HEMODIALYSIS CULTURE OF BACTERIA . _ ’ “ Thesis for the Degree of Ph. D. MICHEGAN STATE UNIVERSITY JOHN EONROE QUARLES, JR.. 1973 ‘w -- r 7 h:‘ L I B R A 14. 1" Michigan State Universrty -.‘ m"“‘ This is to certify that the thesis entitled HEMODIALYSIS CULTURE OF BACTERIA presented by John Monroe Quarles, Jr. has been accepted towards fulfillment of the requirements for Ph . D . degree in Microbiology and Public Health Major professor 0-7 639 y amomc av i HMS & SUNS' L \ 800K BINDERY IN ' UBRA RY amoa ii ”emerge: user “th 4"“; "Mg-nun- if it'll ....Ip4|ul :14! I.."‘ 4 at 5V ABSTRACT HEMODIALYSIS CULTURE OF BACTERIA by John Monroe Quarles, Jr. Hemodialysis culture was implemented and tested as a new technique for growing microorganisms entirely on nutrients obtained by steady- state dialysis with the mammalian blood stream. The technique allowed a culture to grow in direct, continuous communication via an external carotid-jugular shunt with the blood of a living animal, and thus provide a new model system for studying the host-parasite relationship of bacterial septicemia. Two types of culture systems were investigated. The first was comprised of a goat, an artificial-kidney hemodialyzer, and a modu- lar fermentor, and was used to grow large volumes (800 ml) of culture with control of temperature, agitation and aeration. Serratia mar- cescens grew both aerobically and anaerobically in a two-phase growth cycle. Initially the population multiplied exponentially at a maximum rate of about 2 generations/hr. At a cell density of 109.0 to 109.5 viable cells/m1, the culture shifted to a linear growth rate of about 0.2 generations/hr, apparently because diffusion of nutrients through the dialyzer membrane became limiting. The linear growth phase continued to a maximum of about 1010'5 viable cells/ml, at which point the maintenance maximum of the system was reached and viable cell density declined. A toxemia was observed in the host goat, 1 fun.‘ 'i ‘v. John Monroe Quarles, Jr. apparently related to the shift in growth phase or the attainment of a particular cell density. Signs of the toxemia were consistent with those of endotoxemia, including severe leukopenia, pyrexia, and general malaise. Dialysis devices with membranes of nominal molecu— lar weight retentions of 300, 8,000, and 30,000 were used as primary hemodialyzers and as differential dialyzers to obtain cell-free fractionated culture dialysates. The devices which passed the two larger classes of molecules allowed good growth of Serratia, but the 300 molecular weight retention device failed to support growth. When cell—free dialysates of these classes were tested for endotoxic activity by chick embryo lethality and Limulus coagulation tests, the results were equivocal. However, the 8,000 and 30,000 MW dialysates, but not the 300 one, produced profound hypothermia in precooled mice. The artificial kidney-fermentor system did not, however, fully simulate in vivo growth conditions as evidenced by in vitro- like characteristics of Bacillus anthracis grown in the system. This was believed caused at least in part by the exclusion of very low (gaseous) and high molecular weight blood components by the dialysis membranes. A second hemodialysis system, comprised of a newly designed prosthetic hemodialysis culture unit, was also tested. This self- contained unit allowed use of various types of membranes, singly or in combination, and was designed to remain attached to an animal's neck for extended time periods of several days. The unit measured 3.5 x 3.5 x 2.0 cm and contained 3.3 ml of culture. Tests with 4’. iii a v.1. John Monroe Quarles, Jr. B. CUtthracis indicated that this system more closely simulated the in vivo environment than the other system, but restrictions imposed by limited membrane diffusion were believed to prevent a full dupli- cation of in vivo conditions. Although a virulent strain of Trepo- nema pallidum did not multiply in the system, other less fastidious organisms (B. anthracis, S. marcescens, Streptococcus pyogenes, and Blastomyces denmatitidis) did grow. No toxemia was observed with Serratia, probably due to the small volume of the culture and restrictions of diffusion. Trials with protoplasts of Bacillus megaterium, goat peripheral blood cells, and mouse spleen cells demonstrated the prosthetic culture unit has applicability for both delicate microbial and mammalian cells. HEMODIALYSIS CULTURE OF BACTERIA By John Monroe Quarles, Jr. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Microbiology and Public Health 1973 ACKNOWLEDGMENTS I wish to express my thanks to my wife, Joan, for her help, patience, and understanding throughout this work; to the Family Belding for the time and effort they gave; to Dr. Philipp Gerhardt for guidance and the opportunity; and to the fifteen goats for their sacrifices. ii. TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES GENERAL INTRODUCTION GENERAL HISTORICAL REVIEW 4.1 4.2 4.3 4.4 In vitro dialysis culture (1969—1973) 4.1.1 Introduction 4.1.2 Microorganisms 4.1.3 Mammalian cells In vivo dialysis culture (1969-1973) 4.2.1 Introduction 4.2.2 Microorganisms 4.2.3 Mammalian cells Growth of bacteria in vivo vs. in vitro 4.3.1 Introduction 4.3.2 Emperimental models 4.3.3 In vivo growth rates Hemodialysis in medicine 4.4.1 Historical development 4.4.2 Plate-and-frame (Kiil) dialyzer 4 .4 .3 Coil (Kolf'f) dialyzer 4.4.4 Hallow-fiber (Gordie—Dow) dialyzer EXPERIMENTAL RESULTS 5.1 Vascular surgery and prosthesis 5.2 5.1.1 Introduction 5.1.2 Materials and methods 5.1.3 Results 5.1.4 Discussion Fermentor - artificial kidney - goat system 5.2.1 Geat hemodialysis culture of'Serratia marcescens as a model of’septicemia: Growth characteristics (Manuscript) iii Page vi 37 37 37 38 41 41 43 45 (1—20) 5.2.2 TABLE OF CONTENTS (continued) Goat hemodialysis culture of Serratia marcescens as a model of septicemia: Toxemia (Manuscript) 5.3 Prosthetic hemodialysis culture unit 5.3.1 5.3.2 5.3.3 5.3.4 5.3.5 5.3.6 5.3.7 5.3.8 5.3.9 5.3.10 5.3.11 Introduction Materials and methods Prosthetic hemodialysis without culture Prosthetic hemodialysis culture of Serratia marcescens Simulation of in vivo conditions evaluated by growth of Bacillus anthracis in the two hemodialysis culture systems Prosthetic hemodialysis culture of Streptococcus pyogenes Prosthetic hemodialysis culture of Blastomyces dermatitidis Attempted prosthetic hemodialysis culture of’ireponema pallidum Prosthetic hemodialysis maintenance culture of protoplasts Prosthetic hemodialysis maintenance culture of’goat blood cells Prosthetic hemodialysis maintenance culture of’mouse spleen cells 6. GENERAL DISCUSSION 7. GENERAL SUMMARY AND CONCLUSIONS 8. BIBLIOGRAPHY iv Page 46 (1-24) 47 47 49 54 58 64 69 72 75 79 81 84 89 92 96 Section Table 5.1.3 5.2.1 5.2.2 5.3.3.3 5.3.10.3 5.3.11.3 1 1. LIST OF TABLES Title Surgery and maintenance of’arterial-venous shunts in goats Concentrations of’molecular components in goat blood and those in dialysate solution during extended periods of'continuous hemo- dialysis without culture Effect of'passage through the artificial- kidney hemodialyzer on the numbers of blood cells Effect of different membranes on passage of selected blood components during l hr hemo- dialysis Effect of’hemodialysis on peripheral blood cells of the goat Effect of’culture technique on viability and antibody production of‘mouse spleen cells Page 42 45 (15) 46 (16) 56 83 86 Section Figure 5.1.2 5 .2.1 1 2. LIST OF FIGURES Title surgical technique for establishment of vascular shunt. The vessel is permanently ligated distally (right) and temporarily constricted proximally (left). .A vessel tip with attached tubing is inserted (A) and tied in place (B). A double tie is placed around the area of the blood vessel over the inserted tip and cross—tied to the tubing (0). With the vessel tip firmly in place inside the vessel, the temporary constriction is removed to allow blood flow through the tubing (D). Hemodialysis culture system. The major com- ponents of the experimental system shown in the photograph (top) are identified in the tracing (bottom). An artificial kidney is shown, but beaker dialyzers also were employed as the hemodialyzer. Growth curves of'S. marcescens in goat hemo- dialysis culture with an artificial kidney, with a smaller inoculum (A and B) and a larger one (C and D). The results are plotted with both an exponential scale (A and G) and a linear scale (B and D) on the ordinates. Effect of'goat hemodialysis culture of S. marcescens on glucose concentration in the dialysate (open circles) and in the blood (closed circles). Growth curves of'S. marcescens in goat hemo— dialysis culture with highly aerated conditions (A and B) and strictly anaerobic conditions (C and D) in the dialysate—culture circuit. The results are plotted with both an exponential scale (A and G) and a linear scale (B and D) on the ordinates. Growth curves of S. marcescens in goat hemo- dialysis culture with a beaker hemodialyzer having a membrane porosity equivalent to re- tention of'nominal molecular weight 300 (A and B), 10,000 (0 and 0), and 30,000 (E' and F). vi Page 39 45 (16) 45 (17) 45 (18) 45 (19) 45 (20) .1..- a. P» A:\ Section Figure LIST OF FIGURES (continued) Title The results are plotted with both an exponen- tial scale (A, C and E) and a linear scale (B, D and F) on the ordinates. Differential hemodialysis culture system, with three beaker dialyzers of'different porosity inserted in series into the dialysate-culture circuit. In all other respects, the system was the same as that depicted in Fig. 1 of’Gerhardt, Quarles and Belding (1973). Time course of'changes in the responses of a representative animal to hemodialysis without culture (left) and with culture (right): A and E, body temperature; B and F, white blood cells; C and G, red blood cells. In D there is shown the correspond- ing growth curve of’Serratia marcescens in the dialysate-culture circuit. Normalized change in host body temperature as a function of the numbers of’viable cells of’S. marcescens in hemodialysis cultures, averaged from six cultures with four animals. Comparison of’anaerobic with aerobic hemo- dialysis culture conditions on body tem- perature and blood leukocytes of'the animal. Time course of’changes in the response of'a goat to hemodialysis of'purified lipopoly- saccharide endotoxin. In A is shown the body temperature and in B the peripheral leukocyte count. The arrows indicate the time at which the endotoxin was added. Time courses of hemodialysis cultures and host responses with beaker dialyzers of'dif- ferent porosity used as the hemodialyzer, each identified by the porosity threshold in nominal molecular weight of 300 (left, 10,000 (center), and 30,000 (right). vii Page 46 46 46 46 46 46 (17) (18) (19) (20) (21) (22) Section Figure 5.3.2 5.3.4.3 7 1 LIST OF FIGURES (continued) Title Body-temperature response of’precooled mice (average of’10) to injection (0.5 ml, intra- peritoneally) of'salts solution (A), cell- free hemodialysis culture medium (B),.MW 300 beaker-dialysate sample (C), MW 10,000 beaker-dialysate sample (D), and MW 30,000 beaker-dialysate sample (E). Molecular weight distribution of'poly- ethylene glycol M 3,350 before (0) and after (0) equilibrium dialysis through the artificial-kidney membrane. The intersec- tion point represents the porosity exclusion threshold. Prosthetic hemodialysis culture unit. A. The unit is shown mounted on the neck of a goat and connected with tubing to an arterial—venous shunt. B. The unit is shown assembled at the top and beneath it (left to right) are the component parts: blood chamber with silicone rubber tubing, a sheet of cut and punched membrane, a stainless steel support plate, and the dialysate-culture chamber with two sampling extensions. Time courses of'hemodialysis culture and host response to growth of'S. marcescens in the prosthetic unit. viii Page 46 (23) 46 (24) 50 60 ab 5. s -11 3. GENERAL INTRODUCTION In vivo dialysis culture was pioneered in 1896 by Metchnikoff, who implanted a collodion sac containing Vibrio cholera into the peritonal cavity of an animal to demonstrate the production of a diffusible toxin (88). Since then, implanted chambers of various designs have been used for in vivo studies of bacteria, parasites and tissue cells, and diverse other techniques have been developed for culturing organisms in vitro by means of dialysis. These cul- ture systems fundamentally are governed by exchange dialysis: the diffusion of molecules in two directions through a membrane, which separates two solutions with different substances of unequal con— centrations. A culture of microbes growing in a chamber on one side of a membrane is provided nutrients from a chamber on the 0p- posite side of the membrane. As the culture grows, products from it conversely pass through the membrane to the nutrient chamber. The principles and applications of dialysis culture have been re- viewed recently (119) and are updated in another portion of the thesis (Sections 4.1 and 4.2). The determination of growth rates and growth characteristics of bacteria in vivo is difficult. Experimental design and errors inherent in the techniques often make the final results less quan- titative than desired. Some of the techniques used for establishing rates in vivo, and the differences between in vivo and in vitro grown cells, are reviewed in Section 4.3. One of the problems in conducting quantitative studies of bacteria in an animal is that of phagocytic clearing and destruction. Implanted chambers with cell- impermeable membranes provide one means of circumventing this problem. An implanted chamber, however, is limited in use for in vivo culture. The chamber is small, thus limiting the volume of culture, and may be inaccessible or difficult to sample. Because of location inside the host, the entire chamber becomes "walled off” by adher- ing macrophages, and diffusion between culture and host thus becomes reduced or eliminated. The most important limitation is inherent in the fact that the chamber must be implanted in the peritoneal cavity, beneath the skin, or in a similar place where only secondary communication is established with the blood. Direct diffusional communication with the blood stream can be accomplished by means of an "artificial kidney"l, which in recent years has come into routine use in medicine for treatment of chronic kidney failure. Such a hemodialyzer is connected into a bypass of the blood circulation to permit diffusion of body wastes from the blood through membranes to an external reservoir of dialysate solution. The literature on hemodialysis in medicine is reviewed in another portion of the thesis (Section 4.4). "Hemodialysis would seem to offer considerable promise for extracorporeal study of septicemia" (Schultz and Gerhardt, 1969). VI Bi Rafi-unit: hih melementation of this idea was the goal of the research reported in this thesis. The specific objectives were (1) to design and con- struct hemodialysis culture systems entirely supportive of bacterial growth yet atraumatic to the host, (2) to evaluate hemodialysis cul— ture as a model for in vivo septicemic growth of bacteria, and (3) to determine the applicability of the two hemodialysis culture systems that eventuated to diverse organisms, including mammalian as well as microbial cells. 1It should be recognized that the term "artificial kidney" is a misnomer. In practice, these devices commonly do nothing more than dialyze the blood indiscriminately, and thus only partially duplicate the many and complex activities of a healthy kidney. However, it is common practice to use the terms more or less interchangeably and this will also be the case in this thesis. 