Chapter 97 - animal surrogate systems

Shuler, M. L. “Animal Surrogate Systems.”The Biomedical Engineering Handbook: Second Edition.
Ed. Joseph D. BronzinoBoca Raton: CRC Press LLC, 2000 Limitations of Animal Studies • Alternatives to Animal Studies 97.1 Background
Animal surrogate or cell culture analog (CCA) systems mimic the biochemical response of an animal orhuman when challenged with a chemical or drug. A true animal surrogate is a device that replicates thecirculation, metabolism, or adsorption of a chemical and its metabolites using interconnected multiplecompartments to represent key organs. These compartments make use of engineered tissues or cellcultures. Physiologically based pharmacokinetic models (PBPK) guide the design of the device. Theanimal surrogate, particularly a human surrogate, can provide important insights into toxicity and efficacyof a drug or chemical when it is impractical or imprudent to use living animals (or humans) for testing.
The combination of a CCA and PBPK provides a rational basis to relate molecular mechanisms to wholeanimal response.
Limitations of Animal Studies
The primary method used to test the potential toxicity of a chemical or action of a pharmaceutical is touse animal studies, predominantly with rodents. Animal studies are problematic. The primary difficultiesare that the results may not be meaningful to assessment of human response [Gura, 1997]. Because ofthe intrinsic complexity of a living organism and the inherent variability within a species, animal studiesare difficult to use to identify unambiguously the underlying molecular mechanism for action of achemical. The lack of a clear relationship among all of the molecular mechanisms to whole animalresponse makes extrapolation across species difficult. This factor is particularly crucial when extrapolationof rodent data to humans is an objective. Further, without a good mechanistic model it is difficult torationally extrapolate from high doses to low doses. However, this disadvantage due to complexity canbe an advantage; the animal is a “black box” and provides response data even when the mechanism ofaction is unknown. Further disadvantages reside in the high cost of animal studies, the long period oftime often necessary to secure results, and the potential ethical problems in animal studies.
Alternatives to Animal Studies
In vitro methods using isolated cells (e.g., [Del Raso, 1993]) are inexpensive, quick, and have almost noethical constraints. Because the culture environment can be specified and controlled, the use of isolatedcells facilitates interpretation in terms of a biochemical mechanism. Since human cells can be used aswell as animal cells, cross-species extrapolation is facilitated.
However, these techniques are not fully representative of human or animal response. Typical in vitro experiments expose isolated cells to a static dose of a chemical or drug. It is difficult to relate this staticexposure to specific doses in a whole animal. The time-dependent change in the concentration of achemical in an animal’s organ cannot be replicated. If one organ modifies a chemical or prodrug whichacts elsewhere, these situations would not be revealed by the normal in vitro test.
A major limitation on the use of cell cultures is that isolated cells do not fully represent the full range of biochemical activity of the corresponding cell type when in a whole animal. Engineered tissues,especially co-cultures [Bhatia et al., 1998], can provide a more “natural” environment which can improve(i.e., make normal) cell function. Another alternative is use of tissue slices, typically from the liver [Olingaet al., 1997]. Tissue slices require the sacrifice of the animal, there is intrinsic variability, and biochemicalactivities can decay rapidly after harvest. The use of isolated tissue slices also does not reproduce inter-change of metabolites among organs and the time-dependent exposure that occurs within an animal.
An alternative to both animal and in vitro studies is the use of computer models based on PBPK models [Connolly and Anderson, 1991]. PBPK models can be applied to both humans and animals. Because PBPKmodels mimic the integrated, multicompartment nature of animals, they can predict the time-dependentchanges in blood and tissue concentrations of a parent chemical or its metabolites. Although constructionof a robust, comprehensive PBPK is time-consuming, once the PBPK is in place many scenarios concerningexposure to a chemical or treatment strategies with a drug can be run quickly and inexpensively. SincePBPKs can be constructed for both animals and humans, cross-species extrapolation is facilitated. Thereare, however, significant limitations in relying solely on PBPK models. The primary limitation is that aPBPK can only provide a response based on assumed mechanisms. Secondary and unexpected effects arenot included. A further limitation is the difficulty in estimating parameters, particularly kinetic parameters.
None of these alternatives satisfactorily predict human response to chemicals or drugs.
