576. 8 . 097. 29 ( K toxin) : 5 7 6 . 851 . 57 (Cl. welchii)

THE COLLAGENASE ( K TOXIN) OF CL. WELCHII TYPE A

C. L. OAKLEY, G . HARRIET WARRACK and W. E. VAN HEYNMGEN The Wellcome Physiological Research Laboratories, Beckenham, Kent

IN clinical gas gangrene due to infection with Cl. welchii one of the most striking features is the conversion of large amounts of muscle to a soft friable pulp. In 1923 Henry claimed that pieces of live rabbit muscle incubated with Cl. welchii filtrates swelled up, became opaque and underwent changes similar to those occurring in gas gangrene ; in the process it absorbed much of the letha1 factor in the filtrates without producing much effect on the hzmolytic factor. No quantitative experiments were made, and the qualitative results were used only to support the view that the hEmolytic and lethal factors in Cl. welchii filtrates are distinct.

Though these remarkable observations have been much quoted, no one seems to have repeated them until Macfarlane and MacLennan (1945) showed that live human or rabbit .muscle incubated with Cl. welchii type A filtrates fell to pieces, the pieces being unaltered muscle fibres deprived of their reticulin scaffolding (Robb-Smith, 1945). They regard the active substance as a collagenase * and point out its possible importance in the spread of gas gangrene.

Though the muscle-disintegrating activity of the Cl. welchii filtrates used by Maofarlane and MacLennan is neutralised by C1. welchii type A antisera, no evidence is yet available to determine whether this activity is due to any of the toxins of Cl. welchii so far described or whether more than one substance is responsible for it. We have tried to tackle these problems from the immunological standpoint.

Experiments with muacle

We have found that pieces of fresh guinea-pig muscle incubated overnight with Cl. welchii type A atrates squash far more readily under finger pressure than similar pieces incubated with broth and that this muscle-softening activity is neutralised by Cl. welchii type A antisera. Muscle can therefore be used as an indicator to show whether this effect is due to one substance only and whether it is or

* Maschmann (1938) claimed that GI. wekhii filtrates contain 8 collagenase but later (1938-39) he seemed less certain.

220

230 C. L . OAKLEY. G . H . WARRACK AND W . E. VAN HEYNINGEN

is not due to one or other of the recognised toxic substances in Cl. welchii type A filtrates. For these purposes we took one of our sera (R 8531) as standard, ascribed to it an arbitrary value of 180 “uni ts” and tested a number of other sera against it for their capacity to inhibit softening of guinea-pig muscle by several different Cl. welchii type A filtrates produced in very different media.

Method. In a series of Lambeth tubes run out the test dose of the filtrate under test (usually at 2 units), amounts of serum differing by 10 per cent., and sufficient saline to make up to constant volume. After the mixtures have stood for half an hour add to each a piece of muscle (ca. 6 mm. square) from the abdominal wall of a freshly killed guinea-pig, cork the tubes and incubate at 37” C. in a water-bath overnight. In the morning remove the pieces of muscle to the squares of a chequered tray and prod each with the finger to determine whether softening has occurred. The end-point is taken as the tube which contains the maximum amount of antiserum allowing softening to occur.

It is clear that, for these six filtrates at least, the values obtained for each serum agree

TABLE I ComFrison of anti-muacb-softening values of sera determhd

against six different C1. welchii jiltrates

Table I shows the kind of result obtained.

Serum

R 8531 R 5434 A 16836 GGC 3296 LX 251 H 3044 A 24422 R 7843 R 5603 R 7577 H 9564 Ex 770 H 2917

Test dose (a)

Test dose (muscle- softening factor)

AG 835

180 50

110 500 800 500 180 330 900 70 ... ... ...

1-32 ml.E la unit

0.9 m1.G 2 units

Serum values against CZ. wekhii ffltrate

AG 886

180 50

100 460 650 500 160 300 870 65

210 400 700

0.37 m1.f la unit

0.83 m1.f 2 units

AG 893

180 50

130 550 900 500 140 320 650 90

280

800 ...