4. GENERAL HISTORICAL REVIEW 4.1 In vitro dialysis culture (1969-1973) 4.1.1 Introduction The development of procedures and equipment, theory, and ap- plications of dialysis culture were comprehensively reviewed in 1969 by Schultz and Gerhardt (119) and in 1970 by Humphrey (67). The major advantages of growing cells in a dialysis culture system, rather than in a conventional manner, include: the pro— longation of active reproduction, thus allowing the attainment of high densities of viable cells; the stabilization of the maximum sta- tionary phase of the growth cycle; the removal or dilution of dif— fusible inhibitory growth products; the production of culture liquors free of macromolecules from the media; and the capability for study of interactions between separated populations of microbial or mammalian cells. A dialysis system is composed of three functional regions: a reservoir, a dialyzer, and a culture chamber. To achieve maximum efficiency and to allow maximum flexibility and control, the three components are best separated. They may be of almost any size and specific characteristics, depending on the particular goals of the experiment. Generally they are joined through a system of tubing and pumps so that temperature, flow rates, aeration, and agitation of solutions in the nutrient and culture vessels are controlled individually. The dialyzer itself is the key instrument for efficiency in dialysis culture. A plate-and—frame dialyzer, with multiple repeat- ing chambers of membrane separating nutrient from culture solution, was designed and tested by Gallup and Gerhardt (51), Schultz and Gerhardt (119), and Humphrey (67). This "Biodialyzer" was much more suitable for biological work than the chemical dia— lyzers then available. A redesigned version of this dialyzer is in process of manufacture and should soon be available commercially from BioTec AB, Stockholm, Sweden. Plate-and-frame types of arti- ficial kidneys also may be used effectively for in vitro dialysis culture. For example, the Dialung hemodialyzer (Cardiovascular Electrodynamics, Baltimore, Maryland) was applied with good success in dialysis fermentations by Abbott and Gerhardt (1970). Several types of hollow-fiber dialyzers have been introduced recently by commercial sources (Cordis-Dow Corporation, Miami, Florida; Amicon Corporation, Lexington, Massachusetts). These dialyzers are designed primarily for the separation and concentration of mole— cules, but should be applicable to dialysis culture systems. To date, their application to culture of microorganisms has not been thoroughly studied or reported, but preliminary studies by Quarles and Gerhardt (unpublished) have demonstrated the successful use of Cordis-Dow beaker dialyzers and artificial kidneys for dialysis culture of Serratia marcescens and Bacillus anthracis. Schultz and Gerhardt, in their 1969 review (page 37, refer- ence 119) listed 50 different algae, bacteria, fungi, protozoa, C. '11 i4 1;: r9 (7‘ If, and tissue cells which had been studied with dialysis culture in vitro to that time. This review of dialysis culture, including some applications of the technique since 1969, is updated in the follow- ing sections. 4.1.2 Microorganisms Pan and Umbreit (97) used dialysis culture techniques to demon- strate that presumably obligate autotrOphic bacteria will grow on glucose rather than inorganic energy sources if metabolic products are prevented from accumulating by use of dialysis. Nitrosomonas europaea, Nitrobacter agilis, Thiobacillus denitrificans, T. nea- politanus, and T. thioparus were grown on glucose-salts media in the absence of the specific inorganic energy sources. Pyruvic acid was the metabolic product found to inhibit N. agilis, but the toxic product for N. europaea was not identified. Results with the Thiobacillus species indicated that pyruvate (or related keto acids) may be more inhibitory to the organism when growing on glucose than when growing on a specific nutrient. This work was important in that it showed a qualitative as well as a quantitative difference between dialysis and non-dialysis culture, and also showed that redefinition of the term "obligate autotroph" may be necessary. Pan and Umbreit (98) also studied growth of autotrophs and heterotrophs in mixed cultures (both with conventional and dialysis cultures) and found that the effects of autotrOphs on heterotrophs, and the effects of heterotrophs on autotrophs, were highly specific. For example, Escherichia coli and Pseudomonas aeruginosa had can. a ...., a ...M «I: :qs: o~-‘ essentially no effect on Nitrobacter agilis, Streptococcus faecalis caused slight inhibition; Serratia marcescens, Hydrogenomonas eu- tropha and saccharomyces cerevisiae caused slight stimulation; and candida albicans caused significant stimulation on the growth of Nitrobacter. Dialysis culture showed less effect and was considered less successful for the demonstration of interactions than the use of ordinary mixed cultures. Technical considerations such as in— adequate diffusion and aeration may have influenced the results, however. A double dialysis method was used by Edwards (46) to produce antigenic material from the mold that causes "farmer's lung disease." Nutrient medium within dialysis tubing was equilibrated with an aqueous solution of sodium chloride, and mycelia of Micropolyspora faeni were inoculated into the resulting dialysate. After growth of the organism, the growth medium (containing organisms, dialy- zable material from the original nutrient broth, and products of microbial metabolism) was placed in other dialysis tubing and a second dialysis was conducted to separate dialyzable molecules from the mycelium and macromolecular end products. The resultant or- ganisms and macromolecules constituted the antigens for further studies. Dialysis culture was used by Friedman and Gaden (49) to study growth and lactic acid production by Lactobacillus delbrueckii. They confirmed the inhibitory effect of lactate by dialyzing the product away from the culture medium. Inhibition of the culture by lactate occurred after the log phase. By maintaining low lac- tate concentrations at that time, they obtained higher than usual specific growth rates and maximum cell concentrations. Addition— ally, overall acid production in dialysis culture was significantly higher than in conventional non-dialysis batch culture. Wyrick and Gooder (144) grew L-phase variants of Strepto- coccus faecium on membrane filters placed on solid agar to demon- strate that L-elements passed through filters with 0.22 p pores. They also studied reversion of protoplasts, but found that proto— plasts could not pass the filter and form colonies under it. However, in some cases covering the filter with solid L-phase medium before inoculation of protOplasts gave rise to colonies. Protoplasts could not form colonies on the membrane without a covering of agar, thus indicating that the three-dimensional effect of agar was important. In similar studies, Clive and Landman (28) investigated the reversion of protoplasts and L—phase variants of Bacillus subtilis on membrane filters and reported similar results. They also re— ported that the reversion of protoplasts was enhanced by layering wall material on the filters. Several oceanic phytoplankton species were grown in dialysis culture, both in the laboratory and in the field, by Jensen et al. (69). They tested the technique with eleven species of algae and obtained good growth and dense cultures with artificial media and non-enriched sea water. For studying growth characteristics, they constructed a dialysis culture flask based on the flask apparatus of Gerhardt and Gallup (54). A second system consisted of bags of dialysis tubing suspended on a rotary support in apprOpriate media. Sartorius membrane-filters of various porosities and dialysis tubing of regenerated cellulose were used as dialysis membranes. From pre— liminary work, they concluded that dialysis culture was well suited to studying marine microorganisms and could likely be used for ob- taining high densities of algae, assessing the nutritive quality of sea water for phytoplankton production, monitoring biotic and abiotic factors in water, bioassaying pollutants, and studying species interactions. Two important suggested advantages to dial- ysis culture were that it allows cells to accumulate trace sub- stances from large volumes of water and it can eliminate grazing in field studies on phytOplankton growth. In a series of three articles on dialysis fermentation, Abbott and Gerhardt (1,2,3) demonstrated that dialysis culture yields a dramatic increase in the amount of a diffusible product and cell mass. They first studied the conversion of naphthalene to salicylic acid by Pseudomonas fluorescens and found that the cumulative total amount of both cell mass and salicylate continued to increase throughout the 15 days of the experiment. At termination, product formation and total cell mass were still increasing although less than 10% of the cells were viable (1). A kinetic analysis showed the important role of maintenance metabolism in salicylate produc— tion in the system, with maintenance accounting for about 84% of theto 5'5“ iv'iCi' -* «wen i . haul ., In ’2 3933a 10 the total salicylate produced. Dialysis fermentation yielded a 20-fold increase in the production of salicylate compared to a conventional non-dialysis fermentation, and a 2.6-fold increase compared to an optimally recycled non-dialysis control (2). In companion studies (3), an attempt was made to use dialysis to alleviate product control over threonine production by an auxo- troph of Escherichia coli. Abbott and Gerhardt unexpectedly found an inhibition instead of an enhancement. This was explained by depletion of a—e-diaminopimelic acid as the limiting factor rather than threonine inhibition of its own synthesis. The imbalance was not corrected by exogenous replacement of diaminopimelate. Trypanosoma lewisi was cultured in media composed of a "gel phase" of protein, agar, and defibrinated rabbit blood in dialysis tubing by Dusanic (45). He found agar necessary for good growth of the organism. The agar presumably acted as a sink to absorb inhibi- tory products from the medium. In addition to yielding satisfactory growth, the system provided a good means for studying amino acid utilization by the organism. Lysed blood in dialysis tubing was used by Crook et al.(37) as a nutrient source for growing Leishmania mexicana for antigen production. These workers considered dialysis culture the method of choice for eliminating the contamination of antigen by medium components. The technique of growing bacteria on dialysis membranes spread over agar medium was reported to provide a good yield of 11 essentially all the extracellular products of Staphylococcus aureus by Bamen and Haque (10). Although not all products were present in concentrated form, the significant finding was that the entire complement of products could be detected. Concentration of a bacterial growth product via dialysis cul— ture was demonstrated by Bissell et al. (15) and Sarner et al. (116). They grew a sarcina strain (coccus P) in dialysis bags suspended in media to achieve a highly concentrated proteinase. A cell den- sity of about 24 x 108 cells/m1 was obtained in dialysis culture, as compared to 6 x 108 in flasks. The proteinase was retained by dialysis membranes and therefore was concentrated. Chet, Fogel and Mitchell (27) studied the chemical detection of microbial prey by bacterial predators. They found that an exu- date of the phycomycete prey, Pythium debaryanum, was dialyzable. However, a chemotactic exudate of a diatom prey, Skeletonema cos- tatum, was not. In both cases a highly motile pseudomonad species was the predator. In these studies, dialysis was used only to help characterize the exudates of the prey which attracted the predator. Dialysis growth systems were not used, although the possibility seems obvious. The use of dialysates of various protein substrates as media for growth of bacteria in flask culture is common. Pollock (102, 103) found that dialyzable nutrients support growth of mycoplasma. Dialysates of peptone, yeast extract, or hydrolyzed casein were used by Elmer and Nicherson (47) to study the nutritional require— ments of Mucor rouxii. They demonstrated the presence of a u.- 55» I All: 9“ “h in on T) h if“ »‘-1 12 ' which was dialyzable factor, termed Y—factor for "yeast promoting,‘ active in changing the culture from a mycelial phase to a yeast phase. The molecular weight of this substance was estimated as less than 700. A dialysate of brain heart infusion was used by Parisi and Kiley (99) to study chromogenic variants of Staphylococcus aureus. They suggested that differences in amino acid metabolism between parent and mutant account for differences in chromogenic character- istics in at least some cases. 4.1.3 Mammalian cells Dialysis culture techniques have also been used for in vitro studies with mammalian cells and tissues. Rose and colleagues, over a period of some 20 years, have developed and refined equipment for the in vitro dialysis growth of various types of tissue cultures (110, 111). Their equipment has evolved from a single multipurpose chamber to a lZ-chamber circumfusion system which allows control of nutrient and gas flow rates, pressure, agitation and temperature. This system was designed to yield not only growth of tissue cells but also differentiation. Using the system, they observed morpho- logical differentiation and maintenance of the differentiated state in fetal rat ovary cells and thyroid cells. A less elaborate, multipurpose, microperfusion chamber for dialysis culture of microorganisms as well as tissue cells was re- ported by Poyton and Branton (104). This chamber was designed to allow observation of the organisms via phase microscopy while the cells were growing. The chamber was used for studying various or— ganisms which included E. coli, 3 blue-green alga (Nostoc), a 11!. 13 marine amoeba, and several fungi. The article also included refer— ences and a comparison and summary of the most important features of some 14 different perfusion chamber designs. The primary immune response of spleen cells from unimmunized mice was studied by growing the cells on dialysis membranes above a reservoir of medium (82). Three to five days after exposure to sheep or horse erythrocytes, significant numbers of antibody- producing cells were detected. This system had the advantage of allowing the spleen cells to grow undisturbed by agitation or mix— ing, so that foci of antibody-producing cells were not disturbed. In an article with important implications for dialysis and hemodialysis culture of both microbial and mammalian cells, Knazek et a1. (75) reported the use of an artificial capillary device to perfuse both gas and medium into tissue cultures. The capillaries provided a matrix for cell attachment so that secreted products could be harvested without disturbing the culture. The technique of com- bining both nutrient—passing and gas-passing membranes in a single device is a significant step because dialysis bmembranes are essen- tially impermeable to gases such as oxygen and carbon dioxide. The system as devised and tested by these workers allowed cells to attain solid tissue densities in vitro similar to those found in vivo. A variety of inhibitory and stimulatory products of tissues, tissue cultures, and media have been detected and studied by dia- lysis techniques. However, the reported studies did not seem to l4 demonstrate any trend in the probability that a given dialysate or retentate would contain either a stimulatory or an inhibitory fac- tor. Each instance of medium and cells seemed a separate case, and extrapolation from one to another was difficult. For example, Robinson et al. (109) found that a non-dialyzable factor from the perfusion of porcine livers and spleens stimulated colony formation from bone—marrow cells. Metcalf (87) found that dialysates of tissue culture medium from several types of normal cells and leu— kemic cells inhibited colony formation by mouse bone-marrow cells. The inhibition was not species specific, and both leukemic and normal cells produced about equal inhibition. A dialyzable sub- stance from tissue culture medium which could replace the active ingredient of calf serum and stimulate the mitotic activity of BHK cells in vitro was reported by Shodell (124). Taylor et al. (135) found that a dialyzable material from Bacto—peptone could act as a serum substitute for mammalian cells. The dialysate was used as a supplement in conventional medium for growing rabbit myo— cardial cells. The active factor in the dialysate was not identi— fied, but was something other than free amino acids. In all of these cases, the investigators did not use dialysis culture systems but rather studied only dialysates or retentates of dialysis. The theory of dialysis culture and the applications of other workers suggest that at least some of these studies might have produced better results, in terms of increased cell mass and products, if dialysis culture had been used. 15 4.2 In vivo dialysis culture (1969-1973) 4.2.1 Introduction The peritoneal cavity was first used for in vivo dialysis cul- ture (88) and it remains the most widely used today. The most com- mon technique involves implantation of sealed chambers or "sacs" comprised of some type of membrane. Intestinal ligation in situ also has been used for a type of dialysis culture, most frequently for relatively short—term growth experiments and studies on toxin elaboration by bacteria. Examples of this technique are the studies of De et a1. (40, 41) and Kasai and Burrows (71) with Vibrio cholera, Taylor et al. (134) and Smith and Halls (129) with Escherichia coli, and Hauschild et a1. (63) with Clostridium perfringens. However, intestinal ligation appears less suitable than cell-impermeable diffusion chambers for a wide variety of studies, especially those requiring an extended time period. An historical review of in vivo dialysis techniques and a sum- mary of the major achievements up to 1969 is included in the review of Schultz and Gerhardt (119). Some of the more important studies using in vivo dialysis since that review are summarized in the following paragraphs. 4.2.2 Microorganisms A particularly promising area for using diffusion chambers and in vivo dialysis is for the culturing of fastidious or slow growing parasitic microbes. In 1960, Huff et al.