97.2 The Cell Culture Analog Concept
A CCA is a physical replica of the structure of a PBPK where cells or engineered tissues are used in organcompartments to achieve the metabolic and biochemical characteristics of the animal. Cell culturemedium circulates between compartments and acts as a “blood surrogate”. Small-scale bioreactors withthe appropriate cell types in the physical device represent organs or tissues.
The CCA concept combines attributes of a PBPK and other in vitro systems. Unlike other in vitro systems, the CCA is an integrated system that can mimic dose dynamics and allows for conversion of aparent compound into metabolites and the interchange of metabolites between compartments. A CCAsystem allows dose exposure scenarios that can replicate the exposure scenarios of animal studies.
A CCA is intended to work in conjunction with a PBPK as a tool to test and refine mechanistic hypotheses. A molecular model can be embedded in a tissue model which is embedded in the PBPK.
Thus, the molecular model is related to overall metabolic response. The PBPK can be made an exactreplica of the CCA; the predicted response and measured CCA response should exactly match if thePBPK contains a complete and accurate description of the molecular mechanisms. In the CCA all flowrates, the number of cells in each compartment, and the levels of each enzyme can be measured inde-pendently, so no adjustable parameters are required. If the PBPK predictions and CCA results disagree,then the description of the molecular mechanisms is incomplete. The CCA and PBPK can be used in aniterative manner to test modifications in the proposed mechanism. When the PBPK is extended to describethe whole animal, failure to predict animal response would be due to inaccurate description of transport(particularly within an organ), inability to accurately measure kinetic parameters (e.g., in vivo enzyme levels or activities), or the presence in vivo or metabolic activities not present in the cultured cells ortissues. Advances in tissue engineering will provide tissue constructs to use in a CCA that will displaymore authentic metabolism than isolated cell cultures.
The goal is predicting human pharmacological response to drugs or assessing risk due to chemical exposure. A PBPK that can make an accurate prediction of both animal CCA and animal experimentswould be “validated”. If we use the same approach to constructing a human PBPK and CCA for the samecompound, then we would have a rational basis to extrapolate animal response to predict human responsewhen human experiments would be inappropriate. Also, since the PBPK is mechanistically based, it wouldprovide a basis for extrapolation to low doses. The CCA/PBPK approach complements animal studiesby potentially providing an improved basis for extrapolation to humans.
Further, PBPKs validated as described above provide a basis for prediction of human response to mixtures of drugs or chemicals. Drug and chemical interactions may be synergistic or antagonistic. If aPBPK for compound A and a PBPK for compound B are combined, then the response to any mixtureof A and B should be predictable since the mechanisms for response to both A and B are included.
97.3 Prototype CCA
A simple three-component CCA mimicking rodent response to a challenge by naphthalene has beentested by Sweeney et al. [1995]. While this prototype system did not fulfill the criteria for a CCA ofphysically realistic organ residence times or ratio of cell numbers in each organ, it did represent a multi-compartment system with fluid recirculation. The three components were liver, lung, and other perfusedtissues. These experiments used cultured rat hepatoma (H4IIE) cells for the liver and lung (L2) cells forthe lung compartment. No cells were required in “other tissues” in this model since no metabolic reactionswere postulated to occur elsewhere for naphthalene or its metabolites. The H4IIE cells contained enzymesystems for activation of naphthalene (cytochrome P450IA1) to the epoxide form and conversion todihydriol (epoxide hydrolase) and conjugation with glutathione (glutathione-S-transferase). The L2 cellshad no enzymes for naphthalene activation. Cells were cultured in glass vessels as monolayers. Experimentswith this system using lactate dihydrogenase release (LDH) and glutathione levels as dependent parameterssupported a hypothesis where naphthalene is activated in the “liver” and reactive metabolites circulate tothe “lung” causing glutathione depletion and cell death as measured by LDH release. Increasing the levelof cytochrome p450 activity in the “liver” by increasing cell numbers or by preinducing H4IIE cells led toincreased death of L2 cells. Experiments with “liver”-blank; “lung”-“lung”, and “lung”-blank combinationsall supported the hypothesis of a circulating reactive metabolite as the cause of L2 cell death.
