Environ. Sci. Technol. 1992, 26, 2413-2420
Lovley, D. R.; Phillips, E. J. P.; Lonergan, D. J. Environ. Sci. Technol. 1991,25, 1062. De Vitre, R. R.; Buffle, J.; Perret, D.; Baudat, R. Geochim. Cosmochim. Acta 1988,52,1601. Wieland, E.;Wehrli, B.; Stumm, W. Geochim. Cosmochim. Acta 1988,52,1969. Stumm, W.; Wieland, E. In Aquatic Chemical KineticsReaction Rates of Processes in Natural Waters; Stumm, W., Ed.; John Wdey & Sons: New York, 1990;Chapter 13. Zinder, B.; Furrer, G.; Stumm, W. Geochim. Cosmochim. Acta 1986,50,1861. Dos Santos Afonso, M.; Morando, P. J.; Blesa, M. A.; Banwart, S.; Stumm, W. J. Colloid Interface Sci. 1990,138, 74. Sulzberger, B.; Suter, D.; Siffert,C.; Banwart, S.; Stumm, W. Mar. Chem. 1989,28,127. Rickard, T. Am. J . Sci. 1974,274, 941. Pyzik, A. J.; Sommer, S. E. Geochim. Cosmochim. Acta 1981,45, 687. Schwertmann,U.; Taylor, R. M. In Minerals in the Soil Environment, 2nd ed.;Dixon, J. B., Ed.; Soil Science Society of America: Madison, WI, 1989;Chapter 8. Canfield, D. E. Geochim. Cosmochim. Acta 1989,53,619. Frevert, T.; Galster, H. Schweiz. 2.Hydrol. 1978,40,199. Gupta, S.K.Dissertation, Universitiit Bern, 1976. Peters, K.; Huber, G.; Netsch, S.; Frevert, T. GWT, GasWasserfach: WasserlAbwasser 1984,125,386.
(25) Hering, J. G.; Stumm, W. In Mineral Water-Interface Geochemistry;Reviews in Mineralogy 23;Hochella, M. F.,
White, A. F., Eds.; Mineralogical Society of America: Washington, DC, 1990; Chapter 11. (26) Baumgartner,E.; Blesa, M. A.; Maroto, A. J. G. J . Chem. Soc., Dalton Trans. 1982,9,1649. (27) Cornell, R.M.; Schindler,P. W. Clays Clay Miner. 1987, 33,347. (28) LaKind, J. S.;Stone,A. T. Geochim. Cosmochim.Acta 1989, 53, 961. (29) Luther, G. W., III In Aquatic Chemical Kinetics-Reaction Rates of Processes in Natural Waters; Stumm, W., Ed.; John Wiley & Sons: New York, 1990;Chapter 6. (30) Ben-Yaakov, S.Limnol. Oceanogr. 1973,18,86. (31) Berner, R. A. Am. J . Sci. 1970,268, 1. (32) Schoonen, M. A. A.; Barnes, H. L. Geochim. Cosmochim. Acta 1991,55,1495. (33) Luther, G. W., I11 Geochim. Cosmochim. Acta 1991,55, 2839. (34) Urban, N. In Environmental Chemistry of Lakes and Reservoirs; Baker, L. A., Ed.; ACS Advances in Chemistry
Series; American Chemical Society: Washington, DC, in press. Received for review February 12, 1992. Revised manuscript received July 2,1992. Accepted July 13,1992.