(66) showed that the exoerythrocytic stages of malaria could be cultured in implanted l6 chambers within chickens, turkeys, ducks, chick embryos and mice. Chickens were especially capable of encapsulating the chambers, which often is one of the major disadvantages of the system. Another finding was that Plasmodium gallinaceum, although about 0.5 to 1.5 p in size, could pass through 0.45 p Millipore membranes. In later studies with parasites, Petithory and Raussed (101) immu— nized mice to Trypanosoma gambiense via growth in cellulose diffusion chambers. Relatively fastidious bacteria have also been grown in implanted chambers. Rightsel and Wiygul (108) cultured Mycobacterium leprae- murium in cell—impermeable chambers. They found a relationship be- tween growth of the organism and susceptibility of the host. Another important finding was that the organism grew well in chambers in the absence of other tissue cells. That is, living cells were not essential, and multiplication occurred in a cell-free environment. However, inclusion of macrOphages in the chamber seemed to yield better results. The organism had a generation time of 6 to 8 days with macrophages and 11 days in the cell-free environments. Other studies by these workers (143) showed that Mycobacterium leprae- murium grew well in chambers within animals containing cells (human embryonic skin) from a species other than the natural host (mice). In fact, the human skin cells enhanced growth. Chambers without skin cells gave greater yields in mice, the normal host, than in guinea pigs. Less fastidious organisms frequently have been grown in dialysis chambers. Dent et a1. (43) found that bacterial products bub... $61". 5. ' I J, - ~v 17 released from contaminated bursa chambers enhanced antibody produc- tion in chickens. Osebold and colleagues (94, 95, 96) conducted a series of experiments on cellular immunity to Listeria monocyto- genes in mice. They reported that live bacteria must contact macrophages for development of good cellular immunity. Humoral substances produced in response to diffusible antigens were not able to inactivate the organism. In several instances, sham chambers without bacteria increased the host's resistance to later challenge with Listeria, thus indicating that nonspecific resis- tance is associated with the presence of a foreign body. This was a very important finding, in that it showed that resistance studies must be carefully interpreted to separate the role of the chamber from that of the enclosed organism. Osebold et a1. (95, 96) also found that tissue reactions progressed around the chambers to the point that a chronic abscess was formed, and that pleomorphic mutants of Listeria appeared during prolonged cultivation in implanted chambers. Guthrie and Nunez (60) studied delayed hypersensitivity to 1-f1uoro-2,4-dinitrobenzene (DNFB) and Mycobacterium tuberculosis by implanting chambers containing peritoneal exudate cells from guinea pigs sensitized to these agents into unsensitized animals. The recipients developed specific skin-test reactions to DNFB but not to old tuberculin or to purified protein derivative. Apparently there were basic differences in the release of these factors from cells. 18 In an interesting modification of the in vivo dialysis technique, Arko (8) implanted a polyethylene practice golf ball (a hollow plastic ball with numerous holes) in the subcutaneous tissue of rabbits and allowed encapsulation by connective tissue to form an artificial chamber and natural membrane. Neisseria gonorrhoeae grew and retained virulence in these chambers, and the chambers were usable for a period of several weeks. The success of this system raised the possibility of studying in vivo interactions between various factors such as antibody, leukocytes, complement and bacteria. In addition, a relatively simple system was provided for culturing virulent N. gonorrhoeae. It should be noted that the ”membrane" enclosing the organisms was tissue and not an artificial membrane. It seems likely that most bacteria can be grown in some type of in vivo dialysis system, and, indeed, a variety have been cul— tured in implanted cell—impermeable chambers. Three possible lim— itations to the implanted chamber technique should be realized how- ever: (1) chambers may elicit a nonspecific immune response ex- clusive of that caused by the contained bacteria, (2) chambers may be "walled off" to such an extent that diffusion is minimal or eliminated, and (3) chambers may be difficult or impossible to sample without sacrificing the host. 4.2.3 Mammalian cells Normal mammalian cells, antibody-producing cells, and tumor cells have been grown in diffusion chambers. Alekseeva and Yunker (5) implanted 1 x 2 cm envelopes of membrane filters, with 0.1 to 19 0.3 p pores, into animals to grow mouse and rabbit cells. Trans- plantation immunity reactions were investigated by Ambrose et a1. (7), who found that a diffusible toxic antibody was involved in tumor transplantation rejection. This factor appeared to be an IgG immunoglobulin and present only in hamsters immunized specifically to SV40 tumor specific transplants. Adenovirus 31 tumor cells in diffusion chambers were not inhibited in hamsters immunized against SV40 tumor specific transplantation antigen. Borella (19) used diffusion chambers to study the regulating mechanism of IgM and IgG antibody-forming cells. He showed that the expression of IgM memory was inhibited by the appearance of IgG-producing cells as early as 4 to 10 days following primary antigenic stimulation. In other studies (18, 20), Borella showed that the antibody response of spleen cells cultured in diffusion chambers suppressed virus replication. An interesting change in the type of host was used by Tucker (137). He grew a variety of tumor cells in chambers implanted in chick chorioallantois. Growth and differentiation of liver and spleen cell homografts were studied by autoradiography and diffusion chambers in the ab- dominal cavity of newts (58). This work demonstrated that diffu— sion chambers can be implanted into animals other than traditional laboratory animals. Harrison and cadworkers (62) investigated histocompatibility interactions between nixed types of mouse spleen cells by use of diffusion chambers. The effect of radiation on mammalian cells has been studied with implanted diffusion chambers by a number of investigators. 20 Makinodan and co-workers published a series of articles on the effect of radiation on the growth and antibody synthesis of various cells in immunodiffusion chambers (59, 92, 113, 114, 138, 139). The role of thymus cells in restoring resistance to radiated thymectomized mice was studied by Schneiberg et al. (118), who found that thymus- bearing chambers implanted in thymectomized mice restored radiation resistance. Spertzel and Pollard (130) conducted similar studies on spleen cells by implanting membrane-filter chambers in mice. They reported the presence of a humoral factor that increased recovery of the animals. Blood cell cultures have been investigated by several workers. Rasmussen and Hjortdal (107) grew homologous blood and buffy-coat cells for 3 week periods in diffusion chambers in the peritoneum of rats. Microscopic examination showed that neither fibroblasts nor connective tissue fibers deve10ped inside the chambers if con- tamination with extraneous connective tissue was prevented. They concluded, therefore, that the development of fibroblasts in blood and buffy—coat culture is due to contamination with connective tissue cells during sampling of the blood. Benestad (12) maintained mouse bone-marrow cells and blood leukocytes in implanted chambers. He found that granulocyted and macrophages were formed but lympho- cytes and mature end cells were lost. Additionally, the mouse bone-marrow cells synthesized DNA actively when placed within in vivo chambers, as Opposed to a rapid decline reported by others for in vitro studies. Kuralesova (78) also used diffusion 21 chambers to culture and compare bone-marrow cells from mice which had received post irradiation injections of syngenic bonedmarrow cells from mice without these injections. Kitsukawa (74) culture cells of rabbit aortic endothelium in diffusion chambers that were implanted (in l to two weeks) in rabbit abdominal cavities, in an attempt to study the biologic nature of the cells. The cells often showed ring formations of two to three cells, and tended to form "alveolar—like" arrangements around clusters of erythrocytes. Chamber cultures of up to 10 days duration were used by Laerum and Bayum (79) to study the viability of hairlessemouse epidermal cells. They reported that the cells appeared intact for the first 24 hr, but then a significant cell loss occurred. Cell loss out- weighed proliferation under their experimental conditions. In summary, it seems apparent that in vivo growth of tissue cells of various types is successful in implanted diffusion chambers, and that many functions of the cells can be studied by this technique. 4.3 Growth of bacteria in vivo vs. in vitro 4.3.1 Introduction It is a truism that microbes grow differently in vivo than in vitro, but there are very few experimental models for studying these assumed differences. Most studies which purport to show different characteristics fail to withstand close scrutiny. Very few bac- teria actually have definite, easily scored, unequivocal markers 22 that distinguish the two types of growth. Even the determination of a growth rate in vivo becomes a major problem, with an error in counting viable cells greater than that in vitro, because in most instances the errors of in vivo sampling compound the errors of in vitro plate counts. Usually a study on the growth of a bacterium in a host simply uses a given cell density as an original inoculum and uses death or clinical signs at a particular time as the end point, instead of viable cell density in time intervals. The importance of using in vivo grown bacteria to determine the biochemical mechanisms of microbial disease was pointed out by Smith in several significant articles (125, 126, 127). His underlying idea is that virulent strains possess genetic differences from aviru- lent strains. These differences, possibly minor or subtle, are fully expressed only in vivo. Also, the host tissues are continually changing when under microbial attack, and this spectrum of change is not easily reproduced in vitro. These differences in the organism are often difficult to detect and document, but some informative models are available. The following discussion will review the growth of bacteria in vivo in terms of the differences between in vitro and in vivo char— acteristics, and the techniques for determination of in vivo growth rates 0 23 4.3.2 Experimental models Most approaches have involved a search for physiologic or anti— genic differences between virulent and avirulent strains or a search for toxic substances produced by the virulent strain. Segal and Bloch (121, 122) and Bloch and Mizuno (l7) demon— strated that cells of Mycobacterium tuberculosis grown in vitro have biochemical and pathologic differences from those grown in vivo. In vitro grown cells have a higher hydrogen transfer rate, and res- piration is more stimulated by substrates such as glucose, glycerol, lactate, acetate and pyruvate. In vivo grown cells respond to salicylate, which is considered by some workers as correlated with pathogenicity. There are no differences in morphology, staining properties, colonial characteristics, or rate of growth on oleic acid-albumin agar. The degree of aggregation is important in deter— mining virulence: small units (single cells or small aggregates) are less virulent than larger ones. When the two types of organisms are prepared similarly aggregated, the in vivo grown cells are more virulent for mice than the in vitro grown ones. A phenol-killed vaccine prepared from in vitro grown organisms produces better immunity in mice than that from in vivo grown organisms, thus indicating an immunologic difference. One important factor leading to virulence is often that of resistance to phagocytosis. In studies on Yersinia pestis, Burrows and co-workers (24, 25, 26) showed that virulent and avirulent strains are indistinguishable in vitro, but are easily identified r‘ 4 O.. be}. ”5' u." u“, -. , A. 1") : h., A... are i s \ 24 in vivo in mice. Avirulent strains are phagocytosed completely, but virulent strains rapidly become phagocytosis—resistant and kill the host. Later studies indicated that resistant strains can be grown in vitro, and that resistance seems associated at least par- tially with the so-called V- and W-antigens which are produced in vivo. Smith and his colleagues (126, 127) grew Y. pestis in guinea pigs and collected the plasma and body fluids to demonstrate that in vivo grown organisms produce toxin which kills guinea pigs and mice. Fukui et a1. (50) sought to detect metabolic differences between in vivo and in vitro grown Y. pestis cells, but found no significant differences which could account for pathogenicity. However, the in vivo grown cells acquire the ability to oxidize gluconate, which was suggested as posSibly important in the production of virulence antigen. Cells of Bacillus anthracis are both morphologically and meta- bolically different in vivo than in vitro. In vivo grown cells tend to encapsulate to a greater extent, form no chains or only short chains, have a square-ended morphological appearance, and fail to sporulate (55, 125, 127). Smith and Tempest (128) reported that the in vivo grown organism used large quantities of glutamine, threonine, tryptophan, and glycine, but did not use histidine, lysine, tyrosine, phenylalanine, methonine, or alanine for growth. In general, amino acids which are important antimetabolites in vivo are not in vitro. Thus, there seem to be different, but unexplained, utilizations of amino acids in the two environments. 25 The brucellae provide a good example of tissue specificity in vivo caused by the presence of a particular nutrient. Erythritol isolated from fetal and maternal tissues (placentae, fetal fluids, chorions) of susceptible species was shown to be highly stimulatory for growth of the organism (127). Cells of Brucella abortus in vitro preferentially use erythritol instead of glucose even if 1000 times more concentrated. Primary invasion by the organism is a separate problem, not controlled by erythritol. Beining and Kennedy (11) compared Staphylococcus aureus cells grown in vitro on trypticase soy agar or broth to those grown in vivo in guinea pigs. The two types of cells are similar in a number of characteristics: morphology, size, staining, bound and soluble coagulases, bacteriophage type, antibiotic sensitivities, common fermentation reactions, DNA base composition, qualitative tests for hemolysins, deoxyribonuclease, ribonuclease, staphylokinase, pro- tease, lipase, and phosphatase. However, the in vivo grown cells are significantly different in certain other characteristics: respiratory rates, virulence in mice and guinea pigs, agglutina- bility, agar gel diffusion tests, growth on tellurite-glycine agar, and the quantitative production of deoxyribonuclease, alpha- hemolysin, leukocidin and hyaluronidase. Gellenbeck (52) showed that in vitro grown S. aureus cells have a lower exogenous respiratory rate than in vivo grown cells. Metabolic differences between the two types of cells of Streptococcus were reported by Gordon and Gibbons (56). They found 26 that Streptococcus mitis cells that are grown in vivo in gnotobiotic rats and mice have a 3 to 5 times greater glycolytic activity on a per cell basis than those grown in vitro. 