0.4 ml= la unit

0.9 ml.s 2 units

AG 938 ~~

180 50

130 530 700 500 150 300 700

70 210 330 700

3.26 ml.z la unit

1 m1.E 2 units

AG 952

180 50 95

600 700 600 130 340 900 80

210 350 800 -- 0.26 ml.=

la unit

0.6 ml.E 2 units

AG 95

180 70

130 600

500 130 300

80 230 320

...

...

...

1.26 ml la un

D-95 ml 2 unil

very well and it is therefore unlikely that more than one muscle- softening substance is present in these filtrates.*

* When fresh guinea-pig muscle is incubated overnight with some Cl. welchii type A filtrates-eapecially those prepared by growing the organism in peptic digests of horse muscle-a slimy change is produced independent of softening and easily distinguishable from it.

COLLAGENASE OF CL. WELCH11 231

-

Serum -

R 8531 R 5434 A 24422 A 16836 R 6423 H 2917 R 7843 H 9564 R 5603 GGC 3296 R 7577 LX 251 H 3044 Ex 770

If the results obtained by titration of sera against the muscle- softening factor are now compared with the known anti-u, ant i4 and anti-hyaluronidase values of these sera (table 11) it is clear that these

TABLE I1 Comparison of anti-muscle-softening values of sera with their anti-a, a n t i 4 and

anti-hyaluronidase values. The anti-muscle-softening value used i s the average value against several different filtrates

Anti-a value

170

27 57 75

100 140 270 280 370 420 460 520 620

0.2

Anti-6 value

90 190

13.5 500

70 900 130 360

38 11

62 165

0.7

0.2

Anti-hyaluronidase value

640 320

< 20 < 20 320 320

< 20 320 640

1280 < 5 640 320 120

sera do not neutralise the muscle-softening factor in their anti-u, anti4 or anti-hyaluronidase values. I

Anti-muscle- softening value

180 55

150 110 ( 2 750 320 230 800 540

77 700 500 360

iroportion to is therefore

unlikely that the muscle-softening substance is either a toxin, 8 toxin or hyaluronidase : it is probably a new toxin, which we propose to call K (kappa) toxin.

Experiments with collagen

Though testing with pieces of muscle was easy enough-to our great surprise testing to about 10 per cent. offered little difficulty- i t was of little use for large scale work. So, as we knew from Robb- Smith’s work that disintegration of muscle was associated with the destruction of reticulin, we tried collagen as an indicator. Horse tendon pounded with quartz and 1 per cent. saline swells to a jelly which, when spread on glass plates, dries to a substance resembling coarse unglazed paper. Pieces of this collagen “ paper ” were readily disintegrated by Cl. welchii type A filtrates and sera could be tested against a standard for their anti-disintegrating power, using pieces of collagen paper as indicator. Though testing by this method was sometimes easy and the results showed that the substances softening miiscle and disintegrating collagen were the same (i.e. a collagenase), difficulties in preparing uniform collagen paper prevented the develop- ment of a satisfactory test until we hit on the idea of using the commercially available “ hide .powder ”, which is a rich source of collagen. This hide powder was coupled with an azo- dye to give a bright reddish purple indicator (“ azocoll ”). When this was

232 C. L. OAKLEY, G. H . WARRACK AND W . E. V A N HEYNINGEN

incubated with Cl. welchii type A filtrates the disintegration of the collagen set the dye free into the liquid. Sera could then be tested for their power to prevent the diffusion of the dye from the azocoll during incubation with C1. welchii type A filtrates.