A Polar High Molecular Mass Constituent of Bleached Kraft Mill Effluent Toxic to Marine Organisms
Is
Richard M. Higashi,' Gary N. Cherr, Jonathan M. Shenker,+Jeffrey M. Macdonald,* and Donald G. Crosbyg
Bodega Marine Laboratory, University of California-Davis,
Box 247, Bodega Bay, California 94923
A high molecular mass constituent (HMM) of whole bleached kraft mill effluent (BKME),which represents the majority of toxicity to early life stages of marine animals and a plant, has been isolated and partially characterized. BKME was subjected to fractionation coupled with toxicity testing to determine the toxicity of each isolated fraction. The toxic mode of action was also tracked throughout the fractionation using echinoderm sperm motility as an indicator. While most fractions inhibited sperm motility, BKME and HMM did not. Yet, HMM exhibited most of the toxicity of BKME to echinoderm sperm, mollusc embryos, larval sole, and kelp gametophytes. HMM was soluble only in water and appeared to be free of the resin and fatty acids or chlorinated aromatic compounds that are implicated in freshwater acute toxicity of BKME to salmonid fish. Structural analyses indicate that this polar, high molecular mass constituent was devoid of aromatic structure and had other characteristics indicative of lignin breakdown products.
Introduction Bleached kraft mill effluent (BKME) is the combined aqueous waste of a major chemical pulp-making process. A mill typically generates waste water from many of its sections, including pulping, chemical recovery, evaporation and condensation, and multistage bleaching operations. Present address: Department of Biological Sciences, Florida Institute of Technology, Melbourne, FL 32901. Present address: Department of Pharmaceutical Chemistry, University of California, San Francisco, CA 94143. *Department of Environmental Toxicology, University of California, Davis, CA 95616. 0013-936X/92/0926-2413$03.0010
Thus, waste waters from acid, alkaline, chlorine oxidant, and other chemically diverse processes are sewered together to form the whole-mill discharge. The mixing and ensuing reactions of these streams lead to a final effluent that is highly complex. It consists of simple inorganic salts as well as over 250 identified organic and inorganic compounds of low molecular weight (or mass) ( I ) , with probably many more yet to be identified. In addition, an unusual property of BKME is that many of the organic constituents are of high molecular mass (>1 m a ) (2). This material is thought to consist largely of the polar breakdown product(s) of lignin, with lesser amounts of lignin at various degradation stages as well as polysaccharides (2).
The acute toxicity of BKME to a variety of aquatic organisms has been well-documented (3). Studies of sublethal effects of whole effluent range from biochemical and mutagenic to behavioral aspects (3). Partly to obtain more detail for pollution abatement, toxicity studies have been performed on selected mill process streams (cf. ref 2). However, the relevance of these studies to the final effluent toxicity is not clear ( 4 4 3 , probably due to the complex changes that take place upon mixing to form BKME. Therefore, it is essential to first identify and characterize the constituents responsible for toxicity in BKME itaelf, in order to assess environmental hazards of BKME discharges and to develop appropriate pollution control technology. The complexity of BKME has made this identification a nontrivial task. By use of chemical fractionation coupled with toxicity testing and structure identification, the acute freshwater toxicity of nonbiologically treated BKME to salmonid fish has been associated with resin and fatty acids
0 1992 American Chemical Society
Envlron. Sci. Technol., Vol. 26,No. 12, 1992 2413