4.3.3 In vivo growth rates Accurate determinations of growth rates in vivo present a greater problem than determination of differences between in vivo and in vitro grown cells. An in vivo growth rate is actually the net of growth rate less clearance rate. The most obvious and straightforward method to obtain a growth rate is to inject iden- tical inocula into a number of animals and periodically count viable organisms from the entire body of one or more animals. Generally, this requires either the use of germ—free animals or a selective and differential medium to eliminate normal flora. The experimental procedure is usually to remove the skin and feet, grind the weighed carcass in a blender, and plate out a known sample of the suspension. Berry and co-workers used essentially this technique to study salmonella typhimurium in mice (13, 14). They found that approxi- mately 109 viable bacterial cells were recoverable from the animal at time of death. The generation time of the organism in mice is about 58 min, as compared to approximately 22 min in brain heart infusion medium. Another approach involves the removal of only selected organs, e.g., the blood, lungs, liver, spleen, and kidneys. The organ is weighed, ground up to release the bacteria, suspended in a diluent, 27 and then a representative sample is used for viable cell determinations. This technique has been used frequently for both qualitative and quantitative studies. Collins (30, 31, 32, 33, 34) determined growth rates and yields in both normal and immunized or multiply infected animals by re- moval and quantitative culture of selected organs. Several Salmo- nella species were studied in this manner. He found that the bac- terial population in the intestine tended to stabilize at about 104 to 105 organisms after oral inoculation. Oral inoculation was most often used in an attempt to duplicate closely the normal sequence of events during infection. Increasing the size of chal- lenge doses had little influence on the final outcome of the in— fection, but did increase the rate of spread throughout the body. Representative growth rates were difficult to obtain, but the rates were relatively slow. Maximum numbers of organisms per organ were about 103 to 104 after 6 days growth. Several hours per generation was common. Problems of phagocytosis and spreading throughout the body made rate studies more difficult in these experiments. Srivastava and Thompson (131) suggested that best results are obtained if the organism of interest is not mixed with other or- ganisms, as is the case with most total body experiments. Conse- quently they used local lesions in the thighs of mice to enumerate the growth rates of E. coli and to study the effect of streptomycin in the organism in vivo. They stripped off the skin, amputated the thigh, and homogenized the tissue. There was a loss of recov- 7 erable bacteria from 5.8 x 10 viable cells to 1.6 x 107 cells 28 immediately after inoculation, and a l— to l%—hr lag before an overall exponential growth of approximately one generation per 90 min for about 12 hr. A maximum of about 1010 viable cells/ml of homogenate was reached at 12 hr and maintained for 3 days before the viable cell count decreased. Young cells were much more sensitive to streptomycin than old ones. It is obvious that comparing this type of experiment with one involving organs or total body studies is very difficult. The environmental conditions are different and great differences are to be found in rates and maximum numbers attained. Meynell (90) and Maw and Meynell (84) proposed a method for measuring division and death rates of bacteria in vivo. This tech- nique, termed the superinfecting phage method, involves lysogenizing the infecting bacterium with a temperate phage and superinfecting with a differentially marked mutant prophage before inoculation into a host. The superinfecting phage neither lysogenizes most of the bacteria nor replicates during bacterial multiplication. The pro- portion of bacteria with superinfecting phage thus decreases with each generation of bacteria in a predictable manner. Therefore, the proportion of bacteria carrying superinfecting phage can be determined at each generation and a count obtained. The proportion of superinfected cells is determined by inducing vegetative phage growth by exposure to ultraviolet light and plating on bacterial indicator strains specific for the prophage and superinfecting phage respectively. The technique and analytic procedures are somewhat 29 complicated and tedious, but were used for E. coli and S. typhi- murium. In these studies, the division rate of the salmonella in mouse spleen was only 5 to 10% of the maximum in vitro rate. Generation times were 8 to 10 hr in vivo and about 30 min in vitro. Matsuo (83) studied the growth of Mycobacterium leprae in mouse foot pads and found that multiplication of the organism de- pended considerably on the ambient temperature. Multiplication was faster at 20 C than at uncontrolled or partially controlled room temperatures. Growth rates were very slow in all cases, as an original inoculum of 104 organisms per foot pad increased to only about 3 x 106 after 44 wk. His enumeration technique was to mince and grind foot pad tissue, and then to remove and count the acid-fast bacteria. Saymen et al. (117) published a method for quantitatively de- termining viable bacteria in infected tissues or wounds. This method involved placing a known number of bacteria into a stan- dardized surface wound created by surgery. Bacteria were obtained from the wound by surgical removal of the area and washing or swabbing the tissue. Other workers have discussed similar techniques (22, 53, 123). For bacteria which cause a septicemia, determination of viable cells in the blood may be satisfactory for growth rates and total yields. This method is simple and has the advantage of not re- quiring destruction of the animal, so that repeated samples can be obtained from an individual host. Of course, large numbers of bac— teria may be sequestered in various organs and cells of the body 30 and thus missed by sampling the blood. This approach can provide useful information, however, as shown by the studies of Lincoln and his colleagues with Bacillus anthracis (80, 81). They examined growth in separate tissues and whole animals and also conducted quantitative studies on the in vivo growth of the anthrax bacillus by determining the number of organisms per ml of blood. They pre- pared a mathematical model of the septicemia, based on their in vivo studies, which agreed well with actual experimental data from infected animals. 4.4 Hemodialysis in medicine 4.4.1 Historical development The basic concept of hemodialysis was put into medical practice some 60 years ago. In 1913, Abel, Rowntree and Turner (4) published the first article on a hemodialyzer, or "artificial kidney." They envisioned the technique as a means of providing aid in emergencies such as renal failure. The possibilities of using hemodialyzers in chronic cases were unrealized until fairly recently, and relief from acute renal failure remained the main use for many years. In fact, in 1949, one of the leading dialysis centers in Boston suggested that hemodialysis in chronic renal disease was useful only for acute cases or in preparing uremic cases for surgery (85). It was 10 years later before prophylactic or repeated dialysis came into routine use. Frequent dialysis is required for chronic patients because levels of retained products quickly build up to toxic levels. The 31 unsolved problems of the early days of dialysis treatment made re- peated dialysis difficult or impossible. These problems were mainly technical and were related to the inability to establish and main— tain permanent circulatory shunts. Access to the artery and vein required a surgeon to insert and remove the cannulae each time, and bleeding at the surgical site following heparinization often resulted. These factors combined to use up the available blood vessels of patients rapidly and made repeated, long-term dialysis unattractive. Also, the equipment was large, difficult to set up and sterilize, and expensive to use and maintain. Other problems were not technical but rather were a matter of practice and experience. It was soon realized that patients with chronic renal difficulties and slowly progressing renal failure had gradually adjusted to the situation and their systems had compensated to some extent. Rapid, efficient dialysis often actually made them worse than before treatment, as demonstrated in the "disequilibrium syndrome" (136, 141). The situation remained much the same through 1959. That year, two schools of thought were represented by two articles in the Transactions of'the American Society for Artificial Internal Organs. One article reflected the previous years of difficulties and stated the prevailing view that dialysis was not the answer to chronic uremia (115). The other reported on the use of daily hemodialysis via polyvinyl cannulae (93). Cannulae in the latter situation were established in vessels (radial artery and antecubital vein) and maintained patent throughout the treatment period. Thus, repeated 32 dialysis was possible without surgical cannulation each time. More significantly, this allowed prophylactic and not just emer- gency dialysis. That is, the chronic patient could be dialyzed periodically, before a toxic crisis occurred. During the 1960's, techniques and equipment developed and improved rapidly. Quinton et al. (105, 106) published a paper on chronic cannulation for prolonged hemodialysis. Reports were also published on intermittent hemodialysis (120) and indwelling can- nulae and by~passes (64, 91, 105, 142). They and other workers developed the basic procedures which are used with only minor modifications today. Solving the technical and medical problems of hemodialysis made possible the saving of lives which otherwise were certain to be lost. The choice between the goals of either maintenance or rehabilitation of life raised difficult financial, legal, and moral problems which are beyond the scope of this review, but which have not been satisfactorily resolved to date. The technology of equipment and delivery systans also developed rapidly. Multipatient facilities, which allowed dialysis of several patients at a time, were initiated. Central systems for delivery of dialysate and monitoring equipment and patients were developed. Home dialysis became possible as a means of reducing costs and plac- ing fewer restrictions on the patient (16, 38, 86). Techniques for putting shunts in the foot or ankle were developed and tested, thus allowing the patient to conduct self dialysis (9). In all these 33 cases, a combination of improving equipment and procedures had to be combined with increased experience. The first practical clinical dialyzer employed cellophane membranes (44), and similar membranes are still the basic type in use. Although a variety of hemodialyzers have been successfully tested for both chronic and acute dialysis, the most commonly used ones can be classed into three main types: (1) the flat filter press or plate-and-frame type, of which the Kiil dialyzer is proba- bly the best known (72, 73), (2) the coil or membrane-tubing type, with either single or twin coils, of which the Kolff hemodialyzer is the best example (76), and (3) the more recently developed hollow— fiber type, represented by the Cordis—Dow dialyzer (56, 132). Each of these types of hemodialyzers has particular charac- teristics, advantages, and disadvantages. Generally the choice of a dialyzer depends on a combination of features and requirements, including personal preferences of the user. A short description of an example of each of these types is given below: 4.4.2 Plate-and-frame (Kiil) dialyzer (72, 73) A representative Kiil dialyzer is rectangular in shape and 100 x 40 x 20 cm in size. Its weight is about 27 KG. Blood is circulated through parallel channels between sheets of cellophane. A dialysate solution is circulated countercurrently on the opposite side of the membranes. The two solutions flow along the long axis of the dialyzer. Membranes are supported by a series of ridges and grooves in plates on the dialysate side of each membrane sheet. 34 Blood flows through the dialyzer in a thin film to increase the con- tact with membranes and to increase the efficiency of dialysis. This type of dialyzer commonly is made of stainless steel and machined polypropylene. About 400 to 500 m1 of blood is required to fill it and the accompanying tubing, but most of the blood can be returned to the patient at the end of dialysis. The most serious disadvantage is probably the difficulty in assembling and "steri- lizing" the dialyzer. It must be washed, assembled, pressure tested for leaks, liquid "sterilized", and rewashed before use. Proper stretching and placement of previously moistened cellophane membranes on the grooved board requires great care and some practice. About 1% hours work, part of which requires two persons, is usually needed for best results. Dialysis efficiency is good and the unit can be used for chronic or acute dialysis. A two layer system is most often used, but more layers can be added to increase the membrane area. Smaller models, requiring less blood, are available for small patients. 4.4.3 Coil (Kolff) dialyzer (76, 77) This dialyzer is basically a length of cellulose tubing wrapped around a central cylinder and surrounded by a fiber glass or plastic supporting screen. The entire assembly is immersed in a tank of dialysate solution. Two parallel coils are often used to increase membrane area and efficiency, thus reducing dialysis time. Blood is circulated through the tubing, and dialysate is circulated cross- wise to the blood. A blood pump is required to maintain satisfac— tory blood flow. These dialyzers come in three sizes (small, medium, 35 and large) with coil lengths of 5.20, 8.00, and 10.75 meters and dialysis surfaces of 9,000, 14,500, and 19,000 sq cm respectively. Relatively large amounts of blood are required for priming and filling the system (550 to 950 ml). The coil dialyzers are efficient and satisfactory for both chronic and acute dialysis. Coil dialyzers are considered easy to assemble and prepare for use, especially in comparison to Kiil type dialyzers. Coil hemo— dialyzers can be mass produced, presterilized, and discarded after use. Their main disadvantages are in the requirement for mechanical pumping of blood with consequent possible trauma, and in the large priming volume. 4.4.4 Hollow-fiber (Cordis-Dow) dialyzer (57, 132) The hollow-fiber dialyzer is a relatively newly developed type which is very compact and efficient and seems to offer several advan— tages over both Kiil and Kolff types of hemodialyzers. The hollow- fiber dialyzer is a bundle of approximately 11,000 hollow membrane fibers surrounded by a plastic jacket. The fibers are constructed of regenerated cellulose, and are 285 p in diameter (225 p inner diameter) and about 13.5 cm in length. The unit is cylindrical, about 21.6 cm long x 7.0 cm diameter in overall size, and weighs only 680 g, when filled. Blood flows through the inside of the fibers, and dialysate is circulated countercurrently through the outside. The fibers give a total effective membrane surface area of l to 1.3 sq meters, depending on the model. 36 These units are commercially produced, presterilized and stored in formaldehyde, and sold ready for washing and use. Resistance to blood passage is very low and blood pumps are not required. No assembly is necessary prior to use. The dialyzers are considered disposable, but may be flushed, resterilized and reused several times under usual conditions. Only about 100 m1 of blood is required to fill the dialyzer and most of this can be returned to the patient. Advantages of the hollow-fiber dialyzer are that no assembly and only a minimum of preparation (primarily washing away formaldehyde) are required before use, and the units are compact and efficient. Additionally, only a small amount of blood is required to fill the unit, and pumping is not necessary. Significant disadvantages have not become apparent, and the units are coming into widespread use. 5. EXPERIMENTAL RESULTS 5.1 Vascular surgery and prosthesis 5.1.1 Introduction Hemodialysis is dependent on access to a major circuit of the blood circulation, for example shunting the blood supply from an artery through an external bypass to the dialyzer and back into a vein. In human clinical usage, an artery and vein of the wrist or ankle are most commonly used. With experimental animals, vessels of the legs or neck are more convenient. There are two primary techniques for achieving the required access to the blood circulation. The older and most frequently used is the external prosthetic shunt. The other technique involves the creation internally of an arterilized vein (i.e.,an arterial-venous fistula), with access through repeated venipuncture (23). The ex- ternal shunt is more vulnerable to clotting, to possible infection and (especially in experimental animals) to unplanned opening and resultant exsanguination. An external shunt also may restrict activities of the patient or animal. The internal arterial—venous fistula overcomes these limitations and is reported to be preferred by some patients. It does, however, require venipuncture with relatively large needles for each dialysis, and the surgery is more complicated than that required for the external shunt. 37 38 We decided to use the external shunt for hemodialysis culture, recognizing possible shortcomings, for the following reasons: (1) limitations in surgical competence, (2) ease of connecting and disconnecting the variety of equipment to be tested, (3) ease of monitoring and sampling from the shunt, and (4) preferred use of the carotid artery and jugular vein. 5.1.2 Materials and methods The arterial—venous cannula was established by surgical tech- niques like those used for human patients. The shunt was similar to those originally reported and modified by Scribner and co-workers (105, 106). The animal was anaesthesized and an incision of approximately 10 cm made on the right side of the freshly shaved skin of the neck. The incision was parallel to the direction of the major blood ves— sels, about halfway between the animal's head and shoulder. The jugular vein and carotid artery were isolated, with the vagus nerve carefully separated from the artery. Beginning with the artery, each vessel was ligated distally. The vessel was then temporarily constricted proximally to prevent blood loss, and a small opening was made (with scissors) between the temporary constriction and the permanent ligation. A Teflon vessel tip attached to Silastic tubing was inserted into the blood vessel and tied in place. Double or triple ties were used, with cross ties around the vessel and tubing. Other ties were used to attach the process to nearby muscle or fascia. The procedure is shown diagrammatically in Fig. 1. 