THE FERMENTATION OF CARBOHYDRATES IN THE RUMENOF THE SHEEP

BY S. R. ELSDEN, From the A.R.C. Unit of Animal Physiology,at the Department of Biochemistry, Cambridge

(Received 15 May 1945)

(With Six Text-figures)

INTRODUCTIONThe majority of workers are agreed that the digestionof carbohydrate in the rumen is brought about bythe micro-organisms living there, and furthermore,that the microflora plays a much larger part in theseprocesses than the microfauna, for if the Protozoa areremoved by drenching with dilute copper sulphate,the digestive processes continued unhampered (seeVan der Wath, 1942).

The precise role of the microflora is still a matterof controversy. One school of thought stresses thenutritional importance of the volatile fatty acidsfound in the rumen, the end-products of the activityof the microflora. This view is typified by the workof Phillipson (1942), Phillipson & McAnally (1942),Barcroft, McAnally & Phillipson (1944 a, b). Theopposing view is that the volatile fatty acids are oflittle nutritional value to the animal, and that it isthe microflora itself which is important, serving as asource of carbohydrate, protein and the B group ofvitamins. This latter point of view is favoured byBaker who states that' The conclusion has thereforebeen provisionally drawn that it is the substancessynthesized such as microbial protein and polysac-charide, rather than the initial products of decom-position, such as organic acids, that are utilized bythe host animal' (Baker, 19426; see also Baker,1942a). Baker also maintains, on the basis of directmicroscopic observation of rumen contents, that themicroflora converts the starch and fibre of the dietinto bacterial starch. Starch is specified because anumber of the rumen micro-organisms stain bluewith iodine. It is assumed by Baker, though, nevertested experimentally, that these organisms, ladenwith starch, pass through the fore-stomachs into thesmall intestine, where the starch is hydrolysed toglucose by the pancreatic and intestinal amylases.It is equally possible that, during the time taken topass from the rumen to the small intestine, the micro-organisms use up their store of polysaccharide byauto-fermentation, with the production of volatilefatty acids.

That the microflora do act as a source of proteinis bofine out by the fact that part of the protein of

the diet may be replaced by simple nitrogenousderivatives such as urea, and the animal continue tothrive (Owen, Smith & Wright, 1943). The urea isused as a source of nitrogen by the microflora, theprotein of which subsequently becomes available tothe animal. Pearson & Smith (1.943 a, b, c) haveshown that when rumen liquor from a steer with apermanent rumen fistula is incubated in vitro withurea and either glucose or maltose, there is a signi-ficant increase in the amount of protein present; thisis perhaps not entirely unexpected. The synthesis ofthe B group of vitamins in the rumen has been demon-strated- by Wegner, Booth, Elvehjem & Hart (1940).

These opinions are not necessarily contradictory,and it seems certain that the ruminant utilizes boththe products of the activity of the microflora—thevolatile fatty acidsJ—and the microflora itself. Whatis more important *is the relative amounts of eachavailable to the animal. Phillipson (1942) has alsostressed this point. Owing to the difficulty of placinga permanent fistula between the abomasum and theduodenum it has so far proved impossible to estimatehow much, if any, of the bacterial protein and poly-saccharide passes from the fore-stomachs to the smallintestine per diem and how much of this is availableto the animal. Until this surgical problem is solved,and the measurements made, it. will be impossibleto assess with any degree of precision the nutritionalsignificance of the rumen microflora to the animal.

On the other hand, such technical difficulties donot impede the study of the volatile fatty acids in therumen, and recently it has proved possible to assessfairly accurately their importance to the animal. Ithas been shown that considerable quantities of vola-tile fatty acids are produced in the rumen of sheepafter a meal (Phillipson, 1942; Phillipson & McAnally,1942). These workers have also shown that ingestionof diets rich in soluble carbohydrate results in a rapidproduction of volatile fatty acids, with lactic acid asa probable intermediary. Introduction of solutionsof glucose, fructose or sucrose directly into the rumenvia a fistula, produces a similar rapid rise in volatilefatty acids. With diets of hay supplemented by branand oats, or hay alone, there is a much slower rise inthe concentration of volatile fatty acids, and similar

4-2

S. R. ELSDEN

results were obtained when starch or cellulose wereintroduced directly into the rumen. Quin (1943),also working with sheep, has provided comple-mentary evidence, obtained both from in vivo andin vitro studies. He did not follow the productionof volatile fatty acids, but measured the rate of gasproduction in the rumen after feeding a given diet,or introducing a pure carbohydrate directly into therumen.

Phillipson & McAnally (1942) showed that theconcentration of volatile fatty acids is high in therumen and low in the abomasum, and they put for-ward the suggestion that these compounds are ab-sorbed from the rumen, reticulum and omasum.Barcroft et al. (1944a) have shown that, in the sheep,there is a high concentration of volatile fatty acidsin the blood flowing from the rumen, reticulum, andomasum, whilst in the blood from the abomasum,small intestine and systemic circulation the amountsof volatile fatty acid are negligible. The blood fromthe caecal veins also contains significant amounts,and this is in agreement with the observation thatthe intestinal contents are subjected to a secondfermentation in the caecum. Direct measurementof the rate of absorption of volatile fatty acids fromthe rumen and reticulum alone indicates that asignificant proportion of the animal's calorific .re-quirements could be obtained from these compoundsif they be so used, apart from the amounts absorbedfrom the omasum and caecum.

The nature of the volatile fatty acids present inthe rumen is not known with any degree of certainty.Mangold (1929, p. 326) states^that the followingacids have been observed: formic, acetic, propionic,w-butyric, tsobutyric and valeric. . Barcroft et al.(19446), on the basis of the rate of steam distillationof the volatile fatty acids obtained from the rumenof sheep, considered that acetic acid was the principalcomponent, and was associated with small amountsof higher fatty acids; they obtained no evidence forthe presence of formic acid.

