F-actin and Microtubule Suspensions as Indeterminate Fluids

13. 14. '15. 16. 81, 109 (1968). Mean daytime temperature of the plot surface sand was 25.4C (SD, 2.7; n=' 9). The snail can tolerate this, and if either party is at risk, it is probably the parasite. J. J. Gambino,J. Parnsiol. 45, 440, 456 (1959). W. B. Vemberg, F. J. Vernberg, F. W. Beccerdite, Scence 164, 1287 (1969). J. S. McDaniel and J. R. Coggins,J. ElisbiaMitchcU Sci. Soc. 88, 55 (1972). P. J. DeCoursey and W. B. Vemberg, in Symbiosi in F-Actin and Microtubule Suspensions as Indeterminate Fluids ROBERT E. BuXBAiuM, TimOTHY DENNERLL, SERGE WEISS, STEVEN R. HEIDEMANN The viscosity of F-actin and microtubule suspensions has been measured as a func tion of shear rate with a Weissenberg rhoogoniometer. At shear rates of less than 1.0 per second the viscosity of suspensions of these two structural proteins is inversely proportional to shear rate. These results are consistent with previous in vivo measi urments of the viscosity of cytoplasm. This power law implies that shear stres,;s is independent of shear rate; that is, shear stress is a constant at all shear rates less t han 1.0 per second. Thus the flow profile of these fluids is indeterminate, or nearly so. 1[his flow property may explain several aspects of intracellular motility in living aells. Possibl explanations for this flow propet are based on a recent model for semidi]lute suspensions of rigid rods or a cassical friction model for liquid crystals. N CTIN AND MICROTUBULES ARE among the most widely conserved tructural proteins within eukaryotic cells. It has long been suspected that the viscoelastic properties of these polymers (particularly F-actin) play an important role in the maintenance and development of cell shape and in cellular motility (1). Maruyama et al. (2) found that the measured viscosity of F-actin was precisely inversely proportional to the shear rate. These workers reported on the uniqueness of this inverse proportionality and speculated that it may be the result of the formation of an actin network. Subsequent rheological studies confirmed that F-actin suspensions are shear tinning but did not find the same shearviscosity dependence or did not discuss this relationship (3, 4). We confirm the observation of Maruyama et al. for actin and observe a similar inverse proportionality for suspensions of microtubules. Also, we show that as a result of this dependence, the fluid flow profiles for such suspensions are indeterminate. That is, a given force does not fix the velocity of movement in these fluids. Figure 1 is a logarithmic plot of viscosity versus shear rate for two runs of actin at different concentrations, two runs of microtubules with and without taxol [a stimulator of microtubule assembly (5)], and one run of the viscosity standard. The values reported in Fig. 1 and throughout this report are the stable values obtained after shearing the fluids for several seconds. As found previ20 MARCH I987 ously for actin (2) in both actin and mi crotubule suspensions, the torque rises shairply as shear begins but then declines to a st;able value over a time period that depends on the shear rate. In addition to the shear-thinriing behavior characteristic of all polymer suspensions, these plots show that the power law dependency for actin and microtubWdes is nearly -1 for all concentrations. ThaLtis, actin and microtubule suspensions behavre as power law fluids with 1 an = Aln (1) where 9 is viscosity, A is a constant, ; is shear rate, and n is the power law expon,lent. In 12 runs of actin at concentrations fro]m 2 to 6 mg/ml the exponent varied firom -0.85 ± 0.02 (+ standard error of reg,ression coefficient) to -1.15 ± 0.01 witth a mean of -1.00. Both extreme values occurred for actin at 2 mg/ml with the r*xeogoniometer at its lower limit of resolutc ion. In seven microtubule runs the exporlent varied from -0.90 ± 0.01 to -1.03 ± C).03 with a mean of -0.94. The -1 slope (on log-log plots) was found to be nearly iridependent of the shearing history of the s,ample, as-shown in Fig. 2. The slope of the line for increasing shear (-1.00 ± 0.02) is n early identical to that for decreasing si hear (-1.01 ± 0.02). In a run of chrom;atographed actin (3 mg/ml) in which shear was varied at random the power law depende.nce was -0.99 ± 0.09. The power law deE?endence of the microtubule suspension visc-osi- the Sea, W. B. Vernberg Ed. (Univ. of South Carolina Press, Columbia, 1974), pp. 93-109. 