Preparation of azocoll. Sieve 1 lb. hide powder (Baird and Tatlock) to 60 mesh ; about 120 g. sievings should be obtained. To a solution of 0.575 g. benzidine in 100 ml. water containing 3 ml. conc. HC1 cooled in an ice bath add slowly a solution of 0.45 g. sodium nitrite in 10 ml. water. Allow to stand for 10 minutes, then pour the tetrazotised benzidine into a chilled solution of 6.25 g. sodium acetate (CH,COONa. 3H,O) in 500 ml. water. To the mixture add slowly with constant stirring a solution of 1.1 g. R-salt (sodium salt of 2- naphthol-3 : 6-disulphonic acid) in 100 ml. water, followed by 20 ml. N K,CO,. A brick-red dye is formed. In the meantime wash 80 g. of 60 mesh hide powder by repeated suspension in and filtration from 10 litres of water and finally re-suspend in 500 ml. water containing 30 ml. 2N K,CO,. Add the dye to the suspension in 6 equal lots at intervals of 10 minutes. The colour changes from brick-red to purple-red. After all the dye has been added, add 25 ml. 2 N K,CO, and allow the mixture to stand for ten minutes. Centrifuge or filter off the dyed hide powder and wash by re-suspending five times in 5-litre quantities of water. The wash liquors are coloured pink. Then re-suspend in about 300 ml. water, stir constantly and slowly add 1.7 litres of acetone. Repeat washing with acetone until acetone washings are no longer pink, filter off azocoll and remove residual acetone at 37' C. Sieve the azocoll to 60 mesh, dry over P,O, under reduced pressure and keep in 5-10 g. amounts in rubber-capped bottles under nitrogen. Azocoll required for indicator should be suspended as needed in 1 per cent. Manucol IV * (300 mg. to 100 ml.). The Manucol is used to obtain an even suspension of the azocoll ; in addition it may help to prevent the dye diffusing too far into the liquid when azocoll disintegrates.

Make mixtures in Lambeth tubes of the test dose of filtrate a t the level chosen (usually 2 units), amounts of serum differing by 10 per cent., and saline to make up to constant volume. Allow them to stand for half an hour, then add 1 ml. azocoll suspension to each. Incubate over- night in a water-bath at 37" C. (to prevent convection currents the water in the bath should reach the highest liquid level in the tubes). The end-point is taken as the tube which contains the maximum amount of antitoxin showing the development of a well marked red colouring above the azocoll. It has proved easy to use this test as a routine.?

Titration of sera.

Table I11 shows a comparison of the values obtained for sera using all three indicators. For nearly all sera the values obtained using muscle, collagen paper or azocoll as indicator are identical within the limits of the tests, suggesting that the same substance is responsible for the effects produced in all three.

Discordant sera Of the 38 sera examined two-R 5603 and R 6614show significant

discrepancies between the values obtained in muscle and azocoll tests.

* Manucol IV is a polymer of d-mannuronic acid (Allbright & Wilson, Birmingham). Occasionally a filtrate formed a gel with the Manucol ; in such cases isotonic saline was used for suspending the rtzocoll.

+ If sera are tested for very low levels of anti-n activity, the presence of large amounts of serum in the test mixture may lead (in the presence of free K toxin) to bleaching of the azocoll. Mixtures showing bleaching are always underneutrelised.

COLLAGENASE OF CL. WELCH11 233

doses per test dose at 2 units

Sensitivity of serum) value test *

Accuracy of s e ~ m l

There are a t least three possible explanations : ( 1 ) that the muscle- softening and collagen-disintegrating substances are different and that the agreement in serum values using muscle and azocoll as indicators

* per cent.

*20 per cent.

TABLE I11

Comparison of serum values against filtrate AG 835, using three different indicators

Serum values using as indicator Serum

muscle

I R 8531 R 5434 A 16836 A 24422 R 7843 H 3296 H 3044 LX 251 R 5603 R 6514

180 50

110 180 330 500 500 800 900 150

I

Test dose . . 1 0.9 m 1 . ~ 2 K units

collagen " paper "

180 58

100 170 350 600 680 700 900 ...

0.9 m 1 5 2 K units

15

+ 10 per cent. +20 per cent.

(irregular)

azocoU

180 65

105 170 300 400 600 650 360 90

0-5 m1.=2 K units

250

+ 5 per cent.

f10 per cent.