In the same way very little is known of the organ-isms responsible for the fermentation of carbohy-drates in the rumen. Van der Wath (1942) observedthat starch was digested by a coccus which simul-taneously stained blue with iodine; it also fermentedglucose with a similar result. This organism wasisolated in pure culture and was found to fermentboth glucose and starch with the production of un-determined acidic products. Quin (1943) observeda similar organism in the rumen of sheep and, if thediet was rich in soluble carbohydrate, a yeast-likeorganism. The latter fermented glucose rapidly bothin vivo and in vitro and at the same time formed asubstance which caused it to stain brown with iodine.As it multiplied by binary fission Quin classified it asa Schizosaccharomycete, and named it, provisionally,Schizosaccharomyces ovis. It was not isolated in pureculture, nor were its fermentation products identified.A similar organism was found by McDougall (private

communication) to develop in the rumen oJwhen a diet of hay was supplemented by mangolds.

Pochon (1934,1935) claims to have isolated in pureculture a cellulose-fermenting organism from oxrumen contents. The products formed from cellulosewere ethanol, and formic and acetic acids, and laterin the fermentation a little propionic acid. It isemphasized that formic was supposed to account forsome 75 % of the total volatile fatty acids producedand some 64 % of the cellulose used. The productionof such a mixture of compounds by one organismfrom one substrate has no parallel in bacterial fer-mentation, and confirmation of such a unique claimis clearly desirable.

From this brief review of the volatile fatty acidsin the rumen and the agents responsible for theirproduction it is clear that there are many gaps inour knowledge. The present work deals with (1) thenature of the volatile fatty acids present in the rumen;(2) the nature of the bacteria producing them; (3) theeffect of diet on the flora of the rumen.

METHODSRumen contents. The rumen contents used in

these experiments were either taken direct fromsheep with permanent rumen fistulae, or were ob-tained from the animal immediately after slaughterat the abattoir; in the latter case, the samples weretransported to the laboratory in a thermos flask. Thecrude rumen contents were filtered through muslinand the liquor so obtained used both for analysisand for the inocula in the in vitro experiments.

The in vitro fermentation. While the in vitro ap-proach provides a much greater degree of controlover the experiment, it is necessary so to adjustconditions that there is the least possible divergencefrom those which hold in vivo. An outstandingfeature of the rumen is the large inflow of saliva fromthe parotid glands, amounting, in the case of thesheep, to some 3 I./24 hr. (McDougall, private com-munication). This forms the basal medium in whichthe rumen organisms grow, the substrate being pro-vided by the diet. The saliva is approximately o-iNin respect to bicarbonate and contains, in addition,some 30-80 mg. phosphate P/100 ml. (McDougall,private communication). The bicarbonate is the mainbuffer, and the pH. of the rumen contents is main-tained around neutrality. The gas phase consistslargely of CO3 with up to 30% methane, oxygenrarely exceeds 1 % (Mangold, 1929, p. 148). Thefermentation is thus almost anaerobic. The followinginorganic medium was chosen as approximating tothe conditions in the rumen:

o-2MNaHCO3 100 ml.O-I54MKC1 4 ml.o i iMCaCl 2 3 ml.O-I54M KH2PO4 1 ml.oiS4MMgSO4.7H2O 1 ml.o-is6M(NH4)2PO4 s ml.

The fermentation of carbohydrates in the rumen of the sheep^ medium was made up from stock solutions

as required and saturated with pure CO2 immediatelyafter preparation. The fermentation vessel was a150 ml. Buchner flask fitted with a rubber bungcarrying a gas inlet tube closed by a stopcock; thetube was adjusted so as to dip below the surface ofthe medium; and the side arm of the flask was fittedwith a Bunsen valve. 40 ml. of the medium weremeasured into the flask, and to this was added either5 ml. of a 10% solution of the substrate, or, wherean insoluble substrate was used, 5 ml. of distilledwater. The inoculum was 5 ml. of rumen liquor,making a total volume of 50 ml. After inoculationthe flask was stoppered and gassed thoroughly withCO2 and incubated at 380.

The in vivo fermentation. The in vivo fermentationof glucose was studied on two Hampshire Downsheep, aged about 1 year, fitted with permanentrumen fistulae. The animals were fed on the morningof the day preceding the experiment, and food wasthen withheld until the experiment finished 48 hr.later, access to water being allowed at all times.24 hr. after the last meal a sample was withdrawnfrom the rumen, via the fistula, for analysis. A secondsample was withdrawn an hour later and immediatelyfollowing this a dose of glucose (100 g. dissolved in400 ml. distilled water at 380) was introduced, againthrough the fistula. Further samples of rumen con-tents were then taken at suitable intervals during thenext 24 hr. to study the fate of the added glucose.

The substrates. Lactic acid (B.D.H, Analar), ad-justed to pH 7 with NaOH, and glucose (B.D.H.Analar) were used. The cellulose was prepared from•Whatman no. 1 filter paper. It was first thoroughlymacerated in boiling distilled water, filtered off at thepump, and washed with a large volume of distilledwater followed by 95 % ethanol; it was then dried ina steam oven. The washing with ethanol was essential,as it resulted in a light fluffy product which disinte-grated readily on shaking with water. If this stepwas omitted, or if an insufficient volume of ethanolwas used, the product dried in hard lumps whichwould not disintegrate in water. The dried grass usedwas kindly supplied by the Hannah Dairy ResearchInstitute, to whom the author wishes to extend hisgrateful thanks.

Methods of chemical analysis

(1) Volatile fatty acids. These were determinedby the silica gel partition chromatogram (Elsden,1945). By this method the acids were identified,separated quantitatively one from the other, and thefractions titrated with ca. o-oiN NaOH. The volatilefatty acids were separated from the non-volatile acidsby the distillation procedure of Friedemann (1938).

(2) Glucose. The method of Hanes (1929) wasused. Solutions were deproteinized by the additionof ^ vol. of saturated lead acetate, and the excess leadremoved by solid Na2HPO4; after filtration a suitablealiquot was taken for analysis. Experience showed

53that the more usual methods of protein precipitationcould not be used because of the large and variableamounts of bicarbonate encountered. The procedureadopted was found to be adequate and entailed noloss of glucose.

(3) Lactic acid. This was determined on a copper-lime filtrate of rumen liquor, without a preliminaryremoval of protein, by the method of Friedemann& Kendall (1929). 5 ml. of the rumen liquor werepipetted into a 50 ml. volumetric flask; 5 ml. of 20 %CuSO4.5H2O (w/v) were then added, followed inturn by distilled water, 5 ml. 10% suspension ofCa(OH)2 and sufficient distilled water to make up tothe mark. The mixture was allowed to stand for30 min., filtered, and a suitable aliquot taken foranalysis.

