Original article

Fermentation of green alga sea-lettuce (Ulva sp)and metabolism of its sulphate by human colonicmicrobiota in a semi-continuous culture system

M Durand P Beaumatin B Bulman A Bemalier

JP Grivet 2 M Serezat G Gramet M Lahaye 3

1 Laboratoire de nutrition et sécurité alimentaire, Inra-CRJ, 78352 Jouy-en-Josas cedex;2 Centre de biophysique moléculaire, CNRS et université d’Orléans,

IA, avenue de la Recherche, 45071 Orléans cedex 02;3 Laboratoire de biochimie et technologie des glucides, Inra, BP 527,

44026 Nantes cedex 03, France

(Received 17 October 1996; accepted 28 January 1997)

Summary ― The green alga, sea-lettuce (Ulva sp), could be considered as a new source of dietaryfibre. Ulva, however, contains high levels of sulphate, part of which is chemically bound in solublepolymers (ulvan). The purpose of this study was to assess the fermentation characteristics and sulphatemetabolism of Ulva and ulvan by human faecal bacteria fermentation system using a semi-continu-ous fermenter. Ulva and ulvan were poorly fermented, even after adaptation of the microbiota. Only16.6% and 8.9% of Ulva and ulvan organic matter, respectively, were recovered as short chain fattyacids. Nevertheless, 40% of the sulphate in Ulva was dissimilated to sulphide by sulphate-reducingbacteria. Supplementation of Ulva with more fermentable polysaccharides, such as algal xylan andresistant starch, though decreasing the ammonia production originating from Ulva protein degrada-tion, did not significantly reduce the sulphide levels. It is postulated that unless crude Ulva is desul-phated, its daily consumption at a level of 20 g of dry product could stimulate colonic microbialsulphate reduction, which may have detrimental effects for the host.

in vitro fermentation / Ulva / seaweed / colonic bacteria / sulphate reduction

Résumé ― Étude en culture semi-continue de la fermentation de l’algue verte, laitue de mer (Ulvasp), et du métabolisme de ses sulfates par les microorganismes du côlon humain. L’algue verte,laitue de mer (Ulva sp), peut être considérée comme une nouvelle source de fibre alimentaire. Cepen-dant, Ulva contient une teneur élevée en sulfates dont une partie est liée chimiquement à un ensemblede polymères solubles (ulvane). L’objet de cette étude a été d’étudier les caractéristiques fermentaires

* Correspondence and reprintsTel: (33) Ol 34 65 27 10; fax: (33) 01 34 65 23 11.

et le métabolisme du sulfate d’ Ulva et d’ulvane en utilisant un système de fermentation semi-continu.Ulva et ulvane se sont montrés peu fermentescibles, même après adaptation de la microflore. Seuls16,6 % de la matière organique d’ Ulva et 8,9 % de celle d’ulvane ont été retrouvés sous formed’acides gras à chaîne courte. Néanmoins, 40 % des sulfates contenus dans Ulva ont été transformésen sulfure par les bactéries sulfatoréductrices. La supplémentation d’ Ulva par des glucides plus fer-mentescibles tels que du xylane d’algue et de l’amidon résistant, bien que diminuant la productiond’ammoniac due à la dégradation des protéines d’ Ulva, n’a pas limité significativement la formationde sulfure. En conclusion, la consommation journalière de 20 g d’ Ulva brute sèche, pourrait accroîtrela sulfatoréduction par la flore colique, ce qui pourrait avoir des conséquences défavorables pourl’hôte, à moins que ce produit ne soit préalablement désulfaté.

fermentation in vitro / Ulva / algue / flore colique / sulfatoréduction

INTRODUCTION

There is considerable evidence to supportthe view that fermentation of dietary fibresby anaerobic bacteria in the human coloncontributes to the prevention of colonic dis-eases. Short chain fatty acids (SCFA), whichare the principal end products of fermenta-tion, have a variety of recognized physio-logical and clinical properties (Cummings,1995). However, wide ranges of fer-mentability exist among different types ofdietary fibres, and the action of an individ-ual fibre depends on the extent of its break-down, as well as on the pattern of its fer-mentation end products. For example,beneficial effects such as increase in stool

output, dilution of colonic contents and pro-duction of distal colonic SCFAs are mainlyassociated with a low extent, or a slow rateof fermentation (Edwards, 1995). In thisrespect, the green alga, sea-lettuce (Ulvasp), could be considered as a new interestingsource of dietary fibre. This organism con-tains about 30% total dietary fibre of which16.5% are soluble polysaccharides (Lahaye,1991 Recent investigation has shown thatUlva and its soluble fibres, referred to asulvan, were poorly fermented in vitro inbatch culture with faecal bacteria (Bobin-Dubigeon et al, 1996). Their particularchemical structures were suggested toexplain this resistance to fermentation. How-

ever, such short-term studies do not takeinto account the potential capacity of intesti-nal bacterial communities to adapt to breakdown the polysaccharides (Edwards, 1995).It has been shown in rat experiments thatshort ( week) or longer-term adaptations(up to 12 weeks) were necessary to achievethe greatest digestibility or rate of fermen-tation of specific fibres such as gums(Walter et al, 1986; Tulung et al, 1987;Brunsgaard et al, 1995) or fibres fromcanned peas (Monsma and Marlett, 1996). Inhumans, when gum arabic was included inthe diet for 18 days, the proportion of thefaecal flora able to degrade this substrateprogressively rose from 6.5 to more than50%. After the fibre was withdrawn fromthe diet, the proportion of gum arabic fer-menting bacteria returned to the level presentbefore ingestion (Wyatt et al, 1986). Morerecently, the necessity of adaptation ofhuman subjects to xanthan gum, for theirfaecal microflora to achieve maximum

hydrogen and SCFA production, was alsodemonstrated (Daly et al, 1993). Therefore,longer-term studies appear necessary toinvestigate whether adaptation of the floraoccurs with this type of seaweed.

