117 GATA 9(4): 117-123, 1992

Semiquantitative Analysis of X-Linked Gene Expression During Spermatogenesis in the Mouse: Ethidium-Bromide Staining of RT-PCR Products J O H N R. M c C A R R E Y , D O N A L D D. D I L W O R T H , and R. M A R K S H A R P

We have used analysis of ethidium-bromide-stained reverse transcriptase-polymerase chain reaction (RT-PCR) products to assess the effects of X-chromosome inactivation during spermatogenesis in the mouse. RT-PCR was performed on total RNA from eight different spermatogenic cell types, including premeiotic spermatogonia, meiotic spermatocytes, and postmeiotic spermatids, to detect transcripts from five different X-linked structural genes (Pgk1, Zfx, Pdha-1, Hprt, and Phka) and two autosomal genes (Pgk-2 and 13-actin). Relative intensities of ethidium-bromide-stained RT-PCR products representing transcripts from each gene in each cell type were analyzed by densitometry using the Image program (version 1.4, NIH), and normalized against 13-actin values. These results suggest a coordinate inactivation of the X-linked loci at the onset of meiosis, followed by variable rates of decline of corresponding transcript levels reflecting differential mRNA stabilities and~or leaky expression after inactivation. Technically, these results indicate that analysis of ethidiumbromide-stained RT-PCR products can be used to provide a "semiquantitative'" indication of relative levels of specific transcripts in a developing cell lineage without using radioactive probes to quantitate these products.

Introduction Mammalian spermatogenesis represents a dynamic system of cellular development and differentiation. Significant changes in gene expression have been documented in this system by using a variety of techniques, including direct analysis of protein products and/or analysis of mRNAs by Northern blot, in situ hybridization, or the reverse transcriptase-polymerFrom the Department of Genetics, SouthwestFoundation for Biomedical Research, San Antonio, Texas, USA. Address correspondence to Dr. J.R. McCarrey, Department of Genetics, SouthwestFoundationfor BiomedicalResearch, PO Box 28147, San Antonio, TX 78228, USA. Received 2 June 1992; revised and accepted 2 September 1992.

ase chain reaction (RT-PCR) (reviewed in Hecht I1 ]). Notable among these changes is the cessation of expression of X-linked structural genes that has been ascribed to general inactivation of the single X chromosome in XY spermatogenic cells [2, 3]. However, although the occurrence of X-chromosome inactivation (XCI) during mammalian spermatogenesis has come to be accepted as dogma, the number of Xlinked genes that have been directly investigated to document this phenomenon is quite small. Questions remain about whether the exemplary genes studied to date are truly representative of all or most X-linked structural genes, and whether the extent of XCI in spermatogenic cells is similar to that in XX somatic cells. Questions also remain about the exact timing of XCI during spermatogenesis, and how coordinate this inactivation is among affected X-linked loci. In using RT-PCR to analyze the expression of one X-linked structural gene, Pgk-1, during spermatogenesis in the mouse [4], we noted that the intensity of ethidium-bromide staining of the RT-PCR products provided an accurate indication of relative levels of transcript, which was subsequently confirmed by hybridization with a radioactively labeled probe. This led us to test available densitometry programs, including Image (written by Dr. Wayne Rasband, National Institutes of Health, Bethesda, MD) modified from the description by Correa-Rotter et al. [5] to quantitate systematically the intensity of ethidiumbromide-stained RT-PCR products. Here we report that ethidium-bromide staining can be used over a limited informative range to provide a "semiquantitative" indication of relative levels of specific transcripts in a developing cell lineage. We have used this approach to analyze transcript levels encoded by five different X-linked structural genes during spermatogenesis. Our results suggest that these loci become inactivated in a coordinate fashion in spermatocytes at about the time these cells enter meiosis, and that the extent of XCI in spermatogenic cells appears similar to that in somatic cells.