(4) Ethanol. The method of Friedemann & Klaas(1936) was used without modification.

(5) Higher alcohols. These were tested for byoxidizing with bichromate according to Friedemann('938), and the fatty acids produced identified onthe partition chromatogram. No higher alcoholswere in fact observed.

RESULTS

The volatile fatty acids in the rumen. Rumen con-tents from a number of sheep and an ox were analysedfor volatile fatty acids. The results (Table 1) are

Table 1. Volatile fatty acids in the rumenof the sheep and ox

The results are expressed as raM. volatile fatty acid/100 ml. rumen liquor. * indicates that the animal wasat grass, and t that the animal was fitted with a per-manent rumen fistula.

Animal

Sheep F t, Ft, F*t. c, D*t, E, G, W.F.t, W.F.t, 1*, 2*, 3*, 4#

, 5*, 6*, 7*, 8*, 9*

Ox 1

Acetic

S-453-288- i

i-347-687 62 5 2

1 5 1

3'°47-785-255-428-2

n-357-737 98-527-0

4-48

Pro-pionic

1 1 8

1*2

368o-862-142 20 - 7

0-48O-7I2 461-52I-521-824-052 2 3

I 962'l82'260 9 2

Butyric

0-780-842 1 4

i-6

i - 32-440 7

0 2 6

0 6 91 461-481-161-382 6

2-141-672 1

1-940 6 2

Total

7 4 15 3 2

13923-8

I I ' I 2I2'24

3 9 22 2 5

4-441 1 7

8-258 - i

1 1 4

I 7 -8I2-I" • 5 312-81 1 2

6-02

expressed as mM. fatty acid per 100 ml. rumenliquor. Three acids were invariably present: acetic,propionic and a butyric isomer (the method does not

54 S. R. ELSDEN

distinguish between isomers). In addition, a veryfast-moving band, corresponding to a valeric acid,was usually observed on the chromatogram, but, asthe amount was small, the practice was adopted ofcollecting this along with the butyric fraction andtitrating as one. The figure for butyric acid is thusa composite one. It will be observed that there wasconsiderable variation from one animal to anotherand even in the same animal; nevertheless, the fol-lowing conclusions may be drawn. With the excep-tion of one animal, C, acetic acid predominated andamounted to 55-75% of the total; propionic acidwas next, followed by the butyric-valeric fraction.The distillation method adopted for the preliminaryseparation of the volatile fatty acids destroyed formicacid; but even when the oxidizing agent (HgSO4)was omitted, no formic was found. It is of interestthat the same fatty acids, in more or less the sameproportions, were found in the rumen liquors of theox, and it may well be that such a mixture is charac-

prisingly uniform, and under the conditipn^Bedpropionic and acetic acids were the chief volatilefatty acids produced. The position of butyric acid isuncertain, for the amounts produced over and abovethe control were so small as to be almost within thelimits of the analytical procedure. Propionic acidpredominated. This came as a surprise, and its pre-sence was further confirmed in two ways. First bythe microchemical test of Musicant & Kaszuba(1939), which was subsequently used as a routinetest throughout, and secondly by the isolation of thesilver salts of both propionic and acetic acids. Theprocedure for the latter was as follows.

The mixed volatile fatty acids from 40 ml. of fluidfrom a cellulose fermentation were separated by theusual distillation procedure and taken up in 50 ml.chloroform as for the chromatographic method, andthe chloroform extract fractionated on the partitionchromatogram, using ten columns in all. The aceticand propionic fractions were collected separately and

Table 2. Volatile fatty acids, in mM., produced during the in vitro fermentation of celluloseby rumen contents from the sheep

Figures in parentheses refer to the volatile fatty acids in the control experiment incubated without cellulose.XS implies that an excess of cellulose, > 1 g., was used. Those sheep marked f were fitted with a permanentrumen fistula.

Exp.

12

3456789

Sheep

FtFtFtAtBDtEG

Ft

Celluloserhg.

50050050053°XS5O5636XSXS

Total volatileacids

459 (053)3-17 (038)3-91 (060)450 (069)6-75 (091)4-78 (o-o)S'57 (o-o)591 (o-o)5-06 (0-19)

Acetic

2-38 (0-37)1-07 (022)1-79 (0-34)1 95 (o-4i)3-04 (0-53)205 ( - )2-30 (—)2-52 ( - )2-14 (o-io)

Propionic

2 08 (o-i)1-98 (0-07)2-04 (019)2-33 (018)3 23 (016)256 (—)281 (-)2-96 (-)2-78 (006)

Butyric

0-13 (0-06)0-12 (o-og)009 (007)02 (o-i)048 (0-22)o-i7 (—)046 (—)O-43 (—)014 (003)

Timehr.

95i88|88908 i *

1 1 2

95H'i

95*

teristic of all ruminants, though much more workwould have to be carried out before any such generali-zation can be safely made.

The fermentation of cellulose. Preliminary experi-ments iw vitro under the conditions described aboveindicated that cellulose was rapidly fermented withthe production of volatile fatty acids and considerablequantities of gas. About 1 ml. of o-oiN total acidwas produced per mg. cellulose fermented, and inconsequence a considerable quantity of the gasformed must'be COa arising from action of the acidson the bicarbonate buffer. No detailed investigationhas been made of the composition of the gas. Mostof the ammonia N of the medium had disappearedby the end of the fermentation.

Table 2 records the detailed analysis of the volatilefatty acids produced by the in vitro fermentation ofcellulose. The fermentations were not taken to com-pletion nor, in this series of experiments, was theresidual cellulose estimated, so that carbon balancesheets cannot be constructed. The results were sur-

titrated with o-O3iVNaOH. This gave solutions con-taining 76-3 and 55 mg. propionic and acetic acidsrespectively as their sodium salts. Both were treatedin the following way. The solution was made up to200 ml. and acidified with 10 ml. of 2-2VH2SO4 con-taining 10 % (w/v) HgSO4,and sog. of MgSO4.7H3Oadded. The volatile fatty acid was distilled off, heatingbeing continued until the MgSO4 commenced tocrystallize. The distillate was shaken up with excesssolid Ag3CO3, and when the acid had been neutral-ized the residual solid was filtered off. The solutioncontaining the silver salt of the fatty acid was evapo-rated down to incipient crystallization and then leftin the refrigerator. Next day the product was filteredoff and recrystallized from boiling water. The crystalswere filtered off and dried in a vacuum desiccatorover H2SO4. The mother liquors were worked upand a second crop was obtained. In all 96-3 mg. ofsilver propionate, and 49-1 mg. silver acetate wererecovered; the theoretical yields, on the basis of theinitial amounts of each acid, were 189 and 153 mg.