Another concern regarding Ulva is itshigh sulphate content, which can reach 17%of its dry matter (Bobin-Dubigeon et al,1996). Part of this sulphate is chemicallybound as in ulvan, which is a sulphated glu-

curonorhamnoglycan, constituting between12 and 15% of the algal dry matter (Micheland Macfarlane, 1996), and containing about14% sulphate (Ray and Lahaye, 1995). Thisattribute could have significant conse-quences on microbial metabolism in thecolon and potential health implications. Thesupply of free sulphate to the colon resultingeither from a high sulphate intake, eg,> 5 mmol/d (Florin et al, 1991) or from thein situ microbial degradation of sulphatedpolymers (Gibson et al, 1988a, 1991) maystimulate the growth and activity of sul-phate-reducing bacteria (SRB) (Christl etal, 1992). These organisms reduce sulphateto sulphide. When sulphate is abundant, thisreaction can represent an important route ofhydrogen (H2) utilization in situ in the colonand may interact with other microbial path-ways of H2 disposal, namely methanogene-sis and reductive acetogenesis (acetate for-mation from C02 reduction by H2) (Durandet al, 1995). In addition, hydrogen sulphideis toxic towards mammalian cells and it hasbeen suggested that sulphide is a possiblecofactor in the initiation and/or maintenanceof the inflammatory bowel disease, ulcera-tive colitis (Gibson et al, 1991; Roediger etal, 1993). Consequently, knowledge of thefate of the sulphate associated with sea-lettuce is important to predict the potentialhealth implications following its long-termconsumption.

The purpose of the present study was toassess the effect of adaptation on the fer-mentation characteristics of Ulva and ulvan

by human faecal bacteria, using a semi-con-tinuous fermentation system. Emphasis wasplaced on the effect of these products onsulphate metabolism, and in order to ascer-tain more precisely the activities of SRB,the putatively specific inhibitor of sulphatereduction, sodium molybdate, was used. Wealso attempted to clarify if previous adap-tation of the microflora to Ulva would affectits reductive acetogenic and methanogenicactivities.

MATERIALS AND METHODS

Materials

Ulva sp, produced by Algues de Bretagne(France) as Laitue de mer was ground to passthrough a 3 mm screen. It contained 90.4% drymatter (DM) and (% DM): minerals, 19.7; totalnitrogen, 4.9; sulphate, 11.6. Crude ulvan wasprovided by CEVA (Pleubian, France) and fur-ther processed by precipitation by Inra (Nantes,France). It contained (% DM): minerals, 20.8;sulphate, 15.8. The red seaweed Palmaria pal-mata was also processed by CEVA in order toobtain a carbohydrate enriched product (70%DM), mostly composed of xylan, and referredto in this paper as algal xylan. It contained 93.3%DM and (% DM): minerals, 22.8; sulphate, 11.7.Sulphate was formed from sulphuric acid addedduring processing. The sulphate determinationson the above products were performed by Inra,Nantes (Lahaye and Axelos, 1993). Resistantstarch was provided by Inra (Nantes), and con-sisted of retrograded high amylose maize starch(Faisant et al, 1993). ,

Fermentation system

The semi-continuous culture system used in thisexperiment was adapted from the ’Rusitec’designed by Czerkawski and Breckenridge (1977)for rumen fermentation studies, consisting offour independent 1 L capacity vessels. Fresh fae-cal samples were collected from two female sub-jects referenced as subjects A and B. Methanewas never detected in the exhaled breath of sub-

ject A, whereas subject B was a methane excre-tor, but only exhaled moderate levels of methane(about 20 ppm). Freshly voided faeces werestored under anaerobic conditions and processedwithin 48 h. All handling procedures were doneunder anaerobic conditions (under a gas phaseof 100% N2)’

For each vessel, 100 g faeces were homoge-nized in 500 mL C02-saturated mineral solutionsimilar to that described by Michel et al (1996),except for Na2so4l which was reduced by 50%(0.05 g/L). The faecal slurries were then strainedthrough four layers of surgical gauze, and thestrained fluid was introduced into the fermenta-tion vessels. The solid residues were placed intonylon bags (160 x 80 mm, 150 pm pore size)together with 5 g cut pre-fermented straw. This

bag was introduced into the vessel and remainedthere throughout the experiment in order to per-mit bacterial attachment. The volume in the ves-sel was then adjusted to 1 L with a nutritive basalmedium. This contained the mineral solutiondescribed above, together with a mixture oforganic compounds (Sigma, France) (g/L): porcingastric mucin, 2; pancreatin, 1; casein, 2; cys-teine, 0.2; haemin, 0.01; peptone, 0.01; bile salts(50% sodium cholate), 0.05; urea, 0.18; xylan,0.5; pectin, 0.5; amylopectin, 0.5; arabinogalac-tan, 0.5; lintner starch, 0.5 and vitamins (mg/L):riboflavin, 0.5; biotin, 0.012; cyanocobalamine,0.005; folic acid, 0.012; nicotinic acid, pyridox-ine HCI, thiamine HCL and pantothenate, I mg/Leach.