(RFAs) (7). This fiiding was supported, for RFAs as well as chlorinated aromatic compounds (CACs), by similar studies on selected process streams (8-11). The same phenomenon appears to hold in freshwater crustaceans as well (6, 12). However, there currently is little or no evidence as to the source(s) of either acute or chronic toxicity of BKME toward marine species or developing systems. The present paper addresses this gap by coupling chemical fractionation of BKME to marine early life stage chronic toxicity tests and provides preliminary chemical characterization of the toxic constituent. The fractionation was designed to deal with the broad chemical classes found in BKME, and the toxicity tests used early life stages of marine algae, invertebrates, and a vertebrate; some of these tests are considered surrogates for chronic bioassays. Progress related to this paper has been presented in abstract form (13,14),and related analytical methods, being developed for use on other types of mill effluent, have been described (15). The present paper describes in detail the results of research using combined chemical and biological analyses on BKME. Experimental Section Materials. BKME was obtained from a mill lacking secondary treatment in northern California that discharges to marine waters. The mill primarily produced softwood pulp, with a relatively conventional bleaching sequence of chlorine, alkali, hypochlorite, chlorine dioxide, alkali, and chlorine dioxide processes. The grab samples of whole mill effluent were shipped overnight at 2-6 "C to the University of California Bodega Marine Laboratory, and samples were processed upon arrival by adjusting to pH 8 (seawater pH, relevant for marine receiving waters) with NaOH or HC1 to result in "BKME" for experimentation described below. Toxicity tests of BKME were routinely performed within 24 h of receipt. Sources of equipment, reagents, and biological material are cited below upon initial mention. All common reagents and materials for which sources are not mentioned were obtained from Fisher Scientific, Inc. (Pittsburgh, PA), and solvents were of pesticide residue or HPLC grade. All seawater for biological tests was obtained from a sandfiltered system at the Bodega Marine Laboratory and vacuum-filteredon 0.45-pm membrane filters (pH 7.9-8.0). Salinity was determined with a refractometer. Analytical Procedures. Toxicity Tests. BKME and its fractions were assessed for toxicity using early life stages of three marine organisms: purple sea urchin (Strongylocentrotus purpuratus) sperm for fertilization assay, mussel (Mytilus californianus) embryos for development assay, and larval English sole (Parophrys vetulus) for mortality assay. In addition, some fractions were also tested using the gametophyte stage of the giant kelp, Macrocystis pyrifera. Sea urchin sperm cell toxicity testa, which measured the effects of sperm preexposure to toxicants on their fertilizing ability, were conducted using the procedure of Cherr et al. (16) in 2-mL volumes. In some experiments, a slightly modified procedure was employed which predetermined the lowest sperm/egg ratio yielding 95% fertilization in seawater, as control for each experiment. Since excess sperm may mask toxicity, this sperm/egg ratio was then maintained for the toxicity testing. Mussel embryo toxicity tests,which assessed the normality of veliger larval development, were conducted according to Cherr et al. (17) using small-volume (5 mL) test vessels. Larval English sole toxicity tests, which determined lethality (over 96 h) on yolk sac larvae, were conducted as described by Shenker and Cherr (18) in 20-mL volumes. Lastly, giant kelp 2414
Environ. Sci. Technol., Vol. 20, No. 12, 1992
gametophyte toxicity tests were conducted according to Anderson and Hunt (19)to assess the effects of BKME and selected fractions on both zoospore germination and germ tube elongation. These experiments also employed small volumes (10 mL) of test solutions. BKME and its fractions were prepared for toxicity testing in a variety of ways, to compensate for the various physical-chemical properties of the fractions. Hydrophobic (organic solvent soluble) fractions were dissolved in a small amount of HPLC-grade N-methyl-2pyrrolidinone (NMP, Aldrich Chemical, Milwaukee, WI), at a ratio of 4:l extract weight to NMP volume. NMP is an aprotic solvent that has low volatility and a wide range of solvation and was found to be nontoxic to the organisms at test concentrations (I 1.0 ppm). Following solubilization of the extracts in NMP, double-distilled water was