It is now generally accepted that the fermentation of carbohydrates in the rumen gives rise to the formation of large amounts of the lower volatile fatty acids ; but the fact that these acids are absorbed in considerable amounts and at different rates from the rumen into the blood stream has made it clear that in vitro studies of the fermentation processes are necessary in order to ascertain the actual quantities in which they are made available to the ruminant.

The course of carbohydrate dissimilation by the mixed flora and fauna of the rumen is determined primarily by an extraordinarily stable milieu (cf. Marston, 1939, 1948), and so it might be expected that a closely similar environment, andinoculation of the substrate with the natural mixed culture of organisms from the rumen, would be essential prerequisites for the fermentation to proceed in vitro as it does in the rumen.

The study of pure or of ‘purified’ cultures of organisms derived from the rumen may eventually illuminate the intermediate stages in the course of the dissimilations which occur there and the nature of the organisms responsible, but it cannot give precise information concerning products of the fermentation of natural fodders in the rumen. The same may also be said of studies made with related organisms derived from other sources—soil, faeces, etc.—or with fermentations carried out at temperatures or reactions which are outside the limits that obtain within the rumen. Most of the experiments described in the earlier literature on the in vitro fermentation of refractory carbohydrates—cellulose, etc.—must be included in the above categories.

If information cogent to the nutritional physiology of the ruminant is to be obtained, it is essential that some major characteristics of the fermentation as it proceeds in the rumen should be used as criteria by which to gauge the similarity of the courses taken by in vitro and in vivo fermentations of natural foodstuffs. The extent to which certain insoluble carbohydrate fractions of natural fodders are broken down in the rumen, and the amount of methane produced from them, can be measured. If these criteria were met in vitro, under the appropriate conditions of temperature, reaction and anaerobiosis, then the fermentation might reasonably be considered to have taken a course closely similar to that occurring in the rumen, and the fatty acids produced could be taken to represent those actually formed in the rumen.

Samples of wheaten hay and lucerne hay were inoculated with rumen fluid and allowed to ferment in vitro. The breakdown of cellulose and hemicellulose and the production of fatty acids were measured ; in some of the fermentations the amounts of methane and carbon dioxide formed were determined.

Technique

  • (1) Samples of the fodders were inoculated with rumen fluid in Erlenmeyer flasks fitted with inlet and outlet tubes. The air was displaced, either before or after the addition of inoculum, by the passage of carbon dioxide or of nitrogen. The anaerobic state was preserved by closing the outlet tube with a water seal. The flasks were incubated at 38–40° C.

  • (2) In those experiments where the fermentation gases were measured the fodders were placed in a wide-mouthed bottle in a constant temperature water-bath (40° C.). After addition of the inoculum the air above the liquid was displaced by the passage of nitrogen. The gases formed during the fermentation were passed through a drier of ‘Dehydrite’ and the CO2 removed by passage through soda lime or ‘Sofnolite’ guarded by a further drier of Dehydrite. The remaining gas was passed over copper oxide in a furnace (720° C.) for combustion of the methane to carbon dioxide and water. These gases were collected and weighed in absorption tubes. A continuous passage (≏ 30 I./24 hr.) of CO2-free air was maintained through an inlet tube opening into the system immediately after the fermentation bottle.

  • (3) The inocula consisted of rumen fluid drawn from sheep with permanent rumen fistulae. A glass tube, with one end covered by a wrapping of surgical gauze over a layer of wire gauze, was inserted into the rumen. A syringe attached to the other end was used to withdraw the sample. Where small inocula were used, tap water was added to the flask, but with large inocula no other liquid was added.

  • (4) Either calcium carbonate or ammonium carbonate was added with the inoculum to act as buffer during the fermentations.