Each vessel was infused with the basalmedium (BM) at an approximate rate of 1 L/d.When Ulva was tested, nylon bags containing15 g Ulva were introduced into the vessels andremoved 48 h later. The bags were washed withthe mineral solution, and the washings were putback into the fermentation vessels. When algalxylan and resistant starch were used to supple-ment Ulva, they were injected into the vesseltwice per day as a suspension in distilled water(3 g/20 mL and 2.5 g/ 20 mL for algal xylan andresistant starch, respectively). When ulvan wastested, the substrate was solubilized in distilledwater (2.5 g/20 mL) and injected twice daily intothe fermenters. Sodium molybdate and 2-bromoethanesulfonic acid, when necessary, were intro-duced into the basal medium and infused con-

tinuously.

Experiments

Four experiments (studies 1 to 4) were con-ducted to study Ulva fermentation and sulphatemetabolism and one (study 5) to study ulvan fer-mentation and sulphate metabolism. The inocu-lum from subject A (non-methanogenic) wasused in studies 1, 2 and 5 and the inoculum fromsubject B was used in studies 3 and 4. The exper-iments were conducted according to the follow-ing experimental design:

Study 1: fermentation of the basal culturemedium without (one vessel) or with addition ofUlva (one vessel) was followed for 12 days.

Study 2: the aim in this experiment was to con-centrate on sulphate metabolism and its effecton reductive acetogenesis. Two vessels were first

adapted to Ulva for 10 days, then one of the ves-sels received sodium molybdate (20 mM), andthe experiment was run for a further 10 days.

Study 3: Ulva fermentation and sulphatemetabolism were studied for 7 days in the pres-ence (two vessels) or absence (two vessels) ofthe methanogenesis inhibitor 2-bromoethane-sulfonic acid (BES) (20 mM), since the inoculumB was putatively methanogenic. The former ves-sels (Ulva alone) also served as a control forstudy 4.

Study 4: the effect of Ulva supplementation withalgal xylan (two vessels) and resistant starch (twovessels) on the pattern of fermentation and sul-phate metabolism was investigated for 7 days.One vessel of each group received sodium

molybdate (15 mM) from the beginning of theincubation.

Study 5: fermentation and sulphate metabolismin the basal culture medium alone (two vessels)or supplemented with ulvan (two vessels) werestudied for 7 days. One vessel from each groupreceived sodium molybdate as indicated above.

Bacteriological and chemical analyses

Bacterial enumerations

Total anaerobes, methanogenic archaea, sulphate-reducing and acetogenic bacteria were enumer-ated in experiments 2 and 3 in the original faecalsamples (donors A and B, respectively) and inthe fermentation vessels at the end of the incu-bations, following the methods described by Doréet al (1995a) and Bernalier et al (1996a).

Chemical analyses on spent culture media

Effluents from the fermenters were collected

daily. SCFA were analysed by capillary GC afterconversion to tertiary butyldimethylsilyl derivates(Richardson et al, 1989). Ammonia nitrogen(NH3-N) and protein nitrogen (protein-N) weredetermined on three consecutive days at the endof studies 1, 2 and 4. NH3-N was measured bytitration after displacement by magnesium oxide(0.1 M), and protein-N was precipitated bytrichloracetic acid (10% final concentration),centrifuged and determined by the Kjeldahlmethod. Sulphate was determined at intervals in

all experiments except for study 1. Sulphate wasassayed by turbidimetry after treatment with TCAand in the presence of BaC12 and polyethyleneglycol (Bo S6rbo, 1987).

Analyses on the liquid phaseof the fermenters

Samples were taken from the fermenters everymorning just before the change of feed bag ofUlva, or just before the injection of ulvan. Onthese samples, pH was recorded daily and sul-phides were determined at intervals in studies 2and 3 and daily in studies 4 and 5. Sulphideswere assayed by the colorimetric method of Cline(1969) after precipitation of sulphides in 10%(wt/vol) zinc acetate solution (Gibson et al,1988b). In addition, on two different days afteradaptation, pH was recorded at different timesup to 7 h after feeding the substrates.

Collection and analysis of gases

Fermentation gases corresponding to each vesselwere collected daily in a glass cylinder and vol-ume was measured by liquid displacement.Except for studies 1 and 2, the gas was previ-ously bubbled through zinc acetate (10% wt/vol)to trap any gaseous sulphide. Samples of gaswere then analysed for H2, CH4 and C02 by GC(De Graeve et al, 1994) and trapped sulphidewas determined as above.

Measurement of reductiveacetogenic activity by 13C NMR

At the end of studies 2 and 3 the microbiotasfrom each vessel were collected by centrifugation(27 000 g for 30 min at 4 °C) under N2 gas phase.Aliquots were incubated for 16 h in 120 mL peni-cillin flasks in the presence of NaH13C03 with H2as gas phase. The effect of sulphate addition wasassessed by adding 7 mM Na2 S04 to half of theflasks. Incorporation of !3C02 into acetate wasmeasured by !3C-NMR as described by DeGraeve et al (1994).

Statistical analysis

Experimental values were compared by ANOVAand with the Newman-Keuls multiple range test(STATITCF software; ITCF, Paris).

RESULTS

Bacterial enumeration

In the original faeces, the number of totalanaerobes and of reductive acetogens waslower with donor B while those of sulphate-reducing bacteria (SRB) were similar forboth individuals (table I). In agreement withbreath tests, methanogens were not

detectable in faeces from donor A, and theirnumbers were just significant in those fromdonor B.

In the fermentation vessels, despite theten-fold dilution of faeces, the number oftotal anaerobes, SRB and acetogenic bac-teria remained close to that of the originalfaeces, with both inocula. However, thenumber of methanogens from donor B wasgreatly decreased compared to the originalfaeces. Addition of sodium molybdate low-ered the SRB population and that of BESpractically eliminated the methanogens(table I).

