HEPATOLOGY Elsewhere T. JAKE LIANG, EDITOR Gastrointestinal Unit

Massachusetts General Hospital 32 Fruit Street Boston, Massachusetts 02114

GRJ-724

A MARRIAGE OF MICE PRODUCES ANSWERS TO OLD QUESTIONS ABOUT MAN

Callow MJ, Stoltzfus W, Lawn RM, Rubin EM. Expression of human apolipoprotein B and assembly of lipoprotein(a) in transgenic mice. Proc Natl Acad Sci U S A 1994;91:2130-2134.

ABSTRACT

The atherogenic macromolecule lipoprotein( a) [Lp(a) J has resisted in uiuo analyses partly because it is found in a limited number of experimental animals. Although transgenic mice expressing human apolipo- protein (a) [ape( a)] have previously been described, they failed to assemble Lp(a) particles because of the inability of human apo(a) to associate with mouse apolipoprotein B (apoB). We isolated a 90-kilobase P1 phagemid containing the human apoB gene and with this DNA generated 13 lines of transgenic mice of which 11 expressed human apoB. The human apoB transcript was expressed and edited in the liver of the transgenic mice. Plasma concentrations of human apoB, as well as low density lipoprotein (LDL), were related to transgene copy number; the transgenic line with the most copies of human apoB had a >4-fold increase in LDL cholesterol compared with nontrans- genics and a lipoprotein profile similar to that of humans. When human apoB and apo(a) transgenic mice were bred together, plasma apo(a) in mice ex- pressing both human proteins was tightly associated with lipoproteins in the LDL density region. These studies demonstrate the successful expression of human apoB and the efficient assembly of Lp(a) in mice.

COMMENTS Like many other diseases, atherosclerosis is probably

the pathologic end stage brought about by a variety of derangements of normal physiology. Several of the metabolic disturbances that eventually lead to athero- sclerosis begin with disordered hepatic lipid and lipo- protein metabolism. One might even argue that the regulation of lipoprotein metabolism is the most complex and highly differentiated of liver functions (1). Atherosclerosis is a disease that occurs spontaneously only in human beings, and there are numerous differ- ences in lipoprotein metabolism among species. There-

HEPATOLOGY 1994;20:1635-1644. 31/8/60344

ADVISORY COMMITTEE M. SAWKAT ANWER, Boston, Massachusetts

BRUCE R. BACON, St. Louis, Missouri Henry C. BODENHEIMER, New York, New York

JAMES M. CRAWFORD, Boston, Massachusetts NORMAN D. GRACE, Boston, Massachusetts

SANJEEV GUPTA, Bronx, New York JOEL LAVINE, Boston, Massachusetts

fore it has been difficult to find appropriate animal models in which to study the relevant pathogenic factors. The advent of transgenic animal technology is providing a solution to this dilemma. This publication describes the creation of animals that answered several fundamental questions and will provide a model for future animal studies that may be of direct relevance to human beings.

A very large protein, apolipoprotein B (apoB), 550 kDa, is required for the formation of the triglyceride- containing lipoprotein classes (very low density lipo- protein [VLDLI and chylomicrons) by the liver and intestine, respectively (2, 3). In the intestines of all mammalian species, the mRNA encoding the full-length protein apoB,,, is edited by the deamination of cytidine at codon 2153. This results in the creation of a stop signal (UAA) rather than a glutamine (CAA) codon and the formation of a smaller protein called apoB,, (4). The livers of most non-primate species produce both apoB,,, and apoB4*, but in primates only apoB,,, is formed by the liver. This raises the question of whether the lack of editing of apoB mRNA in primate liver is due to a property of human apoB or one of the liver. Further, in most species, most apoB-containing lipoproteins are rapidly cleared from the circulation and little apoB accumulates in the blood, at least while the animal is consuming a low-fat diet. In human beings, however, the large VLDLs are converted to smaller low-density lipoproteins (LDLs), and this apoB-containing lipo- protein becomes the major carrier of cholesterol in human serum even with a low-fat diet. The level of LDL is a strong risk factor for atherosclerosis. If the accu- mulation of LDL in human beings is due to a special property of the human apoB molecule, this is of considerable importance.