  • (5) Analysis of substrates. The residual solids from the fermentations were collected, dried at 100· C., and together with samples of the original fodder, were subjected to analysis. Cellulose was determined by the method of Norman & Jenkins (1933); it will be described subsequently as ‘N. J. cellulose’. To reduce the number of operations, the method was modified slightly by taking less material for analysis and by increasing the strength of the hypochlorite solutions. For the determination of furfuraldehyde-yielding substances the phloroglucide method, as described by the Association of Official Agricultural Chemists (1940), was used; the furfuraldehyde yield of the N.J. cellulose was also determined so that the group could be separated into what have been called for the purpose of this paper, ‘total pentosans’, ‘free pentosans’, and ‘cellulose pentosans’.*

  • (6) Analysis of fatty acids formed. At the end of the fermentation period, after the residual solids had been removed, a suitable aliquot of the liquor was taken for analysis of the mixture of volatile fatty acids present. Analyses were also made of the fatty acids present in the rumen fluid used as inocula to allow for the appropriate corrections to be made in the calculations of fatty acids formed during the fermentation. The liquor was made just alkaline to phenolphthalein and reduced to a small volume. It was transferred to a small flask, 5 g. of magnesium sulphate (MgSO4.7H2O) were added, and the fatty acids liberated by the addition of phosphoric acid, the reaction being adjusted to pH 3–4 (Congo Red). The volume was made up to 7· 5 ml. and, by distillation at constant volume, 50 ml. of distillate containing the pure acids was collected. (In some experiments a double distillation was resorted to, and in such cases formic acid was destroyed in the second distillation by the addition of mercuric oxide to the flask.) The recovery of volatile fatty acids in such a distillation was previously found to be: formic acid 92%, acetic acid 98–99%, propionic and butyric acids 100%. In some of the earlier experiments the acids present in this distillate were determined by the procedure of Gray (1947a). In the majority of experiments a partition chromatographic method was used. This method is described in detail in the Appendix.

Experiments

Series 1 : small inocula

The fermentations of wheaten hay were set up in Erlenmeyer flasks, as described above. Calcium carbonate was included as buffer, and a small inoculum of rumen fluid was used to start the fermentation. The period of fermentation was varied from 16 to 120 hr. The data and results of analyses are summarized in Table 1.

Table 1.

Fermentation in vitro: series 1, small inocula

Fermentation in vitro: series 1, small inocula
Fermentation in vitro: series 1, small inocula

Series 2: large inocula

Two different fodders were used in these experiments—wheaten hay and lucerne hay. The procedure in most cases was similar to that used in series 1, but in those experiments where the production of methane was measured the fermentations were carried out according to the second procedure. The inocula, however, were relatively much larger, and ammonium carbonate was substituted for calcium carbonate as buffer. The period of fermentation was in all cases 48 hr. The details of experimental data and results are set out in Table 2.

Table 2.

Fermentation in vitro: series 2, large inocula

Fermentation in vitro: series 2, large inocula
Fermentation in vitro: series 2, large inocula

The experiments were carried out to investigate the conditions necessary for the reproduction in vitro of the rumen fermentation, and to determine from such in vitro fermentations the amounts of each of the fatty acids made available to the sheep in its rumen.

It has been pointed out that it is desirable to adopt certain criteria from among facts established for the natural fermentation, in order to judge the success of the attempt to reproduce the process in vitro. The criteria which can be employed for this purpose are :

(1) The extent of digestion of the main carbohydrate fractions in the rumen

It has been shown that in a mixture of wheat straw and lucerne hay about 40% of the cellulose is broken down in the rumen (Gray, 1947b). In the digestion of a meadow hay (Marshall, 1949) about the same proportion of the pentosans was fermented.

(2) The production of methane

10–20 1. of methane are produced per kilogram of fodder fed to a sheep ; hydrogen is not present in the fermentation gases.

It will be seen from Table 1 that the series of fermentations in which relatively small inocula were employed did not meet the first of these criteria. Very little cellulose was decomposed even after 5 days, and usually after 2 days the odour of the fermenting mass was foul and not at all characteristic of the rumen fermentation. It will also be seen, however, that an effective fermentation of pentosans took place and that the earliest stage was remarkable for the production of acetic acid almost free from other acids. In the later stages larger amounts of the higher acids, and particularly propionic acid, were formed.