Materials and Methods I s o l a t i o n o f S p e r m a t o g e n i c Cells a n d Preparation of RNA

Highly enriched populations of specific spermatogenic cell types were isolated using Sta Put gradients as described by Bellve et al. [6] and Romrell et al. [7]. Primitive type-A spermatogonia (purity 1>85%) and somatic Sertoli cells (purity >90%) were isolated from prepuberal mice at 6 days of age; type-A plus type-B spermatogonia (combined purity/>85%) and

© 1992ElsevierSciencePublishingCo., Inc., 655 Avenueof the Americas, New York, NY 10010 1050-3862/92/$5.00

118 GATA 9(4): 117-123, 1992

Sertoli cells (purity >90%) were isolated from prepuberal mice at 8 days of age; preleptotene, leptotene plus zygotene, and early pachytene spermatocytes (purities of each ~>85%) were isolated from puberal mice at 17-18 days of age; and late pachytene spermatocytes, round spermatids, and residual bodies (purities of each >90%) were isolated from adult male mice. Purities of the desired cell type in each population were determined by microscopic analysis of cellular morphology. Total RNA was prepared by lysing cells in the presence of guanidinium isothiocyanate and pelleting RNA through 5.7 M cesium chloride as described by Chirgwin et al. [8].

RT-PCR Prior to performing RT-PCR, contaminating genomic DNA was removed from RNA samples by digestion with DNase I as described [9]. RT-PCR was carried out essentially as described [9] using a kit from Perkin Elmer: 3.5 ~g of total RNA was reverse transcribed using 2.5 IxM random hexamer primers, and 350 U reverse transcriptase in reverse transcriptase (RT) buffer (5 mM MgC12 plus 1 × PCR buffer [50 mM KC1 and 10 mM Tris-HC1, pH 8.3]) in a total volume of 140 ILl, at 42°C for 15 min. The products of this reaction were then divided into 35 × 4-1xl aliquots, each representing cDNA derived from 100 ng total RNA. PCR to detect specific transcript sequences was performed on individual aliquots by using 125 ng each of upstream and downstream oligonucleotide primers for each gene (Table l) in a 100-~l reaction containing 2 mM MgC12, 1

J.R. McCarrey et al.

× PCR buffer, and 2.5 AmpliTaq polymerase. Up to 34 cycles of PCR were performed on each sample (each cycle = 94°C, 1 min; 54°C, 1 min; and 72°C, 1 min, with an extension of 5 s/cycle on the 72°C segment). Aliquots (30 txl) of RT-PCR products were removed from each reaction at three different intervals, each differing by three cycles in extent of PCR. The appropriate number of cycles was empirically determined for each gene to provide values that fell into both the linear range of PCR for each reaction and the linear range of detection by ethidium bromide (see below). The [3-actin and Pgk-2 reactions were sampled after 22, 25, and 28 cycles of PCR; the Pgk1 reaction was sampled after 25, 28, and 31 cycles; and the Zfx, Hprt, Phka, and Pdha-1 reactions were sampled after 28, 31, and 34 cycles. In each case, the products from the middle number of PCR cycles run for each gene on RNA from each cell type were then electrophoresed as a set in 1.5% agarose containing 0.2 p.g/ml ethidium bromide. Standards of RT-PCR products derived from 25 cycles of RT-PCR to detect a fl-actin transcript in l0 ng and 100 ng total RNA from male liver were also run on each gel. For determination of the linear range of quantitation based on ethidium-bromide staining, RT-PCR reactions were run on 0, 1, 5, 10, 25, 50, 100, 200, 300,400, 500, and 1000 ng each of total RNA from male liver to detect either a [3-actin or Pgk-1 transcript. Each RT reaction was run as described above except that each was carded out in a total volume of 20 p.1, followed by PCR in a total volume of 100 Ixl, of which l0 p.1 of product was electrophoresed

Table 1. Primers Used for RT-PCR

Gene

Primers (5'-3') upstream/downstream

Product size (bp)