’

End products of fermentation

Results concerning the production of SCFAin the presence of Ulva that were obtained instudies 1 and 2 with inoculum A did notshow any difference with those obtained in

study 3 with inoculum B. The mean pro-duction after 5 days adaptation was70.5 ± 2.1 and 70.9 ± 2.0 mmol/day fordonors A and B, respectively. Therefore,data obtained with both inocula were pooled.

Effect of adaptation on SCFA production

Daily production of SCFA during 12 daysadaptation to the basal medium was steadyfrom day 4 to day 9, before decreasing by10% (fig 1). During the same period, Ulvaaddition resulted in a gradual rise in SCFA,which levelled off after day 5 of the incu-bation, showing that about 5 days wererequired for full adaptation of the flora to

Ulva (fig 1 Consequently, the durations ofstudies 3-5 were reduced to 7 days.

Supplementation of Ulva by algal xylan(XYL) and resistant starch (RS) greatlyincreased total SCFA production from thefirst day of incubation. Then, it remainednearly constant with resistant starch andincreased slightly with xylan (fig 1). Whenulvan was the substrate, SCFA productionfirst decreased by 15% from day 1 to day 3and then stabilized at a level only slightlyhigher than that of the basal medium alone(fig 1 ).

Balance of end products

The balance of end products was calculatedas the mean of the production on days 5, 6and 7 (table II). Compared to the basalmedium, Ulva addition significantlyincreased all SCFAs and had only a smalleffect on their molar proportion (table II):there was a small decrease in acetate ratiosand a slight but significant rise in valerateand the branched chain fatty acids (BCFA)

isobutyrate and isovalerate. Calculation ofSCFA production owing to Ulva alone (Ulvavessels minus basal medium vessels)showed that (weight by weight), 9% of thecrude product or 16.6% of its organic mat-ter were recovered as SCFA.

The increase in SCFA effected by car-bohydrate supplementation was significantlymore marked with algal xylan than withresistant starch. Compared to Ulva alone,algal xylan elevated the molar proportionof propionate and lowered that of butyrate,whereas resistant starch slightly increasedacetate. Both carbohydrates decreased therelative proportions of isovalerate and valer-ate (table II). In addition, resistant starchresulted in the formation of some longerchain fatty acids such as caproic and hep-tanoic acids (1.7 and 0.7 mmol/day, respec-tively). SCFA formed from algal xylan andresistant starch alone (carbohydrate supple-mented vessels minus Ulva vessels)amounted to (weight by weight) 34 and 37%of the crude products, and to 47.9 and 41.4%of their organic matters, respectively.

Compared to the basal medium, the addi-tion of ulvan increased slightly (< 10%),although significantly, total SCFA produc-tion, and this increase was entirely due to arise in acetate (table II). The amount ofSCFA produced from ulvan (ulvan vesselminus basal medium vessel) representedonly 6.4% of crude ulvan and 8.9% of itsorganic matter.

Carbon dioxide (C02) was the main fer-mentation gas produced. Compared to thebasal medium, addition of Ulva and its fur-ther supplementation with algal xylan orresistant starch greatly increased the vol-ume of C02 released (from about 200 mL upto 800 mL/day). Ulvan addition also resultedin a significant increase in C02 (+ 34%)compared to the basal medium. Hydrogenwas detected at very low levels except onthe last two days of incubation in thepresence of resistant starch (mean:2.2 mmol/day), and following sodiummolybdate addition (see below). Only traceamounts of methane were detected in thevessels from studies 3 and 4, which wereinoculated with faeces from donor B. There-

fore, the addition of BES in study 3 had noeffect at all on the fermentation end productsfrom Ulva (results not shown).

Both protein- and ammonia-N outputwere increased by Ulva addition. Algalxylan and resistant starch supplementation ofUlva increased protein-N but decreasedammonia-N (table II).

The initial (before feeding) pH of the fer-menters receiving the basal medium, Ulvaalone and ulvan were not significantly dif-ferent and remained near neutral (6.8-6.9)(table II). When both carbohydrates wereadded, the initial pH valves were signifi-cantly lower (- 0.1 unit). After feeding, thepH rose briefly with Ulva, decreased within2 h with xylan and remained stable withresistant starch (fig 2).

Sulphate metabolismand the effect of sodium molybdate

Sulphate metabolismin the absence of molybdate

The basal medium contained very low lev-els of sulphate (0.35 mmol/L). Neverthe-less, sulphides were detected in culture ves-sels at a higher concentration than that ofsulphate, demonstrating that some sulphidesmust have originated from other sources.

Ulva feeding increased the daily sulphateinput by 15 mmol. However, irrespectiveof the inoculum, sulphate recovery in theeffluent was low, while sulphide concen-trations in the vessels rose to about6 mmol/L (6.0 ± 0.6 for inoculum A, and6.7 ± 0.6 for inoculum B). In table III, theresults of both inocula were pooled.

Ulva supplementation with resistantstarch did not change the sulphate input,whereas algal xylan added 6.8 mmol/day.It can be observed from figure 3 that withresistant starch, sulphate disappeared fromthe culture effluents within the first 3 days ofincubation, whereas with algal xylan, itincreased up to day 3, before declining. Withboth carbohydrates, sulphide concentrationsin the vessels rose up to day 4, and thenremained almost constant (6 to 7 mmol/L)(fig 3). Finally, on the last days of incuba-tion, the sulphide concentrations were notsignificantly different for all vessels fedUlva (table, III).Some gaseous sulphide was produced

from the vessels fed Ulva, but the amountswere low (mean values for the three lastdays: 0.56 ± 0.05, 0.95 ± 0.10 and 0.68 ±0.03 mmol/day for Ulva alone, + RS and +XYL, respectively).