In the rumen itself the fodder meets at once with very large numbers of organisms, and so, after the failure of these first experiments, it was decided to increase the size of inocula. Furthermore, it was noticed that the CaCO3 was an inefficient buffering agent in that the reaction changed to about pH 5 in many of the fermentations; ammonium carbonate, which was found to maintain the reaction between pH 6 and 7, was therefore substituted for it. It is evident from Table 2 that the fermentations carried out under these conditions have met the criteria laid down, and there is therefore good reason to claim that the products are similar to those formed in the rumen itself from the same fodders.

(3) The fatty acids

From the mean values of fatty acid production in Table 2 it is calculated that a sheep eating 1 kg. of fodder per day would have available from the rumen fermentation the amounts of acids shown in Table 3.

Table 3.
graphic
graphic

In Table 4 the molecular proportions in which these acids were formed in vitro are contrasted with the proportions in which they occur in the rumen fluid. The differences between the two groups imply that the rates of absorption of the individual acids differ considerably. The ratio of acetic to propionic acid formed in vitro from wheaten hay is very much less than the ratio of the same two acids in the rumen fluid. It must therefore be concluded that propionic acid is more rapidly absorbed than acetic acid when this fodder is fermented in the rumen. A similar examination of the remaining data in this table suggests that, if the products found in vitro are in fact produced in the rumen, then the relative rates of absorption of the acids must fall in the following series:

Table 4.

Fatty acids in the rumen fluid and in the in vitro fermentations

Fatty acids in the rumen fluid and in the in vitro fermentations
Fatty acids in the rumen fluid and in the in vitro fermentations

A discussion of the earlier literature relating to the rates of absorption of the fatty acids from the rumen, and a report of observations which provide confirmation of the findings from the present series of in vitro experiments is given in Part II of these studies (Gray & Pilgrim, 1951).

Accumulation of products at the site of fermentation

In the sheep the fatty acids are removed continuously by absorption through the rumen wall, and failure to reproduce this phenomenon in vitro may have some influence on the rate of fermentation, and on the proportions in which the acid products are formed. A recent report on the fermentation of cellulose with and without removal of acid products from the site of fermentation (Louw, Williams & Maynard, 1949) showed that there was a reduction in the rate of fermentation, but no change in the proportions of acids formed when these products were allowed to accumulate in the fermentation vessel. In this regard it may be noted that it was necessary in the present work to continue the fermentation for 48 hr. in order to arrive at the same extent of substrate breakdown that was known to take place in vivo in 24 hr.

The gases

In a large number of experiments conducted in metabolism chambers in this laboratory the production of methane from a variety of fodders has been accurately measured, and it may be said that from 10 to 20 1. of methane are produced by the sheep from a kilogram of the fodders used in the present series of fermentations. This volume was produced in vitro, within 48 hr. ; no hydrogen was formed.

The production of CO2 cannot readily be measured in vivo because of the difficulty of separating the fermentation product from the respiratory gases. The amount formed was measured in some of the in vitro experiments of the present series, but the data have not been presented as the subject of gas production will be dealt with in a subsequent communication.

Other products

No attention was paid to other products of the rumen fermentation, such as the non-volatile acids and the various neutral bodies.

Microscopic appearance of the fermentation

Insufficient is known of the bacteria responsible for the fermentation to allow them to be used as criteria in the way that chemical changes have been used, and so no detailed microscopic examination of the mixed population of micro-organisms in these fermentations was undertaken. But the protozoa, whatever part they play in these processes, are characteristic members of the population, and since they do not occur outside the normal rumen it is probable that their continued existence in vitro would point to a close reproduction of the conditions obtaining in the rumen. In the first, unsuccessful, series of fermentations the protozoa did not survive but in the second series where relatively large inocula were employed they continued to thrive.