Further, in human beings and in New World primates (but in no other species), a lipoprotein particle called LP(a) accumulates. This particle is the size of LDL and contains apoB,,,, to which another protein, apo(a), is covalently linked. The level of LP(a) is a risk factor for atherosclerosis independent of other factors, including LDL and high-density lipoprotein levels. The question of where and how LP(a) is formed has not been resolved; nor are there suitable nonprimate models in which to study this lipoprotein (5).

This work is the result of collaboration between two laboratories that are leaders in their fields. Eddie Rubin’s laboratory at the University of California,

1635

1636 HEPATOLOGY Elsewhere HEPATOLOGY December 1994

Berkeley, is a leader in transgenic technology; and Dick Lawn’s laboratory at Stanford University has pioneered LP(a) molecular biology. These investigators now provide partial answers to some of the questions raised above.

Because of its size and other, yet-unexplained prop- erties of the apoB gene, it has been difficult to express this gene in vivo although it has been partially expressed in cell culture (6). Recent advances in technology allowing cloning of large genomic fragments enabled Rubin’s laboratory and one other group to obtain very large genomic clones of the apoB gene (7). This achievement was accomplished by the use of a P1 vector that can be used to clone genomic libraries composed of large DNA segments. In this study, the investigators were able to isolate a DNA fragment of 80,000 to 90,000 bp that contained the entire apoB coding region (about 45,000 bp), as well as large 3‘ and 5’ flanking segments of DNA, which it was anticipated would have the information governing the expression and regulation of the gene.

These large fragments were introduced into the nuclei of fertilized mouse ova, and 11 lines of transgenic mice expressing human apoB were obtained. Between one and 15 copies of the gene were incorporated into the founders of the various transgenic mouse lines. The amount of human apoB in the blood of the mice appeared to be proportional to the number of genes present.

Strikingly, despite the introduction of this large fragment into the mouse genome, apoB was expressed only in the liver and not in the intestine. This implies that one or more DNA sequences that are very remote from the coding region of the apoBV gene may be required for intestinal expression of the gene. Fur- thermore, in the livers of mice, both human apoBlo0 and human apoB,, were expressed. Thus human liver must lack the RNA-editing capability; the human gene can clearly be edited by mouse liver, implying that the human mRNA has the appropriate information to allow editing.

The expression of human apoB in the transgenic mice led to an increase in the serum cholesterol that was due almost exclusively to the accumulation of an LDL-like particle in the sera of the transgenic mice. One impli- cation of this observation is that something intrinsic to human apoB protein may be responsible for the lack of rapid clearance of VLDL that results in the accumu- lation of an LDL-like particle. The authors ascribe this to poor binding of human apoB to the mouse LDL receptor and thus to an intrinsic property of human apoB. This particular point will, however, require further experimentation; it is possible that accumu- lation is due to overproduction of apoB per se, which could lead to saturation of the VLDL clearance mechanism. To resolve this, it will be necessary to overexpress mouse or rat apoB to a comparable extent and show that it does not accumulate and to demon- strate that mouse apoB is not also accumulating in the LDL fraction of human apoB expressors. Interestingly, under most circumstances, the amount of apoB mRNA

does not correlate with the rate of lipoprotein production (3). Therefore the mechanism of the effect of the gene dose seen in thse mice also remains to be elucidated.

In the second part of the paper, the LP(a) story is addressed. Some years ago, Lawn and his colleagues at Genentech cloned the gene for apo(a) (8). They made the startling discovery that this molecule had a high degree of homology to plasminogen. The latter molecule is composed of five regions called kringles (because their shape resembles a Danish pastry called a kringle). Apo(a) is composed of almost precise copies of the protease portion and kringle 5 of plasminogen and multiple tandem copies of kringle 4 of plasminogen. Depending on the number of copies of kringle 4, the molecular weight of apo(a) varies between 300 and 800 kDa. In general, the level of LP(a) in the blood is inversely related to the size of the apo(a) being produced, but other, yet- unknown regulatory factors also exist. Lawn and col- laborators in Dallas (9), had earlier produced transgenic mice that expressed apo(a). These mice did not, however, produce LP(a) unless human LDL was infused (9). This suggested that mouse apoB did not couple with apo(a) to form LP(a). This raised the possibility that formation of LP(a) was either human apoB specific or required a factor present in primate liver or on the human particle. Interestingly, even though these mice did not have LP(a), they were unusually susceptible to atheroscle- rosis, confirming the potential role of apo(a) in athero- genesis (10).