17. My thanks to M. R. Carriker and[L.E.Hurdfor valuable comments on the manuscript and to A. M. Barse for field assistance. Partial fund was provided by a Biomedical Research Support Grant fromn the University of Delaware. 4 September 1986; accepted 28 Januarv 1987 ty was similarly insensitive to shear history. Our measuremcnts varied only slightly. The small standard errors of regression coefficients indicate little variation of points around the calculated line. The slope from run to run also varied little at nominally the same conditions. In five runs of actin at 6 mg/ml from three different actin preparations the mean slope was 1.01 ± 0.04 (SEM). The possibility that our results are due to slipping at rheometer surfaces is eliminated by the finding that actin and microtubule suspensions pour easily and are well mixed after the shearing process in the rheometer. The observed viscosities for suspensions of these two filamentous constituents of the cytoplasm are consistent with previous in vivo measurements of cytoplasmic viscosity. Figure 3 shows a comparison of data obtained from various cytoplasmic studies with that of F-actin suspension at 3 mg/ml. This similarity of the viscosity function in vitro and in vivo is particularly interesting in view of the lack of true crosslinks within purified F-actin (6) and the highly cross-linked structure of cytoplasm (7). Power law observations in this range are significant because they contradict Graessley's classic random-coil-entanglement explanation that suggests the minimum n value is -9/11 (8). Typically, data show n between -0.4 and -0.85 for polymer solutions (9). Significantly, since shear stress is viscosity multiplied by shear rate (Amn)j =Al(n+'1) (2) unit where is the shear stress (force per area). Our observation that n = -1 means that the shear stress is independent of shear rate. In this case, fluid motion is completely indeterminate. In Fig. 4 we plot shear stress data for actin, microtubules, and the Newtonian fluid standard. One's intuition, for example, on the basis of flooded rivers, is that increased shear increases shear stress. As shown in Fig. 4, the Newtonian standard T = 'O = r R. E. Buxbaum, Department of Chemical Engineering, Michigan State University, East Lansing, MI 48824. T. Denncril, S. Weiss, S. R. Hcidemann, Department of Physiology, Michigan State University, East Lansing, MI 48824. REPORTS I511 Downloaded from on March 30, 2016 10. E. J. Pearson and T. C. Cheng,J. Invert. Patlo. 46, 239 (1985). 11. Snails could have been present during the day but buried. On three daytime tides after evicting snails I scratched 4 to 5 cm into the entire plot. A few snails were unearthed but never enough to approach nighttime counts. Moreover, snails tend to unbury wtien submerged by tides. Never was there a marked unburying after daytime low tides. 12. C. W. Schaefer, P. W. Milch, N. L. Levin, Nautilus follows the intuitive expectation. However, for both actin and microtubules the shear stress remains constant at all shear rates tested. This constant is the valueA in Eqs. 1 and 2. In these experiments we controlled the shear rate and measured the force. In most biological (and other) situations, however, the interaction is reversed: a pump, a gravitational potential, or a cell supplies a particular shear stress, and the flow pattern accommodates it. However, as shown in Fig. 4, a force alone cannot fix the flow patterns for actin and microtubule (that is, indeterminate) suspensions. For any geometry, only one force pattem gives reasonable flow rates; push with less force and there is no motion, push with more force and the fluid velocity is limited by inerta alone. Actin and microtubules are widely regarded as important in determining the physical properties of cytoplasm (1). Energy-dependent movement of partides through the Fig. 1. Viscosity as a function of shear rate at 23°C on a log-log scale for suspensions of cytskeletal proteins and a viscosity standard. Actin was purified from rabbit skceletalmuscl by the meth- od of Spudich and Watt, or by their method and fiuther purification by chromatography on Sephacryl S-200 (Phannacia, Piscataway, NJ) (19). Punifled actin was 4 maintained in dialysis at40C against buffer A [2.0 mM tris-HCI, pH 8.0, 0.2 mM 10,000 cytoplasm occurs in all eukaryotic cells, and brane-cortex structures or, in the case of cytoplasm itself is capable of