Bases spanned

Reference

~-actin

GCGGACTGTTACTGAGCTGCGT GAAGCAATGCTGTCACCTrCCC

453

1209-1661

19

Pdha-i

CAAGTGTTGAAGAATTAAAG TTC AAGCCTITFGTTGTCTG

287

991-1278

18

Pgk-1

AAGCGCACGTCTGCCGCGCTGTTCT GTTGGCTCCATTGTCCAAGCAGAAT

238

19-256

20

Phka

AATTCACTACTGCCCAGGGCITCAAC GCTTCAGCTCAGCTGGGTTATAGTAT

233

1-232

21

Hprt

CGAGGAGTCCTGTrGATGTTGC CTGGCCTATAGGCTCATAGTGC

172

687-860

22

Zfx

CAGTTGTCATCCAGGATGTC TCGTTGTCCATAGTCAGTCC

504

212-716

17

Pgk-2

AGGAGATACTGCTACTTGCTGCGCC GATGATGACAGAATTAAGACTrGCT

300

1119-1418

16

© 1992 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010

119 GATA 9(4): 117-123, 1992

X-Linked Gene Expression in Spermatogenesis

1

in 1.5% agarose and analyzed by ethidium-bromide staining. ngRNA:

Analysis of RT-PCR Data

01

5

12 05

1 2 3 4 5 0 5 0 0 0 0 00 0 0 0 0 0 00

Actin -

Each gel was photographed onto positive/negative Polaroid Type 55 film under ultraviolet illumination. The negative image was then captured by using a digital camera, which consisted of a Philips 56470series CCD imaging module (Philips Components, Slatersville, RI) and a video Cosmicar automatic iris ES series lens (Asahi Precision, Saitame-ken, Japan) capable of resolving 256 levels of light at each of 610 (horizontal) x 490 (vertical) pixels. These data were analyzed using Image software (Version 1.44), a public-domain program that is useful for capturing, manipulating, and analyzing digital images. Image requires an Apple Macintosh with at least 2 megabytes of memory, a monitor with the ability to display 256 colors or shades of gray, and a floating-point coprocessor or the PseudoFPU Init, which can emulate a missing coprocessor. A digital image of each EthBr gel was captured from light transmitted through the photographic negative that was placed over a diffused light source. A densitometric plot for each electrophoretic lane was generated from the digital images by using a modified version of a procedure outlined in the manual that comes with the Image software. Since the amplitude of the density peaks are proportional to the size of the plot windows inside Image, the plot windows were fixed to the same size so that the areas measured underneath the density peaks would accurately reflect the true density. The screen pixel is the unit in which the areas are expressed by Image. Quantitation of samples was based on an external reference standard curve derived from the density peak areas from a single gel with lanes containing PCR products from 0, 1, 5, 10, 25, 50, 100, and 200 ng of fl-actin RNA. Quartic equation coefficients were derived by polynomial regression of the density peak areas from the reference fl-actin gel by using a four-parameter model. These coefficients were then used to estimate the amount of ~-actin RNA equivalents represented by the specific density peak areas for each cell-type-gene combination. Internal standards for each gel consisted of two lanes with PCR products from 10 and 100 ng of ~-actin RNA, respectively. A gel-specific correction factor was calculated from these two lanes by averaging the ratio of the estimate of fl-actin RNA equivalents to the amount actually loaded into each lane. Specifically, gel-specific correction factor = ([El0 + [El00 + I00]) + 2

10]

A

Pgkol -

B Fisure 1. Calibrationof ethidium-bromide-stainedRT-PCR signals. RT-PCR analysisof [3-actin(A) and Pkg-1 (B) transcripts in varyingamountsof total RNA isolatedfrom male mouse liver tissue. The [3-actinamplificationproduct is 453 bp, and the Pgk1 amplificationproduct is 238 bp. Amountsof template RNA in nanograms are indicatedabove each lane.

where E l 0 and El00 represent estimates of fl-actin RNA equivalents for the lanes containing PCR products from 10 and 100 ng fl-actin RNA. Initial estimates of ~-actin RNA equivalents provided by the quartic equation coefficients were multiplied by the gel-specific correction factor to obtain the final estimate. The average value of this ratio was 1.262 ( ± 0 . 1 5 SD).