Lawn and Rubin mated their mice, and lo and behold the progeny produced a molecule indistinguishable from human LP(a). This lends strong support for the hy- pothesis that the ability to produce LP(a) is a specific property of primate apoB. These mice will now be an important experimental model in which to address the questions of precisely how LP(a) is formed and what controls its levels.

Thus the creation of human apoB transgenic mice and their marriage with apo(a) transgenic mice answered several unresolved questions in hepatic lipoprotein physiology and will provide an important set of animal models to facilitate future research on atherosclerosis, an important disorder of hepatic metabolism.

ALLAN COOPER, M.D. Research Institute Department of Medicine Stanford University Palo Alto, California

REFERENCES 1. Cooper AD. Hepatic lipoprotein and cholesterol metabolism. In:

Zakim D, Boyer T, eds. A textbook of liver disease. Philadelphia: Saunders, 1990:96-114.

2. Young SG. Recent progress in understanding apolipoprotein B. Circulation 1989;80:219-233.

3. Dixon JL, Ginsberg HN. Regulation of hepatic secretion of apolipoprotein B-containing lipoproteins: in formation obtained from cultured liver cells. J Lipid Res 1993;34:167-179.

4. Powell LM, Wallis SC, Pease RJ, Edwards YH, Knott TJ, Scott J.

HEPATOLOGY Vol. 20, NO. 6, 1994 HEPATOLOGY Elsewhere 1637

A novel form of tissue-specific RNA processing produces apolipo- protein-B48 in intestine. Cell 1987;50:831-840.

5. Lawn RM. Lipoprotein(a) in heart disease: a remarkable protein that transports cholesterol and binds with blood clots can raise the risk of a heart attack: comparisons between it and other blood proteins may explain why. Sci Am 1992;266:54-60.

6. Yao Z, Blackhart BD, Johnson DF, Taylor SM, Haubold KW, McCarthy BJ. Elimination of apolipoprotein B48 formation in rat hepatoma cell lines transfected with mutant human apolipo- protein B cDNA constructs. J Biol Chem 1992;267:1175-1182.

7. Linton MF, Farese RV J r , Chiesa G, Grass DS, Chin P, Hammer RE, Hobbs HH, et al. Transgenic mice expressing high plasma concentrations of human apolipoprotein BlOO and lipoprotein (a). J Clin Invest 1993;92:3029-3037.

8. McLean JW, Tomlinson JE, Kuang WJ, Eaton DL, Chen EY, Fless GM, Scanu AM, et al. cDNA sequence of human apolipoprotein(a) is homologous to plasminogen. Nature 1987;330:132-137.

9. Chiesa G, Hobbs HH, Koschinsky ML, Lawn RM, Maika SD, Hammer RE. Reconstitution of lipoprotein (a) by infusion of human low density lipoprotein into transgenic mice expressing human apolipoprotein (a). J Biol Chem 1992;267:24369-24374.

10. Lawn RM, Wade DP, Hammer RE, Chiesa G, Verstuyft JG, Rubin EM. Atherogenesis in transgenic mice expressing human apolipo- proteida). Nature 1992;360:670-672.

EFFECTS OF HIGH-NORMAL AND LOW-NORMAL SERUM POTASSIUM LEVELS ON HEPATIC

OR ARTIFACTS?

Zavagli G, Ricci G, Bader G, Mapelli G, Tomasi F, Maraschin B. The importance of the highest normo- kalemia in the treatment of early hepatic encepha- lopathy. Miner Electrolyte Metab 1993;19:362-367.

