak remained a fter the reaction with HDI, which corresponded to a peptide reactivity index of 7. The reaction products of HOI and LTL were probably represented by additional peak s with retention time s between approximately 2 and 4 min (figure 2, chromatogram 0) .
No detect able reaction was noted betwe en eth ylene oxide and LTL at a molar ratio of 4.0. A clear decrea se in the LTL peak area was, howe ver , obs erved when the ratio was increased to 450 and the reaction was monitored for 3 h. Under these conditions, an amount of LTL co rrespondi ng to approximatel y 15 % of the original amount was reacted per hou r.

Chemicals
Several well-defined chemicals were allowed to react with LTL for 10 min at 37°C and at a molar ratio of 4. The same procedure was also used for three reactive textile dyes collected directly from a dye house.  The result s, expressed as the peptide react ivity index, are shown in tab le I. The industrially used isocyanate prod ucts were tested at a concentration of 2 mg/ml, ie, a hypothetical molecular weight of 500 was assumed. The results, grouped acco rding to the type of isocyanate product, are shown in table 2. The blocked isocyanates were dissolved in tetrahydrofuran (20 mg/ml) as a first step and then diluted with acetonitrile to 2 mg/m!. Monomeric isocyanates had been found previously to elicit the same peptide reactivity index regardless of whether Table 2. Res ults obtained for industrially used isocyanate products. In the t est, 50 111 of the isocyanate solution (2 mg/ml in acetonitrile) was added to 500 111 of the L-Iysy l-L·ty ro sy l-Llysine (LTL) solution (0.1 urnotrml), The isocyanates are grouped according to the type of product. (N = the number of iso c yanate products tested w ithin each gr oup, HOI = he xamethylene d i- they were directly dissolved in acetonitrile or tetrahydro furan as a first step .

Discussion
Chemical reactivity is a relative concept. Industrially used chemica ls may be reactive in organic solvents, at elevated pressures and temperatures, at extreme pH values, in the presence of catalysts and initiating substances, etc. The react ivity we wanted to estimate was that relevant to biological con ditions, ie, aqueo us solutions, neutral pH, and a temp erature of 37°C . Several reactive industrial chemicals have been found to exert sensitizing and to xic effects, presumably by interaction with mucosal surface proteins. An in vitro test system simulating this in vivo protein interaction was designed . We wanted to work with diluted solutions, since the concentration of inha led chemicals is generally low. However, the analytical difficulties and the influence of trace impurities in solvents and buffers increase when low concentrations are to be analyzed . We found an LTL concentration of 0. 1 umol /rnl to be a suitable compromise.
For statistica l reasons, the decrease in the LTL peak area for a single injection of a sample had to be at least approximately 5 070 to be considered significant on the 99 % confidence level (normal distribution, one-taile d test) . We wanted to report the results on a scale on which each increment represented a significant difference, and a scale from 0 to IO was chosen.
2II Table I clearly shows that LTL was not affected by simple acids, bases, and solvents. Isocyanates, anhydrides, and chloramine-T, on the other hand, elicited peptide reactivity indexes in the range of 5 to 10.
It should be noted that not all substances known to sensitize the respiratory tract could be shown to react with the LTL peptide at a molar ratio of 4. For instance, the three reactive dyes tested did not. In this case the explanation may be that such dyes act as haptens without any covalent binding to a protein at neutral pH. Reactive dyes are normally activated by alkali before they become reactive and are able to bind covalently to textile fibers. Several reactive dyes have been shown to bind to human serum albumin and human 19G through adsorption (57)(58)(59), and we have previously suggested that covalent bonds may not always be necessary for the in vivo formation of dyeprotein conjugates (24). Thus reactivity does not seem to be a prerequisite for hapten sensitization.