The fermentation of carbohydrates in the rumen of the sheep 55f. The compounds analysed as follows.

(Found :C, 20-2 ;H, 3-1; C3H5O2Ag requires C, 19-9;H,a-8. C, I4'8;H, i-8; C2H3O2Ag requires C, H'4;

H o \

The fermentation of glucose. Table 3 gives the de-tailed analysis of the volatile fatty acids produced bythe in vitro fermentation of glucose. Acetic, pro-pionic and butyric acids were formed. Propionicacid predominated as in the case of cellulose but, incontrast to the latter, significant amounts of butyricacid were formed. Phillipson & McAnally (1942)showed that when glucose was fermented in therumen there was a transient appearance of lactic acid,and in view of this finding the course of the in vitrofermentation of glucose was followed in more detail.

cally in Fig. 1. It will be noted that the glucose wasused up within 24 hr. and that large quantities oflactic acid and volatile fatty acid, chiefly propionicand acetic, were then found to be present. Duringthe next 48 hr. the lactic acid disappeared, and therewas a corresponding increase in the volatile fattyacids. Propionic and acetic acids were the mainproducts, but it is important to notice that somebutyric acid was also produced.

Comparison of the fatty acids of the rumen liquorwith those produced from the in vitro fermentationof both cellulose and glucose reveals a qualitativesimilarity; quantitatively, however, there is a dis-crepancy. Thus, acetic acid predominated in vivo,whereas propionic acid was the main product in vitro.

Table 3. Votatile fatty acids produced during the in vitro fermentation of glucose by rumen contentsfrom the sheep. 500 mg. (2-78 mM.) were fermented in each experiment

Figures in parentheses refer to the volatile fatty acids in the control experiment incubated without glucose. Thesheep marked t were fitted with a permanent rumen fistula.

Exp.

1234567

Sheep

FtAtCBDtEG

Total volatileacids

4-54 (o-6o)3-25 (0-69)330 (00)441 (0-91)4-80 (o-o)3-04 (o-o)2-79 (o-o)

Acetic

1-56 (0-34)I-O2 (041)o-57 (—)1-41 (0-53)I-5° (-)091 (—)039 (—)

Propionic

S-22 (0-19)•83 (0-18)•50 (—)93 (0-16)

*-3° (—)18 (—)

•30 (—)

Butyric

0-72 (0-07)0-40 (010)123 (—)1-07 (022)0-94 (—)o-9S (—)i-10 (—)

Timehr.

889088£8ti

11295

The fermentations were conducted on ten times theusual scale mentioned above. Immediately followingthe addition of glucose a sample was removed fromthe mixture for analysis, and the remainder gassed

Fig. 1.

with C02 and placed in the incubator at 380. Atsuitable intervals of time samples were taken and themain solution re-gassed with C02 prior to replacingin ^he incubator. The results are expressed graphi-

It became necessary, therefore, to study further thein vivo fermentation of. carbohydrate, and in par-ticular the nature and amounts of the volatile fattyacids produced.

For this purpose two Hampshire Down sheep,W.F. and B.F., with permanent rumen fistulae, wereselected; B.F. was used.mainly to confirm the obser-vations made with W.F. At the time this series ofexperiments were initiated W.F. was on a diet ofpoor-quality hay and was not putting on weight,though the diet appeared to be adequate calorifically.Reference has already been made to the generalmethod of conducting the experiments. The resultsof the first experiment are given in Fig. 2. There wasa very slow utilization of glucose, so that 9 hr. afterdosing there was still a significant amount remainingin the rumen liquor. This slow rate of glucose fer-mentation was also reflected by the almost insigni-ficant increase in lactic acid and total volatile fattyacids; at no stage was a peak observed. This was incontrast to the experiments of Phillipson & McAnally;(1942). The experiment was repeated at intervalsover a period of 2 months with similar results.

By this time the animal was in such poor conditionthat the diet was changed to one consisting of good-quality clover hay, and after 14 days on this diet theexperiment was repeated. As will be seen fromFig. 3 a considerable change had occurred. Theanimal was now able to ferment glucose rapidly, and

S. R. ELSDEN

the bulk of the glucose had disappeared within 6 hr.of dosing; there was also a significant increase involatile fatty acids, and lactic acid made its appearancein quantity. Parallel to the chemical analyses, obser-

trength

Fig. 2.

10 12 14 16 18 20 22 24Time hr.Fig. 3-

vations were made on the micro-organisms through-out the experimental period. For this purpose 5 ml.of rumen liquor were taken and mixed with 1 ml. of50% (v/v) formalin as a preservative; the sample

could then be examined at leisure. Prior ^microscopic examination, 1 ml. of double strengthGram's iodine was added in order to reveal the pre-sence of iodophile organisms. The samples takenprior to the introduction of the glucose were palebrown in colour, and microscopic examination, whilstrevealing the presence of Protozoa and numerousbacteria, amongst which could be seen large diplo-cocci and streptococci, showed no preferential stain-ing with iodine. On the other hand, samples taken1 hr. after dosing with glucose and treated withiodine were a brownish purple in colour, and underthe microscope could be seen many of the largediplococci and streptococci now stained a deep bluewith the iodine. Similar organisms were observedattached to pieces of plant debris, and it is clear thatthe presence of iodophile organisms does not neces-sarily imply that the polysaccharide store, revealedby the iodine, was produced from the material towhich the organism was attached. At this stage theProtozoa were filled with iodophile cocci: this hasalso been observed by Van der Wath (1942).

