We have adapted a non-radioactive technique to detect localized mRNAs in whole-mount Xenopus embryos. Synthetic antisense RNA transcribed in the presence of digoxygenin-UTP is used as a probe and is detected via an anti-digoxygenin antibody. We show that localized mRNAs can be detected from late gastrula to tadpole stages and that high as well as low abundance RNAs can be detected. The method was tested on muscle actin and α-globin RNAs, whose localization has previously been characterized. In addition, we used the method to determine the distribution of XA-1 RNA, an anterior ectoderm-specific RNA, which we show is expressed in the periphery of the cement gland as well as in the region of the hatching gland. The sequence of an XA-1 cDNA predicts a protein rich in proline and histidine.

Three techniques have been used to analyze localized gene expression in Xenopus embryos. Since the embryos are large, microdissected tissues can be assayed biochemically for the presence of specific RNAs (Mohun et al. 1984); skilled dissection can yield a high degree of spatial resolution (Hopwood et al. 1989b) but cannot provide information at the level of single cells. Protein products can be localized by immunological methods, and analysis of embryos by whole-mount immunohistochemistry provides both high resolution and three-dimensional information (Dent et al. 1989; Hemmati-Brivanlou and Harland, 1989); however, after isolating a gene of interest it takes considerable time and effort to raise and purify specific antibodies. In situ hybridization is a powerful technique for examining the spatial expression of RNAs in embryos, but in situ hybridization to sectioned Xenopus embryos is laborious and not reproducible. The method has not been sufficiently sensitive to detect rare transcripts, such as those from many homeobox genes, so that only moderately abundant RNAs have been analyzed. Even then an exposure time of weeks (Weeks and Melton, 1987; Sato and Sargent, 1989) or even months (Ruiz i Altaba and Melton, 1989) is often necessary. Since in situ hybridization has been carried out on sectioned tissue, two-dimensional information must be reconstituted’into three dimensions; this must be done either by the imagination of the investigator or by computer (Wilkinson et al. 1987).

Recently a sensitive, non-radioactive in situ hybridization method has been developed for the localization of specific RNAs in whole-mount Drosophila embryos (Tautz and Pfeifle, 1989). We show that a modification of this technique can be successfully used to detect localized RNAs in embryos of the frog Xenopus laevis. In contrast to in situ hybridization of radiolabelled probes to tissue sections, the non-radioactive method is rapid, sensitive and allows staining of whole embryos.

As well as confirming that the method works with genes such as muscle-specific actin and α-globin whose transcripts have previously been localized, we have analyzed the expression of the XA-1 gene (Sive et al. 1989). From dissection experiments, XA-1 is known to be expressed in the cement gland and non-brain head ectoderm of embryos, but the cellular distribution of this transcript has not been determined.

The in situ hybridization procedure was adapted from Tautz and Pfeifle (1989) and Kintner and Melton (1987).

Embryos

Xenopus laevis were obtained from the Berkeley colony maintained by the laboratory of J. C. Gerhart. Ovulation of females and in vitro fertilization were carried out as described by Condie and Harland (1987). Developmental stages were determined according to Nieuwkoop and Faber (1967).

The vitelline membranes of dejellied embryos were loosened by treatment with proteinase K (5 μgml−1) for 5 to 10 min and were then manually removed. Digestion was monitored under the microscope and when the membranes were seen to lift from the surface the embryos were washed in 1/3 MR (lxMR=100mM NaCl, 1.8mM KC1, ImM MgCl2, 2 HIM CaCl2, 50μgml−1 gentamycin, buffered to pH6.9 with 5mM Hepes). Selected embryos were transferred to a 5 ml screw cap glass vial filled with distilled water. After the embryos settled, the water was removed and the vials were filled to the brim with MEMFA (0.1M Mops pH7.4, 2OIM EGTA, ImM MgSO4, 3.7% formaldehyde). Embryos were allowed to fix at room temperature for 1.5 to 2h on a Labquake rotator (Labindustries, Inc). MEMFA was made freshly from a stock of 10x salts and 37% formaldehyde. Surprisingly, the use of paraformaldehyde at this stage led to higher background staining. The fix was removed and replaced with methanol. After a few minutes of equilibration with the methanol, the embryos were stored at —20°C.