Results

Quantitation of Ethidium-Bromide-Stained RT-PCR Products To determine the extent of the linear range of detection of RT-PCR products by ethidium-bromide staining, and the ability to infer relative input amounts of specific mRNAs from output values of corresponding ethidium-bromide-stained amplification products, we carried out RT-PCR on a series of increasing input amounts of RNA to detect [3-actin or Pgk-1 mRNAs (Figures 1 and 2). Increasing amounts of input RNA yielded correspondingly increasing amounts of RT-PCR amplification product representing ~-actin (Figure 1A) or Pgk-1 (Figure 1B) transcripts. The informative range over which differences in input amounts of RNA produced detectable differences in intensities of ethidium-bromidestained products appeared to span 1-200 ng of input RNA (Figure 2). Above this level, the response plateaued and was no longer informative. Values determined for Pgk-1 transcript levels were quite similar to those for ~-actin in the range of 1-100 ng of input RNA (Figure 2). Thus, we ran all other RT-

© 1992 ElsevierSciencePublishingCo., Inc., 655 Avenueof the Americas,New York, NY 10010

120 G A T A 9(4): 117-123, 1992

J.R. McCarrey et al.

1600 1400 1200 < 1000 C

[ ----~-

u~ 800 ¢

Actin

600

that more accurately reflected known differences in input values of template RNA than did the unconverted values determined directly from densitometry of the ethidium-bromide-stained gels. Thus, we routinely used this conversion on RT-PCR results for all other genes analyzed.

Analysis of X-Linked Gene Transcript Levels in Spermatogenic Cells

t~ 400 W ._~ ~. 200 O 0 0

500 Input RNA (ng)

1000

Figure 2. Dilution curves of [3-actin and Pkg-1 calibration series. Data shown in Figure l were quantified as described under Materials and Methods and plotted to show output values (expressed in pixels) versus input values (nanograms of template RNA).

PCR reactions at appropriate cycle numbers to yield amounts of product that fell in this "linear range." Differences in input amounts of template RNA were reflected in differences in the intensities of ethidium-bromide staining of the amplification products over the informative range described above. However, the relationship between differences in increments of input amounts of template RNA and that of corresponding output amplification product was typically less than 1:1. Thus, we tested the use of the dilution curve produced by the analysis of/3actin transcripts in the titration experiment shown in Figures 1 and 2 to convert values deduced from ethidium-bromide-stained products of the Pgk-1 titration experiment back to corresponding input amounts (Table 2). This conversion consistently yielded values

We used amplification primers (Table 1) specific for each of five X-linked genes (Pdha-1, Phka, Pgk-1, Zfx, and Hprt) and two autosomal genes (/3-actin and Pgk-2) to detect corresponding transcripts in RNAs from somatic Sertoli cells, premeiotic spermatogonia (primitive type A and type A plus type B), meiotic spermatocytes (preleptotene, leptotene plus zygotene, early pachytene, and late pachytene), postmeiotic (round) spermatids, and residual cytoplasmic bodies (Figure 3). Images of the ethidium-bromidestained amplification products representing the Xlinked gene transcripts were quantified by densitometry and normalized against the/3-actin control values to yield the values represented in Figure 4. Four of the five X-linked gene transcripts showed a notable decline relative to levels of/3-actin transcript beginning in meiotic primary spermatocytes (preleptotene-pachytene) and continuing in postmeiotic spermatids and residual bodies (Figure 4). Levels of transcripts from the fifth gene, Hprt, did not appear to decline below those of/3-actin during spermatogenesis. However, there was an initial decline in levels of this transcript in early spermatocytes, followed by a transient increase in late pachytene spermatocytes and then a return to levels similar to those of/3-actin transcripts in spermatids and residual bodies.

Table 2. R T - P C R calibration Raw data

Relative increase

Converted data

Actin RNA (ng)

Actin

Pgk-1

0

0

0

1

0

0

5 10 25 50 100 200

54 74 279 476 683 1076

126 294 419 486 683 922

Actin

Pgk-I

Actual

0.0 0.0 6.1 8.7 25.3 48.3 94.3 164.3

0.0 0.0 14.1 26.5 39.6 50.0 94.3 153.7

. . . 2.0 2.5 2.0 2.0 2.0

Pgk-1

Raw . . .

Conv. . . .

1.4 3.8 1.7 1.4 1.6

© 1992 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010

. . . 1.4 2.9 1.9 2.0 1.7

Raw

Conv.