No attempt was made to count the numbers ofiodophile cocci owing to their uneven distributionin the fluid. Qualitatively, it may be said that thestaining reaction remained constant as long as glucosepersisted in the rumen liquor, and thereafter dimin-ished steadily, so that by the 18th hour it. had dis-appeared; but organisms, similar in size and shapeto those which had given the reaction, were to beseen in plenty. The sheep B.F., which had beenmaintained on the same poor-quality hay, but supple-mented with bran and crushed oats, showed a similarseries of changes to those recorded in Fig. 3, and themicroflora showed an intense iodine reaction within30 min. of introducing the glucose. The same typesof micro-organisms, as far as could be judged bymicroscopic examination alone, were present.

Forty-two days after changing the diet of W.F.,the animal's response to glucose was again tested.The results are given in Fig. 4. Maximum concen-tration of lactic acid was reached 1 hr. after dosing,earlier than in the previous experiment, and it hadentirely disappeared by the 6th hour. Glucose wasused much more rapidly than before, and had beencompletely utilized by the 2nd hour. There was alarge increase in the concentration of total volatilefatty acids, reaching a peak 3 hr. after dosing.

Fig. s shows the changes occurring in the indi-vidual volatile fatty acids during this experiment.The important feature is the increase in propionicacid as compared with acetic acid; this point isfurther emphasized by Table 4, in which the ratio ofacetic to propionic acid is given for the various stagesin the fermentation. Some, butyric acid was alsoproduced.

As in the previous experiment, rumen liquordrawn prior to the administration of glucose showedno marked coloration on treating with iodine, butsubsequent to the dosing it was coloured a dark

The fermentation of carbohydrates in the rumen of the sheep 57

5-Ui

oa1

— 4.c

E 3 .

6

JUs

I .

7dv-rt

Glucose

^ ^

110 12 14 16

Time hr.16 20 22 24

Fig- 4-

flora. Even before the addition of glucose, numerouslarge, oval, colourless, yeast-like organisms could beseen, similar to the schizqsaccharomycete describedby Quin (1943). Subsequent to the addition ofglucose, these organisms stained a deep brown withiodine, and many were observed to be in a state ofactive multiplication by binary fission. There werevery few iodophile cocci present, and it appeared as

2 3 4Time hr.

Fig. 5-

Table 4. The concentration of acetic and propionicacids, and the ratio of acetic to propionic in the rumenliquor of sheep W.F. at various times during the fer-mentation of glucose. The values recorded are ex-pressed as mM. acidj 100 ml. rumen liquor

Timehr.

01

36 2 5

17-524

Acetic

2-OI2-682'341-911-27o-8o

Propionic

0-47I'OI

1-791-320-580-25

Acetic/propionic

4 2 82 6 6

i"3i-452 1 93 2

brown. This reaction persisted up to the 6th hour andthen steadily diminished so that by the 18th hour ithad disappeared. Microscopic examination revealed

profound change had occurred in the rumen

36 48 60

Time hr.Fig. 6.

though they had been supplanted by the yeast-likeorganism.

With sheep W.F. it was possible to compare thein vivo fermentation of glucose with the in vitro, andto this end an experiment, similar to that recordedin Fig. 1, was performed. The results are given inFig. 6. The in vitro experiment reflects, on the whole,the in vivo. Both resulted in the production of acetic,propionic and butyric acids, with propionic acid pre-dominating; and in both lactic acid appeared as an

S. R. ELSDEN

intermediary. The fact that there were considerablequantities of propionic acid appearing in each typeof experiment is of importance, and will be com-mented on later when the origin of propionic acidis discussed.

The similarity of the chemical changes occurringboth in vivo and in vitro was paralleled by the simi-larity in the microflora which developed. In vitrothere were very few iodophile occi, but on the otherhand large numbers of the schizosaccharomycetedeveloped, and these stained a dark brown withiodine during the first 24 hr. of the experiment.Subsequently this ability was lost, presumably dueto the removal of the polysaccharide store by auto-fermentation. No ethanol was detected at any stage

The origin of acetic acid. It will have been ^ pthat propionic acid was the major component of thevolatile fatty acids produced in vitro from glucose,cellulose and lactic acid; whereas in vivo, on thenormal mixed diet, acetic acid predominated. Thequestion arose, therefore, was this discrepancy dueto the fact that, in vivo, a very mixed substrate wasfermented, and the nature of the substrate condi-tioned the nature of the end-products ? For it is wellknown that, even in pure culture studies, the pro-,portions of the end-products vary with the type ofsubstrate fermented. Or were the fatty acids pro-duced in vitro an artefact in the sense that the in vitroconditions were such as to favour the developmentof a completely different flora from that in vivo,

Table 5. Volatile fatty acids, in mM., produced during the in vitro fermentation of lactic acid, added assodium lactate, by rumen contents of the sheep. $oomg. (5*55 mM.) lactic acid were used in each experiment

Figures in parentheses refer to the volatile fatty acids in the control experiment incubated without lactic acid.Those sheep marked t were fitted with a permanent rumen fistula.

Exp.

12345

Sheep

FtAtCBG

Total volatileacids

5-20 (060)3:38 (0-69)3-53 (o-o)378 (091)2-95 (o-o)

Acetic

2 1 0 (0-34)0-73 (0-41)i-36 ( - )092 (053)028 (—)

Propionic

276 (019)2-37 (0-18)2-02 (—)1-37 (0-16)••52 ( - )

Butyric

034 (070)0-28 (010)015 ( - )1-49 (0-22)i-i5 ( - )

Timehr.

889088J8ii

141*

Table 6. Volatile fatty acids, in mM., produced during the in vitro fermentation of dried grass by rumencontents from the sheep. An excess of dried grass was used and at the termination of the experiment there wasa considerable residue

The figures in parentheses refer to the volatile fatty acids in the control experiment incubated without grass.Those sheep marked t were fitted with a permanent rumen fistula.

Exp.

12345

Sheep

CDtEGFt

Total volatileacids

505 (o-o)906 (o-o)978 (o-o)7-45 (o-o)9-65 (019)

Acetic

2-22 (—)474 (—)572 ( - )408 (—)508 (o-io)

Propionic

i-95 (—)3-0 ( - )264 (—)2-54 (—)3-72 (0-06)

Butyric

o-88 (—)i-32 (—)142 (—)0-83 (—)0-85 (003)

Timehr.