Ulvan addition supplied daily a total of7.4 mmol sulphate. Some sulphate wasrecovered in the culture effluents but sul-

phide concentrations in the fermentersremained similar to those of the basalmedium (table III).

Effect of molybdate additionon sulphate metabolism

With the basal medium and with ulvan,molybdate had only a slight depressingeffect on sulphide concentrations

(- 0.4 mmol/L). Conversely, in all the ves-sels fed Ulva, molybdate significantlyincreased sulphate output and stronglydecreased sulphide concentrations. How-ever, the increase in sulphate output wasless marked in the presence of algal xylan(fig 3, table III).

Effect of molybdate additionon end products of fermentation

Whatever the treatments, total SCFA pro-duction was decreased by molybdate. Thiswas generally due to a reduction in acetate,except with xylan, where propionate wasreduced by 30°70 (table IV). In all the vesselsfed Ulva, molybdate addition resulted inproduction of formate and hydrogen (fig 4).With Ulva alone, these productions werelow and transient, whereas they were moreabundant and lasted longer with addition ofcarbohydrates, especially with resistantstarch (fig 4).

Reductive acetogenesisof the microflora adapted to Ulva

Despite adaptation of the microflora to Ulva,which was rich in sulphate, synthesis of

13C_acetate from 13C02 was intense in theabsence of added sulphate (table V). In bothcases the ratio of double-labelled acetate tototal labelled acetate represented about 40%,and on average, the labelling was equallydistributed among the methyl and carboxylgroups of acetate.

Adaptation of the microflora to molyb-date (study 2) reduced both total acetate pro-duction by the culture and the proportion oflabelled acetate in total acetate (table V).

Addition of sodium sulphate to the bathcultures reduced total acetate and the syn-thesis of acetate from !3C02 in study 2 butnot in study 3. Here, one of the vessels thatshowed a high 13C02 incorporation intoacetate was not negatively affected by sul-phate addition in contrast with the other two.This explains the particularly high valuesof standard errors (table V).

DISCUSSION

Fermentation pattern

In this study, we used continuous rather thanbatch cultures because we hypothesized thatadaptation of the colonic bacteria to the sea-weed Ulva and its main soluble constituentulvan was necessary to optimize their fer-mentation, since these substrates are notroutinely part of the diet. The results weobtained on SCFA production from Ulvado not show strong adaptative effects (fig 1 ).

The first period of adaptation may have beendue to enzyme induction, which is a com-paratively rapid process (Salyers, 1985;Macfarlane et al, 1990). Then the numberof bacterial species able to degrade Ulvamay have slightly increased. This slightadaptation of the microflora to Ulva was

not shown with ulvan. On the contrary,inhibitory effects on SCFA productionseemed to occur from day 1 to day 3 of theincubation. This observation is consistentwith the partial inhibitory effect of ulvanover cellulase activity that has been recentlydetected (Bobin-Dubigeon, 1996).

Ulva was fermented to a very limitedextent compared to algal xylan and resis-tant starch, which provided three times moreSCFA per gramme of organic matter. Inaddition, the small amounts of SCFA formedfrom Ulva did not appear to originate fromits soluble carbohydrates: ulvan. These datasupport the previous findings of Bobin-Dubigeon et al (1996) who, using batch cul-ture, showed that insoluble fibres containedin Ulva were more fermentable that the sol-uble ones (ulvan). This resistance wasmainly attributed to the particular chemicalstructure (linkage and sequence of sugars) ofthese soluble polymers (Bobin-Dubigeon etal, 1996) and to the lack of the specificdepolymerases in colonic microorganisms(Michel and Macfarlane, 1996).

Addition of Ulva to the basal mediumincreased the relative proportion of theBCFA, isobutyrate and isovalerate. Similarobservations were made in batch cultures

(Bobin-Dubigeon et al, 1996). These BCFAare generated almost exclusively from aminoacid fermentation, especially from valineand leucine, and are therefore good markersof protein breakdown (Macfarlane and Cum-mings, 1991). From the results of productionof SCFA from casein and bovine serumalbumin obtained in batch-cultures, Mac-

farlane et al (1992) assumed that approxi-mately 30% by weight of the protein bro-ken down is converted to SCFA and thatBCFA constitute 20% of these SCFA. Thesedata enabled us to estimate the protinaceousmaterials that were broken down during ourincubations. The calculated values were2.4 g/day for the basal medium alone and2 g/day for Ulva alone. This would meanthat all the proteinaeous materials of thebasal medium and about half those of Ulvawere degraded which is in line with theamount of ammonia produced. In vivo, it isnot known whether and to what extent Ulva

proteins are absorbed in the upper part ofthe intestine. Their contribution to the for-mation of SCFA or of ammonia in the largeintestine cannot therefore be estimated.

Yields of SCFA obtained from the fer-mentation of algal xylan and resistant starchwere relatively high (48 and 41 % organicmatter, respectively) and comparable withthe figures given for cell wall polysaccha-rides (35-54%) by Cummings (1995). Thesesubstrates provided energy for microbialgrowth as shown by the clear-cut increase inprotein formation following their addition.We can estimate that 2.3 and 2.6 g microbial

protein-N were produced by 100 g OM fer-mented for algal xylan and resistant starch,

respectively. This corresponds approxi-mately to an efficiency of microbial syn-thesis of 30% (weight by weight), a valuethat is comparable with that calculated byCummings (1984), from excretion of fae-cal bacterial mass. The decrease in ammoniaobserved with these polysaccharides couldresult either from inhibition of amino aciddeamination, or from a greater ammoniauptake for microbial protein synthesis. Thelatter was more likely under our conditionsbecause the absolute levels of BCFA didnot change with carbohydrate supplemen-tation. Ammonia formation has a numberof possible harmful health implications(Macfarlane and Cummings, 1991), there-fore, if part of Ulva proteins reach the largeintestine, supplementing this product withsuch polysaccharides should be beneficialfor the host.