rapid, energy- algae, by cytoplasmic walls. The flowing dependent movement in some cells (10, 11). cytoplasm and the containing cytoplasm are Our data suggest that the velocity of move- fundamentally similar in their content of ment is not limited by force, and the low water and proteins. Flow indeterminacy Reynolds number characteristic of living may explain how these two regions of simisystems (12) suggests that inertia is not lar chemistry have such different velocities responsible. By deduction then, the ob- while bearing exactly the same stresses. Posserved velocities are fixed by the chemical sibly, the cell pushes the cytoplasm at conkinetics of the biological motors. For exam- stant force in some part of the cell and drags ple, structures within cells move suddenly on the cytoplasm in the rest of the cell over moderate (-30 ,m) distances at con- (necessarily with the same force per unit stant velocity, and they halt equally suddenly area). If the area pushing is smaller than the (10). One explanation is that no motion area dragging, there will be no cytoplasmic occurs until a threshold force is reached, flow. If it is larger, flow is rapid. Thus, for then a constant velocity is maintained by the finite flows and equal shear on the cytostable kinetics of the mechanochemical mo- plasm and membrane, small differences in tor. The unusual fluid dynamics of nearly in- 10,000 determinate fluids may also explain the flow properties that characterize cytoplasmic streaming in various cells (11). In these cells 1,000 shearing flows are contained by cell mem- + 0 - 10 + S 1,000 10+ iL 1. j 0.I 101 oC 100 o M ATP (adenosine triphos- > k phate), 0.5 mM mercaptoethanol, and 0.2mM CaC12] for no more than 6 days. 10 Microtubule protein was purified from bcef brain in the presence of glycerol according to the method of Shelanski et at. (20) with the 1I modification that 0.1M Pipes (piperazine ethanesul0.1 0.01 0.001 I fonic acid), pH 6.6, was Shear rate (sec-1) used in all reassembly buffers. A model R.18 modified (21) Weissenberg rheogoniometer was used for all rheometry experiments. Actin was polymerized by adding KCI to a final concentration of 50 mM and MgSO4 to a final concentration of 2 mM to G actin at various concentrations in buffer A; the mixtur was applied quickly to the bottom cone (2°, 5 cm in diameter) of the rheogoniometer, and the actin was polymerized for 1 hour at 23°C before measurements were made. Microtubule protein suspended in 0.lM Pipes,pH 6.9, 1 mMMgCl2, 1 mM EGTA, 0.1 mM EDTA, and 0.5 mM GTP (guanosine tiphosphate) was polymerized at 37°C for 20 minutes and applied to the cone of the rheogoniometer. A humidified atmosphere was maintained around the cone and plate assembly by adding a small pool of water to the floor of the instrument's temperature control chamber. For each sample of protein loaded into the rheometer, 10 to 14 shear rates (drive settings) were chosen to obtain torque readings. After each torque reading the drive was turned off and the torque allowed to return to baseline before we reset the drive for the next reading. Measurements on a single sample of protein took approximately 2 hours. The rheogoniometer was calibrated with a Newtonian viscosity standard (R12000, Cannon Instiment Company). Measurements ofviscosity and shear were calculated from rheogoniometer drive settings and torque readouts by standard formulas. (O) F-actin, 6 mg/ml; slope, -1.00 ± 0.02; (0) chromatographed F-actin, 2 mg/ml; slope, -0.98 ± 0.02; (A) polymerized microtubule protein, 12 mg/mil; slope, -1.00 ± 0.01; (A) microtubule protein, 12 mg/ml, polymerized in the presence of 5 pM taxol; slope, -0.95 ± 0.03; (0) viscosity standard, slope, 0.00 ± 0.01. 1512 100 06 .2 0.01 Shear rate 0.1 (.c-1) Fig. 2. Two shear protocols for chromatographed F-actin, 3 mg/ml, from low to high rates of shear (0), slope, -1.00 ± 0.02, and from high to low shear (0), slope, -1.01 ± 0.02. 10,000' 1-X. 1,0004 0. 0 a Q 10 1004 a" 1" 1"1% l0o 1%1V 1' 0.001 . . .. .