Comparison between the Time Course of Changes in Nerve Growth Factor Protein Levels and Those of Its Messenger RNA in the Cultured Rat Iris*

In previous experiments, it has been demonstrated that, in rat irides in culture, a rapid increase in nerve growth factor (NGF) levels occurred (see Barth, E.-M., Korsching, S., and Thoenen, H. (1984) J. Cell Biol. 99, 839-843). We have now determined the levels of mRNANGF in rat irides as a function of time in culture as well. After an initial lag period of 2 h, mRNANGF levels were transiently increased, so that after 12 h, they had increased 35-fold with respect to zero time. In contrast, poly(A)+ RNA levels dropped to 55% of the zero time values within 5 h, recovered to 85% after 24 h, and remained constant until the end of the observation period. Total ribosomal RNA was found to remain constant, indicating that there was no nonspecific decline of overall metabolic function. Actinomycin D prevented the increase in mRNANGF without reducing the basic mRNANGF levels over a 5-h time period, indicating that the enhanced synthesis of NGF in the rat iris in culture is primarily mediated by an augmented production of mRNANGF. The increases of mRNANGF, cellular NGF, and NGF released into the medium were found to be strictly sequential. Monensin selectively abolished the increased production of mature NGF (see Barth et al.) but not of mRNANGF, suggesting that the processing of NGF precursor is prevented.

In previous experiments, it has been demonstrated that the density of sympathetic innervation was positively correlated with the levels of NGF' and its mRNA in peripheral sympathetic target organs (1,2,17). In contrast, the high NGF content of sympathetic ganglia was not accompanied by equivalent levels of mRNANGF, suggesting an NGF accumulation in sympathetic ganglia resulting from retrograde axonal transport rather than from local synthesis. This interpretation was further supported by the observation that blockade of retrograde axonal transport by 6-hydroxydopamine (selective destruction of sympathetic nerve terminals) and colchicine (disassembly of microtubules) resulted in a rapid NGF decrease in sympathetic ganglia (t, 4-5 h) and a rapid concurrent NGF increase in the corresponding peripheral target organs (3). These observations raised the question whether the NGF increase in the target organs was exclusively due to an elimination of the efficient removal of NGF by retrograde axonal transport or whether there was also an enhancement of NGF ~ ~ * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

T o whom reprint requests should be addressed.
The abbreviation used is: NGF, nerve growth factor. synthesis. In the rat iris, levels of NGF were increased 10 days after sensory or sympathetic denervation (4). Subsequent tissue culture experiments with rat irides had demonstrated that after a lag period of 4-5 h, a rapid increase in NGF levels occurred which was prevented by blockade of protein (cycloheximide) and RNA (actinomycin D) synthesis (5). We have now investigated whether the enhanced NGF formation results from an enhanced synthesis or an enhanced processing of the NGF precursor, which is not recognized by antibodies directed against the end product (6). Therefore, we have developed a sensitive assay for the absolute quantification of mRNANGF using RNANGF probes complementary to endogenous mRNA. In addition, recoveries have been determined by adding a shortened RNANGF fragment with a sequence identical to that of endogenous mRNA as an internal standard. We show that the increase in NGF levels in the iris is preceded by a nearly equivalent increase in mRNANGF, suggesting that the increased NGF synthesis results mainly from an augmented formation of mRNA.