Algal xylan and resistant starch showeddifferent fermentation patterns. Resistantstarch (retrograded amylomaize starch) hada slow rate of fermentation, as shown by thestability of the pH after feeding. These char-acteristics were already observed when wecompared the same resistant starch to lac-tulose (Dore et al, 1995b). Algal xylan fer-mented rapidly, induced a pH decreaseshortly after feeding, and produced highproportions of propionic acid, thus corrob-orating previous observations of Lahaye etal (1993).

Metabolism of sulphate from Ulvaand possible interactionswith methanogenesis and acetogenesis

Our study clearly showed the importanceof sulphide production in all the vessels fedUlva. About 40% of Ulva sulphate was dis-similated to sulphide. The effect of theinhibitor, sodium molybdate (Mo), whichgreatly decreased sulphide concentrationand increased sulphate output from the ves-sels, further demonstrated the high level ofactivity of sulphate-reducing bacteria in the

presence of Ulva. The formation of low lev-els of sulphide, either in the basal medium,or in Mo-supplemented vessels originatedmost probably from microbial degradationof the sulphur amino acids, cysteine andmethionine in which sulphur exists in itsmost reduced state. In the rumen, cysteine isknown to be degraded into NH3, pyruvateand H2S in the presence of sulfhydrase andfeeding methionine to sheep increased therumen sulphide pool (Doyle and Moir,1979). We had first expected that supple-mentation of Ulva with fermentable carbo-

hydrates such as resistant starch wouldreduce sulphide concentrations by decreas-ing the pH and/or by stimulating assimila-tory sulphate reduction in relation to theincrease in biomass synthesis. Our resultsdid not support this prediction. First, prob-ably because of the high concentration ofammonia, the pH did not decrease below6.7 and dissimilatory sulphate reduction wasshown to be reduced only below pH 6.5(Gibson et al, 1990). Further, the rise inbiomass synthesis did not support high lev-els of sulphate incorporation. If we assumethat the value of S/N ratio in the colonicflora is similar to that reported for rumenmicroorganisms (S/N = 0.06) (Durand andKomisarczuk, 1988), we can approximatean increase of S incorporation into biomassof 0.2-0.4 mmol, corresponding to the200 mg protein-N induced by resistantstarch. This value is negligible compared tothe observed sulphide concentrations.

Our results showing that sulphide pro-duction from ulvan was negligible confirmthe recent observations of Michel et al

(1995). These authors concluded that resis-tance to desulphation appeared to be thelimiting factor in ulvan sulphate metabolismby SRB. However, our particular ulvanpreparation contained some free sulphate(about 30%) (Lahaye, personal communi-cation) which was recovered in the cultureeffluents in the presence as well as inabsence of Mo. This would indicate an

inhibitory effect of ulvan itself on dissimi-latory sulphate reduction.

In our studies, inhibition of sulphidereduction by Mo resulted in release of for-mate and hydrogen. Gibson et al ( 1988a)also observed that inhibition of SRB by Moresulted in accumulation of hydrogen inlarge quantities. In our experiments, how-ever, the production of hydrogen and for-mate were greater in the presence of resistantstarch than with Ulva alone, or when sup-plemented with algal xylan, despite similarlevels of sulphide production in the absenceof Mo. This indicates that electron donorsfor sulphate reduction varied depending onthe nature of the polysaccharides being fer-mented. Low molecular weight organicacids such as lactate, propionate or acetatecan be used as electron donors in addition to

H2 and formate (Gibson, 1990). Sodiummolybdate was not entirely specific againstsulphate reduction. Although it did notapparently decrease the total number ofanaerobes, it always reduced, whatever thetreatment, production of SCFA. This reduc-tion was most apparent with resistant starch(-18%). In this case, inhibition of acetateproduction might have been the consequenceof hydrogen accumulation, which is knownto inhibit regeneration of reduced electroncarriers involved in the oxidation of pyruvateto C02 and acetyl-CoA in a number of bac-terial species. This explanation is not valid,however, for the basal medium, or for Ulva,where H2 production was very low. Ourresults of 13C02 incorporation into acetatealso clearly demonstrate an inhibitory effectof Mo on reductive acetogenesis. Molyb-date being a competitive irihibitor, shouldbe added at the equivalent level of ambientsulphate dissolved in the system. A level of20 mM molybdate is routinely employed inhuman faecal incubations (Gibson et al,1988a; Strocchi et al, 1993) without notableinhibitory effects. It is possible that somemicrobial strains are more susceptible toMo in continuous rather than in batch cul-tures. Some inhibitory effects of 20 mM Mo

on methane production have been recorded(Oremland and Capone, 1988). Mo is knownto form molybdosulphide complexes whichcan bind available sulphide ions and cause adeficiency in sulphur availability.