2.3 1.4 1.2 1.4 1.3

1.9 1.5 1.3 1.9 1.6

. . .

121 X-Linked Gene Expression in Spermatogenesis

6dSd6d8dP S S G G L

L Z

E P

L P

R T

R N N BT R

GATA 9(4): 117-123, 1992

10100 A A

~ Acfin

1.6

r-

-

~c - i

1.2

--~ ~

0.4

~

o.o

Pgk-1

o

D

Hprt

-

B

C

Pgk-1

~ - Zfx "0

-

Pdha-1

Phka

A

Phka

Actin

--~---*--

o'~

Pdha.1.

~

D.

-J

W

~

n-

Cell Type Figure 4. X-linked gene transcript levels in spermatogenic cells. RT-PCT data from gels shown in Figure 3 were quantified as described in the text and plotted as percent of the corresponding [3-actin transcript level in each cell type. Specific gene transcripts measured are identified by gene symbol in the inset. Cell types: 6-day Sertoli cells (6dS), primitive type-A spermatogonia (6dG), type-A plus type-B spermatogonia (8dG), preleptotene spermatocytes (PL), leptotene plus zygotene spermatocytes (LZ), early pachytene spermatocytes (EP), late pachytene spermatocytes (LP), round spermatids (RT), and residual cytoplasmic bodies (RB).

Zfx-

E

Hpn.

Pgk-2

ity of our approach to detect differences in relative levels of specific transcripts in spermatogenic cell types.

Discussion

-

G Figure 3. RT-PCR analysis of specific genes in spermatogenic cells. Total RNA from two testicular somatic cell types, 6-day (6dS) and 8-day (8dS) Sertoli cells, and eight spermatogenic cell types, primitive type-A spermatogonia (6dG), type-A plus typeB spermatogonia (8dG), preleptotene (PL), leptotene plus zygotene (LZ), early pachytene (EP) and late pachytene (LP) spermatocytes, round spermatids (RT), and residual cytoplasmic bodies (RB), was analyzed by RT-PCR for transcripts from two autosomal genes, B-actin (Actin) and Pgk-2, and five X-linked genes, Pkha-1, Phka, Pgk-1, Zfx, and Hprt. Sizes of corresponding amplification products for each gene are described in Table 1. No-template (NT) and no reverse transcriptase (NR) controls were run for each gene. Amplification products from RT-PCR for ~-actin in 10 ng (10 A) and 100 ng (100 A) were also run with each set.

As previously reported [3], Pgk-2 transcripts were not detected during spermatogenesis until the onset of meiosis, and then showed an increase in postmeiotic spermatogenic cell types (Figure 3). This confirmed further the integrity of the RNAs from postmeiotic spermatogenic cell types, and the valid-

We have shown that detection and analysis of ethidium-bromide-stained RT-PCR products can be used to follow changes in levels of specific gene transcripts during the development and differentiation of a particular cell lineage. Our approach was not designed to determine exact amounts of transcript molecules per cell, but does provide a reasonable estimate of relative levels of transcripts in different cell types. Hence we have termed this analysis "semiquantitative." We believe that for many purposes this approach will provide sufficient information regarding relative levels of specific transcripts in different cell types, without the need for blotting the gels and hybridizing with a radioactively labeled probe, and without the need for constructing a specific internal standard for each gene to be assayed. Several sources of experimental error could potentially compromise the accuracy of any PCR-based method to detect specific gene transcripts, especially considering that any minute initial error will be amplified by this process. The data in Table 2 demonstrate the extent of the reliability of our approach. The differences among relative values for each Pgk1 transcript amplification product measured directly by densitometry were typically less than the actual