added to make up the appropriate BKME equivalents. In contrast, the aqueous samples (BKME, lyophilized composite BKME, water-only soluble, >10 kDa, 1 kDa, and go% motility), those with decreasing motility were scored as 3 (50-75% motility), 2 (25450% motility), and 1 (>0-25% motility), and glutaraldehydefixed sperm were scored as 0 (0% motility). Gas-liquid chromatography (GC) used a Varian Model 3300 gas chromatograph (Varian Instruments, Palo Alto, CA) equipped with a flame ionization detector and a 0.4-pm film, 0.18 mm i.d. X 40 m DB-1 stationary-phase open-tubular column (J&W Scientific, Folsom, CA). Sample preparation for GC analysis followed a general procedure for the simultaneous analysis of compounds with COOH, OH, NH, and SH functional groups (21, 22). Solvent-free residues of fractions were derivatized for 10 h at room temperature in a 1:l mixture of acetonitrile+methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide (MTBSTFA, Regis Chemical, Morton Grove, IL),and then analyzed by direct injection into the GC. GC analysis parameters included the following: H2 carrier gas at 60 cm/s, injector at 260 "C, detector at 320 OC, column at 60 OC held for 2 min, ramped to 150 "C at 20 "C/min, and then ramped to 300 "C at 10 "C/min; splitless injector purge valve held for 1.0 and then valve open at 100 mL/min flow for the duration of the run. To estimate recovery of RFAs, BKME was fortified to 1 ppm with 0-methylpodocarpate (OMPC, Helix Biochemical Co., Richmond, BC, Canada), using methanol as carrier, and processed through the fractionation procedure described below. Recovery of OMPC was quantified by GC analyses as described above. GC-mass spectrometry (GC-MS) employed a Hewlett-Packard Model 5890 gas chromatograph (HewlettPackard Analytical Products, Palo Alto, CA) equipped with a 0.25 mm i.d. x 30 m DB-5 open-tubular column (0.25-pm film; J&W Scientific) and interfaced to a VG Model Trio-2 quadrupole mass spectrometer (VG Masslab,
Altrincham, UK). Sample preparation and chromatographic conditions were identical to the GC method described above, except for He carrier gas at 35 cm/s and transfer line at 250 "C, and MS conditions employed 70-eV electron ionization and a source temperature of 200 "C. High-pressure liquid chromatography (HPLC) employed an Isco (Lincoln, NB) Model 2360 gradient programmer, Isco Model 2350 pump, and Isco Model V, UVvisible detector. In addition to analytical information obtained in conjunction with the preparative HPLC procedures described below, we also performed analytical size-exclusion chromatography (SEC-HPLC) using a 250 mm X 4.6 mm i.d. column packed with Shodex OH-Pak KB-805 (Showa Denko K.K., Tokyo, Japan). Molecular mass calibrations were carried out using a protein molecular weight kit (MW-GF-100, Sigma Chemical Co., St. Louis, MO) and a polysaccharide standard kit (Pullulan P-82, Showa Denko K.K.). The detection was at 208 nm, and the mobile phase was isocratic using 10 mM Tris buffer at pH 8 and 0.6 mL/min flow. UV-visible spectrophotometry used a Bausch & Lomb Model 2000 double-beam or a Hewlett-Packard Model 8524A diode-array instrument (Hewlett-Packard Analytical Products). Samples were diluted to maintain the maximal response to less than 2.0 absorbance units full scale, and all measurements were made in quartz cells. IH NMR spectroscopy was performed on a GE/Nicolet Model GN-500 FT-NMR spectrometer (General Electric NMR Instruments, Fremont, CA). Dry samples were reconstituted in 99.8 atom % D20 (Aldrich Chemical Co.) and placed in a 5-mm NMR tube. Spectra were recorded at 500 MHz using 2048 one-pulse transients acquired with 90" pulse, an interpulse delay of 2 s, and a sweep width of 5000 Hz with digital resolution of 8K. Residual HDO was suppressed during interpulse delays, frequency was locked to DzO,spectra were referenced to HDO, which was assigned a chemical shift value of 4.76 ppm, and sample temperature was at 20 "C. Fractionation of BKME. All procedures were conduded under room light and temperature, unless otherwise indicated. BKME was filtered at 4 "C on 8-pm cellulosic filters, divided into 200-mL aliquots in Teflon beakers which were immersed in liquid Nz until