This prototype system [Sweeney et al., 1995] was difficult to operate, very non-physiologic, and made time course experiments very difficult. An alternative system using packed bed reactors for the “liver”and “lung” compartments has been tested [Shuler et al., 1996; Ghanem, 1998]. This system successfullyallowed time course studies, was more compact and simpler to operate, and was physiological with respectto the ratio of “liver” to “lung” cells. While liquid residence times improved in this system, they still werenot physiologic (i.e., 114s vs. an in vivo value of 21s in the liver and 6.6s vs. in vivo lung value of about1.5s) due to physical limitations on flow through the packed beds. Unlike the prototype system, noresponse to naphthalene was observed.
This difference in response of the two CCA designs was explained through the use of PBPK models of each CCA. In the prototype system, the large liquid residence times in the liver (6 min) and the lung(2.1 min) allowed formation of large amounts of naphthol from naphthalene oxide and presumably theconversion of napthol into quinones that were toxic. In the packed bed system, liquid residence timeswere sufficiently small so that the predicted naphthol level was negligible. Thus, the PBPK provided amechanistic basis to explain the differences in response of the two experimental configurations.
Using a very simple CCA, Mufti and Shuler [1998] demonstrated that response of human hepatoma (HepG2) to exposure to dioxin (2,3,7,8-tetrachlorodibenzo-p-dioxin) is dependent on how the dose isdelivered. The induction of cytochrome p450IA1 activity was used as a model response for exposure todioxin. Data were evaluated to estimate dioxin levels giving cytochrome P450IA1 activity 0.01% of maximal induced activity. Such an analysis mimics the type of analysis used to estimate risk due tochemical exposure. The “allowable” dioxin concentration was 4 × 10–3 nM using a batch spinner flask,4 × 10–4 nM using a one-compartment system with continuous feed, and 1 × 10–5 nM using a simpletwo-compartment CCA. Further, response could be correlated to an estimate of the amount of dioxinbound to the cytosolic Ah receptor with a simple model for two different human hepatoma cell lines.
This work illustrates the potential usefulness of a CCA approach in risk assessment.
Ma et al. [1997] have discussed an in vitro human placenta model for drug testing. This was a two- compartment perfusion system using human trophoblast cells attached to a chemically modified poly-ethylene terephatholate fibrous matrix as a cell culture scaffold. This system is a CCA in the same senseas the two-compartment system used to estimate response to dioxin.
The above examples are the first that attempt to mimic circulation and metabolic response to model an animal as an integrated system. However, others have used engineered tissues as a basis for testingthe efficacy of drugs or toxicity of chemicals. These tissues are important in themselves and could becomeelements in an integrated CCA.
97.4 Use of Engineered Tissues or Cells
for Toxicity/Pharmacology
The primary use of engineered tissues for toxicity testing has been with epithelial cells that mimic thebarrier properties of the skin, gut, or blood brain barrier. The use of isolated cell cultures has been ofmodest utility due to artifacts introduced by dissolving test agents in medium and due to the extremesensitivity of isolated cells to these agents compared to in vivo tissue. One of the first reports on the useof engineered cells is that by Gay et al. [1992] reporting on the use of a living skin equivalent as an in vitrodermatotoxicity model. The living skin equivalent consists of a co-culture of human dermal fibroblastsin a collagen-containing matrix overlaid with human keratinocytes that have formed a stratified epider-mis. This in vitro model used a measurement of mitochondrial function (i.e., the colormetric thiazolylblue assay) to determine toxicity. Eighteen chemicals were tested. Eleven compounds classified as non-irritating had minimal or no effect on mitochondrial activity. For seven known human skin irritants,the concentration that inhibited mitochondrial activity by 50% corresponded to the threshold value foreach of these compounds to cause irritation on human skin. However, living skin equivalents did notfully mimic the barrier properties of human skin; water permeability was 30-fold greater in the livingskin equivalent than in human skin. Kriwet and Parenteau [1996] report on permeabilities of 20 differentcompounds in in vitro skin models. Skin cultures are slightly more permeable (two- or threefold) forhighly lipophilic substances and considerably more permeable (about tenfold) for polar substances thanhuman-cadaver or freshly excised human skin.
The above tests are static. Pasternak and Miller [1996] have tested a system combining perfusion and a tissue model consisting of MDCK (Madin-Darby canine kidney) epithelial cells cultured on a semi-porous cellulose ester membrane filter. The system could be fully automated using measurement of trans-epithelial electrical resistance (TER) as an end point. A decrease in TER is an indicator of cell damage.