88J11295

141*95*

of the experiment, and thus, if the organism producedethanol, it must have been removed as fast as it wasformed. The similarity existing between the floradeveloping in vitro and that occurring in vivo is afurther indication of the physiological nature of thein vitro conditions.

The fermentation of lactic acid. The experimentson the fermentation of glucose described in the pre-vious section showed that lactic acid was producedand that it subsequently gave place to volatile fattyacids. A more detailed study of the lactic acid fer-mentation was made using the in vitro method. Theresults are given in Table 5. Acetic, propionic andbutyric acids were produced, and, as in the case ofthe other substrates studied, propionic acid pre-dominated.

with the result that different end-products wereproduced ?

The studies on the fermentation of glucose re-corded above are evidence in favour of the view thatthe conditions were physiological, and therefore anydifferences between the in vitro and in vivo fermen-tations were due primarily to the nature of the sub-strates fermented. To test this further a series ofexperiments was performed in which dried grasswas fermented in vitro. Table 6 records the resultsobtained. It will be seen that large amounts of volatilefatty acid were produced consisting of acetic, pro-pionic and butyric acids; acetic acid was the majorconstituent. These experiments, therefore, supportthe hypothesis that the predominance, in vivo, ofacetic acid is due to the nature of the substrate^At

The fermentation of carbohydrates in the rumen of the sheep

^ time it is realized that a much more detailedstudy must be made before this is firmly established.Such a study requires, in the first place, a muchmore detailed knowledge of the carbohydrate com-ponents of animal feeding stuffs than at presentexists, particularly of the so-called hemicellulosesand pentosans; and in the second place, a plentifulsupply of such compounds in a relatively pure form,with which to study their behaviour both in vivo andin vitro.

The origin of propionic acid. The observation thatpropionic acid was normally present in the rumen,and that it was produced during the in vitro fermen-tation of cellulose, glucose and lactic acid, particu-larly the last-mentioned, suggested at once thatmembers of the genus Propionibacterium were playinga part in the fermentations in the rumen. Attemptswere therefore made to isolate these organisms bothfrom rumen contents and from the in vitro fermen-tations.

The procedure was as follows. Tubes of Stephen-son's inorganic medium, pH 7-4, and containing inaddition 0-4% Difco yeast extract (w/v) and 1%(w/v) sodium lactate, were inoculated with a drop ofthe liquid under examination. Incubation was at 3 8°for 10 days under an atmosphere of N2 containing5 % COa. At the end of this time there was a heavygrowth at the bottom of the tubes. Two or threesubcultures were made on the liquid medium, andfinally the cultures were plated out on the samemedium solidified with 2% agar. After replating anumber of times pure cultures were obtained of whatwere considered to be members of the genus Pro-pionibacterium. They had the following properties:small, Gram-positive cocci, sometimes in pairs;catalase-positive; colonies 1-2 mm. diameter, dome-shaped and cream-coloured; anaerobes, with goodgrowth at the bottom of yeast extract, lactate agarstabs, but with no surface growth; fermented lacticacid and glucose with the production of propionicand acetic acids and CO2. These properties are con-sistent with the view that these organisms are mem-bers of the genus Propionibacterium. In all purecultures were isolated from the rumen contents ofthree sheep and an ox, and from the in vitro fermen-tations of cellulose, glucose, lactic acid and driedgrass by the rumen liquor from four other sheep.The isolation of these organisms makes it highlyprobable that they are responsible for the productionof the propionic acid found in the rumen, and thatpart of the acetic acid must also be ascribed to them.

DISCUSSION

As stated in the Introduction, the present communi-cation deals with three different problems: (1) thenature of the volatile fatty acids in the rumen andthe quantities of each present; (2) the organismsresponsible, either directly or indirectly, for the pro-

59duction of volatile- fatty acids and the compoundsfrom which they are produced; (3) the effect of dieton the composition of the rumen microflora. A com-plete answer has been obtained to problem (1), anda partial answer has been obtained to problems (2)and (3).

Acetic, propionic and butyric acids were invariablyfound in the rumen, and frequently a higher acid,probably a valeric isomer. The question of formic isstill undecided. It could not be found on the occa-sions it was sought for, but a failure to find it on afew occasions does not imply that it is never present.It must also be pointed out that the methods availablefor the detection and estimation of this compoundare unreliable, due to a lack of specificity. To se.ttlethis point new analytical procedures are needed, andthe two most promising appear to be (1) the utiliza-tion of formic hydrogenlyase, an enzyme found inE. coli and certain other members of the entero-bacteriaceae which splits formic acid quantitativelyinto H2 and CO2; this method has already been usedby Woods (1936). (2) The method for the separationof the volatile fatty acids, introduced by Schicktanz,Steele & Blaisdell (1940), based on the fractionaldistillation of the azeotropes of volatile fatty acidsand aryl hydrocarbons.

Of the vast numbers of micro-organisms found inthe rumen only three species have been definitelyassociated with the fermentation of carbohydrate.They are: (1) the untyped, iodophile coccus, isolatedin pure culture by Van der Wath (1942), and observedbut not isolated by Quin (1943) and by the author;(2) the yeast-like organism, first recorded by Quin(1943) and named by him Schizosaccharomyces ovis,which is also recorded in the present paper; (3) un-typed members of the genus Propionibacterium, whoseisolation from rumen contents and whose role in-thebreakdown of carbohydrate in the rumen are nowreported for the first time.

The iodophile coccus was shown by Van der Wath(1942) to play a part in the breakdown of glucose andstarch; Quin showed that it was connected with thefermentation of glucose, and similar, observations arenow recorded. The precise part played by thisorganism is still a matter of conjecture, and theproducts produced from these compounds are un-known; Van der Wath (1942) found that acid wasproduced from both, but this acid has not as yet beenidentified. As the organism is a coccus, it is probablethat lactic acid is one of its end-products. It is certainthat the organism converts part at least of itssubstrate into a polysaccharide resembling starch.Further study of this organism is a problem for thefuture.