MATERIALS AND METHODS
Preparation of Cultured Iris-Wistar rats (150-200 g, both sexes) were killed by cervical dislocation, and the irides were dissected and kept in culture as described previously (5). The time periods between killing and start of culture were always less than 5 min. The irides were cultured in Dulbecco's modified Eagle's H21 medium with 10% rat serum at 37 "C. At the end of each incubation period, the irides were frozen and kept at -70 "C until further use.
Isolation of RNA and Blotting Procedures-Total RNA was prepared according to Melera and Rusch (7) using bentonite ( A 1 2 0 3 ) as a protein adsorbent and RNase inhibitor. The extracted RNA was treated with RNase-free DNase I (200 pglml) (Boehringer Mannheim) and proteinase K (200 pg/ml) (Boehringer Mannheim) essentially according to Tullis and Rubin (8). Electrophoresis of glyoxylated RNA was performed as described previously (9). Gels were run at 10 mA for 12 h. For the quantification of mRNANGF, the total RNA of four irides (19.2 f 1.4 pg, all values are expressed in terms of means f S.E.) was applied to the wells, whereas for the quantification of 18 S rRNA, only one-tenth of the total RNA from a single iris was loaded together with 1 pg of carrier t-RNA. Dot blots were prepared with the multifold apparatus (Schleicher & Schull) using 10 X SSC (1.5 M NaC1,0.15 M sodium citrate) and 0.1% sodium dodecyl sulfate as sample buffer. The amount of total RNA applied in the serial 2fold dilutions started with 10 pg.
Preparation of RNA Probes-The large PstI-cleaved fragment of a mouse cDNANGFinsert (0.92 kilobase) obtained from J. Scott (Medical Research Center, Harrow, England) (10) and a mouse 18 S rRNA-cDNA insert (1.95 kilobases) obtained from I. Grummt (University of Wurzburg, Federal Republic of Germany) (11) were recloned into plasmids pSP6 in both orientations to the SP6 promotor of SP6 RNA polymerase (12). The recombinant plasmid templates were used for in vitro run-off RNA synthesis, and the polarities of the asymmetric transcripts were either complementary (RNANGF+, 18 S rRNA+) to or identical with (RNANGF-) the corresponding cellular RNA. Labeled RNANGF+ and/or 18 S rRNA+ was used as a hybridization probe, whereas unlabeled RNANGP transcripts were used as standards (see "Results"). In the case of NGF-transcripts, special care was taken to ensure purity. Removal of DNA by DNase I (20 pg/ml) treatment was found to be complete by addition and subsequent degradation to mononucleotides of a radioactively labeled DNA standard. In subsequent experiments, ~~-~'P-labeled UTP was added after DNase treatment, and the removal of mononucleotides was monitored in the precipitations using 2 M ammonium acetate and 70% ethanol. The Aseo (1 unit = 40 pg tRNA) was subsequently read from the spectrum obtained by using a DU-8 spectrophotometer (Beckman Instruments). The in uitro synthesized RNA was glyoxylated, run on an agarose gel, and stained with acridine orange (30 pg/ml) (13). Single bands of the predicted size demonstrated the specificity of synthesis and the absence of template DNA. After addition of carrier tRNA (5 pg/ml), the in vitro synthesized RNA was stored in aliquots as ethanol precipitates at -70 "C. Labeled RNANGF+ and 18 S rRNA+ probes (specific activity 6 X 10' dpmlpg) were prepared as described previously (2).
Hybridization and washing conditions for RNA and oligo(dT)12-18 probes were as described previously (2) using 50% formamide except that hybridization temperatures for RNANCF+, 18 S rRNA+ and oligo(dT) were 65, 70, or 4 "C, respectively. The differences in the optimal hybridization temperatures were mainly determined by the differences in the length and G-C content of the probes.
Densitometric Quantification of mRNA-2-Fold serial dilutions of standard RNANGF-(0.92 kilobase) were applied to the 1.5% agarose gels in parallel with the test samples. After Northern blot hybridization, the areas obtained from the densitometric scanning of the RNANGF-bands were plotted as standard curves. In most cases (depending on the intensity of the signal), the densitometrically obtained data were cross-checked with those from liquid scintillation counting of dissolved nitrocellulose bands cut from the appropriate region of the filter. All mRNANGF values were corrected for the RNA recoveries.
The films (Fuji RX) were exposed to the Northern blots or dot blots for time periods that produced linear signals of the densitometric scanning a t the given intensity of the bands.

RESULTS
Quantification Procedure for mRNANGF-For the determination of the absolute quantities of mRNANcF/iris, we used in vitro synthesized unlabeled RNANGF as a calibration standard. The blotting and hybridization efficiencies were assumed to be identical for in vitro and in vivo synthesized RNANGF because of their similar size (920 uersus 1350 bases) and identical sequence.
The SP6 system enabled us to synthesize an RNA fragment of suitable length which was used for the estimation of the recovery of RNA from individual preparations. The SP6 NGF-template plasmid was cut with the restriction enzyme NcoI so that the template directed the synthesis of an RNANCF-fragment of about 510 bases. A known amount of this short NGF-fragment (50 pg, if not otherwise stated) was added to the irides during thawing. The recovery of the RNA during the preparation was estimated from a calibration standard of this short fragment and was also included in the gels in separate lanes (data not shown). In Fig. 1 RNA decreased to levels between 55 and 60% of the zero time values (see below); and therefore, the mRNANCF increase calculated in terms of poly(A)+ RNA would be about 60-fold.
Time Course of 18 S rRNA and Total Poly(A)+ RNA-In contrast to the marked changes in the mRNANCF levels, those of 18 S rRNA, which were determined as a measure for total cellular RNA, remained essentially unchanged over the whole time period investigated (Fig. 2). However, the poly(A)+ RNA content in total RNA decreased to 55% of the zero time values within the first 5 h (Table I)  Comparison between the Time Courses of mRNANcF, Cellular NGF, and NGF Released into the Culture Medium-As reported earlier, the NGF release into the culture medium was preceded by a corresponding increase in NGF tissue levels (5). As shown in Fig. 3, the increase in tissue NGF was preceded by an increase in mRNANGF. The half-maximal level of mRNANGF was reached after 4 h, that of tissue NGF after 9 h, and that of NGF released into the culture medium after 12 h (5).

Effect of Monensin and Actinomycin D on mRNANGF-
Previous experiments have demonstrated that the carboxysodium ionophore monensin, which blocks the transfer of peptides through the Golgi apparatus (14), prevents both the  increase in tissue NGF and NGF released into the culture medium (5). We therefore investigated the effect of monensin on the levels of mRNANGF. In contrast to the production and release of NGF, the increase in mRNANGF was not affected by monensin using parallel cultures for the measurements of NGF and its mRNA. Actinomycin D, which has previously been shown to block the NGF increase of rat irides in culture (5), also prevented the increase in mRNANGF (Table 11). It is noteworthy that actinomycin D prevented the rapid increase but did not reduce the basic levels of mRNANGF.