© 1992 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010

122

GATA 9(4): 117-123, 1992 differences in input amounts of RNA template. By calibrating these values against the standard curve generated from the [3-actintranscript results, we were able to estimate relative differences that more accurately reflected the actual input differences. The most accurate estimates were obtained in the range corresponding to the fl-actin and Pgk-1 transcript signals detected in 50-100 ng of input RNA; thus most of our subsequent measurements of transcript levels in spermatogenic cell RNAs were made in this range. Further verification of this approach was provided by our measurements of specific gene transcripts in spermatogenic cells, which showed expected changes in relative transcript levels during spermatogenesis. Our results for the two Pgk transcripts (Pgk-1 and Pgk-2) showed a similar pattern of expression (although the actual levels were somewhat different) to those deduced by Singer-Sam et al. [10], who used RT-PCR with gene-specific internal standards and detection by hybridization with a radioactively labeled probe. The levels of transcript that we detected in this experiment encoded by the Pgk-1 gene were similar to those we detected by probing a blot of a separate RT-PCR experiment with a radioactively labeled oligonucleotide probe [4]. It is difficult to pinpoint precisely the exact time of XCI in spermatogenic cells on the basis of these results, due to the variable persistence of cytoplasmic RNAs depending on the relative stability of each transcript. However, these results are consistent with the idea that XCI occurs at about the time that spermatogenic cells enter meiosis, during the transition between spermatogonia and spermatocytes. We observed a slight decrease in transcript levels from several of the X-linked genes in spermatogonia between 6 and 8 days after birth (primitive type-A spermatogonia versus type-A plus type-B spermatogonia). The most consistent decline occurred in primary spermatocytes, however, especially in preleptotene, leptotene plus zygotene, and early pachytene spermatocytes (Figure 4). The variability in rates of decline of each X-linked gene transcript after XCI is most likely due to differential stabilities of each mRNA. Thus, it would appear that Pdha-I transcripts are the most labile of those assayed here, followed in order of increasing stability by Phka, Pgk-1, Zfx, and Hprt transcripts. Transient increases in transcript levels during spermatogenesis (for example, Hprt and Zfx) could be the result of leaky expression of these genes after XCI. We [4] and others [11] have previously noted that, unlike XCI in female somatic cells, XCI in

J.R. McCarrey et al.

spermatogenic cells does not appear to be followed by an increase in methylation of CpG dinucleotides. Thus, we have suggested that the transcriptional repression induced by XCI in spermatogenic cells is not as stringent as that in somatic cells [4]. The absence of stable repression could result in leaky expression, leading to the transient increases in transcript levels observed here. In the case of Hprt, this suggestion is supported by the observation of a relative increase in corresponding enzyme activity in spermatids as reported by Allsop and Watts [12]. Four of the X-linked loci investigated here span a region from 23 to 48 centi-Morgans from the centromere on the mouse X chromosome, while the position of the fifth X-linked locus, Pdha-1, is undetermined [13]. This suggests that this entire region is subject to XCI, and that the extent of the X chromosome affected by XCI in spermatogenic cells is similar to that in somatic cells. The function of XCI in spermatogenic cells remains an enigma at this point, but the effect of this process poses an interesting problem to differentiating spermatogenic cells. Specifically, these cells must cope with the loss of transcription of X-linked genes encoding potentially critical products. We previously described one process by which the loss of Pgk-1 gene product has apparently been compensated for by expression of the autosomal Pkg-2 gene [3]. We provided evidence that the Pgk-2 gene evolved as a functional retroposon of the Pgk-1 locus, and suggested that the subsequent selective advantage conferred by specific expression of this autosomal gene in spermatogenic cells contributed to its conservation I14]. Of course, compensation for the loss of ongoing transcription of a particular X-linked gene would not be necessary if the product of that gene is sufficiently stable to maintain necessary levels throughout spermatogenesis [15]. In this regard, it is interesting that transcripts encoded by the Hprt gene, for which no autosomal counterpart exists, appear to be the most stable among the five X-linked gene transcripts that we studied, while three of the four less stable transcripts are encoded by genes for which autosomal counterparts have been described (Pgk-1 [16], Zfx [17], and Pdha-I [18]). Interestingly, it has been suggested that each of the autosomal counterparts for these three genes evolved as a retroposon from its X-linked counterpart, and it has been shown that each is specifically expressed in the late stages of spermatogenesis. This compensatory expression can be visualized by comparing the patterns of the Pgk-1 and Pgk-2 transcripts in Figure 3 (D versus G). Sim-