The system was tested using nonionic surfactants and predicted the relative occular toxicity of thesecompounds. The perfusion system mimics some dose scenarios (e.g., tearing) more easily than a staticsystem and provides a more consistent environment for the cultured cells. The authors cite as a majoradvantage that the TER can be measured throughout the entire exposure protocol without physicallydisturbing the tissue model and introducing artifacts in the response.
Probably the most used cell-based assay is the Caco-2 model of the intestine to determine oral availability of a drug or chemical. The Caco-2 cell cultures are derived from a human colon adenocar-cinoma cell line. The cell line, C2Bbel, is a clonal isolate of Caco-2 cells that is more homogeneous inapical brush border expression than the Caco-2 cell line. These cells form a polarized monolayer withan apical brush border morphologically comparable to the human colon. Tight junctions around thecells act to restrict passive diffusion by the paracellular route mimicking the transport resistance in the intestine. Hydrophobic solutes pass primarily by the transcellular route and hydrophilic compounds bythe paracellular route. Yu and Sinko [1997] have demonstrated that the substratum (e.g., membrane)properties upon which the monolayer forms can become important in estimating the barrier propertiesof such in vitro systems. The barrier effects of the substratum need to be separated from the intrinsicproperty of the monolayers. Further, Anderle et al. [1998] have shown that the chemical nature ofsubstratum and other culture conditions can alter transport properties. Sattler et al. [1997] provide oneexample (with hypericin) of how this model system can be used to evaluate effects of formulation (e.g.,use of cyclodextrin or liposomes) on oral bioavailability. Another example is the application of the Caco-2system to transport of paclitaxel across the intestine [Walle and Walle, 1998]. Rapid passive transportwas partially counter-balanced by an efflux pump (probably P-glycoprotein) limiting oral bioavailability.
Another barrier of intense interest for drug delivery is the blood-brain barrier. The blood-brain barrier is formed by the endothelial cells of the brain capillaries. A primary characteristic is the high resistanceof the capillary due to the presence of complex tight junctions inhibiting paracellular transport and thelow endocytic activity of this tissue. Acceptable in vitro models have been more difficult to formulate butone commercial system marketed by Cellco, Inc. is available that uses a co-culture of endothelial cellsand astrocytes to form a barrier.
A recent example of a different in vitro system is described by Glynn and Yazdanian [1998] who used bovine brain microvessel endothelial cells grown on porous polycarbonate filters to compare the transportof nevirapine, a reverse transcriptase inhibitor to other HIV antiretroviral agents. Nevirapine was themost permeable antiretroviral agent and hence may have value in HIV treatment in reducing levels ofHIV in the brain.
These isolated cultures mimic an important aspect of cell physiology (oral uptake or transport into the brain). In principle, they could be combined with other tissue mimics to form a CCA that would beespecially useful in testing pharmaceuticals.
97.5 Future Prospects
Significant progress is being made in the construction of tissue engineered constructs [Baldwin and Saltz-man, 1996]. These efforts include highly vascularized tissues, which are especially important for toxicologicalor pharmacological studies in the liver. Liver constructs, often based on co-cultures, have become increas-ingly more normal in behavior (see [Griffith et al., 1997]). For toxicological studies multiple test systemsoperated in parallel are desirable which suggests the need for miniaturization to conserve cells, reagents,and time. Bhatia et al. [1998] have described a microfabricated device for hepatocyte-fibroblast co-cultures.
A CCA based on the concepts described here and incorporating these advanced-engineered tissues could become a powerful tool for preclinical testing of pharmaceuticals. While drug leads are expandingrapidly in number, the capacity to increase animal and human clinical studies is limited. It is imperativethat preclinical testing and predictions for human response become more accurate. A CCA should becomean important tool in preclinical testing.
Defining Terms
Animal surrogate:
A physiologically based cell or tissue multi-compartmented device with fluid cir- culation to mimic metabolism and fate of a drug or chemical.
Engineered tissues:
Cell culture mimic of a tissue or organ; often combines a polymer scaffold and Physiologically based pharmacokinetic model (PBPK):
physiology by subdividing the body into a number of anatomical compartments, each compart-ment interconnected through the body fluid systems; used to describe the time-dependent distri-bution and disposition of a substance.