Probes

The linearized DNA templates shown in Table 1 were used for in vitro transcription reactions in the presence of digoxigenin-11-UTP (Boehringer Mannheim 1209 256) with either T3, T7 or SP6 RNA polymerase (as described by Melton et al. 1985). Linearized template DNA (2.5μg) was transcribed in 50μ1 of 40 mM Tris-HCl pH 7.9, 6mM MgCl2, 2mM spermidine-HCl, 10 mM DTT, 1 mM ATP, CT’P and GTP, 0.33 mM digoxygenin-11-UTP (BMB 1209 256), 0.66 mM UTP, with 10μCi of 32P-CTP (400 Ci mmol−1) and 90 units of the appropriate enzyme for 2 to 3h. After standard DNAsel treatment the reaction was diluted to 100μ1. After taking 1 μ1 to determine the total counts available, the remainder of the reaction was spun through a 1ml column of Sephadex equilibrated in 0.3 M sodium acetate pH5.5, 0.1% SDS. Following ethanol precipitation, the RNA pellet was resuspended in 50μl of 40 mM sodium bicarbonate, 60 mM sodium carbonate and incubated for 35-50 min at 60°C to generate 200-500 nucleotide RNA probes (Lynn et al. 1983). The RNA was again precipitated with ethanol and the final pellet was resuspended in 400 μl hybridization buffer; 50% formamide. 5xSSC, 500μgml Torula RNA (Calbiochem), 50μgml heparin and 0.1 % Tween 20. A second sample was then taken to determine the counts incorporated. The theoretical maximum yield (100% incorporation) is 66 μg.

Table 1.

Summary of plasmids used with the efficiency of synthesis and amounts used in in situ hybridization

Summary of plasmids used with the efficiency of synthesis and amounts used in in situ hybridization
Summary of plasmids used with the efficiency of synthesis and amounts used in in situ hybridization

In situ hybridization

Fixed embryos in methanol were gradually rehydrated with ME (90% methanol, 10% 0.5M EGTA) and PTw (lx PBS+0.1% Tween-20) by 5 min incubations in ME, 75% ME+25 % PTw, 50 % ME+50 % PTw, 25 % ME+75 % PTw, 100% PTw. They were then washed three times, 5 min each, in PTw. Unless otherwise stated, in all steps of the protocol the vials were filled almost completely with liquid and rocked on a nutator (Clay Adams). The embryos were then incubated at room temperature in 10μgml−1 proteinase K (in PTw) on the nutator. The time of incubation must be adjusted for different batches of proteinase K but 15-30 min has given good results. Early stage embryos (prior to stage 20) are more sensitive to damage and should be monitored carefully.

Embryos subsequently swell but most stay intact through the procedure. At the end of the proteinase K treatment, embryos were washed twice, 5 min each, in PTw and were refixed for 20 min with 4% paraformaldehyde in lx PBS at room temperature. Paraformaldehyde yielded marginally better results than formaldehyde. The refix was followed by five washes, 5 min each in PTw. Since the embryos are somewhat delicate after protease treatment, the rocking steps are done by laying the vials horizontally on the nutator and filling the vials to reduce turbulence. Where solutions are more precious and volumes are smaller, the vials are placed vertically in a rack on the nutator.

Hybridization buffer (1 ml) was added to a near empty tube and the embryos allowed to settle and equilibrate for a few minutes. The hybridization buffer was changed and the embryos were prehybridized for 1 h at 50 °C. The prehybridization buffer was replaced with fresh hybridization solution containing the probe (5-10 μg ml−1) and embryos were incubated overnight at 50°C. At the end of the hybridization the probe was saved. We find that the probe can be recycled up to 3 times. Hybridization was also tested at 65 °C with no obvious improvement in background but with loss of morphology, particularly in later stage embryos where the endodermal mass enlarged in comparison to the other tissues.

The embryos were brought from hybridization buffer to 2xSSC gradually by 1.5 ml changes of the following solutions: 75% hyb+25% 2xSSC, 50% hyb+50% 2xSSC and 25% hyb+75 % 2xSSC each wash for 10min at 37°C in a shaking waterbath at between 0 to 60 revs min−1. This was followed by two washes 20 min in 2xSSC at 37°C. The non-hybridized excess RNA was removed by incubation of the embryos in a solution of 2xSSC containing 20μgml−1 RNase A and 10 unitsml−1 RNAse T, at 37°C for 30min. From 2xSSC the embryos were transferred to 0.2xSSC and washed twice in 0.2xSSC for 30min, at 55°C; and stepped back to PTw, 5-10min changes of 75% 0.2xSSC+25% PTw, 50% 0.2xSSC+50% PTw, 25% 0.2XSSC+75% PTw and 100% PTw.