~~~1 0.01 0.1 Shear rate I 10 (sec-1) Fig. 3. Viscosity of chromatographed F-actn, 3 mg/ml (0) compared with viscosity measurements of cytoplasm by various in vivo methods: (0) Valberg and Albertini (22), (O) Nemoto (23), (A) Sung et at. (24), (A) Hirmato (25), (M) King and Macklem (26), (0) Sato et al. (27), (V) Crick and Hughes (28), and (V) Yagi (29). The linear regression for the in vivo data (----) gives a slope of -1.02 0.18. SCIENCE, VOL. 235 Fig. 4. Shear stress as a function of shear rate for chromatographed F-actin, 3 mg/ml (0); poly- 20 merized tubulin, 6 mg/mi (A); and the R12000 viscosity standard (O). C~ 10 0 0 0.001 a 0 0.01 0 Shear rate chemistry can produce enormous differences in flow. This viscosity relation may also play a role in slow axonal transport, which involves a flow of cytoskeletal elements down the neural axon at a variety ofdifferent rates characteristic of the different cytoskeletal elements (13). If the fluid flow is nearly indeternminate and the axonal pressure drop is constant, slight changes in the pushing pressure or entrance region pressure drop will greatly affect the transport rate down the axon. Another possibility from shear-tinning behavior that does not require flow indeterminacy concerns orgaiisms that locomote upon a slime trail. In most instances this slime is a polymer solution that would be shear thinning; most crawling results from sustained oscillations of contraction and extension. Two models are possible: (i) an asymmetric oscillation in space so that the moving area is always smaller than the stationary area or (ii) an asymmetric oscillation in time both with the condition that the drag on the front of the organism over the substrate is deternined by Newtonian viscosity (in water, for example) and the drag on the rear of the creature is determined by shear-thinning slime. Then, as in Fig. 1, if the cell extends faster than it contracts, a net force will move the organism forward. Snail slime possesses "solid-like" properties at low shear rates (14) in concert with a shearthinning fluid model. The rheology of actin suspensions has generally been thought to be the result of network destruction by shearing (2, 6, 15). Although microtubule suspensions do not form gels or the looping entanglements required by network theories (4, 8), they have the rheological properties of actin suspensions. Further, the limiting power law exponent for entangled polymers is calculated to be -9/11 (8). Our data (Fig. 2) also indicate that shearing history has a relatively small effect on the viscosity of actin and microtubule suspensions. These observa20 MARCH 1987 ,1 ~~~~~~1 (sec-1) tions suggest that network destruction is not primarily responsible for the rheology of these filamentous proteins. We have formulated two alternative models for these viscosity observations. One model follows a suggestion of Zaner and Stossel (3) for actin suspensions and assumes that semidilute suspensions of rigid rods are at concentrations below the liquid crystal formation (that is, an isotropic distribution of the rods in space before shearing). In this model, as treated by Doi and Edwards (16) and by Jain and Cohen (17), a large obstructional effect is caused by the constraint that rods cannot pass through each other. Assuming that the frequency of rod-rod obstructions is due to shear-independent Brownian motion, Doi and Edwards (16) found regions where shear stress is independent of shear rate. To the extent that this model is appropriate, the phenomenon that is typically called "gelation" (true solidification) in the biological literature is seen as the "glass" (liquid with solid properties) transition induced by this obstructional constraint. This glass transition explains why the viscosity of cytoplasm and actin suspensions is so similar, that is, why crosslinkers in cytoplasm have so little effect on viscosity, as shown in Fig. 3. It also explains why actin "gels" behave as liquids. This "log jam model" can predict the observed -1 power law dependence of the viscosity and also that the power can be smaller than -1 over small ranges of shear. We propose a second model to explain the observations with relatively flexible actin filaments and to extend the applicable concentrations above those for liquid crystal formation. This model differs from the network model and rigid rod model in that the fluid is locally anisotropic. Fluid motion is not continuous but occurs among discrete domains in a classical analogy to solid sur- face friction or superfluidity. Anderson (18) has proposed this model to describe largescale density waves in smectic B liquid crys- Fig. 5. Polarization micrographs of suspensions of actin and microtubules (Leitz Ortholux I Pol microscope, x1200). (A) F-actin at 25 mg/mi. (Similar domains were observed at concentrations of 8 mg/nl, but the birefringence was too dim to photograph