39 FIG. 1. Surgical technique for establishment of vascular shunt. The vessel is permanently ligated distally (right) and temporarily constricted proximally (left). A vessel tip with attached tubing is inserted (A) and tied in place (B). A double tie is placed around the area of the blood vessel over the inserted tip and cross-tied to the tubing (C). With the vessel tip firmly in place inside the vessel, the temporary constriction is removed to allow blood flow through the tubing (D). 40 The Silastic tubing was aligned parallel to the vessels for about 5 cm and then brought outside the body through openings in the skin about 1 cm to either side of the incision. The arterial tube was above the incision and the venous tube, below. The incision was closed with sutures and allowed to heal. The two externalized tubes were connected with a Teflon connector or drug infusion "T" to establish extracorporeal arterial— venous circulation. In practice, an infusion "T" was usually left in place for 24 to 48 hr after surgery to allow periodic flushing of the shunt, and was then replaced with a connector for the duration of the shunt. Several hours after surgery, heparin anticoagulant therapy was initiated and maintained as long as the shunt remained patent. About 8,000 to 10,000 USP units of heparin (as a sterile solution of sodium heparin, Upjohn Company) were given subcutaneously every 12 hr to maintain the blood clotting time about 3 times normal. The shunt was put into use for hemodialysis by clamping shut the Silastic tubing on each side of the connector, removing the connector, and attaching a hemodialyzer to the arterial and venous tubes. The clamps were removed and blood was allowed to flow from the animal, through the dialyzer, and back into the animal. Heparin was continuously infused into the hemodialyzer as required to prevent clotting. The experimental animals were all young-adult female goats of mixed breed, in weights from 50 to 125 lb. The animals were 41 maintained unrestrained indoors, in rooms approximately 10 x 12 ft. Prior to surgery, the goats were checked for parasites and the general state of health, and various normal values (clotting times, blood cell numbers) were determined. Following surgery, the shunt was covered with a light dressing of gauze and tape, and the entire neck was covered with a canvas collar for protection of the shunt. 5.1.3 Results A summary of surgical results and maintenance of the arterial- venous shunt for all the experimental animals used in the study is given in Table 1. The weight for each animal is the weight at time of surgery. 5.1.4 Discussion Goats reportedly are poor anaesthetic risks during surgery (48) and our results reflected this observation. Three of the fifteen goats died during surgery, apparently due to the anaesthetic. In our experience, the usual anaesthetic of choice, sodium pentabarbitol, must be used with caution. Better and very promising results were obtained with an experimental drug designated CI-744, recommended and obtained through the courtesy of Dr. R. W. Coppock of the Parke- Davis Company. According to the company ("CI-744 Brochure for Investigators," Feb. 29, 1972), the drug is a combination of two ingredients: tiletamine hydrochloride (CI-634) and a diazepinone (CI-716). Tiletamine is a central nervous system depressant which produces analgesia and cataleploid anaesthesia. The diazepinone is a tranquilizer and CNS depressant with anticonvulsant and anti- anxiety activity. 42 TABLE 1. Surgery and maintenance of'arterial-venous shunts in goats Animal Weight Duration di§l§gis Number Reason for number (1b) 0f shunt time Of termination (days) (hr) dialyses l 125 3 0 0 Clotted 2 68 69 6 7 Exsanguinated 3 88 69 68 8 Exsanguinated 4 74 -- -- - Died, surgery 5 58 46 66 7 Clotted 6 51 -- -- - Died,surgery 7 8O 47 39 4 Exsanguinated 8 72 58 20 3 Clotted 9 50 16 4 2 Clotted 10 52 -- -- - Died, surgery 11 52 65 92 14 Exsanguinated 12 100 23 46 6 Clotted 13 50 24 39 6 Clotted 14 8O 32 66 6 Clotted 15 60 51 209 8 Clotted 43 After the prosthesis was established, the primary problems were in preventing accidental exsanguination via unplanned opening of the shunt and in preventing clotting in the shunt. The first problem was solved reasonably well by use of protective canvas collars over the shunt, hobbles on the animal's legs, and removal of all protruding objects from the room. The second problem was not completely solved, despite anticoagulant therapy with heparin, and clotting remained somewhat a matter of chance throughout the study. However, the mini- mum time the shunts remained patent was only once less than three weeks (the first attempt is excluded) and was usually four weeks or greater. This was considered an acceptable duration. 5.2 Fermentor - artificial kidney - goat system The hemodialysis culture system as initially designed was com- prised of an animal, a commercial artificial kidney, and a modular fermentor, connected through rubber and tygon tubing. A biological pump (Maisch metering pump, Tuthill Pump Company, Chicago) circulated the culture, and the animal's heart circulated the blood. This system allowed the use of several hundred milliliters of culture with precise control over temperature, aeration and agitation of the culture suspension. The design was similar in concept to the fermentor-dialysis system described for in vitro dialysis culture by Schultz and Gerhardt (119). A Dialung artificial kidney (obtained through the courtesy of Dr. W. G. Esmond, Baltimore, Maryland) was initially tested for use in the hemodialysis culture system. This plate—and—frame type of 44 dialyzer was designed for use with human kidney patients. In our use with the goat, however, it caused hemolysis and rapid clotting of the blood to the extent that blood flow in the dialyzer ceased. It was thus considered unacceptable for our work. A Cordis-Dow hollow-fiber artificial kidney (Cordis Corpora- tion, Miami, Florida) then was tested and successfully used for hemodialysis with the goat. Therefore, it was chosen as the hemo- dialyzer for development of the system. The basic characteristics of the fermentor — artificial kidney - goat system, its application as a model of septicemic growth, and the demonstration in vivo of a dialyzable toxic product of Serratia marcescens are described in the following manuscripts. The manu- scripts were written in the format for and will be submitted for publication in Infection and Immunity or a comparable journal. 45 5.2.1 Goat hemodialysis culture of'Serratia marcescens as a model of septicemia: Growth characteristics. OManuscript) PHILIPP GERHARDT, JOHN M. QUARLES and RALPH C. BELDING Manuscript for submission to Infection and Immunity (ABSTRACT) Hemodialysis was employed to simulate growth conditions in mammalian blood but without phagocytosis. The blood stream was shunted surgically via prosthetic tubing from a carotid artery through the hollow—fiber membranes in an artificial-kidney hemo- dialyzer and back into a jugular vein. Culture in the dialysate solution concurrently was pumped from a modular fermentor through the hemodialyzer jacket outside of the membranes and back into the fermentor. Hemodialysis between the two circuits was maintained continuously. With the goat and Serratia marcescens selected as a host-parasite model, this new culture system allowed the inoculum initially to multiply at the maximum exponential rate and then at a lesser linear rate, equally well under aerobic or anaerobic con- ditions. Beaker hemodialyzers were equally effective with membrane porosity equivalent to nominal molecular weights of 10,000 or 30,000, but not 300. -1- (INTRODUCTION) We have devised a new way to grow organisms by in vivo dialysis, with the experimental rationale derived from the fact that the in- terior milieu in animals is maintained essentially by circulation of the blood. Dialyzers connected directly with the circulatory system are in common clinical use as artificial kidneys for humans. It appeared that such a hemodialyzer could be used to establish con- tinuous communication between the blood stream of an animal and a fixed volume of dialysate solution inoculated with an organism. In this way, dialyzable molecular constituents of the blood would dif- fuse into the culture and so feed it, yet the blood cells and macro- molecules would be too large to diffuse through the membrane, so phagocytosis and immunological reactions against the culture would be prevented. Conversely, metabolic products of small molecular size from the culture would diffuse through the membrane barrier into the blood and so relieve the feedback inhibition that often limits growth in a closed culture system. Also, diffusible toxins cxruld exert effects. Furthermore, membranes of different porosity could be employed in the system, for example, to distinguish between the exchange of macro- and micromolecules in hemodialysis. The system would be analogous to the batch fermentor - continuous reservoir mode of operating an in vitro dialysis culture (see Fig. 8, reference 7), but with the blood supply of an animal used as the nutrient reservoir. Historically, in vivo dialysis culture first was attempted by implanting a collodion membrane sac into the peritoneal cavity-— in 1896, Metchnikoff grew cholera bacteria in this way and demonstrated -2- the production of a diffusible toxin (4). Dialysis chambers subse- quently were implanted in the peritoneal cavity and elsewhere for a number of purposes with a number of organisms, including animal cells and tissues, as reviewed by Schultz and Gerhardt (7). The use of implanted chambers, however, is limited by the different environment in the peritoneal or other body fluid than in the blood itself, the occluding growth of macrophages on the membrane surface, the restricted size, and the difficulty in sampling. We conceived the use of hemodialysis to offset these limita- tions and essentially to transpose septicemic conditions extra- corporeally. The feasibility of hemodialysis culture was demon- strated in tests with the domestic goat and Serratia marcescens selected as the host-parasite model. The methodology, an unusual bimodal pattern of bacterial growth, and the effects of three porosities of membrane in hemodialysis culture are described in the present paper. The toxemic host response is reported in a succeeding one. -3- MATERIALS AND METHODS The hemodialysis culture system is depicted and diagrammed in Fig. 1. Blood from an experimental animal was circulated by the heart through one side of an artificial—kidney hemodialyzer (the blood circuit), and dialysate culture from a modular fermentor was circulated by a pump through the opposite side (the dialysate- culture circuit). The goat was selected as the experimental animal because of its convenient size, long neck for accessibility in vascular surgery, placid disposition, and hardiness. The goat has disadvantages in its relative sensitivity to anaesthesia and hemolysis (1). Short- haired domestic goats of mixed breed were used. All were females, 1 to 3 years of age, and weighed from 50 to 100 pounds. They were maintained indoors in stalls, unrestrained except for a rear-leg hobble. During dialysis trials the goat was partially harnessed into a caged platform but was free to stand, lie, eat, drink, and excrete wastes. A permanent external prosthetic shunt was established between a carotid artery and jugular vein in the neck by vascular surgery, based on the techniques of Quinton et a1. (5,6). Shunts, vessel tips, and tubing were made of medical grade Teflon, Tygon, or silicone rubber (Cobe Laboratories, Inc.), as used for humans. Sodium heparin (Upjohn Co.) was used to prevent the blood from clotting, with subcutaneous injections of 15,000 to 20,000 USP units/24 hr. During dialysis, about 1,500 USP units/hr were -4- continuously infused into the arterial shunt (Sage Infusion Pump, Model 240; Orion Instrument Co.). Ethyl isobutrazine hydrochloride was administered as a tranquilizer in early experiments, but later was obviated by training the animal. Neither drug affected growth of the test bacterium. The possibility of blood clotting in the hemodialyzer was monitored visually and by a thermometer in the effluent tubing, where the temperature went down if the flow was reduced by clotting. A given goat became usable for hemodialysis experimentation about a week after surgery and remained so for about 1 to 3 months, when accidental exsanguination by the goat or the formation of a blood clot in an artery or vein (despite heparinization) caused termination. A hollow-fiber artificial kidney was selected as the hemodialyzer (Cordis-Dow Model 2 or 3, Cordis Laboratories Inc.). The jacketed cylindrical unit measures 7.0 cm in diameter and 21.6 cm in height, and contains a bundle of 11,000 hollow-fiber membranes fabricated from regenerated cellulose, through which the blood is circulated. Each fiber is 13.5 cm in length and 225 pm in inside diameter, and the bundle provides about 1 m2 of total membrane surface area. The dia- lyzer was sterilized with 1.5% formaldehyde and rinsed thoroughly with water before use. The culture-dialysate was circulated outside the hollow-fiber membranes, through the dialyzer jacket. Beaker chemical dialyzers with a looped bundle of hollow- fiber membranes (Cordis Laboratories Inc.) also were employed as -5- hemodialyzers. The beakers measure 7 cm in diameter and 14 cm in height, and provide a total membrane surface area of about 1,000 cm2. Beaker dialyzers with three different retention porosities were used, identified in terms of the nominal molecular weight (MW) threshold for dialysis: MW 300 ("Osmolyzer", Model b/HFU-l), MW 10,000 ("Dia— lyzer", Model b/HFD-l, with the same porosity as the artificial- kidney dialyzer), and MW 30,000 ("Ultrafilter", Model b/HFU—l). The beaker dialyzers first tested caused hemolysis and toxemia, and so were unsatisfactory for use as hemodialyzers. However, the embedding material used to secure the fiber bundles subsequently was changed in manufacture, and more recent products proved satisfactory as hemo- dialyzers. A modular 1-liter glass fermentor with control of temperature, agitation and aeration (Microferm Model MP 102, New Brunswick Scien- tific Co.) was used to contain the dialysate-culture suspension. Anaerobic conditions were achieved by sparging the fermentor contents with sterile argon, which was scrubbed free of oxygen by passage over hot copper wire kept reduced with a stream of hydrogen gas. Aerobic conditions were achieved by sparging the fermentor contents with filter—sterilized air at a rate of 2.4 1/min and by driving the impellor at a rate of 300 rev/min. A sterilizable gear pump (Maisch metering pump, Tuthill Pump Co.) was used to circulate the dialysate-culture from the fermentor through -6- the tubing to the dialyzer and back. Thick—walled rubber tubing was used for the dialysate-culture circuit, but the tubing in the blood circuit was medical-grade Tygon or silicone rubber. The dialysate—culture circuit routinely was charged with 800 m1 of a glucose-salts solution approximately balanced in makeup to that of goat blood (1). The solution was constituted as follows per 100 ml: 50 mg glucose, 310 mg Na+ (as NaCl and Na acetate), 15 mg K+ (as KCl), 11 mg Ca++ (as CaClZ) and 3.7 mg Mg++ (as MgClz). The solution volume represented about 40 to 50% of the total blood volume of a goat. S. marcescens strain 8UK was selected as the test organism be- cause of its previous use in developmental studies on in vitro culture (2,3). This gram-negative bacterium occurs mostly as single cells, grows rapidly on synthetic or natural media under either aerobic or anaerobic conditions, and is red pigmented when grown on most media at about 30 C. S. marcescens commonly is thought of as saprophytic, but is quite capable of parasitic growth in the body fluids, produces endotoxin and may cause serious diseases in man. The inoculum was prepared from an aerated exponential culture at 35 C in trypticase soy broth (BBL), which was sedimented and resuspended in the dialysate solution to an appropriate concentration. Hemodialysis was initiated and the system allowed to equilibrate for 1 hr prior to introducing the inoculum. The culture was main- tained at 39 C (the approximate normal temperature of goats) and con- sequently was not pigmented. Samples of the culture were removed periodically by syringe-and-needle through a self-sealing rubber 'm e were are for PIOE lact -7- diaphragm positioned in the circuit tubing. Cell populations were measured by optical density and by viable cell counts that were performed by surface plating on trypticase soy agar (BBL). Cell—free samples of the culture for biochemical analyses were obtained by membrane filtration. Samples of blood were collected in ethylenediamine tetraacetic acid (EDTA) to obtain plasma, or were allowed to clot to obtain serum. The plasma and serum samples were immediately stored at -20 C or -70 C until assayed. Assays for the following routinely were made on the samples with a