RNA hybrids were detected by immunohistochemistry in a procedure similar to the one described previously (Hemmati-Brivanlou and Harland, 1989). The PTw was replaced with PBT (PBS+2mgml−1 BSA+0.1% Triton X-100), and the vials were rocked at room temperature for 15 min. The PBT was removed and replaced with 500 μl of fresh PBT+10% heat-inactivated goat serum (Gibco; the serum is heated at 56°C for 30min). This step saturates non-specific immuno-globulin-binding sites. Even though the commercially available digoxigenin antibody is raised in sheep and therefore the optimum blocking serum should be sheep serum, we found that goat serum worked adequately. Vials were rocked vertically at room temperature for 1h. This solution was replaced with 500 μl fresh PBT+10 % goat serum containing a 1/1000 dilution of the affinity-purified sheep anti-digoxigenin coupled to alkaline phosphatase antibody (BMB cat. no. 1093 274). Tubes were rocked vertically overnight at 4 °C. To remove excess antibody, the embryos were washed at least 3 times 1 h each with PBT at room temperature (filled up vials and horizontal rocking).

For the chromogenic reaction with alkaline phosphatase, embryos were washed 3 times, 5 min each at room temperature with 100mM Tris pH9.5, 50mM MgCl2, 100mM NaCl, 0.1 % Tween 20 and 1 mM Levamisol added freshly to inhibit endogenous alkaline phosphatase. The last wash was replaced with 1 ml of the same solution and 4.5 μl nitro blue tétrazolium (NBT, Sigma; 75 mg ml−1 in 70% dimethylformamide) and 3.5 μl 5-bromo-4-chloro-3-indoyl phosphate (BCIP, Sigma; 50 mg ml−1 in 100% dimethylformamide) were added directly to the tube. Embryos were rocked at room temperature (it is not critical to protect the reaction from light). The color reaction was visible from 5 min to 60 min after treatment. When satisfactory signal and background were observed, the solution was replaced with PBS. After a total of 2 washes with PBS (5min each at room temperature), the specimens were dehydrated for 3 min in methanol and mounted in 2:1 benzyl benzoate: benzyl alcohol (BB/B A) for observation. Although satisfactory for intense signals, such as the actin signal shown in Fig. 2A, this medium dissolves less intense stains. Therefore, a limitation to the sensitivity of the method is the solubility of the stain. The actin signal was photographed after 1 month in BB/BA whereas the globin signal faded in one week. For epidermal RNAs, the specimens can be mounted in aqueous media, but for deeper stains, such as those in the nervous system, the embryos must be cleared. If the background associated with a particular probe is not substantial, the NBT/BCIP reaction can be allowed to proceed for 6 h; under these conditions the blue stain remains highly localized and is sufficiently intense to remain insoluble. To circumvent the solubility problem, we have also tested the Vectastain II kit as suggested by Dent et al. (1989). This kit gives satisfactory results, though it is not as sensitive as NBT/BCIP and the contrast is inferior. HRP-conjugated secondary antibodies were also tested and are relatively insensitive; however, in cases where mRNA is abundant and NBT/BCIP stains too intensely, the HRP-conjugated anti-body is a useful alternative.

In general we have used albino embryos in these experiments. However, the blue stain contrasts well with pigment so that the procedure can be carried out on pigmented embryos (see Fig. 2). We have not attempted to stain embryos bleached with hydrogen peroxide (Dent et al. 1989), nor have we tested whether the blue stain is resistant to bleach after the staining is complete.

Northwestern Blot

Radiolabelled and dig-11 UTP labelled synthetic RNA was fractionated by electrophoresis in formaldehyde agarose gels and transferred to a nylon filter. The filters were exposed to X-ray film to estimate the quality and quantity of the synthetic RNA.

The relative efficiencies of incorporation of dig-UTP by SP6, T7 and T3 RNA polymerases were determined as follows: Sense transcription templates driven by the SP6, T7 and T3 promoters and containing the chloramphenicol acetyl transferase gene, (CAT), were prepared. The transcribed sequences were identical except for short polylinker regions. An aliquot of RNA produced from identical transcription reactions was quantitated by 32P-CTP incorporation, and 10 ng of each was run out on a formaldehyde agarose gel. An autoradiograph of the gel was scanned with a Hoefer densitometer and the peaks integrated to determine the relative quantities of RNA loaded per lane.