* Accuracy to which the end-point of a single titration can be read.

is fortuitous ; (2) that the muscle-softening substance disintegrates collagen, but that another collagenase is present having no muscle- softening properties ; (3) that R 5603 and R 6514 are non-avid and that the finely divided azocoll competes with them for toxin more effectively than pieces of muscle.

We therefore attempted to neutralise the muscle-softening substance i n filtrate AG 952 with R 5603 and R 6514 and determined the values of sera against the partly neutralised filtrate, using azocoll as indicator. Table IV shows that there is no evidence for the existence of a second collagenase, for if there were, it would have been expected that the values of some of the sera against it would have been different from their anti-muscle-softening values. Nor has it been possible to demonstrate non-avidity in R 5603 or R 6514, for tests at higher levels show no tendency for the muscle and azocoll values to approximate.

The discrepancies in the values of R 5603 and R 6514 are therefore unexplained, for we find it difficult to believe that so many sera would

234 C. L. OAKLEY, G. H. WARRACK A N D W. E. VAN HEYNINGEN

show concordant values in muscle and azocoll tests if the substances effective in these tests were different.

TABLE IV

Serum values (wing azocoll as indicator) against unneutralised and partially-neutralised AG 952

Serum

R 8531 R 5434 R 6712 R 6743 R 7843 GGC 3296 H 2917

I Test dose .

Serum value against

flltrate AQ 952 AG 952 partially neutralised with R 5609

180 50

120 320 300 570 800

180 55

100 310 300 500 700

0.33 ml.=2 K units 0.8 ml.El K unit

AG 952 partially neutralised with R 6514

180 45

120 330 250 580 700

0.8 ml.ZEl K unit I

Can u toxin disintegrate muscle ?

It is easy to show that Cl. welchii type A filtrates in which all the u toxin has been neutralised still disintegrate muscle, while filtrates in which all the K toxin has been neutralised have no muscle- disintegrating power, though large amounts of a toxin may still be present.

Histological examination of pieces of muscle or liver incubated with C1. welchii type A filtrates shows that, if K toxin is present, the reticulum surrounding the muscle fibres or liver trabecuk is destroyed whether u toxin is present or not. If all the K toxin is neutralised, no effect is produced on reticulum, however much u toxin is present.

Do types B, C and D of C1. welchii produce K toxin?

A few strains of types B, C and D have been examined. Of these only type C produced measurable amounts of K toxin (0.83 ml. of filtrate+ K unit by azocoll testing), while filtrates of type B and D cultures contained only traces.

Summary

1. Cl. welchii type A filtrates contain at least one additional toxin immunologically distinct from u toxin, 8 toxin and hyaluronidase. This substance ( K toxin) is a collagenase which breaks down muscle by attacking its collagen and reticulin scaffolding and may therefore be responsible for the pulping of muscle seen in human gas gangrene.

COLLAGENASE OF CL. WELCH11 235

2. The anti-rc values of sera can be determined using muscle, A method for preparing

3. A few sera show unexplained discrepancies between their anti-K

collagen " paper " or azocoll as indicator. azocoll is given.

values determined against muscle and azocoll as indicators.

We should like to express our thanks to Mr A. T. Glenny for constant help and advice, to Dr H. J. Rogers for the anti-hyaluronidase estimations and to Dr R. G. Macfarlane, Dr J. D. MacLennan and Dr A. H. T. Robb-Smith for allowing us to use their findings before publication.

REFERENCES

HENRY, H. . . . . . . . 1923. This Journal, xxvi, 497. MACFARLANE, R. G., AND MAC- 1945. Lancet, ii, 328.

MASCHMANN, E. . . . . . 1938. Biochem. Z. , ccxcvii, 284.

R O B B - S ~ H , A. H. T. . . . . 1945. Lancet, ii, 362.

LENNAN, J. D.

3, . . . . . 1938-39. Ibid., CCC, 89.