The reactivity of ethylene oxide seems to be too low to induce hypersensitivity reactions when inhaled. We found no specific IgE antibodies among 20 workers who worked in a plant producing ethylene oxide and were suspected of having work-related respiratory symptoms (Wass & Belin, unpublished data). However, hypersensitivity reactions to ethylene oxide have occurred among dialysis patients when the gas has had direct access to the blood in the dialysis filter. These circumstances might be reflected in the peptide test, in which no reaction was noted at a low ethylene oxide concentration (molar ratio 4, 10 min), but the substance revealed its reactivity clearly when the molar ratio was increased to 450 and the reaction was monitored for several hours.
One major application of this study concerns isocyanates, a group of reactive chemicals which has achieved ever-increasing technical importance in the last couple of decades. They are the starting materials for polymers called polyurethanes. Polyurethanes are used as flexible foam, rigid foam, electrical insulation material, lacquers, elastomers, glues, etc. The health hazards associated with isocyanates have been known for many years, and the use of these substances is regulated in Sweden and many other countries.
In spray painting, isocyanate-based coatings are used for automobiles, airplanes, railroad cars, furniture, etc. The necessary spray equipment is easier to produce and handle if a one-component paint is used (all ingredients are premixed in the paint) than if two-component systems (the isocyanate is mixed with the other ingredients during the spray operation) are used. The search for one-component systems that are stable enough to be stored for several months has led to products in which the reactive isocyanate (NCO) groups are blocked at ambient temperatures. Caprolactam, butanone oxime, and malonic acid diethylester have been used as blocking agents for isocyanates (60). Temperatures above IOO°C are needed to release the blocking of the NCO 212 groups. Our data (table 2) indicate that isocyanates blocked with these three agents are inactive at 37°C. This finding suggests a significantly reduced risk of respiratory reactions when such blocked isocyanates are handled at room temperature, compared with prepolymers or monomers of isocyanates. From a theoretical point of view, an enzymatic degradation of blocked isocyanates in vivo to the corresponding free isocyanate (with NCO groups) may be claimed but, as far as we know, this hypothesis has not been supported by either in vivo or in vitro experiments.
The blocked isocyanate with phenol as the blocking agent, on the other hand, showed considerable reactivity. We tested two different batches of this product with the same result, peptide reactivity index 3; phenol itself did not react with LTL. Isocyanates blocked with phenol represent an old type of product that has been used mainly for wire coatings. It is interesting to note that the only report, as far as we know, of respiratory symptoms in regard to an occupational exposure to blocked isocyanates describes a man who sprayed with an adhesive containing a phenolblocked TDI (61). Isononyl phenol, on the other hand, seemed to be an efficient blocking agent.
We found that prepolymers of MDI were considerably more reactive than prepolymers of TDI or HDI. This difference was obviously due to the high content of monomers in the MDI prepolymer, in which monomer concentrations of up to 45 0,10 may exist.
Isobutyl chloroformate (ClCOOCH 2Cl-I(CH3 h, CAS 543-27-1), a chemical used in analytical chemistry for the activation of carboxyl groups (62,63), was found to have a peptide reactivity index of 7. This finding suggests a potential risk of respiratory symptoms if the substance is inhaled. We know of no such reports, but the substance is probably handled mostly by experienced chemists using fume hoods and other protective procedures.
We think that a positive test with the present peptide method strongly suggests that the tested chemical has, when inhaled, the potential to act as a hapten or to cause nonimmunologic disturbances (eg, interference with receptor proteins or enzymes). A negative test, on the other hand, does not exclude this possibility. The test result itself is not enough to judge a substance regarding its health hazards but may provide an important piece of information.
Several of our results could have been anticipated from what was known of the chemical properties of the substances. For instance, as phthalic anhydride was known to be a reactive substance, it was natural to expect that the same would apply to trimellitic anhydride. The peptide method can still be useful for substances with well-known chemical properties, however, because the method makes it possible to quantify the reactivity. Furthermore, the method can be applied to the testing of both complex mixtures and new chemicals with largely unknown properties.