The incorporation of digoxigenin in each of the bands was estimated as follows: The filter was washed for 15 min in PBS and then washed in PBT for 30 min at room temperature. This was followed by an incubation in PBT+10% goat serum for one hour. The anti-digoxigenin antibody (BMB) was diluted 1:2000 in PBT+10% goat serum and incubated with the filter for 2h at room temperature. The filter was then washed in PBT for an hour with 4 changes. PBT was replaced by two changes of 100mM Tris pH9.5, 50 mM MgCl2, 100 mM NaCl and 0.1% Tween 20 for 15 min. 4.5 μl ml−1 of NBT and 3.5 μl ml−1 BCIP were added to the last wash and the purple color resulting from the chromogenic reaction was detected after 5 min. The wet filter was then scanned and the peaks integrated to quantitate each band; the relative efficiency of dig-UTP incorporation was then determined based on the masses of RNA present in each band. We have not attempted to determine the absolute number of digoxygenin groups per RNA molecule.

DNA cloning and sequencing

An apparently full-length cDNA clone (800 bp) of XA-1 was isolated from a stage 33 Xenopus cDNA head library cloned into lambda Zap (Stratagene) employing a probe previously isolated by subtractive cloning (see Sive et al. 1989). The plasmid contained within the phage (pBStSK-, into which the original insert was cloned) was excised according to the manufacturer.

Unidirectional exonuclease III deletions of the insert were constructed according to Henikoff (1987). After size selection, the sense strand of individual clones was sequenced (as single-stranded DNA) using the reverse primer and the T7 Sequenase kit (United States Biochemical Corporation). The antisense strand was sequenced after subcloning the entire insert into pBStSK+, employing primers based on the previously sequenced strand.

Sequences were analyzed employing the Genepro program version 4.2 (Riverside Scientific Enterprises). Genbank and PIR databases were searched for significant homologies (lod score of at least 100, requiring a match of 10β0 amino acids).

Comparison of the efficiency of incorporation of digoxigenin-UTP into synthetic RNA using T3, T7 and SP6 RNA polymerase

Most transcription vectors currently in use contain unique RNA polymerase promoters on either side of a polylinker, allowing the use of experimental antisense probes and control sense probes for in situ hybridization. The comparison between the probes is valid only if the relative incorporation of dig-UTP by the RNA polymerases used is approximately equal, and therefore produces probes of equal specific activities.

We have tested SP6, T3 and T7 RNA polymerases on similar templates in identical reactions containing α32P- CTP and dig-UTP. 32P-CTP incorporation was used to quantitate the amount of RNA produced. Samples of each RNA were run on a gel and blotted onto nylon membrane. An autoradiograph of the blot is seen in Fig. 1A. The digoxigenin in each band was detected immunologically and is shown in Fig. IB. The auto-radiograph and the blot were scanned to quantitate the relative amounts of label present, and we find that the different RNA polymerases incorporate dig-UTP with the following relative efficiencies: if SP6 is defined as 1.0, T3 is 1.4, and T7 is 1.7. Thus all polymerases incorporate the nucleotide with similar efficiencies. Furthermore, the RNA products are full length, showing that the modified nucleotide does not cause premature chain termination.

Fig. 1.

Comparative efficiency of digoxigenin-U I P incorporation using T3, T7 and SP6 RNA polymerases. (A) Autoradiogram of 32P-CTP labeled CAT RNA transcripts blotted onto a nylon membrane. The RNA polymerases used were SP6, T3, and T7, lanes 1, 2, and 3, respectively, under identical reaction conditions. (B) Anti-digoxigenin-ll-UTP antibody detection of the RNA on the same blot as shown in (A). The transcripts are just over 900 bases in length.

Fig. 1.

Comparative efficiency of digoxigenin-U I P incorporation using T3, T7 and SP6 RNA polymerases. (A) Autoradiogram of 32P-CTP labeled CAT RNA transcripts blotted onto a nylon membrane. The RNA polymerases used were SP6, T3, and T7, lanes 1, 2, and 3, respectively, under identical reaction conditions. (B) Anti-digoxigenin-ll-UTP antibody detection of the RNA on the same blot as shown in (A). The transcripts are just over 900 bases in length.

Fig. 2.