the BKME was frozen, and then lyophilized using a cold trap at -70 "C to highly concentrate the sample (23). The resulting residue, a fine brown powder, was stored desiccated at -20 "C until use. To obtain a representative BKME composition and toxicity, as well as to provide a constant and stable stock of BKME, three lyophilized samples were mixed in equal quantities to create a single composite sample, which represented an average BKME over 2 months of discharge; this composite sample was used throughout. For toxicity testing, BKME was reconstituted by adding the appropriate mass (2.5 g/L) of lyophilized composite BKME residue to distilled water and dispersing it for 5 rnin with the aid of a 100-W sonic bath. The material dissolved completely and was not subjected to further manipulation before use. In addition to the above, lyophilization was itself the first step of the fractionation process, as it removed the volatile compounds from BKME in a rigorous fashion. The second step was to remove nonpolar compounds from the dry residue using organic solvents. The dry residue was subjected to three sequential methylene chloride (MeClZ) extractions at 33 mL/g of residue, each performed by 1 h of sonication in a 100-W bath at 40 "C followed by 7 h standing at room temperature. At the end of each extraction, the residue was recovered on 0.22-pm Teflon
filters in a stainless steel filter assembly. The MeClz-insoluble residue was then subjected to a second series of three extractions, except that acetonitrile (MeCN) was used at 800 mL/g of residue. The organic solvent extracts were then pooled and concentrated on a rotary evaporator followed by a stream of Nz. The concentrates were further fractionated using preparative HPLC on the instrument previously described. For both extracts, an Alltech (Deerfield, IL) Econosphere CN cartridge column (250 mra X 10 mm i.d., 5-pm particle size) with CN guard column was used. For fractionation of the MeC12 extract, a multilinear ternary gradient was used, consisting of a S m i n hold at 99:l:O hexaneMeClZ-MeCN followed by a 15-min linear gradient to 1:990, a 15-min hold, a subsequent linear gradient over 15 min to 0:1:99, and a hold at this level for 20 min. For fractionation of the MeCN extract, the following gradient program was used: 15-min hold at 991:O MeCl2-MeCNHzO followed by a 15-min linear gradient to 1:99:0, a 20min hold, a subsequent 20-min linear gradient to 0:50:50, and a hold a t this level for 40 min. In each case, three major HPLC fractions, 1 (1-25 rnin), 2 (25-50 rnin), and 3 (50-85 min) were collected for analysis by toxicity and mode-of-action testing; to avoid photolytic breakdown of constituents, the detector was turned off during fraction collections. All HPLC fractions were concentrated using a rotary evaporator and N2 stream. The residue that remained after organic solvent extraction was soluble only in water. Therefore, we turned to ultrafiltration to further fractionate this residue. Toxicity data and UV-visible spectra were used as criteria to test the effectiveness of several membrane types and "molecular weight cutoff" (MWCO) ratings, with respect to both toxicity and chemical separation. It was found that the Amicon (Danvers, MA) YM-10 membrane, which is of relatively hydrophillic composition (24) and a 10-kDa MWCO, proved to be the most effective. To perform the separations, the water-only soluble residue was reconstituted to its original volume with HPLC-grade water and placed in a 43-mm-diameter stirred cell with a YM-10 membrane installed. The ultrafiltration step, essentially an extraction with water, used the following procedure: samples were concentrated under 10 psi Nz to 25% of original volume, brought back up to original volume, and concentrated to 25% volume again,with the concentration procedure repeated three more times. All ultrafiltration operations were conducted at 1-5 "C. BKME itself (that is, prior to any lyophilization or solvent extraction) was also subjected to identical ultrafiltration procedures to yield coarse high molecular mass fractions of BKME. This material was used to compare toxicity characteristics with the material as isolated above.