ENCEPHALOPATHY: FACTS, HALF-FACTS

ABSTRACT

An inverse relation is known to link blood potassium with renal synthesis and the release of ammonia. Given the liability of hyperammonemia for precipitating hepatic encephalopathy (HE), 28 patients affected by stage I HE were equally divided into two groups and maintained up to their death at the highest (5.4-5.5 mEq/l) or the lowest (3.5-3.6 mEq/l) normokalemia levels. When compared with the lowest normokalemia group, the highest one showed an early, albeit tran- sient, improvement in the mental state (as assessed by both EEG and psychiatric investigations) and to a lesser extent in hepatic functions (as assessed by the variations in serum bilirubin, GPT, GGT and plasma prothrombin time). In the highest normokalemia group the survival was also prolonged. The cause of this improvement may be related to the induced decrease in blood pH, the consequent depression of renal ammo- niagenesis and the rise in the arterial and urine NH+,JJ3N3 ratios. These factors reduce the entry of ammonia into the cells and enhance the urinary excretion of this metabolite, respectively.

COMMENTS

A spoonful of sugar makes the medicine go down.

Mary Poppins Cirrhotic patients, especially those with alcoholic

cirrhosis, are prone to potassium depletion resulting from loss of potassium through vomiting, diarrhea, deficient dietary intake and the use of kaliuretic diuretic

drugs. Furthermore, secondary hyperaldosteronism, which is common in decompensated cirrhosis, prevents the body from repleting the lost potassium. Indeed, in our experience decompensated alcoholic cirrhotic pa- tients, on admission to the hospital, have metabolic alkalosis (mean arterial pH, 7.49) and hyperammonemia (mean arterial ammonia level, 242 rJ.g/dl; upper limit of normal, 150 pg/dl); a mean serum potassium level of 3.0 mEq/L; cumulative body potassium deficit of 400 to 2,000 mEq) (1; Conn HO, Unpublished observations). It is ironic that physicians, who are about to admit such patients to the hospital, apparently have an uncon- trollable urge to administer furosemide just before admitting them, resulting in the renal excretion of any residual potassium stores.

The association of diuretic-induced hypokalemia, kaliopenic metabolic alkalosis and hepatic encepha- lopathy (HE) was elucidated 25 years ago in an in- sightful series of studies by Imler and his associates in Strasbourg (2, 3). They demonstrated that chlortha- lidone, an early loop diuretic agent (100 to 200 mg/day for about 1 wk), administered to 31 edematous, cirrhotic patients, increased mean arterial ammonia levels from 74 2 112 pg/dl, decreased mean serum potassium con- centrations from 4.0 to 2.7 mEqL and increased mean arterial pH from 7.45 to 7.53. In eight of these patients, about one fourth of the gruop, early HE developed during the first week of treatment. The mean ammonia concentration in these eight patients was appreciably higher than that of the 23 who did not become encephalopathic (141 vs. 94 p,g/dl, p < 0.001). A close correlation between the increase in arterial ammonia levels and the decrease in serum potassium levels was observed. These biochemical aned clinical abnormalities were prevented or reversed by the administration of potassium chloride. These alterations were associated with an increased arterial-femoral venous ammonia difference, indicating that the increased arterial am- monia concentration is not the result of a decrease in muscle uptake of ammonia (Fig. 1). In fact, it occurred despite a small increase in arteriovenous muscle uptake. Hepatic uptake of ammonia remained constant.

Similar results were observed with other kaliuretic diuretics such as chlorothiazide (41, mercury derivatives (51, furosemide and ethacrynic acid (3). Furthermore, similar patterns were observed in metabolic hy- pokalemic alkalosis induced by diarrhea, decreased dietary potassium intake, hyperaldosteronism and kalio- penia-inducing cation-exchange resins (3) or in experi- mental animals with potassium depletion (6, 7).

When acetazolamide (Diamox), a carbonic anhydrase inhibitor, was administered (750 to 100 mglday), a different pattern was observed: Although the serum potassium decreased and the arterial ammonia levels increased as a result of increased renal ammoniagenesis, the arterial pH decreased, indicating that the pH effect was a consequence of pure carbonic anhydrase inhibition (8).

Spironolactone (Aldactone) given in doses of 300 to 800 mg/day for 1 to 4 wk, which also induced HE in about