The number of sulphate-reducing bacte-ria (SRB) was relatively high and similar inboth original faeces, though donor B wasmoderately methanogenic and harboured asignificant number of methanogens in herfaeces. This absence of a clear-cut inverse

relationship between faecal SRB andmethanogens is in agreement with the recentresults of Dore et al (1995a). However, afteradaptation to Ulva in the fermenters inocu-lated with the flora from donor B, the num-ber of methanogens decreased and only traceamounts of CH4 were produced. This inhi-bition of methanogenesis was, most proba-bly, related to the high sulphate content ofUlva which led to competitive exclusion ofmethanogens by SRB, as previouslyobserved by Gibson et al ( 1988a). Here, thisexclusion was probably facilitated by thefact that the inoculum was provided by alow CH4 excretor and harboured a moderatenumber of methanogens. Contrary tomethanogenesis, neither the concentrationof acetogens nor the reductive acetogenicactivity were affected by the adaptation toUlva. Pure cultures of acetogenic bacteriafrom the human colon have been shown

recently to be extremely versatile withregard to their substrate utilization patterns(Bernalier et al, 1996b), which could facil-itate their growth in the presence of activesulphate reducers. The addition of sulphateat a level of 7 mM which is what could be

expected in the colon of human subjectsconsuming about 15 g dry Ulva, counter-acted reductive acetogenesis in one experi-ment out of two, demonstrating that SRBpresent in some individual inocula can com-

pete with acetogenic bacteria for H2. Thisis consistent with previous observationsshowing fluctuating responses to sulphateaddition (Bernalier et al, 1996a).

In conclusion, we confirmed that Ulva,and particularly its soluble polymer, ulvan,are to a large extent resistant to fermenta-tion, and we have shown that adaptation ofthe microbiota occurs to a very limited

degree. Our results also clearly demonstratethat Ulva, though producing low amountsof SCFA, stimulates sulphide formation bymethanogenic and non-methanogenicmicrofloras, owing to its high free-sulphatecontent. Its supplementation with more fer-mentable polysaccharides, though probablybeneficial for the host by decreasing ammo-nia, should not significantly reduce sulphideproduction. Therefore, it can be postulatedthat unless crude Ulva is previously pro-cessed in order to remove its free sulphate,its regular consumption at a level of 15 or20 g of dry product daily could support highactivities of sulphate reducing bacteria inthe colon, which might interact with othermicrobial pathways of H2 disposal and bedetrimental to the health of the host.

ACKNOWLEDGMENTS

Part of this work was supported by the EC con-tract AIR I CT 92 518. The authors thank GMacfarlane and C Michel for their helpful adviceduring this study.

REFERENCES

Bernalier A, Lelait M, Rochet V, Grivet JP, GibsonGR, Durand M (1996a) Acetogenesis from H2 and

C02 by methane- and non-methane-producinghuman colonic bacterial communities. FEMSMicrobiol Ecol 19, 193-202

Bernalier A, Rochet V, Leclerc M, Dore J, Pochart P(1996b) Diversity of H /CO2 utilizing acetogenicbacteria from feces of non-methane-producinghumans. CurrMicrobio132, 1-17

Bo S6rbo (1987) Sulfate: turbidimetric and nephelo-metric methods. Meth Enzymol 143, 3-6

Bobin-Dubigeon C (1996) Caractérisation chimique,physico-chimique et fermentaire de produits ali-mentaires a base d’algues. These de Doctorat, Uni-versit6 de Nantes, France

Bobin-Dubigeon C, Lahaye M, Barry JL (1996) Humancolonic bacterial degradability of dietary fibresfrom sea-lettuce (Ulva sp). J Sci Food Agric 73,149-159

Brunsgaard G, Bach Knudsen KE, Eggum BO (1995)The influence of the period of adaptation on thedigestibility of diets containing different types ofindigestible polysaccharides in rats. Br J Nutr 74,833-848

Christl SU, Gibson GR, Cummings JH (1992) Role ofdietary sulphate in the regulation of methanogene-sis in the human large intestine. Gut 33, 1234-1238

Cline JD (1969) Spectrophotometric determination ofhydrogen sulfide in natural waters. GimnolOceanogr 14, 454-458

Cummings JH (1984) Microbial digestion of complexcarbohydrates in man. Proc Nutr Soc 43, 35-44

Cummings JH (1995) Short chain fatty acids. In:Human Colonic Bacteria. Role in Nutrition, Phys-iology and Pathology (GR Gibson, GT Macfar-lanne, eds), CRC Press, Boca Raton, USA, 101-130

Czerkawski JW, Breckenridge G (1977) Design anddevelopment of a long-term rumen simulation tech-nique (Rusitec). Br J Nutr 38, 371-384

Daly J, Tomlin J, Read NW (1993) The effect of feed-ing xanthan gum on colonic function in man: cor-relation with in vitro determinants of bacterialbreakdown. Br J Nutr 69, 897-902

De Graeve KG, Grivet JP, Durand M, Beaumatin P,Cordelet C, Hannequart G, Demeyer D (1994)Competition between reductive acetogenesis andmethanogenesis in the pig large intestinal flora. JAppl Bacteriol76, 55-61 1

Dore J, Pochart P, Bernalier A, Goderel I, Morvan B,Rambaud JC (1995a) Enumeration of Hz-utilizingmethanogenic archaea, acetogenic and sulfate-reducing bacteria from human faeces. FEMS Micro-biol Ecol 17, 279-284

Dore J, Durand M, D’Agay-Abensour L, Pochart P,Flourié B, Bernalier A, Champ M, Rambaud JC(1995b) In vivo and in vitro fermentation of resis-tant starch and lactulose by human colonicmicroflora. Reprod Nutr Dev 35, 590-591 1

Doyle PT, Moir RJ (1979) Sulfur and methioninemetabolism in sheep. II. Quantitative estimates ofsulfur metabolism in the sheep’s stomach. Aust JBiol Sci 32, 65-76

Durand M, Komsarczuk S (1988) Influence of majorminerals on rumen microbiota. J Nutr 118, 249-260

Durand M, Bernalier A, Doré J (1995) Hydrogenmetabolism in the colon. In: Dietary Fibre and Fer-mentation in the Colon, COST92 Workshop, EspooFinland