flame photometer and autoanalyzer (Model SMA 12, Technicon Corp.) by use of standard procedures: sodium, potassium, calcium, magnesium, inorganic phosphate, glucose, urea nitrogen, uric acid, total protein, albumin, bilirubin, cholesterol, alkaline phosphatase, lactic dehydrogenase, and glutamic-oxaloacetic transaminase. -3- RESULTS During hemodialysis in the absence of culture, dialyzable mole- cules in the blood reached and then maintained equilibrium with the dialysate solution. In Table 1 are listed the greatest and least values obtained from periodic sampling of the blood and dialysate during 9 hr of control dialysis. All of the values remained essen- tially within normal ranges. The lower values in the dialysate solu- tion usually were those in the original dialysate, before equilibrium was established. Large molecules in the blood, such as the proteins, did not pass through the dialysis membrane. Subsequently it was demonstrated that the equilibrium was attained within 1 hr. The original glucose-salts solution was inadequate to support growth of S. marcescens, mainly because of the absence of a nitrogen source. Even after attainment of equilibrium with blood, the dialy- sate solution supported only a slight amount of bacterial multipli- cation if samples of the dialysate were withdrawn into separate culture tubes and inoculated. However, when hemodialysis was maintained continuously for an extended period, an inoculum of S. marcescens in the dialysate cir- cuit of the system multiplied extensively and in a characteristic bimodal pattern (Fig. 2). Multiplication of the inoculum proceeded, usually after a short lag period, at an exponential rate (about 9.5 2 generations/hr) until a population density of about 10 viable cells/ml was reached (Fig. 2A). At this juncture, the multiplication rate sixtl 9331 from line Enlt Sist catj tier ”(jig Eran rESU -9- rate changed from an exponential to a linear function at about one— sixth the generation time (Fig. 2B). The shift in growth kinetics was explained by the bacterial population reaching a density for which the nutrient demand exceeded the steady-state supply of nutrients from hemodialysis. Because the diffusion of nutrients occurred at a linear rate, bacterial multiplication was limited similarly. The effect of bacterial multiplication on glucose concentrations in the dialysate and in the blood is shown in Fig. 3. The dialysate glucose decreased rapidly to low values as the culture approached a density of about 109 viable cells/ml and became undetectable by the time a density of 1010 viable cells/ml was reached. The blood glucose also decreased, but not so greatly. When a larger inoculum was employed, the bimodal pattern of multiplication also occurred, but with the exponential rate per- sisting only for a short period (Fig. 20). The rate of multipli- cation soon became linear (Fig. 2D), and the population continued to increase in this fashion for several hours. A maximum was reached at about 1010'6 viable cells/ml, followed by a decline in the population. At first the cultures were managed with highly aerobic condi- tions in the fermentor because of knowledge that relatively little oxygen from the blood can diffuse through a cellulose dialysis mem- brane (2,7). However, strictly anaerobic conditions in the fermentor resulted in the same bimodal growth pattern and supported essentially -10- the same exponential and linear rates of multiplication (Fig. 4). Apparently the limitation on growth rate of S. marcescens by oxygen supply (2,3,8) occurs only in artificial media or at population densi- ties in excess of those attained with hemodialysis culture. The rate of exponential multiplication in hemodialysis culture, about 2.0 generations/hr, was essentially as great as the maximum that had been attained by any other means of culture or type of medium. The artificial-kidney hemodialyzer was available only with one membrane porosity, which is equivalent to retention of molecules larger in nominal molecular weight than about 10,000. However, chemical beaker dialyzers, which were constructed similarly with a bundle of hollow membrane fibers and available with three different types of dialysis membrane, were found satisfactory for use as a hemodialyzer. In Table l are included the concentrations of some blood components in the beaker dialysates after equilibrium hemo- dialysis without culture. The growth response in goat hemodialysis culture of S. marcescens with beaker dialyzers of three different membrane porosities is shown in Fig. 5. The MW 300 beaker dialyzer did not support a significant change in bacterial population over an extended time period (Fig. 5A and B) although dialysate glucose decreased slightly during the trial, indicating metabolic activity by the culture. In vitro dialysis with the MW 300 beaker and trypticase soy broth medium in the reservoir also failed to support growth. The MW 10,000 beaker, which is constructed of the same membrane as in the artificial kidney, and the MW 30,000 one -11- both supported a bimodal pattern of growth (Fig. 5C and D, Fig. 5E and F, respectively) much the same as that with the artificial kidney (compare with Fig. 2). Representative rates were 1.7 generations/hr during exponential growth and 0.2 generations/hr during linear growth with the MW 10,000 beaker, and 1.8 generations/hr (exponential) and 0.3 generations/hr (linear) with the MW 30,000 beaker. DISCUSSION A bacterial septicemia in vivo, in terms of growth kinetics, usually represents a steady state between parasite multiplication in the blood serum and clearance by phagocytosis, with any net growth considerably less than the maximum rate. Only when the host defense mechanisms are overwhelmed can the parasite population increase rapidly and exponentially. Such a fulminating septicemia was simulated by the conditions of hemodialysis culture, in which assimilable nutrients from the serum were continuously available to the bacterial culture but phagocytes were excluded. The results during the primary growth phase of S. marcescens demonstrated that even an organism capable of rapid multiplication can attain its maximum exponential rate from the continuous supply of dialyzable blood constituents. The secondary phase of linear growth, i.e., first-order kinetics, ensued when the rate of nutrient demand by the bacterial population exceeded the steady-state rate of nutrient diffusion through the membrane, which then became limiting. This situation seems unlikely -12.. t0 happen with an in vivo septicemia, and would be delayed in hemodialysis (1111 tures with smaller inocula (compare Fig. 2B and 2D) or those with Organisms that multiply more slowly. The onset of linear growth is Primarily a function of the effective surface area of membrane relative t0 the amount of organisms. Only a limited number and type of hollow-fiber membranes were avail— able for use in hemodialysis culture. The MW 300 beaker membrane proved Jincapable of supporting growth, apparently because essential nutrient molecules were retained. The MW 10,000 membrane in the beaker or arti— fic ial kidney and the MW 30,000 beaker membranes both supported a similar Pattern of growth. Both regulate the passage of much the same size class of Small molecules from the serum, and both retain serum proteins. Experiments usually were terminated after 12 to 15 hr of con- tinuous hemodialysis culture because the animals showed signs of acute physiological stress, such as pyrexia and leukopenia, which were presumed to result from toxins produced by the culture and con- tinIJously diffused into the animal. This toxemia is characterized in a succeeding paper. 7 -13- ACKNOWLEDGMENTS We gratefully thank Josephine Belding, R.N., for surgical assistance and T. C. Belding, D.V.M., for surgical anaesthesia. The work was supported by the U. S. Public Health Service via grant AI-09760 fI’Om the National Institute of Allergy and Infectious Diseases, a general research support grant to the College of Veterinary Medicine at the Uni- Versity, and training grant GM—01911. -14- LITERATURE CITED Fletcher, W. S., A. L. Rogers, and S. S. Donaldson. 1964. The use of the goat as an experimental animal. Lab. Animal Care 14:65-90. Gallup, D. M., and P. Gerhardt. 1963. Dialysis fermentor system for concentrated culture of microorganisms. Appl. Microbiol. 11:506-512. Gerhardt, P., and D. M. Gallup. 1963. Dialysis flask for con- centrated culture of microorganisms. J. Bacteriol. 86:919-929. Metchnikoff, E., E. Roux, and T. (Aurelli-) Salimbeni. 1896. Toxine et antitoxine cholerique. Ann. Inst. Pasteur 10:257-282. Quinton, W., D. Dillard, J. J. Cole, and B. H. Scribner. 1961. Possible improvements in the technique of long term cannulation of blood vessels. Trans. Amer. Soc. Artif. Internal Organs 7:60-63. Quinton, W. D., D. Dillard, and B. H. Scribner. 1960. Cannula- tion of blood vessels for prolonged hemodialysis. Trans. Amer. Soc. Artif. Internal Organs 6:104—113. Schultz, J. S., and P. Gerhardt. 1969. Dialysis culture of microorganisms: design, theory, and results. Bacteriol. Rev. 33:1—47. Smith, C. G., and M. J. Johnson. 1954. Aeration requirements for growth of aerobic microorganisms. J. Bacteriol. 68:346-350. t:.....(\ \.. I}... .s..\‘ . .t... .. .\ ~_\.~A -15- .mumNsHmHeoame seamen euHs u: «H on OH was smeuHa HmHUHHHuum nuHs u: a .muwoo aoanmx mo wmmofiammamnu Q can one .mufiao woouumah . I O O J 4 I § I \ 307 (n J .1 III 0 . u 20> .J 0 15 > a no i I o J oziééIoIzoééé‘eIoné HOURS OF CULTURE FIG. 2. Growth curves of S. marcescens in goat hemodialysis culture with an artificial kidney, with a smaller inoculum (A and B) and a larger one (C and D). The results are plotted with both an exponential scale (A and C) and a linear scale (B and D) on the ordinates. -13- LOG VIABLE CELLS/ML cu- uh- ‘- .- di- '0' IOO 90 80 70 60 50 4O 30- 20- lOb O I I I I I O 2 4 6 8 IO l2 HOURS OF CULTURE I GO 1 I DIALYSATE GLUCOSE (MG/IOO ML) FIG. 3. Effect of goat hemodialysis culture of S. marcescens on glucose concentration in the dialysate (open circles) and in the blood (closed circles). -19- AEROBIC ANAEROBIC I'IITITfirIIII LOG VIABLE CELLS/ ML -n- ur- db ult- db .J 2 B o I?) 430» -~ - .1 LL] 0 320'. di- -4 m S > '0 JI- q 0’ 9 l l L l l 1 l 0246 8|Ol2024 6 8|Ol2 HOURS OF CULTURE FIG. 4. Growth curves of S. marcescens in goat hemodialysis culture with highly aerated conditions (A and B) and strictly anaerobic conditions (C and D) in the dialysate-culture circuit. The results are plotted with both an exponential scale (A and C) and a linear scale (B and D) on the ordinates. -19- AEROBIC ANAEROBIC I I I I r 1* r’ I r’1* r I I _’ -I I: '\ (D .J _J d LLJ U U .1 H (D S > 0 -I O _.l l q I -I _.l :2 a H .1 30 I. -II- ..J U U m I- I- -I _J 2() ‘ (D S > IO - .. r m 9 ()L 1 l L 1, l 1 I. L 0246 8|Ol2024 6 8|Ol2 HOURS or CULTURE FIG. 4. Growth curves of'Sh marcescens in goat hemodialysis culture with highly aerated conditions (A and B) and strictly anaerobic conditions (C and D) in the dialysate-culture circuit. The results are plotted with both an exponential scale (A and C) and a linear scale (B and D) on the ordinates. -20- II MW 300 MW I0,000 MW 30,000 I I I I I U I I I T I l l I T I I r .J z '0'. ‘b «I- q \ (D ._I d o 9 b CIV- db q I.” 5’ E ‘ 8 " III- I... _‘ E >. 1 £9 3 7 b {P 4)- -4 6 :- qu- -n- q l l l l j l l l 1 l L l l L l A l l U T r I I T I I I I T T U T T T T 1 '0 b «I- -I d . _, B D F : \ b ‘1 fit. q I (I) P I .J I at i o 6 b db 1 In i a. 4 I- 4 l s I > I In 2" -' I 9 I I oi .xxanoo—¢.1LL_J_J O 2 ‘IIB 6 KDIZ OIZI‘I 6 6»K3l2 O 2244 6 8 "BIZ HOURS OF CULTURE FIG. 5. Growth curves of'S. marcescens in goat hemodialysis culture with a beaker hemodialyzer having a membrane porosity equivalent to retention of’nominal molecular weight 300 (A and B), 10,000 (C and D), and 30,000 (E and F). The results are plotted with both an exponential scale (A, C and E) and a linear scale (B, D and F) on the ordinates. 46 5.2.2 Geat hemodialysis culture of Serratia marcescens as a model of septicemia: beemia. (Manuscript) JOHN M. QUARLES, RALPH C. BELDING, TEOFILA C. BEAMAN, and PHILIPP GERHARDT Manuscript for submission to Infection and immunity (ABSTRACT) Serratia marcescens grown by continuous hemodialysis in a fermentor - artificial kidney system caused a general toxemia with acute pyrexia and leukopenia in the host goat, more so with aerobic than anaerobic culture conditions. A large amount of purified S. marcescens endotoxin in the system produced similar effects, but only transiently and to much less an extent. The dialyzable toxic culture material depressed the body temperature of precooled mice, even after boiling the sample, but produced equivocal results with chick-embryo lethality and Limulus coagulation tests. The use of beaker dialyzers with different membrane porosities indicated that the size of the dialyzable toxic material was equivalent to a nominal molecular weight between approximately 300 and 10,000. By analysis of the membrane diffusion threshold, the maximum molecular size was further defined relative to a rigid globular protein of 15,000 in molecular weight and 1.9 nm in hydrodynamic radius or a flexible fibrous polyglycol of 5,500 in molecular weight and 2.6 nm in hydrodynamic radius. (INTRODUCTION) Hemodialysis culture is a way to obtain growth conditions in vitro simulating those in mammalian blood, but unrestrained by phago- cytosis. The blood stream is surgically shunted via prosthetic tubing from a major artery through the hollow-fiber membranes of an artificial— kidney hemodialyzer and back into a major vein of the animal. Concur- rently, the dialysate culture is pumped from a modular fermentor through the hemodialyzer jacket outside the membranes and back into the fermentor. In this way, the culture and blood are in continuous diffusional communication but the cells of each are separated. With the goat and Serratia marcescens selected as a host-parasite model, this new system allows the bacterial inoculum initially to mul- tiply at the maximum exponential rate. Eventually the rate of diffusion becomes limiting, and the multiplication of bacteria then changes to a lesser and linear rate. Populations in the order of 1010.5 viable cells/ml are attained in a 12—hr period (2). The response of the goat host to this simulated fulminating septi— cemia with S. marcescens was examined in the studies reported below. Because the membrane in the artificial kidney prevents the passage of molecules larger in nominal molecular weight than approximately 10,000, including macromolecular toxins, a toxemic response was not expected and did not occur initially as the culture grew. However, an acute toxemia with signs of pyrexia and leukOpenia developed secondarily. This host response to culture was compared with that to purified lipo- P°1ysaccharide endotoxin in the same situation, the dialyzable toxic -2- material was separated and its heat stability and biological activity were assayed with tests considered indicative of endotoxin, and the maximum size of the dialyzable toxic material was defined relative to two types of molecules. MATERIALS AND METHODS The hemodialysis culture system, with a modular fermentor and either an artificial kidney or a beaker dialyzer, was the same as de- scribed previously (2) except for the use in certain trials of three beaker dialyzers also in the dialysate-culture circuit (Fig. 1). In this arrangement, three separate samples of dialyzable culture prod— uct were obtained simultaneously while the animal was hemodialyzed with the culture. Unless otherwise stipulated, the culture in the fermentor was aerated and stirred. Vascular surgery and management of the goat for hemodialysis were the same as described previously (2). The body temperature was measured with a rectal thermometer as frequently as needed. Samples of blood for biochemical analyses were prepared and analyzed as be- fore (2). Samples of blood for cell counts were treated with EDTA to prevent clotting. Leukocyte and erythrocyte counts then were made by microscopic examination of the cells by use of counting cham- bers, and differential cell counts by use of Wright-stained smears. Tests for endotoxin-like activity were made by chick-embryo le- thality (4) and Limulus coagulation assays (5,6). Activity also was assayed by the effect on the temperature regulation of precooled mice held at 4 C, with 0.5 ml of sample injected intraperitoneally. Rectal -3- temperatures were taken every 30 min for 3 hr (YSI model 428C Tele- Thermometer with physiological probe, Yellow Springs Instrument Co.). Ten mice were used for each sample, and the average temperature change was determined. Purified lipopolysaccharide endotoxin was studied by allowing the hemodialysis culture system (with the artificial kidney) thoroughly to equilibrate for 4 hr and then introducing the endotoxin into the fer- mentor at a final concentration of 0.05 mg/ml. The endotoxin was a commercial preparation from S. marcescens (Lipopolysaccharide W, Difco Laboratories). Membrane diffusion thresholds of the Cordis—Dow hollow-fiber artificial kidney were determined by conducting dialysis of poly- ethylene glycols (PEG, Union Carbide) of known molecular sizes and molecular