Results Isolation of Toxic Fraction. In this study, BKME was found to be relatively toxic according to the marine testa used (see results below). Identification of the toxic components of this BKME was accomplished by chemically fractionating BKME (Figure l), coupled with toxicity testing to measure the toxicity of each fraction. Thus, development of the fractionation procedure was guided by the toxicity data, which was supplemented by sperm motility data to confirm whether the mode of action of isolated fractions was identical to BKME. Additional data from UV-visible spectrophotometry and GC analysis for resin and fatty acids were used to help improve chemical separation for some steps. BKME was adjusted to pH 8 to ensure constituents were in protonation states relevant to marine receiving waters. Environ. Sci. Technol., Vol. 26, No. 12, 1992
2415
LYOPHILIZED BKME
4
FILTRATIoN LYOPHILIZATION
Table I. Toxicity of BKME Fractions in Early Life Stage Marine Tests
WHOLE BKME
I MeCI, EXTRACTION
MeCI, EXTRACT
HPLC
SUB-
(nonpolar)
NORMAL PHASE
FRACTIONS
MeCNEXTRACT HPLC (intelmdlale P O ~ ~ W I NORMAL PHASE
SUBFRACTIONS
EXTRACTED BKME
I MeCN EXTRACTION
1 1 WATER-ONLY SOLUBLE BKME
I
fraction
composite BKME entire fractn 2.3 (lyophilized) MeCl, extract entire frmtn 44.2 HPLC Id >100.0 HPLC 2 70.4 19.2 HPLC 3 12.1‘ MeCN extract entire fractn HPLC 1 25.9 HPLC 2 X0.0 HPLC 3 40.7 water-only soluble entire fractn 2.5‘ >10 ma 1 kDa 30.0
1.4 1.8 14.V
-
8.9 12.3 >30.0 -
‘EC,, effective concentration of fraction that results in 50% effect. The total amount of the chemical fraction present in BKME is normalized to 100% for direct comparability; thus these numbers are in ‘BKME equivalents”. “verage of three replicates unless otherwise indicated. -, not determined. 10 kDa
4 5 0
test concn test concn (factor above level of (% BKMEY motilitp 40
17.39
4
50 20 40
1.13 1.65 13.79
1 2 4
0
*Concentration of fraction used to obtain the presented motility
(%BKME) ,o
more. The numbers are in ‘BKME equivalents” as explained in LARVALSOLE
8-
S4-
-
2-
0-
VARIOUS SAMPLES OF BKME
2. T o m ot MMdual B K M samplm and W k anespondhg iyophillzed residws. EC, valm are expressed as percent of BKME h assay SOLmn. Ths graphs are tor sampies assayed with saa uchin sperm cell (top) and with larval sob (bonom). The iyophilirsd resldw 01 each sample was necessariiy tested on a different day using inaL dual tesl organkm dierent hom moSe used for assessing the hesh B K M . Samples were obtained from the same source and pocessBd as oescrioed in the text.
The next step, lyophilization, resulted in a fine brown powder residue accounting for 2.5 g / L of effluent. Most samples experienced very little loss of toxicity upon lyophilization (Figure 2), validating the first step of fractionation while demonstrating that volatile constituents did not contribute significantly to the toxicity of RKME. Fiwre 2 also shows that toxicity varied from sample to sample. Therefore, we mmbined three lyophilized samples over 2 month9 of discharge to form a single composite 2416
Envlron. Sci. Technol.. Voi. 26. No. 12. 1992
Table I. “Test concentrstion divided by ECw from Table I. cLevel of motility was scored on a scale of (t4: control were scored as 4 (>W%motility), those with decreming motility were m r e d a(l 3 (56-7570 motility), 2 (25-50% motility), and (>(t25% motility), and elutaraldebvde-fixed merm were scored as 0 (0% matilitvl.