Detection of specific mRNAs by whole-mount in situ hybridization. (A) Detection of muscle actin at different stages of Xenopus development. Embryonic stages 13 (dorsal view), 24 and 36 (lateral views) are presented from top to bottom. The blastopore of the stage 13 embryo is arrowed (bp). In addition to somites the heart (h) of the stage 36 embryo is stained. (B) Localization of the globin mRNA in the blood islands of a stage 32 tadpole. The staining is detected on the ventral side of the embryo and is bilaterally symmetrical along the anteroposterior axis. (C) Distribution of the XAG-1 transcript in the cement gland of a stage 23 embryo. Note that under saturating conditions of staining some non-specific staining is also detected on the dorsal side of the embryo. (D-F) Localization of the XA-1 transcript. (D) and (E) represent two different views of a stage 23 embryo; (D) lateral view, (E) anterior view. (F) A stage 36 tadpole viewed in the same plane and magnification as (E). All embryos presented in this figure are albinos with the exception of (B) where a pigmented embryo was used. (A,B and C) have been cleared in benzyl-benzoate/benzyl alcohol to see deep staining; (D,E and F) have not been cleared.

Fig. 2.

Detection of specific mRNAs by whole-mount in situ hybridization. (A) Detection of muscle actin at different stages of Xenopus development. Embryonic stages 13 (dorsal view), 24 and 36 (lateral views) are presented from top to bottom. The blastopore of the stage 13 embryo is arrowed (bp). In addition to somites the heart (h) of the stage 36 embryo is stained. (B) Localization of the globin mRNA in the blood islands of a stage 32 tadpole. The staining is detected on the ventral side of the embryo and is bilaterally symmetrical along the anteroposterior axis. (C) Distribution of the XAG-1 transcript in the cement gland of a stage 23 embryo. Note that under saturating conditions of staining some non-specific staining is also detected on the dorsal side of the embryo. (D-F) Localization of the XA-1 transcript. (D) and (E) represent two different views of a stage 23 embryo; (D) lateral view, (E) anterior view. (F) A stage 36 tadpole viewed in the same plane and magnification as (E). All embryos presented in this figure are albinos with the exception of (B) where a pigmented embryo was used. (A,B and C) have been cleared in benzyl-benzoate/benzyl alcohol to see deep staining; (D,E and F) have not been cleared.

Detection of muscle actin RNA

As a first test of the method we set out to detect muscle-specific cardiac actin mRNA, which is abundant and highly localized in the somites and later in the heart (Mohun et al. 1984; Dworkin-Rastl et al. 1986; Kintner and Melton, 1987; Hopwood et al. 1989a). The AC100 plasmid (Dworkin-Rastl et al. 1986; Kintner and Melton, 1987) was linearized with PvuII (see Table 1) and was transcribed with SP6 RNA polymerase. Fig. 2A shows the pattern of expression of muscle actin message in Xenopus embryos from late gastarla to the swimming tadpole stage. The earbest stage in which specific staining was observed was the late gastrula, where two bands of staining correspond to the developing somites (Kintner and Melton, 1987). Some distortion of early embryos often takes place during the procedure; in this case the dorsal view makes the blastopore resemble the neural groove of later embryos, but observation of the embryo in different planes makes identification as a late gastrula stage unambiguous.

The detection of muscle actin RNA at this stage allows us to compare the sensitivities of the two in situ hybridization methods. Using radiolabelled probes on sections, Kinter and Melton (1987) detected muscle actin at stage 14. Therefore, the two methods appear to be comparable in sensitivity, detecting 4x106 tran-scripts per embryo. However, where the radiolabelled probe required a 5 day exposure, the muscle actin stain shown in Fig. 2 was developed in 10 min.

A problem with the staining of these early stages can be seen in the background staining of the archenteron floor of the stage 13 embryo. We also see background staining of the blastocoel roof of earlier embryos (not shown); this kind of background has been noted previously by Hopwood et al. (1989a). Nevertheless, the detection of muscle actin RNA at this early stage illustrates the potential of the whole-mount method. In particular the dorsal view of the stage 13 embryo (Fig. 2A) shows how extensive the presumptive head region is, and how far posteriorly somites are forming at this stage. At later stages the background is usually lower and in the stage 24 embryo the RNA is only detected in the somites. At later stages such as the stage 36 embryo shown here (Fig. 2A), the muscle actin message is also detected in the heart (Hopwood et al. 1989a) and branchiomeric muscle (Dworkin-Rastl et al. 1986). These results show that the whole-mount method is sensitive and has sufficiently low background to apply to other mRNAs. In addition the advantage of observing whole embryos, rather than reconstructing an image from sections, is readily apparent.