weights. Polyethylene glycol 4000 (41.0 osmolal), with number average molecular weight (HQ) 3350 was dialyzed against 20.5 osmolal PEG 20,000 (M; 17,500). Polythylene glycol 1540 (M; 1540) and PEG E 9000 Ofih 9,500, Dow Chemical Co.) also were dialyzed against the high molecular weight glycol of the dialysis membranes to maintain osmotic stability. The volume of glycol was 400 m1 and dialysis was conducted for 4% hr at 23 C. The change in concentration of polyethylene glycol was measured by refractometry (Bausch and Lomb Precision Refractometer, Model 33-45-01, sodium light source) and by dry weight determinations. The molecular weight distribution of PEG 4000 before and after dialysis W38 determined using Biogel P-lO (Biorad Laboratories) by the method of Scherrer and Gerhardt (7). The mass of the polymer in each elution fraction (1.5 ml) was determined by dry weight measurements. -4- RESULTS w/‘\ 1% ,IA Eemgdialysisflwithout culture. During hemodialysis in the absence of culture, body temperature of the goat remained within a normal range, 102 to 103.5 1:1 E (Fig. 2A). However, total leukocytes in blood samples from the animal increased in number during the first few hours of control hemodialysis and then remained essentially constant at the higher level (Fig. 2B). Erythrocytes remained at an approximately constant level from the outset (Fig. 2C), and biochemical constituents also were unaffected control hemodialysis (see Table l in reference 2). Blood samples also were removed from the inlet and outlet of the artificial kidney at various times during control hemodialysis. The results (Table 1) indicated the hemodialyzer itself did not significantly affect the numbers of erythrocytes, leukocytes or total cells. Similar studies with differential cell counts indicated that the passage of blood through the dialyzer did not have a selective effect on either of the two most numerous types of leukocytes (neutro- phils and lymphocytes). aEESEEE£Z§£§,31EElEElEEEE’ When an inoculum of S. marcescens was introduced into the dialysate solution, after the usual equili- bration period, the host goat at first remained quite normal. This lack of host response persisted for a number of hours, depending on the size of the inoculum, even though the bacterial population in the dialysate-culture circuit may have risen to 109 viable cells/ml. An acute episode of fever then ensued with dramatic suddenness. The general picture of this pyrexia is shown in Fig. 3, in which the -5- average temperature change in six trials was normalized as a function of the number of viable bacterial cells in the dialysate culture. The time courses of this acute fever and other host responses in a representative trial are shown in Fig. 2, together with control responses of the same goat during hemodialysis without culture. The body temperature (Fig. 2E) started to rise after about 7 hr of hemo- dialysis culture, which coincided with the time when a population of 109 viable cells/ml was exceeded (also see Fig. 3) and when multipli— cation changed from an exponential to a linear rate in the culture (Fig. 2D). An acute reduction in the number of peripheral blood leukocytes was a second main sign of host toxemia (Fig. 2F), and typically the onset of leukopenia preceded that of the pyrexia by about 2 hr. The number of erythrocytes remained within the normal range (Fig. 2G). Among the other physiological parameters monitored, only the concentration of blood glucose changed significantly with hemodialysis culture (see Fig. 3 in reference 2). In addition to and at about the same time as these host responses in the second phase of hemodialysis culture, the goat evidenced general signs of toxemia with violent shivering and lethargy that persisted for 2 hr or more. In some instances bloody urine was passed, but in other cases kidney function appeared blocked. Feeding and drinking also stopped during an episode of toxemia. The fever and leukopenia observed with aerated hemodialysis cultures were less apparent with cultures maintained under strictly anaerobic conditions, although the cultures multiplied similarly in -5- the two situations (Fig. 4). The reason for the difference in host responses was not apparent. It might be thought that the host responses were caused by some nonspecific effect of bacterial growth. However, goat hemodialysis culture of Bacillus anthracis (the avirulent Sterne vaccine strain) did not result in a toxemic host response even though massive bac- terial growth occurred with either aerobic or anaerobic conditions in the fermentor. EEQEEEEIXE£§»9§“EEESESEEEJ The host response to hemodialysis culture of Shmarcescens was compared with that to a large amount of purified lipopolysaccharide endotoxin from the same bacterium (Fig. 5). After the equilibration period and about 1 hr after the introduction of endotoxin into the dialysate solution, the body temperature in- creased by about l.5 F and the number of peripheral leukocytes de— creased by about 3000/mm3. The animal also showed signs of discomfort, but not as marked as with hemodialysis culture of S. marcescens. These effects with pure endotoxin persisted for less than 2 hr, then re- turned to normal and remained constant for the rest of the trial. The small and transient nature of the responses suggested either that the endotoxin preparation was not pure or that biologically active endo— toxin represents a wide distribution of molecular sizes, some of which are dialyzable. Apparently only a small part of the total amount-par- ticipated in the reaction. -7- activit . Beaker dialyzers with three different porosities were em- ployed to separate and estimate the molecular size of the dialyzable toxic material produced during hemodialysis culture of S. marcescens. Although the artificial-kidney hemodialyzer was available only in the one porosity of membrane [maximum retention equivalent to nominal molecular weight (MW) of about 10,000], chemical beaker dia— lyzers were available not only in the same membrane porosity but also in a finer one (MW 300) and a coarser one (MW 30,000). Each type was substituted for the artificial kidney in the hemodialysis culture sys- tem, and the results are shown in Fig. 6. The total surface area of membrane in each beaker dialyzer was only one-tenth of that in the artificial kidney, and consequently the time courses of the culture and the host responses were somewhat different than before. The MW 300 beaker as a hemodialyzer failed to support any multi- plication of the inoculum (Fig. 6A), and no significant responses were observed in the host (Fig. 6B and C). The MW 10,000 beaker as a hemodialyzer supported much the same patterns of culture growth (Fig. 6D) and host responses (Fig. 6E and F) as did the artificial kidney (compare with Fig. 2). A longer lag period than before was observed in the culture, and the fever response of the goat occurred about in the middle of the exponential phase of culture growth. The MW 30,000 beaker as a hemodialyzer also supported a typical pattern of culture growth (Fig. 66), although the fever (Fig. 6H) and leuko— penia (Fig. 6I) occurred earlier in the time course than usual, —8- probably because a high bacterial p0pulation density was reached earlier as the result of a larger inoculum in this particular trial. The differential dialysis system shown in Fig. 1 then was em- ployed to collect three different size classes of culture product in the beakers simultaneously with artificial—kidney hemodialysis. After the goat evidenced acute toxemia, the three beaker-dialysate samples were removed and assayed, as follows. Injection of the MW 10,000 beaker—dialysate sample into another goat (3 ml, intravenously) partially duplicated the results obtained by direct hemodialysis culture with either the artificial kidney or the same beaker dialyzer. Even though the amount of toxin was much less than that transferred over a period of time in continuous hemo- dialysis, the MW l0,000 beaker sample evoked pyrexia, but without measurable leukopenia. Similar results were obtained with the MW 30,000 beaker sample. However, the MW 300 beaker sample (and also the original salts solution, employed as a control) did not evoke any measurable host response. The three beaker-dialysate samples also were tested by intra- peritoneal injection into precooled mice and measurement of body temperature depression, a test which is considered indicative of endotoxin (1,3). The results are shown in Fig. 7. Neither the original salts solution (Fig. 7A) nor the dialysate sample from the MW 300 beaker (Fig. 7C) had a significant effect. The cell-free hemodialysis culture medium had the greatest ef- fect, and lowered the temperature in the mice by about 2.5 C (Fig. 7B). -9- A depression of about 1 F was obtained with the MW 10,000 beaker sample and about 2 F with the MW 30,000 beaker sample. Diarrhea occurred in each case during the time of temperature depression. All of the mice were observed for at least 48 hr afterwards, without further signs of illness. The beaker-dialysate solutions appeared free of bacterial growth throughout the culture trial and were sterile when tested by inoculation of 0.5 ml samples into trypticase soy broth at termination, so clearly the results were attributable to dialyzable toxic molecules. The three beaker-dialysate samples also were assayed by two other tests considered indicative of endotoxin, those for chick-embryo le— thality (4) and Limulus coagulation (5,6). Only marginal effects occurred, and the results were considered equivocal. The toxic material was found to be heat stable by heating the active beaker-dialysate samples at 100 C for 5 min and then measuring the body— temperature response of precooled mice to injections of the heated sample. Results like those in Fig. 7D and E were obtained. “W The results from use of the beaker dialyzers with different membrane porosities indicated that the size of the dialyzable toxic material was equiva- lent to a nominal molecular weight more than about 300 and less than about 10,000. The latter definition of size is relative to the molecular weight of globular proteins that reportedly just diffuse through the membrane in the artificial kidney and MW 10,000 beaker dialyzer (Bulletin No. l75-ll87-7l, Dow Chemical Co.). 0f the proteins employed, cytochrome c (presumably of bovine heart origin, -10- molecular weight 13,370) lies closest within the dialysis threshold of the membrane and has an equivalent hydrodynamic radius (rES) of 1.88 nm, determined with the Einstein-Stokes formulation and based on an experimentally determined diffusion coefficient of D20,w = 11.4 x 107 (8). Myoglobin (presumably of horse heart origin, molecu- lar weight 16,890) lies just outside the dialysis threshold of the membrane and has an rES = 1.90 nm, based on 020,w = 11.3 x 107 (8). Consequently, the MW 10,000 membrane has a threshold porosity equivalent to a rigid globular molecule approximately intermediate in size between the above two proteins, i.e., molecular weight 15,000, rES = 1.89 nm. However, the dialyzable toxic material from S. marcescens might be more fibrous and flexible in its molecular configuration, and its size might be better related to the dialysis threshold of a more comparable molecule. A long flexible molecule Cflke a poly- ethylene glycol) in solution coils randomly and loosely into a spherical form that is large relative to its molecular weight. Scherrer and Gerhardt (7), using polyglycol samples, have devised a distribution-analysis method to determine the porosity threshold of bacterial cell walls. By use of this new method, the exclusion threshold of the artificial-kidney membrane was successfully determined with a polydisperse polyethylene glycol sample of 3,350 mean in number- average molecular weight (ML), whereas smaller and larger polyglycol -11- samples proved unsatisfactory. The results (Fig. 8) showed that the size of a just-excluded molecule, determined from the inter- section point of the two distribution curves, was equivalent to a quasi-monodisperse polyethylene glycol of Mh = 5,500 and rES = 2.6 nm. Thus, the maximum molecular size of the dialyzable toxic mate- rial from S. marcescens was further defined relative to two different types of molecules. If its molecular configuration were like a rigid globular protein, the toxic material was less than molecular weight 15,000 and rES = 1.9 nm. If like a flexible fibrous poly- glycol, it was less than molecular weight 5,500 and rES = 2.6 nm. DISCUSSION The nature of the dialyzable toxic material demonstrated by goat hemodialysis culture of S. marcescens is unclear, but the most likely possibility seemed that of an endotoxin or endotoxin—like substance. Bacterial endotoxins produce diverse responses in animals, including fever, transient leukOpenia (followed by leukocytosis), hyperglycemia, circulatory system disturbances, altered resistance to bacterial infections, hemorrhage, and (in sufficiently large doses) irreversible shock and death. The fever, leukopenia, and general physiologic response of the goat, and the thermostability of the toxic material, all were consistent with endotoxemia. The hypothermic -12- reaction of mice obtained with the dialyzable toxic material was considered especially characteristic of endotoxin (1,3), and the results were similar to those reported by Dennis (l) for Vibrio fetus endotoxin. Native endotoxin usually is considered to be comprised of a macromolecular complex of protein, lipid and polysaccharide with a nominal molecular weight of about 1,000,000. Molecules of this size could not permeate the membranes of any of the dialyzers used in this study. However, the smallest moiety of endotoxin that yields some or all of the typical reactions is unclear. Milner and Finkelstein (4) reported that the state of dispersion of endotoxin particles profoundly affects pyrogenicity and chick—embryo lethality tests but does not in- fluence mouse lethality tests. In the present experiments, continuous hemodialysis allowed the passage of relatively small molecules in a natural state. It is possible that these unaltered molecules then aggregated or became adsorbed to blood constituents, and thereby had activities which would be lost or altered during the traditional chemical and physical procedures used in the purification of endotoxin. It is possible that the dialyzable toxic material produced in hemodialysis culture of SL marcescens was not endotoxin. Proliferat- ing bacteria produce many types of metabolic products, including pyro- genic ones of small size. It seems improbable that an organic acid or other ordinary metabolite would cause the range or severity of reactions that were observed. But it is possible that a small moiety of the lipo- polysaccharide molecule caused the results, e.g., lipid A or conjugated -13- or simple protein (9,10). This also would account for the similar but lesser response obtained by hemodialysis of purified endotoxin. The argument also might be advanced that the results were caused by intact bacterial cells passing through an undetected opening in the membrane. However, in control experiments in vitro with artificial- kidney dialysis cultures and a reservoir of nutrient medium, the reservoir remained sterile for at least 24 hr. Furthermore, in the control hemodialysis trials, blood proteins did not pass into the dialysate circuit. ACKNOWLEDGMENTS We gratefully thank Josephine Belding, R.N., for surgical assist- ance, and T. C. Belding, D.V.M., for surgical anaesthesiology. The work was supported by the U.S. Public Health Service via Grant AI—O976O from the National Institute of Allergy and Infectious Diseases, 3 general research support grant to the College of Veterinary Medicine, and Training Grant GM-019ll. -14- LITERATURE CITED Dennis, S. M. 1972. Hypothermia in mice due to Vibrio fetus endotoxin. Cornell Vet. 62:296-300. Gerhardt, P., J. M. Quarles, and R. C. Belding. 1973. Goat hemodialysis culture of Serratia marcescens as a model of septicemia: growth characteristics. Infec. Immun. Halberg, F., and W. W. Spink. 1956. The influence of Brucella somatic antigen (endotoxin) upon the temperature rhythm of intact mice. Lab. Invest. 5:283-294. Milner, K. G., and R. A. Finkelstein. 1966. Bioassay of endotoxin: correlation between pyrogenicity for rabbits and lethality for chick embryo. J. Infect. Dis. 116:529-536. Reinhold, R. B., and J. Fine. 1971. A technique for quanti— tative measurement of endotoxin in human plasma. Proc. Soc. Exp. Biol. Med. 137:334-340. Rojas-Corona, R., R. Skarnes, S. Tamakuma, and J. Fine. 1969. The Limulus coagulation test for endotoxin. A comparison with other assay methods. Proc. Soc. Exp. Biol. Med. 132:599-601. Scherrer, R., and P. Gerhardt. 1971. Molecular sieving by the Bacillus megaterium cell wall and protoplast. J. Bacteriol. 107:718-735. Smith, M. H. 1970. Molecular weights of proteins and some other materials including sedimentation, diffusion and frictional coefficients and partial specific volumes. In Handbook of bio- chemistry. Selected data for molecular biology. Sober, H. A. (ed.). Chemical Rubber 00., Cleveland, Ohio. 10. -15- Wober, W., and P. Alaupovié. 