sample. In addition, BKME can be stored in this compact, dry form without loss of toxicity for at least 1year (data not shown), so that samples of BKME that have been stored serve as a “library” of samples. As the compceite was a dry residue, a wide polarity range of solvents (even water-miscible ones) could be used to extract the formerly aqueous sample. Three extractions each of two solvents, nonpolar MeCI4and a water-miscible solvent of intermediate polarity, MeCN, were used sequentially to remove compounds of nonpolar and intermediate polar character, respectively. As expected, the RFAs and CACs were found across both solvent extracts (GC-Ius analysis data not shown). The solvent procedures appeared to be reasonably efficient for these types of compounds; we obtained high recovery (>93%, by GC analysis) of OMPC in these fractions. Furthermore, there were no detectable RFAs or CACs in the last of the six sequential solvent extracts (GC-MS analysis data not shown). Although these organic solvent procedures efficiently extracted the RFAs known to be acutely toxic to juvenile salmonid fish in freshwater (7),the extrack did not contribute significantly to BKME toxicity in the marine testa
1.0AU-
Table 111. Toxicity of a BKME Sample Subjected to Only Ultrafiltration”
INDULIN AT
sample BKME >10 kDa 10 kDa (retentate) and 1-kDa fraction, the >lO-kDa fraction, harbored the majority of toxicity in all three bioassays (Table I). Among all membranes tested, this 10-kDa MWCO membrane also yielded the largest difference in the UV-visible spectra between retentate (>lo kDa) and filtrate (lO-kDa retentate accounted for 0.24 g/L of effluent; therefore, more than 90% of the residue mass, which consisted primarily of simple inorganic salts, was removed by this step without significant loss in toxicity. Table I11 shows the results of sea urchin sperm and mussel embryo toxicity tests on ultrafiltration fractions (10-kDa MWCO) of freshly obtained BKME that was not subjected to either lyophilization or solvent extraction. As this trend is similar to that in Table I for both bioassays, it appears that lyophilization and solvent extraction procedures did not affect the toxic nature of the high mo-
sample lyophilized BKME >10 kDa >7.0(75%)
0.25 0.49 8.0
“This table was generated from the lyophilized residue of the same sample of effluent as Table 111. *The numbers are in “BKME equivalents” as explained in Table I. ‘The germination end point for the Macrocystis gametophyte bioassay is shown here in order to show the ECm. The same trend in the distribution of toxicitv was observed for the germ tube growth end Doint.
lecular mass constituents. The mode of action of the >lO-kDa retentate (Table 11) further substantiated this conclusion. In an attempt to further resolve the molecular mass ranges that harbored the toxicity, multiple molecular mass fractionation of the water-only soluble residue was also attempted with progressively higher MWCO membranes, which revealed a progressive reduction in toxicity retention. This seemed to suggest that some common character important to toxicity was present in high molecular mass substances throughout its size range. Unfortunately, a quantitative comparison with the data in Table I would not be valid, as the available membrane materials differed in their MWCO ratings. The composition of the membrane is of great importance as some membranes with MWCO of 30 kDa can still efficiently (>80%) reject the passage of some small molecules (((1 kDa) found in pulping effluents, such as RFAs (24).This severe deviation from MWCO ratings applied primarily to low molecular mass compounds, which are typically not tested (cf. refs 25-27). The deviation was largest with hydrophobic membranes and hydrophobic molecules and the least with hydrophilic membranes and hydrophilic molecules (24). In our work (a) we employed a hydrophilic membrane (Amicon YM series) that deviated the least from the MWCO-rated behavior among those compared (24); (b) the sample being processed was the solvent-extracted,very hydrophilic (water-only soluble) fraction; and (c) the retentate was washed through four successive ultrafiltration concentration cycles. Therefore, it seemed reasonable that our conditions of ultrafiltration yielded a retentate fraction that was apparently free of low molecular mass compounds, and this was additionally supported by the SECHPLC data presented below. Lastly, the same BKME sample from Table I11 was lyophilized and coarsely fractionated using ultrafiltration alone (10-kDa MWCO) to see if similar toxicity patterns are exhibited by kelp gametophytes, which is a recently introduced marine bioassay (19). The kelp gametophytes showed the pattern similar to that of the sea urchin sperm (Table IV) and qualitatively similar to that for the bioassays in Tables I and 111. Thus, the retentate material Environ. Sci. Technol., Vol. 26, No. 12, 1992
2417
PROTEIN CALIBRATION PULLULAN CALIBRATION
I
I
I
kDa
4
2
I
m
2
I
3 4 5 ELUTION VOLUME (rnL)
1
d
d
&A&
b
md
0
kDa
"N-
'I
I
6
I
7
I
8
4
I
I
I
I
I
1 0 1 1 1 2 1 3 1 4
Figure 4. SEC-HPLC tracing of the >lO-kDa fraction. The ordinate Is relative absorbance at 208 nm, which is the ,A, of the >lO-kDa fraction. The abscissa is the elution volume, with the corresponding molecular mass calibration curves shown at the top of the figure. The partla1 tracing in the right half of the chromatogram is recorded at 0.002 absorbance unL full scale (AUFS) at 208 nm, demonstrating that there were only very low levels of low molecular mass substances present in the > 10kDa fraction. Chromatographic conditions are ghren in the text.