A detailed study of the schizosaccharomycete isalso required. Quin (1943) associated this organismwith the very rapid fermentation that occurs whensheep are fed on a diet rich in soluble sugars. Healso showed that the organism could be partiallyseparated by fractional centrifugation, and that sus-

6o S. R. ELSDEN

pensions of the organism in bicarbonate buffer fer-mented glucose very rapidly as judged by the rateof gas formation. More than this cannot be said atthe moment, for no information was forthcoming onthe nature of the other substances formed, but fromthe experiments recorded in the present paper andfrom preliminary experiments with suspensions pre-pared by fractional centrifugation, it seems likelythat ethariol is not produced. The isolation of thisorganism in pure culture and a detailed study of itsmetabolism is a problem of immediate importance,though it is probable that useful information couldbe obtained from suspensions prepared from rumenliquor by fractional centrifugation.

The work described in the present paper suggestsmost strongly that unidentified members of the genusPropionibacterium. are functional members of therumen flora. The evidence can be summarized asfollows: (i) members of this genus have been isolatedfrom the rumen contents of a number of sheep andan ox; (2) lactic acid appears during both the in vivoand the in vitro fermentation of glucose, but givesplace to volatile fatty acids, mainly propionic andacetic acids, though under some conditions more orless butyric acid may be formed. The ability toconvert lactic acid into propionic and acetic acids isdiagnostic of the genus Propionibacterium. This evi-dence makes it certain that these organisms areresponsible for the formation of the propionic acidwhich is invariably found in the rumen. The com-ponents of the diet from which propionic acid iseither directly or indirectly produced can be referredto only in general terms at this stage. The solublesugars of the diet are sources of propionic acid;cellulose is also a possibility, for the in vitro experi-ments with this substance suggest that the propionicacid bacteria play a part in its breakdown; but it stillremains to be shown that the in vivo fermentationof cellulose resembles the in vitro.

Under the conditions used- the fermentation ofcellulose appears to be a two-stage process, a primarybreakdown of cellulose by the specific organisms,followed by a secondary fermentation of the productor products thus formed by the propionic acid bac-teria. The evidence for this is that members of thegenus Propionibacterium were isolated from the invitro fermentations of cellulose, and they showed noability to ferment cellulose. It is possible that themethod of isolation was such as to cause the loss ofthis property; the former hypothesis seems, however,to be more reasonable. If it be accepted that thefermentation of cellulose to volatile fatty acids, awidespread phenomenon in nature, involves two ormore types of organisms, then the wide variety ofvolatile fatty acids produced by allegedly pure cul-tures of cellulose fermenters is explained, for thenature of the end-products will be conditioned bythe type of secondary organisms. In a similar mannerthe breakdown of soluble sugars in the rumen appearsto be a two-stage process, the first group of organisms

producing lactic acid, the second group, ^which are the propionic acid bacteria, ferment lacticacid to volatile fatty acids.

The presence of propionibacteria in the rumen isconsistent with the older work on this group. Theirnormal habitat was considered to be dairy produce,and the problem was to explain how they arose there.Burri (1911) demonstrated that the faeces of cowscontained large numbers of propionic acid bacteria;and Thoni (1906)1 demonstrated their presence innatural rennet, and, what is more important, fromthe abomasum from which the rennet was made.This work established the fact that this group oforganisms do occur in the alimentary canal of thecow, and in consequence their occurrence in therumen is not unexpected.

The origin of acetic acid has not as yet been satis-factorily explained. Part, at least, must arise throughthe activities of the propionic acid bacteria, but thefact that, in vivo, some 55-75 % of the total volatilefatty acid of the rumen liquor is acetic acid cannotbe readily explained in terms of these organismsalone. The experiments with dried grass suggest thatthe diet conditions the composition of the flora,Which in turn conditions the nature of the end-products; and it may well be that the acetic acidarises directly from polysaccharides such as hemi-celluloses and pentosans, which, as was shown byMcAnally (1942), are more readily fermented in therumen than cellulose. The protein of the diet mayalso give rise to some acetic acid.

The presence of propionic acid in the rumen havingbeen discussed, it is now necessary to consider theorigin of butyric acid. Like propionic acid it is pro-duced from glucose and lactic acid; it was not formedin significant amounts from cellulose under in vitroconditions. The fact that it is produced from lacticacid provides a possible clue to the type of organismresponsible. Recently, Barker & Haas (1944) showedthat the anaerobic organisms, isolated from humanintestinal contents by Lewis & Rettger (1940) andby King & Rettger (1942), fermented lactic acid toa mixture of acetic and butyric acids along withtraces of a higher fatty acid, probably caproic acid.It seems possible that a similar type of organismmay function in the rumen.

So far no mention has been made of the methanewhich is produced in the rumen. Formerly it wasascribed, without convincing evidence, to the actionof the cellulose fermenters, but the recent work ofBarker (1936, 1939) has demonstrated conclusivelythat methane is produced by the reduction of CO2,with the simultaneous oxidation of primary or se-condary alcohols, or the lower fatty acids, dependingon the type of organism. The conditions in therumen, anaerobiosis, a plentiful supply of CO2 andvolatile fatty acids, are ideal for the growth of thisgroup of organisms. So far no search has been madefor them, but, if a complete picture of the events inthe rumen is to be obtained, a detailed study of this

The fermentation of carbohydrates in the rumen of the sheep 61

i will have to be made, and the classical work ofBarker will be an invaluable guide.

Finally comes the question of the effect of diet on,the composition of the rumen flora. Van der Wath(1942) demonstrated by direct counts that thenumber of bacteria in the rumen is conditioned bythe composition of the diet. Thus when teff hay,consisting largely of fibre, was fed, the total count,as compared with animals on a mixed diet, was low;when this was supplemented by protein or urea, thenumber of micro-organisms was increased—thus Nwas one of the limiting factors. When the diet wasfurther supplemented by starch or molasses, thecount was again increased, and when a third supple-ment was given in the form of bone meal or inorganicphosphate, the total bacterial count was again raised.The technique of total counts, whilst providinguseful information, is limited in its scope; thus littleinformation of the number of species present, or thenumbers of each species is obtained, for unless anorganism possesses a highly characteristic shape orspecific staining reaction, it is impossible to dis-tinguish one species from another. The method ofviable counts is also of little use here, and it is clearthat a completely new approach is needed.