1971. Studies on the protein moiety of endotoxin from gram-negative bacteria. Characteriza- tion of the protein moiety isolated by phenol treatment of endotoxin from Serratia marcescens 08 and Escherichia coli 0 l4l:K85(B). Eur. J. Biochem. 19:340-356. Wober, W., and P. Alaupovié. 1971. Studies on the protein moiety of endotoxin from gram—negative bacteria. Characteriza- tion of the protein moiety isolated by acetic acid hydrolysis of endotoxin from Serratia marcescens 08. Eur. J. Biochem. 19:357-367. -l6- .uwsouwo mummkamam onu GH ouauaoo usonoHB mammamfipoewz mo u: m umumm cogs» muomamuommma o>fiumunmmouemm 6 SN ooo.omo.m oo~.qa Amsocm>v umauao Hm ooo.omq.m ooo.mH Aamfiuounmv uoHaH owwvo> A EE\.ooV AmaB\.ocv o mamm 00 H moumuounuhum mouhooxsoq H m an Hamouwmxumm omNNmo woomo k0 mamasxe we“ :0 ammmmeemoeme mmrweanewow%woso we“ emxogxo mmcmmom k6 oofi%%m .H mam :1 8 m 3 .. .. .1 o 7 .. - I I I I I I I I I I T I I06 "’ A "’ E " 5 UI TEMP (F) .. F . 6‘ 2 2 3‘ ‘9 o In 3 6h 4- dh ‘ q '1’ qu- uI- III- Ilb RBC (IOB/MM3) (D l 8' ‘W‘ 7 . . O 2 4 6 8 IO I2 0 2 4 6 8 IO I2 TIME (HOURS) FIG. 2. Tune course of'changes in the responses of'a repre- sentative animal to hemodialysis without culture (left) and with culture (right): A and E, body temperature; B and F, white blood cells; C and G, red blood cells. In D there is shown the corre- sponding growth curve of’Serratia marcescens in the dialysate- culture circuit. -19- 1: +3 - ‘ n. E ._ +2 - ‘ .2. w +l - ‘ (D 2: <1 I: t) C) - ‘ I 1 l l 6 7 8 9 IO LOG VIABLE CELLS/ML FIG. 3. Normalized change in host body temperature as a funption of the numbers of’viable cells of'S. marcescens in hemodialysis cultures, averaged from six cultures with fbur animals. -20- 3’ AEROBIC ANAEROBIC \ I l l I I I I I I I I w I l «- j A -II- D .- lu IO- 0 I- - In 9- ‘ Ed - . s. 8’ > 7 I- III- a- 8 -. r -J I I I I I .L uh- - E - [I O. "" .. 2 -W ‘ LIJ F d d- I I I 1 fl; "" -I "' -I 5 \ III I0 —- 9 d o “b q co ; "" .4 6 - .- , 4 - 1 1 - J l l I I I l I i O 22 4» 6 £3 K)|2 C) 2 ‘4 £5 8 ICIIZ IWMEIHOURS) FIG. 4. comparison of'anaerobic with aerobic hemodialysis culture conditions on body temperature and blood leukocytes of the animal. -21- [05- I I I I I I d A A 2: I04— ~ ‘2‘ w IOS- - *— I02- -I ”A '6'. T I r I I r d 2 2 B n\ l4-— _I O 3 |2- — co IO ADDET _ a: L, I l l I l o 2 4 6 8 IO I2 TIME (HOURS) FIG. 5. Iime course of’changes in the response of’a goat to hemodialysis of purified lipopolysaccharide endotoxin. In A is shown the body temperature and in B the peripheral leukocyte count. The arrows indicate the time at which the endotoxin.wus added. -22- ; 1 1”": 300 ' MW l0.000 MW 30.000 \ I ' ' I I U I I I I I I I v T g I A D G u 0 I "' - U 9.. q)- ~0- 5 d 2 8" «I- ... 4 S 7- «II- '4'- -I . 900000—04, l 8 6 qr- .. q ‘ —’ A: # i i : l I l l 1 1 1 A I l 1 l '07_ B v .. v u r I . . v v . v I I E l" H ‘ A I06- “ -- - 2: I05- .. -- . g IO4- .. .. - III-J 'OBW .. q... .. '02- l 1 n l l .0. .0- ‘ Ie- C -- F -- I - "’2 I6- «- ~~ - 5 I4- -- «- - n 2 l2- cu- dI- -I g mm -- - - 3 8- 4- .. - 6- «r -- d 4’- 1 l l l l l l l l l l l I. l l l J l 1 2468I0|202468l0I202468l0|2 TIME (HOURS) FIG. 6. Time courses of'hemodialysis cultures and host responses with beaker dialyzers of’different porosity used as the hemodialyzer, each identified by the porosity threshold in nominal molecular weight of 300 (left), 10,000 (center), and 30, 000 (right). -23- I I I I T I A- SALTS SOLUTION 0W°\o—o. -| .. J -2 .- 1 1 l l ‘ +2 h I I I I I d + | B-CULTURE MEDIUM 0 -+ -| I. - -2 _ -4 +2 - i 1 I d _ C - MW 300 DIALYSATE OM- l L 1 1 4L 1 I I I f f f 0- MW l0.000 DIALYSATE CHANGE IN TEMPERATURE (C) + +|- o d —| .. .. -2b 1 l l L- l I 0 30 60 90 I20 I50 I80 TIME (MINUTES) FIG. 7. Body-temperature response of'precooled mice (average of 10) to injection (0.5 ml, intraperitoneally) of'salts solution (A), cell-free hemodialysis culture medium (B),.MW'300 beaker-dialysate sample (0), MW 10,000 beaker-dialysate sample (D), and MW 30,000 beaker-dialysate sample (E). -24- Io‘ MOLECULAR WEIGHT, I?" I 2 3 4 5 6 7 8 9 IO T T T TTrTTT DRY WEIGHT (mg) 50 55 50 415 :0 35 30 25 ELUTION FRACTION no. FIG. 8. Molecular weight distribution of’polyethylene glycol Mk 3,350 before (0) and after (a) equilibrium dialysis through the artificial-kidney membrane. The intersection point represents the porosity exclusion threshold. 47 5.3 Prosthetic hemodialysis culture unit 5.3.1 Introduction The studies with the fermentor - artificial kidney - animal system indicated that in vivo conditions were not fully simulated by hemodialysis. This shortcoming was considered caused mainly by two factors: (1) the artificial-kidney membrane was impermeable to molecules of nominal molecular weights greater than about 10,000, including most blood proteins, and (2) the membrane was poorly per- meable to gases, including CO2 and 02. Membranes that pass larger molecules and gases are available in the form of hollow-fiber dia- lyzers, but these are not designed for hemodialysis and are limited in the availability of different types of membranes. Furthermore, the design of the system and the associated equipment (fermentor, pumps, and tubing) is complex. Finally and importantly, the re- quirement for constant attendance with the animal restricted dialysis culture with an artificial kidney to about 15 hr. Consequently, considerable effort was spent in the conception of a workable design and eventually in the construction and testing of prototype models of a hemodialysis culture unit which would allow the use of a nutrient—passing membrane simultaneously with a gas- transport membrane, which could be mounted on an animal, and which would enable dialysis culture over an extended time period. The aid of B. M. Stutsman in constructing the prosthetic hemodialysis unit is gratefully acknowledged. 48 Because prosthetic shunts of Teflon and Silastic rubber tubing were maintained for periods of several weeks without causing clotting or lysis of the animal's blood, we incorporated this design concept for the blood flow channel of the new hemodialysis culture unit. That is, the blood flowed through a single piece of Silastic rubber tubing, with an opening in one side of the tubing exposed to the membrane. This design had the advantage of causing minimal physical restrictions in the path the blood followed as it passed from the artery, through the dialyzer, and back into the vein. Also, the flow of the blood through the dialyzer swept across the membrane, thus helping keep it free of deposits and assuring more effective dialysis. Two separate "windows" (2 x 7 mm) of membrane were aligned longitudinally between the blood and dialysate, and these allowed molecular communication between the two regions. The design was such that the two windows may be of the same or different membrane material. Only a small amount of membrane (about 28 mmz) was in contact with the blood. However, this membrane area was sufficient to supply nutrients to the culture chamber because the unit was designed for continuous dialysis with a small volume of culture (about 3.3 ml). The size of the entire unit was made so that it could easily be mounted on a goat's neck when attached to the arterial-venous shunt and impose only minimal restrictions on the animal's movements. The design employed also eliminated the need for continuous infusion of anticoagulant and the requirement for continuous monitoring of the animal and equipment. Satisfactory agitation of the culture was achieved via the animal's normal movement. 49 This prosthetic hemodialysis culture unit likely will be changed and improved in the future. The results with the prototype unit reported in the following sections are incomplete and are intended as a progress report to aid in further develOpment of the device and application of it in hemodialysis culture. 5.3.2 Materials and Methods The prosthetic hemodialysis culture unit in present state of development is depicted in Fig. l. The assembled dialyzer consists of four main pieces: (1) the blood chamber, consisting of a plastic block fitted with silicone rubber tubing with a slotted opening, (2) membranes, (3) supporting metal plate, and (4) dialysate- culture chamber. The unit is self—contained, with the blood, mem— branes and culture all in close proximity. The two chambers were machined from Lucite plastic, and measured about 2 cm x 3.5 cm x 3.5 cm when assembled. The dialysate or culture chamber was 3.5 cm x 3.5 cm x 0.9 cm in outside dimensions. Inner dimensions of the chamber were 0.7 cm x 2.1 cm x 2.1 cm, which provided a working volume of approximately 3.3 ml. Sampling from the dialysate-culture chamber was achieved in a first model by puncture of a self-sealing rubber vaccine-bottle stopper, and in a subsequent model by a similar but smaller stOpper at the end of tubing extending from the sides of the chamber. This change made sampling easier while the unit was attached to the ani- mal and also protected the membranes from possible puncture. In use, one of the two adaptors was used for sampling and the other for 50 FIG. 1. Prosthetic hemodialysis culture unit. A. The unit is 8720er mounted on the neck of a goat and connected with tubing to an cmaterial-venous shunt. B. The unit is shown assembled at the top and beneath it (left to right) are the component parts: blood chamber l071th silicone rubber tubing, a sheet of cut and punched membrane, a Stainless steel support plate, and the dialysate—culture chamber with two sampling extensions. 51 equilization Of air pressure. Three non-threaded holes were drilled in each Of two sides Of the culture chamber, through which screws passed into corresponding threaded holes in the blood chamber block. The blood chamber block was approximately 0.9 cm x 3.5 cm x 3.5 cm in size. An angled hole was drilled through the block from two of the (Opposite) smaller faces so that exit and entry ports Ivere formed on these sides and a slot was formed on one of the Llarger faces. Silicone rubber tubing was threaded through this hole érnd firmly attached to the block with Dow-Corning Silastic medical aidhesive silicone Type A (catalog #891). The tubing was then txrimmed flush with the face of the block, thereby forming a sili- cuone rubber-lined blood flow channel with.a longitudinal Opening (ssee Fig. l). The membrane covered this Opening when the unit meals in use. A metal supporting plate was constructed of polished 24 gauge EStLainless steel (about 0.05 cm in thickness). Two separate holes (12 mm x 7 mm and 4 mm apart) in the center of the plate were aligned ILCDngitudinally over the slotted Opening Of the blood channel in the blood block. This steel plate provided support for the membrane Eirld the two holes allowed the use Of two different types of nuEmmbrane. Any type Of sheet membrane material which is compatible with bJLood may be used in the hemodialyzer, either alone or in combina- tZion with another type. Examples Of tested materials include dialy— ESIius membranes of cellophane (DuPont 315), filter membranes Of 52 cellulose acetate (Millipore Ha 0.45 p and Gelman GA 8), filter membranes Of polycarbonate (Nuclepore GE 10, 0.1 p) and gas- transport membranes Of silicone rubber (Dow-Corning Silastic). Membranes Of two different types of material were prepared by cut- ting a hole in one (properly aligned with the blood channel and hole in the supporting plate) and covering the Opening with the second material. Dow-Corning Silastic medical adhesive was used to join the two membranes. The adhesive is non toxic and produces a strong bond between any combination Of the membranes listed above. The unit was connected to the external arterial—venous shunt by means Of Teflon connectors. For dialysis trials, the assembled unit was attached to the animal's neck in a variety Of ways. At first it was held on the side Of the neck in a tape—and—gauze sling, which allowed the shunt and connector tubing to flex and bend with- out crimping and reducing blood flow. The unit also was mounted on the back of the neck by use Of longer connecting tubing, but this increased the resistance to blood flow. The last method was to tie the unit to a supporting platform constructed of rubber-covered fiber mesh, which was sutured to the animal's neck. This provided a firm but flexible support from which the unit could be quickly removed and re-attached. The Open mesh nature Of the support allowed Q-1rculation Of air and prevented irritation and infection. The unit as presently constructed was not steam sterilizable but instead had to be sterilized with ethylene oxide (Bard Steri— lizer System 2279, D. R. Bard, Inc.). Prior to use, the dialyzer 53 was loosely assembled with the membrane in place and exposed to the gas for 6 to 18 hr. However, the unit could be machined of steel, polycarbonate, or other steam sterilizable material if desired. Routinely, a solution of 0.85% NaCl was used as the initial solution in the chamber. A saline solution was used because it is non-nutritive and allowed determination of electrolytes dialyzed into the chamber from the animal's blood, while also preventing hemolysis of erythrocytes. In some instances whole blood or tissue unis placed in the chamber. With the artificial kidney - fermentor saystem, it was necessary to use a more complex, balanced solution Euecause the large volume used in the dialysis chamber (800 ml) would 111>set the animal's electrolyte balance if saline were the dialysate. TFIIe small volume (3.3 ml) Of the plastic hemodialysis culture unit EiJLlowed the use of a less complex solution. For periodic sampling, as in growth studies, specimens of 0.1 ml “Weare removed and the volume replaced with saline. For equilibrium 531:1niies, specimens Of varying amounts up to the entire contents Of t211e3 chamber were removed as required. This was necessitated pri- Inarilyby the fact that the autoanalyzer used for biochemical tests reCluired 4 ml specimens, and generally not greater than about 1 to 5 dilutions, for maximum reliability. Biochemical tests, hematologic E’t34£lies, and growth curves were conducted as described for the arti- ficial kidney - fermentor system except in the instances as given in the materials and methods sections of the particular trials listed below. 54 5.3.3 Prosthetic hemodialysis without culture 5.3.3.1 Introduction The prosthetic hemodialysis unit allowed the use Of membranes composed of essentially any material, with the limitation only that the material be comparable with blood and in sheet form. In order to Obtain information on the types Of molecules from blood which traverse the three main types Of membrane (dialysis, filter and gas- transport membranes) under conditions Of equilibrium hemodialysis, samples Of these materials were tested in the unit without culture. frhese experiments were designed to provide both qualitative data on tihe types Of molecules which passed (or were retained by) the membranes and also a rough comparison of rates and amounts. 5.3.3.2 Materials and methods The dialysate chamber Of the miniature hemodialysis unit was ifiilled with sterile 0.85% NaCl and the unit was connected to the Eilrterial—venous shunt. Hemodialysis was allowed to proceed for 1 hr (2hr in the example of DuPont 315 dialysis membrane), and then title resulting dialysate solution was removed and tested for the blood constituents as listed in Table 1. Biochemical tests were <2Onducted by standard procedures on an SMA 12 autoanalyzer (Labora- t:Ofli'jyof Clinical Medicine, Lansing, Mich.). In each trial, prior t3c> 14se, the hemodialyzer was sterilized with ethylene oxide gas, with the designated membranes in place. Combination membranes were prepared as described in Section 5.3.2. 55 5.3.3.3 Results The results of trial dialyses with the several membranes are shown in Table 1. 5.3.3.4 Discussion The silicone rubber membrane (Dow-Corning Silastic) is a gas- transporting material without porosity (119) , and consequently none of the blood components assayed for in this trial dialyzed through this membrane (Table 1). Therefore, it should be recognized that in dialysis trials using silicone rubber in combination with a nu- trient passing membrane, the effective membrane area for nutrient transfer is reduced by 50%. Dialysis membranes, generally manufactured from regenerated Cellulose, have an average rated pore size of about 5 nm diameter. Thus, they retain large molecules (e.g., albumin and other serum Proteins or enzymes) but pass small molecules (e.g., glucose or Salts). The membrane used in these trials retained albumin or the other proteins but allowed calcium, phosphorus, glucose, and urea '30 pass. Glutamic—oxalacetic transaminase values for this trial, and the other trials, in the approximate range Of 10 to 15 units are considered Of marginal significance because Of variability in Several negative control specimens. The filter membranes tested, manufactured of cellulose acetate or polycarbonate, contain larger pores (in the range of 0.1 to 0.45 p) and therefore pass many proteins and other large molecules. As Shown in Table l, Gelman, Millipore, and Nuclepore membranes passed 56 aaaonam umooxm .Ha ooa\wa mm commoumxm moowumuuaoucou .muamumnH menu GH mam%HMfio a: 039 a .A.=an mafia: doaumx mm omMGHEanmuu van .A.D.