harbors the majority of toxicity under four marine developmental bioassays representing very diverse taxa. Preliminary Characterization of Toxic Fraction. A representation of the molecular mass range contained in the retentate fraction was obtained by SEC-HPLC (Figure 4);coupling SEC-HPLC fractions with a toxicity determination was not feasible due to the small scale of the technique. The SEC-HPLC conditions were devised such that the calibration curves of two very different polar macromolecular substances, proteins and polysaccharides, came the closest to each other. These standards were chosen for their polyionic and hydrogen-bonding properties as well as low aromatic content, which are all characteristics expected of the more polar high molecular mass constituents from kraft mill effluents (2628). In addition, the chromatogram of Figure 4 was generated at the absorbance maximum of the retentate fraction (208 nm). Under these conditions, the molecular mass of the retentate fraction was measured to be 40-400 kDa. Note that the fraction appears to be devoid of low molecular mass components, as shown by the 0.002 absorbance unit fullscale overtracing of Figure 4. As noted above, the UV-visible spectrum (Figure 3) of the toxic retentate fraction, dissolved in solvent HzO, had = 208 nm. Although the 208-nm band probably a A,, represents a mixture of several functional groups, they are most likely carbonyl groups, including carboxylate, monoconjugated carboxylate, and aldehyde or hemiacetal groups, as the wavelength is too short to be typical of other carbonyls such as ketones, amides, or esters (29). Otherwise, the spectrum is devoid of features in the 245-290-nm range, indicating that structures in this toxic fraction lacked the aromaticity of BKME (Figure 3). Instead, absorbance in the aromatic region present in BKME was observed in the low molecular mass filtrate. The 'H NMR spectrum of the toxic retentate fraction in D20 is shown in Figure 5. The resonance at 6 8.5 ppm may be indicative of aldehydefhemiacetal groups, and the NMR spectrum also supports a general lack of aromatic structures by the lack of resonances in the 6 6-8 ppm region (29). Resonances from 6 3-6 ppm may reflect chlorinated or hydroxylated aliphatic and alkene structures, with the strong resonances at 6 3.5-3.8 ppm probably arising from methoxyl groups, and 6 0.5-3 ppm represent aliphatic structures. 2418
Environ. Sci. Technoi., Vol. 26, No. 12, 1992
I
10
9
8
7
6
1
1
5
4
I
3
2
1
0
CHEMICAL SHIFT (ppm) F w r e 5. 'H NMR spectrum of the >10kDa fraction, recorded at 500 MHr wlth sample dissolved in 99 atom % D,O and wlth resMuai HDO suppressed. The ordinate is relative intensity of resonance, and the abscissa is based on nonsuppressed HDO resonance, assigned a chemical shift value of 6 4.76 ppm. The spectrum is discontinuous at 6 4.6-5.0 ppm as reslduai HDO resonance was removed from the spectrum for clarlty. Spectrum baseline corrections (each correction was