successfully.) (B) Microtubules at a concentration of 6 mg/ml. tals. If one domain slides over another, two roughnesses encountering one another must either be completely overwhelmed by the forces driving them or must stop the motion entirely. A roughness cannot provide a continuous source of linear dissipation because any roughness that causes a purely elastic deformation does not dissipate energy. Instead, it returns as much mechanical energy to the lattice as was present in the original deformation. Viscous stress is thus viewed as the (shear rate independent) force necessary to substantially kink or break those protein strands that extend from one solidlike domain to another. The effect that biologists call gelation is, presumably, liquid crystal domain formation. This solid friction analogy is supported by the finding of ourselves and others (2) that the initial shear stress (analogous to static friction of solids) is always greater than the steady-state stress (analogous to sliding friction of solids). Additional evidence for this model is shown in Fig. 5 in which actin and microtubule protein suspensions reveal distinct domains under polarized light in contradiction to the assumptions of the rigid rod model described above. REFERENCES AND NOTES 1. T. P. Stossel, J. Cel Bidl. 99, lSs (1984); K. R. Porter, ibid., p. 3s; D. L. Taylor and J. S. Condeelis, Int. Rev. Cytol. 56, 57 (1979). 2. K. M. Maruyama, M. Kaibara, E. Fukuda, Biohm. Bio*py. Acta 371, 20 (1974). 3. K. S. Zaner and T. P. Stossel, J. Biol. Chem. 258, 11004 (1983). 4. M. A. Rockwell, M. Fechheimer, D. L. Taylor, CeU Motil. 4, 197 (1984); M. Sato, G. Leirnbach, W. H. Schwartz, T. D. Pollard, J. Biol. Chem. 260, 8585 (1985). 5. P. B. Schiff, J. Fant, S. B. Horwitz, Nature (London) 277,665 (1979). 6. K. S. Zaner and T. P. Stossel,J. CeU Bid. 93, 987 (1982). REPORTS I513 7. J. J. Wolosewick and K. R. Porter, ibid. 82, 114 (1979); J. Heuser and M. W. Kirschner, ibid. 86, 212 (1980); M. Schliwa, K. B. Pryswansky, J. Van Blerkom, Philos. Tram. R. Soc. London Sff. B 299, 199 (1982). 8. W. W. Graessley,J. Chem. Phys. 43, 2696 (1965); ibid. 47, 1942 (1967). 9. R. B. Bird, R. C. Annstrong, 0. Hassenagen, Dynmics of Polymeric Liquids (Wiley, New York, 1977), vol. 1. 10. L. I. Rebhun, Int. Rev. Cytol. 32, 92 (1972). 11. N. Kamiya, Annu. Rev. Plant Phsiol. 32, 205 (1981). 12. E. M. Purcell, Am. J. Phys. 45, 3 (1977). 13. M. Willard, W. M. Cowan, P. R. Vagelos, Proc. Nati. Acad. Sci. U.SA. 71, 2183 (1974); P. N. Hoffinan and R. J. Lasek, J. Cell Biol. 66, 351 (1975). 14. M. W. Denny and J. M. Gosline, J. Exp. Med. 88, 375 (1980). 15. S. MacLean-Fletcher and T. D. Pollard, J. CeU Biol. 85, 414 (1980). 16. M. Doi and S. F. Edwards, J. Chem. Soc. Faraday Trans. 2 74, 560 (1978). 17. S. Jain and C. Cohen, Macrmolculs 14, 759 (1981). 18. P. W. Anderson, Basic Noiiom in Coensed Matter PhAsi (Benjamin-Cummings, Menlo Park, CA, 1984), pp. 159-164. 19. J. A. Spudich and S. Watt,J. Biol. Chem. 246, 4866 (1971); S. MacLean-Fletcher and T. D. Pollard, Biochem. Biop ys. Res. Commun. 96, 18 (1980). 20. M. L. Shelanski, F. Gaskin, C. R. Cantor, Proc. Nati. Acad. Sci. U.SA. 70, 765 (1973). 21. J. Meissner, J. Appl. Pdym. Sci. 16, 2877 (1972). 22. P. A. Valberg and D. F. Albertini,J. Cdl Biol. 101, 130 (1985). 23. I. Nemoto, IEEE Trams. Biomed. Eng. 29, 745 (1982). 24. K. P. Sung ct al., Bioph. J. 39, 101 (1982). Two Mammalian Genes Transcribed from Opposite Strands of the Same DNA Locus JOHN P. ADELMAN, CHRIS T. BOND, JAMES DOUGLASS, EDWARD HERBERT This report describes the characterization of a genomic locus in the rat that encodes overlapping genes occupying both strands of the same piece of DNA. One gene (strand) encodes gonadotropin-releasing hormone (GnRH). A second gene, SH, is transcribed from the other DNA strand to produce RNA ofundefined function. The RNAs transcribed from each DNA strand are spliced and polyadenylated, and share significant exon domains. GnR.H is expressed in the central nervous system while SH transcripts are present in the heart. IThus, the genome of a mammalian organism encodes two distinct genes by using both strands of the same DNA. I NHERENT TO THE CONCEPT OF A EU- within the first Gart intron and transcribed DNA strand opposite the transcribed strand is to serve as a necessary conduit for double-stranded, semiconservaiive DNA replication. Likewise, although the discovery of introns brought about modifications in the concept