DISCUSSION
The results of this study show that in rat irides in culture, after a lag period of 2 h, the mRNANGF increased rapidly to a TABLE I1

Effects of monensin and actinomycin D on the levels of mRNANGF
After dissection, the irides were put into cultures (for 5 or 24 h) containing the drugs in the concentrations indicated. The total RNA was extracted, and the densitometric values obtained from the Northern blot analysis are given in relative units (1 unit = 0.65 pg of mRNANGF/iris). maximal 35-fold level at 12 h. This rapid increase was followed by a gradual decline to an 11-fold level at 24 h and thereafter remained constant at a 2-3-fold elevated level between 48 and 72 h. In contrast to the rapidly changing mRNANGF levels, the 18 S rRNA remained constant over the whole observation period of 72 h. The quantitative determination of 18 S rRNA was used as a measure for total tissue RNA (28 S and 18 S rRNA are derived from a common 45 S precursor RNA (15) and account for more than 90% of total cellular RNA). The fact that the ribosomal RNA remained constant over a 48-h cultivation period is a further indication that an impaired overall function of the iris in culture is not responsible for the observed changes in mRNANGF. This is in agreement with previous experiments in which no changes in the incorporation of [36S]methionine into proteins over similar cultivation periods had been observed (5). The drop of poly(A)+ RNA to 55% of zero time values within the first 5 h therefore seems unlikely to reflect a general deterioration of the iris preparation, although the mechanism of such an apparent decrease of poly(A)+ RNA remains to be determined. Using denervated rat adrenals, a similar transient drop in total poly(A)+ RNA has been previously observed (16). Thus, in cultured iris, the decrease in total poly(A)+ RNA serves to emphasize the specificity of the increase of mRNANCF. In terms of absolute values/iris, mRNANGF levels underwent a maximal increase of 35-fold after 12 h. Relative to poly(A)+ RNA, the increase is 60-fold after 12 h and approaches the 200-fold increase in NGF protein shown in previous experiments (5). Thus, a minor post-transcriptional contribution to NGF synthesis, e.g. by stabilization of the NGF protein, cannot be excluded.
The fact that actinomycin D abolished the rapid increase in the first 5 h, without reducing the mRNANGF level below the base line, indicates that the increase in mRNANGF levels predominantly results from an enhanced synthesis of mRNANCF rather than from a stabilization of pre-existing mRNANGF. Recently, Shelton and Reichardt (17) reported an initial experiment of changes in mRNANGF levels in rat irides in culture. Both the absolute values and the increases in culture are different from the present results in that the zero time mRNANGF values, are 13-fold lower and the maximal increase after 5-10 h in culture is only 7-fold. The lower absolute zero time values may be explained by differences in the RNA extraction procedure used. Since Shelton and Reichardt did not measure recovery, selective loss of mRNANGF is also possible. Their culture conditions were very similar to ours with the exception of the use of fetal calf serum in place of rat serum. In previous experiments, however, the differences in NGF tissue levels and NGF released into the culture medium were very few when cultures supplied with fetal calf serum were compared with those containing rat serum (5).
In previous studies (5), we have demonstrated that monensin prevented the increased production and release of NGF in rat irides in culture. In the present study, we have demonstrated that the changes in mRNANGF levels were not affected by monensin. This observation is in agreement with the previous interpretation that monensin blocks the transit of the NGF precursor through the Golgi apparatus and with that the processing to mature NGF (5). This dissociation of changes in mRNANGF from those of formation of the end product is in contrast to the observations made for tropoelastin (18), where monensin prevents not only the processing of this molecule, but also the enhanced formation of its mRNA. The mechanism(s) responsible for the changes in mRNANGF can at the moment be only a matter of speculation. The rapidly increasing levels of mRNA reaching a maximum after 12 h are suggestive of a mechanism resulting from the release of constituents stored in nerve terminals. These nerve terminals are disconnected from their cell bodies by the culture procedure and therefore degenerate. The degeneration process starts more rapidly the shorter the peripheral nerve stump (19) and therefore begins virtually immediately in the cultured iris system. At the present time, experimental efforts are in progress aimed at the identification of the substance(s) responsible for the rapid increase in mRNANGF. Interestingly, this increase is prevented by the addition of cycloheximide (5 pg/ml) to the culture medium,' suggesting that a protein synthesis-dependent mechanism is involved.
The stably elevated mRNANGF levels between 48 and 72 h could suggest that under physiological conditions, the NGF production in the target cells is repressed by the presence of the innervating neurons. Such an assumption is in agreement with preliminary experiments in the sciatic nerve where we have observed an at least 10-fold increase within 6-h after nerve transsection in segments distal to the cut, which contain the degenerating peripheral 8x011s.~ Thus, the peripheral part of the sciatic could resemble the cultured iris, where all the innervating nerve terminals degenerate. Experiments are in progress aiming at the identification of the cells involved in the synthesis of NGF by in situ hybridization.