The authors wish to record their thanks to Mr H. R. Marston, F.R.S., Chief of the Division, for his interest and encouragement in the experimental work and for his criticism and help in the preparation of the manuscript.

The separation of volatile fatty acids on a ‘celite’ partition chromatogram

The following procedure avoids some of the difficulties associated with Elsden’s original method (1946) and is in effect an adaptation from that method and the method of Peterson & Johnson (1948). The difficulties inherent in the preparation of silica gel, and the variable behaviour which it exhibits are avoided by the use of celite. A more important gain, however, is that acetic acid is readily eluted from the column and can therefore be determined directly by titration. If other acids, moving more slowly on the column, are present in the mixture analysed, Elsden’s method does not separate them from acetic.

Reagents

  1. Celite diatomaceous silica analytical filter aid (Johns-Manville). Dried overnight at 105° C. before use.

  2. Moir’s indicator solution. A solution of Moir’s (modified methyl orange) indicator, 0·75–1·50 mg./ml. in CO2-free aqueous solution.

  3. BB1. Thiophene-free benzene containing 1% by volume of pure n-butanol.

  4. BB7·5. Benzene with 7·5% by volume of n-butanol. Immediately before use in the preparation or development of columns, these developer solutions are shaken with CO2-free water and filtered through a pad of cotton-wool.

  5. Standard N/10 NaOH. Protected from CO2.

  6. Standard N/100 NaOH. Protected from CO2.

  7. Phenol red indicator. 0·1% phenol red in 20% alcoholic solution.

  8. Potassium bisulphate. Finely powdered, dried overnight at 105°C. before use.

Preparation of solution of acids for analysis

An aliquot sample of aqueous distillate containing 2–3 mmol, of total acid is pipetted into a suitable flask or beaker. Total acid is estimated by adding 1 drop of phenol red solution and titrating to a faint pink end-point with N/10 NaOH.

A further 1 drop of N/10 NaOH is added and the solution evaporated just dry on a water-bath. A minimal amount of water (2–4 drops) is added to redissolve the residue and 3 g. of dry KHSO4 is added. A bent stirring rod is used to break up the mixture to a homogeneous pink powder and to ensure that no drops of moisture remain on the walls of the flask.

The mixture is extracted six or seven times with about 4 ml. of dry BB 7· 5 solution, the extracts being filtered through a small pad of glass-wool into a 25 ml. standard flask and made up to volume. Of this solution 1 ml. is subsequently titrated for total acid.

Preparation of the column

The tubes are 35 cm. long, 12 mm. internal diameter, constricted at the bottom to support a glass bead or sintered glass disk, and spun out to a funnel at the top. A cork fits in the top, containing a tube which leads, through a 2-way tap, to a reducing valve on a nitrogen cylinder. A tightly packed pad of glass-wool covers the glass bead to a depth of in., but this is not required when a sintered glass disk is used.

For each column 5 g. of dry celite is mixed thoroughly in a mortar with 3 ml. Moir’s indicator solution. A slurry of this is made with wet BB 1, and added to the tube small amounts at a time. Rapid packing may be effected by using a piece of glass tubing, one end of which is covered by silk or linen supported by a circle of metal gauze. The covered end should fit snugly into the chromatograph tube.

Separation of acids

The rate of flow through the column is adjusted to about 1 ml./min., a pressure of 3–5 Ib./sq.in. being necessary at the surface.

The level of excess BB 1 is taken down until it just enters the surface of the column. A i ml. aliquot of the BB 7·5 solution of the acids being estimated is added, taken down to the surface and washed in with 1 ml. BB 1. A further 10 ml. of BB 1 is added and run through the column. By this means the butyric and propionic acids are moved away from the acetic acid which remains near the top of the column. The effluents are collected in 16×4 cm. Pyrex boiling tubes. Collection of the butyric effluent is begun before the butyric band reaches the bottom of the tube, and collection of propionic begun when the propionic band is almost ready to emerge.