Edwards CA (1995) Dietary fibre, fermentation andthe colon. In: Dietary Fibre. Mechanisms of Action

in Human Physiology and Metabolism (C Cherbut,JL Barry, D Lairon, M Durand, eds), John LibbeyEurotext, Paris, France, 51-60

Faisant N, Champ M, Colonna P, Bul6on A, Molis C,Langkilde AM, Schweizer T, Flouri6 B, GalmicheJP (1993) Structural features of resistant starch at theend of the human small intestine. Eur J Clin Nutr47,285-296

Florin T, Neale G, Gibson GR, Chfistl SU, CummingsJH (1991) Metabolism of dietary sulphate: absorp-tion and excretion in humans. Gut 32, 766-773

Gibson GR (1990) Physiology and ecology of the sul-phate-reducing bacteria. J Appl Bacteriol 69, 769-797

Gibson GR, Cummings JH, Macfarlane GT (1988a)Competition for hydrogen between sulphate-reduc-ing bacteria and methanogenic bacteria from thehuman large intestine. J Appl Bacteriol 65, 241-247

Gibson GR, Cummings JH, Macfarlane GT (1988b)Use of a three-stage continuous culture system tostudy the effect of mucin on dissimilatory sulfatereduction and methanogenesis by mixed popula-tions of human gut bacteria. Appl Environ Microbiol54,2750-2755

Gibson GR, Cummings JH, Macfarlane GT, AllisonC, Segal I, Vorster HH, Walker ARP (1990) Alter-native pathways for hydrogen disposal during fer-mentation in the human colon. Gut 31, 679-683

Gibson GR, Cummings JH, Macfarlane GT (1991)Growth and activities of sulphate-reducing bacteriain gut contents of healthy subjects and patients withulcerative colitis. FEMS Microbiol Ecol 86, 103-112 2

Lahaye M (1991) Marine algae as sources of fibres:determination of soluble and insoluble dietary fibrecontents in some sea vegetables. J Sci Food Agric54,587-594

Lahaye M, Axelos M (1993) Gelling properties ofwater-soluble polysaccharides from proliferatingmarine green seaweeds (Ulva spp). CarbohydrPolym 22,261-265

Lahaye M, Michel C, Barry JL (1993) Chemical,physicochemical and in-vitro fermentation charac-teristics of dietary fibres from Palmaria palmata(L) Kuntze, Food Chem 47, 29-36

Macfarlane GT, Cummings JH (1991) The colonicflora, fermentation, and large bowel digestive func-tion. In: The Large Intestine: Physiology, Patho-physiology and Disease (SF Phillips, JH Pemberton,RG Shorter, eds), Raven Press Ltd, New York,USA, 51-92

Macfarlane GT, Hay S, Macfarlane S, Gibson GR(1990) Effect of different carbohydrates on growth,polysaccharidase and glycosidase production byBacteroides ovatus, in batch and continuous cul-

ture. J Appl Bacteriol 68, 179-187

Michel C, Macfarlane GT (1996) Digestive fates ofsoluble polysaccharides from marine macroalagae:involvement of the colonic microflora and physio-logical consequences for the host. J Appl Bacte-riol 80, 349-369

Michel C, Willis CL, Gibson GR (1995) Fermentationof soluble sulphated algal, polysaccharides byhuman colonic bacteria ecological, significance inrelation to hydrogenotrophic species. Soc GenMicrobiol Abs Booklet, P30

Michel C, Lahaye M, Bonnet C, Mabeau S, Barry JL( 1996) In vitro fermentation by human faecal bac-terial of total and purified dietary fibres from brownseaweeds. Br J Nutr 75, 263-280

Monsma DJ, Marlett JA (1996) Fermentation of car-bohydrate in rat ileal excreta is enhanced with cecalinocula compared with fecal inocula. J Nutr 126,554-563

Oremland RS, Capone DG (1988) Use of specificinhibitors in biogeochemistry and microbial ecol-ogy. Adv Microb Ecol 10, 285-393

Ray B, Lahaye M ( 1995) Cell-wall polysaccharidesfrom the marine green algae Ulva rigida (Ulvales,Chlorophyta). Extraction and chemical composi-tion. Carbohydr Res 274, 251-261 1

Richardson AJ, Calder AG, Stewart CS, Smith A(1989) Simultaneous determination of volatile andnon-volatile acidic fermentation products of anaer-obes by capillary gas chromatography. Lett ApplMicrobiol 9, 5-8

Roediger WEW, Duncan A, Kapaniris 0, Millard S(1993) Reducing sulfur compounds of the colonimpair colonocyte nutrition: implications for ulcer-ative colitis. Gastroenterology 104, 802-809

Salyers AA (1985) Breakdown of polysaccharides byhuman intestinal bacteria. J Environ Pathol Toxicol

Oncol 5, 211-231 1

Strocchi A, Ellis CJ, Levitt MD (1993) Use ofmetabolic inhibitors to study H2 consumption byhuman feces: evidence for a pathway other thanmetanogenesis and sulfate reduction. J Lab ClinMed 121, 320-327

Tulung B, R6m6sy C, Demign6 C (1987) Specificeffects of guar gum of gum arabic on adaptationof cecal digestion to high fiber diets in the rat. JNutr 117, 1556-1561 1

Walter DJ, Eastwood MA, Brydon WG, Elton RA(1986) An experimental design to study colonicfibre fermentation in the rat: the duration of feeding.Br J Nutr 55, 465-479

Wyatt GM, Bayliss CE, Holcroft JD (1986) A changein human faecal flora in response to inclusion of

gum arabic in the diet. Br J Nutr 55, 261-266