Quin (1943) established that the composition ofthe rumen microflora is far from constant, and isconditioned by the nature of the diet. He employedwhat may be termed the biochemical method ofassaying the flora. He studied the in vivo and thein vitro fermentation of glucose by the rumen liquorfrom sheep on different diets; the rate of gas forma-tion was taken as an index of the rate of glucosefermentation. Sheep fed on veld grass hay, a diet oflow nutritional value, fermented glucose very slowly;whereas, when the diet was changed to lucerne, eitherfresh or as hay, the rate of fermentation was speededup. Coincident with the change in diet was a changein the composition of the microflora. First, the iodo-phile coccus established itself as a dominant organismonly to be succeeded by the schizosaccharomycete.This latter organism was associated with the mostrapid type of fermentation. These observations havebeen fully confirmed by the present work, though,instead of following the rate of gas evolution, the rateof glucose fermentation was measured directly, andthe products formed determined.

It is clear from all this work that a diet deficient insoluble sugars or starch does not favour the develop-ment of a flora capable of using them; it wouldbe curious were it otherwise: and therefore when

glucose is introduced into the rumen of an animal,fed on such a diet, it is fermented very slowly. Fromthe work of Van der Wath (1942) it is evident thata starch-rich diet produces a rumen flora in whichthe iodophile coccus is a dominant organism, whereasa diet rich in soluble sugars, whilst resulting in atemporary dominance of the iodophile cocci, ulti-mately leads to the schizosaccharomycete establishingitself, accompanied by the almost complete dis-appearance of the cocci.

The importance of further studies of the factorscontrolling the composition of the microflora of therumen cannot be over-emphasized, for it is certainthat a vigorous microflora is one of the conditionsessential for the health of the animal. There are alsobroader issues, for such studies open the way to theinvestigation of bacterial ecology, a field which hasbeen as yet poorly developed.

SUMMARY

1. Acetic, propionic and butyric acids are themain volatile fatty acids in the rumen of the sheep.Acetic acid accounts for 55-75 % of the total.

2. Cellulose, glucose and lactic acid are rapidlyfermented in vitro by rumen contents with the pro-duction of acetic, propionic and butyric acids. Pro-pionic acid is the major component in all cases: verylittle if any butyric acid is formed from cellulose.

3. The in vitro fermentation of glucose closelyresembles the in vivo.

4. The in vitro fermentation of dried grass yieldsthe same three acids, but with acetic acid pre-dominating.

5. Members of the genus Propionibacterium havebeen isolated from the rumen, and evidence is pre-sented to show that these organisms are responsiblefor the production of the propionic acid found inthe rumen.

6. The dietary history of the animal is shown toinfluence the rate at which glucose is fermented inthe rumen, and the composition of the rumen micro-flora.

The author wishes to express his most gratefulthanks to Sir Joseph Barcroft, F.R.S., Prof. A. C.Chibnall, F.R.S., and Dr Marjory Stephenson,F.R.S., for their interest and encouragement in thiswork, and to his. colleague, Dr A. T. Phillipson, forhis helpful advice and co-operation throughout.

REFERENCES

BAKER, F. (1942a). Nature, Lond., 149, 220.BAKER, F. (19426). Nature, Lond., 150, 479.BARCROFT, J., MCANALLY, R. A. & PHILLIPSON, A. T.

(1944a). J. Exp. Biol, 20, 120.BARCROFT, J., MCANALLY, R. A. & PHILLIPSON, A. T.

£19446). Biochem. J. 38, iii.

BARKER, H. A. (1936). Arch. Mikrobiol. 7, 404.BARKER, H. A. (1939-40). Ant. van Leeuwenhoek, 6, 201.BARKER, H. A. & HAAS, V. (1944). J. Bad. 47, 301.BURRI, R. (1911). Ta'tigkeitsber. d. Schweiz. Milchwirtsch.

u. Bakt. Anstalt Bern-Liebefeld, p. 467. (Quoted fromvan Niel, C. B. (1928), Thesis, Delft, p. 17.)

62 S. R. ELSDENELSDEN, S. R. (1945). Biochem. J. (in the Press).FRIEDEMANN, T. E. (1938). J. Biol. Chem. 123, 161.FRIEDEMANN, T. E. & KENDALL, A. I. (1929). J. Biol.

Chem. 82, 23.FRIEDEMANN, T. E. & KLAAS, R. (1936). J. Biol. Chem.

" 5 . 47-HANES, C. S. (1929). Biochem. J. 23, 99.KING, J. W. & RETTGER, L. F. (1942). J. Bact. 44,

301.LEWIS, K. H. & RETTGER, L. F. (1940). J. Bact. 40,

287.MCANALLY, R. A. (1942). Biochem. J. 36, 392.MANGOLD (1929). Handbuch der Ernahrung und des

Stoffviechsels der Landwirtschaftlichen Nutztiere, 2.MUSICANT, L. & KASZUBA, F. J. (1939). J. Amer. Chem.

Soc: 6i, 2974.OWEN, E. C , SMITH, J. A. B. & WRIGHT, N. C. (1943).

Biochem. J. 37, 43.PEARSON, R. M. & SMITH, J. A. B. (1943 a). Biochem. J.

37. 142.

PEARSON, R. M. & SMITH, J. A. B. (19436).37. 148.

PEARSON, R. M. & SMITH, J. A. B. (1943c). Biochem. J.37. 153-

PHILLIPSON, A. T. (1942). J. Exp. Biol. 19, 186.PHILLIPSON, A. T. & MCANALLY, R. A. (1942). J. Exp.

Biol. 19, 199.POCHON, J. (1934). C.R. Soc. Biol., Paris, 199, 983.POCHON, J. (1935). Ann. Inst. Pasteur, 55, 676.QUIN, J. I. (1943). Onderstepoort J. Vet. Sci. 18, 91.SGHICKTANZ, S. T., STEELE, W. I. & BLAISDELL, A. C.

(1940). Industr. Engng Chem. (Anal, ed.), \2, 320.THONI, J. (1906). Landw. Jahrb. Schweiz, pp. 1-62.

(Quoted from van Niel, C. B. (1928). Thesis, Delft,P- I3-)

VAN DER WATH, J. G. (1942). Thesis, University ofPretoria.

WEGNER, M. I., BOOTH, A. N., ELVEHJEM, C. A. & HART,E. B. (1940). Proc. Soc. Exp. Biol., N.Y., 45, 769.

WOODS, D. D. (1936). Biochem. J. 30, 515.