© 1992 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010

123

GATA 9(4): 117-123, 1992

X-Linked Gene Expression in Spermatogenesis

ilarly, the decrease in Zfx transcripts is compensated for by increasing expression o f the Zfa gene, which is the likely source o f the smaller amplification product visible in Figure 3E. Some X-linked genes m a y encode products that are not required in late spermatogenic cells or spermatozoa, so that their inactivation m a y not need to be compensated for during spermatogenesis. One possible example is the Phka gene, which encodes phosphorylase kinase. Since sperm use fructose as their primary energy source [19], it is not apparent that they need an e n z y m e involved in glycogen metabolism. This may explain w h y no autosomal counterpart has evolved for this gene despite the fact that it encodes a transcript that is relatively unstable (Figure 4). Final confirmation o f the exact time o f X C I in spermatogenic cells and the subsequent impact on transcript levels will require the simultaneous analysis of nuclear run-off and steady-state cytoplasmic transcripts in these cells. This experiment is now under way in our laboratory. However, here we have used ethidium-bromide staining o f R T - P C R products to show that, in most cases, X C I leads to a decline in levels o f X-linked transcripts in meiotic and postmeiotic spermatogenic cells. This work was supported by the National Science Foundation (DMB 92458). J.R.M. is the recipient of a research career development award from the National Institutes of Health (HD 00829).

References 1. Hecht NB: In Cell and Molecular Biology of the Testis. (Desjardins C, Ewing L, eds: The Testis, vol 5.) Oxford, Oxford University Press, 1992 (in press)

2. Lifschytz E, Lindsley DL: Proc Natl Acad Sci USA 69:182186, 1972 3. McCarrey JR, Thomas K: Nature 326:501-505, 1987 4. McCarrey JR, Berg WB, Paragioudakis SJ, Zhang PL, Dilworth DD, Arnold BL, Rossi JJ: Dev Biol 154:160-168, 1992 5. Correa-Rotter R, Mariash CN, Rosenberg ME: Biofeedback 12:154-158, 1992 6. Bellve AR, Cavicchia JC, Millette CF, O'Brien DA, Bhatnager YM, Dym M: J Cell Biol 74:68-85, 1977 7. Romrell LJ, Bellve AR, Fawcett DW: Dev Biol 49:119-31, 1976 8. Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ: Biochemistry 18:5294-5299, 1979 9. Dilworth DD, McCarrey JR: PCR Methods Appl 1:279-282, 1992

10. Singer-Sam J, Robinson MO, Bellve AR, Simon MI, Riggs AD: Nucleic Acids Res 18:1255-1259, 1990 11. Venolia L, Cooper DW, O'Brien DA, Millette CF, Gartler SM: Chromosoma 90:185-189, 1984 12. Allsop J, Watts RWE: Differentiation 32:144-147, 1986 13. Davisson MT, Lalley PA, Peters J, Doolittle DP, Hillyard AL, Searle AG: Cytogenet Cell Genet 58:1152-1189, 1991 14. McCarrey JR: Nucleic Acids Res 18:949-955, 1990 15. Handel MA, Hunt PA, Kot MC, Park C, Shannon M: Ann NY Acad Sci 637:64-73, 1991 16. Boer PH, Adra CN, Lau YF, McBurney MW: Mol Cell Biol 7:3107-3112, 1987 17. Mardon G, Luoh SW, Simpson, EM, Gill G, Brown LG, Page DC: Mol Cell Biol 10:681-688 18. lannello RC, Dahl HHM: Biol Reprod 47:48-58, 1992 19. Tokunaga K, Taniguchi H, Yoda K, Shimizu M, Sakiyama S: Nucleic Acids Res 14:2829, 1986 20. Mori N, Singer-Sam J, Lee CY, Riggs AD: Oene 45:275280, 1986 21. Bender PK, Laller PA: Proc Natl Acad Sci USA 86:9,99610,000, 1989 22. Konecki DS, Brennand J, Fuscoe JC, Caskey CT, Chinault AC: Nucleic Acids Res 10:6763-6775, 1982

© 1992 Elsevier Science Publishing Co., Inc., 655 Avenue of the Anlericas, New York, NY 10010