of genes as linear discrete units along the chromosome, introns have been regarded as raw genetic material susceptible to high levels of evolutionary drift with regard to nucleotide sequence and even to size fluctuation. Exons have been regarded as the biologically relevant portions of a gene, existing under varying degrees of selective pressure, reflective of the relative advantage they confer upon the organism possessing the particular allele. However, a number of recent reports have indicated a need to reexamine these concepts. The Gart locus of Drosophila melanogaster, known to encode three purine pathway enzymatic activities, has been shown to contain an entire gene encoding a cuticle protein, "nested," cuticle protein gene itself is divided by an intron. Another Drosophila locus, encoding dopa decarboxylase, has been shown to share an 88-bp region at its 3' end with the 3' end of a transcript arising from an unknown gene on the opposite strand (2). A comparable situation has been described for kayotic gene is that the function of the from the opposite DNA strand (1). The The Institute for Advanced Biomedical Research, Oregon Health Sciences University, 3181 S.W. Sam Jackson Park Road, Portland, OR 97201. *This work is Herbert. I5I4- dedicated to the memory of Edward &r^rn"ea ftltO11rP a rlgonll or mousc nNA L cells, BALB/cTS-A-.3T3, which encodes two 3' overlapping transcripts of undetermined functions (3). The data presented here exerrormiia.o tend these observatiions to describe the vir2 700- 500- 3 4 25. Y. Hiramato,Exp. Cel Res. 56, 201 (1969). 26. M. King and P. T. Macklem,J. Appl. Physil. Resir. Environ. Evercics Phrsiol. 42, 797 (1977). 27. M. Sato, T. Z. Wong, R. D. Allen, J. Cell Biol. 97, 1089 (1983). 28. F. H. C. Crick and A. F. W. Hughes, Exp. CcU Res. 1, 37 (1950). 29. K. Yagi, ERp. Biochem. Physiol. 3, 73 (1961). 30. We thank J. Sethna and D. Berry for alerting us to important modeling work, K. Jayaraman for the use of his equipment, C. Cohen and J. Condeelis for stimulating conversations, and J. R. McIntosh for comments on the manuscript. S.R.H. thanks J. Condeelis for the hospitality m his laboratory. Supported by NSF grant BNS 8401904. S.R.H. is a recipient of a Research Career Development Award from the National Institutes of Health. 21 August 1986; accepted 30 December 1986 tual cohabitation of a genomic locus within mammalian DNA by two distinct genes. Both DNA strands serve as templates for transcription, and the resulting polyadenylated [poly(A)+] RNAs contain a significant amount of shared exonic sequences. The gonadotropin-releasing hormone (GnRH) gene is a single-copy gene exhibiting a similar arrangement of four exons in both rats and humans (4). In the maturation ofhuman placental preproGnRH messenger RNA (mRNA), the first intron is not removed and thus the human placental GnRH mRNA possesses an exceptionally long 5' untranslated region (5). During experiments to characterize the hypothalamic form of GnRH mRNA, human placental GnRH complementary DNA (cDNA) was nicktranslated and used as a hybridization probe to screen rat hypothalamic cDNA clones. DNA sequence analysis demonstrated that ten of the clones showing hybridization to this probe represented the hypothalamic form of preproGnRH mRNA. Surprisingly, Fig. 1. Northern analysis of rat hypothalamic and heart poly(A)+ RNA. Duplicate samples of heart (lanes 2 and 4) and hypothalamic (lanes 1 and 3) poly(A)+ RNAs were prepared as a Northern blot and probed with an SP6-generated GnRH (lanes 1 and 2) or SH (lanes 3 and 4) strand-specific probe. In control hybridizations to Ml3 template DNAs of defined sequence the SP6 probes were shown to be completely strand-specific. Size markers are indicated to the left. Total cellular RNA was prepared from the indicated rat tissues by the guanidinium/GsCl method (27) and poly(A)+ RNAs selected on oligo(dT)-cellulose. Glyoxal-denatured poly(A)+ RNA (2.5 ,ug per lane) was fractionated on a 1.4% agarose gel, run in 10 mM NaPO4 (pH 6.7) (28), and subsequently transferred to Nytran. SP6 strand-specific RNA probes were generated as described (7). Probes were labeled to a specific activity of 5 x 0l8 counts per minute per microgram of RNA. Blots were probed at 60'C in standard hybridization solution in the presence of 50% formamide. Filters were washed in .07x standard saline citrate/5 mM EDTA/0.5% SDS at 700C and exposed to x-ray film for 10 hours. SCIENCE, VOL. 235