Tissue slice:
A living organ is sliced into thin sections for use in toxicity studies; one primary organ can provide material for many tests.
References
Anderle P, Niederer E., Werner R., Hilgendorf C, Spahn-Langguth H, Wunderu-Allenspach H, Merkle HP, Langguth P. 1998. P-Glycoprotein (P-gp) mediated efflux in Caco-2 cell monolayers: theinfluence of culturing conditions and drug exposure on P-gp expression levels. J. Pharm. Sci. 87:757 Bhatia SN, Balis UJ, Yarmush ML, Toner M. 1998. Microfabrication of hepatocyte/fibroblast co-cultures: role of homotypic cell interactions. Biotechnol. Prog. 14:378 Baldwin SP, Saltzman WM. 1996. Polymers for tissue engineering. Trends Polym. Sci. 4:177Connolly RB, Andersen ME. 1991. Biologically based pharmacodynamic models: tool for toxicological research and risk assessment. Annu. Rev. Pharmacol. Toxicol. 31:503 Del Raso NJ. 1993. In vitro methodologies for enhanced toxicity testing. Toxicol. Lett. 68:91Gay R, Swiderek M, Nelson D, Ernesti A. 1992. The living skin equivalent as a model in vitro for ranking the toxic potential of dermal irritants. Toxic. In Vitro 6:303.
Ghanem A. 1998. Application of a novel packed bed cell culture analog bioreactor and corresponding pharmacokinetic model to naphthalene toxicology. Ph.D. Thesis. Cornell University, Ithaca, NewYork.
Glynn SL, Mehran Y. 1998. In vitro blood-brain barrier permeability of nevirapine compared to other HIV antiretroviral agents. J. Pharm. Sci. 87:306.
Griffith LG, Wu B, Cima MJ, Powers M, Chaignaud B, Vacanti JP. 1997. In vitro orgarogensis of vacularized liver tissue. Ann. N.Y. Acad. Sci. 831:382.
Gura T. 1997. Systems for identifying new drugs are often faulty. Science 273:1041Kirwet K, Parenteau NL. 1996. In vitro skin models. Cosmetics & Toiletries 111 (Feb):93Ma T, Yang S-T, Kniss DA. 1997. Development of an in vitro human placenta model by the cultivation of human trophoblasts in a fiber-based bioreactor system. Am. Inst. Chem. Eng. Ann. Mtg., LosAngeles, Ca, Nov. 16-21.
Mufti NA, Shuler ML. 1998. Different in vitro systems affect CYPIA1 activity in response to 2,3,7,8- tetrachlorodibenzo-p-dioxin. Toxicol. in Vitro 12:259.
Olinga, P, Meijer DKF, Slooff, MJH, Groothuis, GMM. 1997. Liver slices in in vitro pharmacotoxicology with special reference to the use of human liver tissue. Toxicol. In Vitro 12:77.
Pasternak AS, Miller WM. 1996. Measurement of trans-epitheial electrical resistance in perfusion: poten- tial application for in vitro ocular toxicity testing. Biotechnol. Bioeng. 50:568.
Sattler S, Schaefer U, Schneider W, Hoelzl J, Lehr C-M. 1997. Binding, uptake, and transport of hypericin by Caco-2 cell monolayers. J. Pharm. Sci. 86:1120.
Shuler ML, Ghanem A, Quick D, Wong MC, Miller P. 1996. A self-regulating cell culture analog device to mimic animal and human toxicological responses. Biotechnol. Bioeng. 52:45.
Sweeney LM, Shuler ML, Babish JG, Ghanem A. 1995. A cell culture analog of rodent physiology: application to naphthalene toxicology. Toxicol. In Vitro 9:307.
Walle UK, Walle T. Taxol transport by human intestinal epithelial Caco-2 cells. Drug Metabol. Disposit.
Yu H, Sinko PJ. 1997. Influence of the microporous substratum and hydrodynamics on resistances to drug transport in cell culture systems: calculation of intrinsic transport parameters. J. Pharm. Sci.
86:1448.

Source: ftp://ftp.kemt.fei.tuke.sk/KEMT537_LE/_materialy/09-Biomedical%20Engineering%20Handbook,%20The%20-Volumes%201%20&%202/ch097.pdf