The flow may be stopped at any stage by opening the tube to air through a twoway tap.

Development with BB 7· 5 is begun when the leading edge of the propionic band reaches within 2 cm. of the bottom of the column and continued throughout the collection of propionic and acetic acids. The elution of acetic acid is continued until 8–10 ml. has been collected in addition to that required to leave no trace of pink on the column.

If the formic acid concentration is sufficiently small, the formic band will stay near the top of the column, which may then be used again. Columns should be left with the surface covered and the lower end sealed.

It is necessary to discard the first analysis made on a newly prepared column.

Titration of acids

Titrations are carried out in an atmosphere of CO2-free N2, with vigorous mechanical stirring. The N2 is bubbled and the mixture agitated for 3 min. before titrating. Standard alkali is run in from a 10 ml. CO2-protected burette.

N/100 NaOH is standardized against N/100 HC1, using 5 ml. N/100 HC1, 15 ml. CO2-free water, 15 ml. BB 1 and 2 drops of phenol red.

The total acid in the BB 7·5 solution being analysed is estimated by titrating 1 ml. in the presence of 15 ml. BB 1 and 20 ml. CO2-free water. The effluents containing the individual acids are titrated after adding 20 ml. of CO2-free water. Two drops of phenol red indicator are used in all titrations and the end-point, which must be approached slowly, is taken as the first appearance of a definite, orange-free, pink colour.

  1. When wheaten hay and lucerne hay were fermented by organisms from the rumen of the sheep it was necessary to employ a large inoculum of rumen fluid in order to reproduce the rumen fermentation in vitro. With a small inoculum the fermentation did not conform to the known characteristics of the natural process.

  2. Products per kilogram of wheaten hay fermented in vitro were: fatty acids 200–250 g.—acetic acid 41 %, propionic acid 43 % and butyric acid 16% (by weight) ; methane 15 1.

    Products per kilogram of lucerne hay were : fatty acids 250–300 g.—acetic acid 53%, propionic acid 29% and butyric acid 18% (by weight); methane 20 1.

  3. The findings support the view that, owing to the more rapid absorption of propionic than of the other acids from the rumen, the proportion of this acid remaining in the rumen fluid is considerably less than the proportion actually formed in the fermentation.

Association of Official Agricultural Chemists
(
1940
).
Official and Tentative Methods of Analysis
, 5th ed.
Washington
:
Association Official Agricultural Chemists
.
Elsden
,
S. R.
(
1946
).
Biochem. J
.
40
,
252
.
Gray
,
F. V.
(
1947a
).
J. Exp. Biol
.
24
,
11
.
Gray
,
F. V.
(
1947b
).
J. Exp. Biol
.
24
,
15
.
Gray
,
F. V.
&
Pilgrim
,
A. F.
(
1951
).
J. Exp. Biol
.
28
,
83
.
Louw
,
J. G.
,
Williams
,
H. H.
&
Maynard
,
L. A.
(
1949
).
Science
,
110
,
478
.
Marshall
,
R. A.
(
1949
).
Brit. J. Nutr
.
3
,
1
.
Marston
,
H. R.
(
1939
).
Ann. Rev. Biochem
.
8
,
557
.
Marston
,
H. R.
(
1948
).
Biochem. J
.
42
,
564
.
Norman
,
A. G.
&
Jenkins
,
S. H.
(
1933
).
Biochem. J
.
27
,
818
.
Peterson
,
M. H.
&
Johnson
,
M. J.
(
1948
).
J. Biol. Chem
.
174
,
775
.
*

The fact that these substances have, for convenience, been called pentosans is not meant to imply any exact knowledge of their nature. The group of furfuraldehyde-yielding substances present in such plant tissues is a wide one: it includes pentosans, polyuronides and other substances not so well defined. In general, losses of furfuraldehyde-yielding substances have been taken as an approximate measure of the fermentation of the whole hemicellulose group.