Detection of α-globin mRNA

Before proceeding with RNAs whose localization has not been determined precisely, we also tested an α-globin probe. The β-globin message is a good test candidate because it is less abundant than actin at tailbud stages and it has been shown to be localized to a different region of the embryo, the blood islands which are part of the ventral mesoderm. It is also useful to develop a sensitive assay for globin RNA that could be used to investigate the formation of ventral mesodermal cell types in explanted tissues. A cDNA encoding a portion of the Xenopus αTl-globin (Widmer et al. 1981) was subcloned into pGEM-blue to produce poGTldB (M. Boice unpublished). Transcription with T7 RNA polymerase makes an antisense RNA (see Table 1). Fig. 2B shows the pattern of expression of the globin transcript in a Xenopus tadpole (stage 32). In agreement with previous observations, we find this mRNA to be present exclusively in the ventral mesoderm at this stage of development. The blood islands bifurcate anteriorly and are fused posteriorly. The sense probe done in parallel did not show any specific hybridization (not shown).

Spatial distribution of XA-1

The XA-1 gene is expressed in the presumptive cement gland and non-brain head ectoderm from late gastrula to post-hatching stages (Sive et al. 1989; Sive et al. 1990). XA-1 is an interesting gene in that its expression marks the presumptive head region of the late gastrula (stage 12) before any morphological boundary is manifest (Sive et al. 1989). This gene has also been useful in allowing dissection of the hierarchy of inductions that occurs during anteroposterior axis formation (Sive et al. 1990).

XA-1 encodes a poly(A)+ RNA of approximately 800 bases. An apparently full-length cDNA clone was sequenced on both strands (see Materials and methods) (Fig. 3). The longest open reading frame specifies a protein of 198 residues, Mr approximately 22000. There are 3 in-frame methionine residues, with the first in the most favorable context for translation initiation (GCC ATG T). The projected protein is rather hydrophilic, except for an 18 residue stretch at the amino end. This may be a signal peptide, suggesting that XA-1 may be a secreted or integral membrane protein. The protein is very proline-(15 %) and histidine-(12%) rich, and on this basis can be further divided into a more proline-rich amino domain and a basic carboxyl domain. The high proline content throughout the protein makes it unlikely that XA-1 can form long stretches of unbroken alpha helix. An interesting motif in this gene is a twice-repeated 24 amino acid sequence, beginning at residues 73 and 132, with 20/24 identical residues between the repeats. No striking similarities to previously identified open reading frames were apparent (see Materials and methods).

Fig. 3.

Sequence of XA-1. Nucleotide and conceptual protein sequence of the longest open reading frame are shown. Nucleotides and amino acids (in parentheses) are numbered at the end of each line. The first ATG, the first stop codon, the poly A addition signal (AATAAA) and a 24 amino acid repeated sequence (see text) are underlined. Two potential glycosylation sites are present in the carboxyl domain, at amino acids 169 (NGS) and 181 (ASS). The poly A tail begins immediately after the last nucleotide shown.

Fig. 3.

Sequence of XA-1. Nucleotide and conceptual protein sequence of the longest open reading frame are shown. Nucleotides and amino acids (in parentheses) are numbered at the end of each line. The first ATG, the first stop codon, the poly A addition signal (AATAAA) and a 24 amino acid repeated sequence (see text) are underlined. Two potential glycosylation sites are present in the carboxyl domain, at amino acids 169 (NGS) and 181 (ASS). The poly A tail begins immediately after the last nucleotide shown.

The localization of XA-1, determined from dissection and RNA blot analysis was of low resolution. We therefore used the whole-mount in situ hybridization technique to localize the transcripts at single-cell resolution. Antisense RNAs were hybridized to embryos of different stages, and after visualizing the hybridized RNA with either NBT/BCIP or the vectastain II kit, the embryos were cleared for observation. Since no deep staining was apparent, the embryos were remounted in PBS for observation of the ectodermal staining. Figs 2D and E show two views of a stage 23 embryo. The transcript is present in cells that form a ridge starting behind the otic vesicle, extending along the dorsal midline, bifurcating in front of the eyes and reaching the cement gland. The periphery of the cement gland expresses XA-1, in contrast to other genes, which are expressed in the main mass of cement gland tissue (Jamrich and Sato, 1989). Since it was possible that the peripheral staining of the cement gland was due to failure of the probe to penetrate cement gland tissue, we carried out a control hybridization with XAG-1 (Sive et al. 1989). XAG-1 is a suitable candidate since dissection experiments show that it is almost exclusively expressed in the cement gland at stage 23 (Sive et al. 1989). The in situ hybridization shown in Fig. 2C confirms this, and shows that the whole of the cement gland can be stained intensely after hybridization. This result, coupled with the detection of actin transcripts in deep tissues, illustrates that the RNA probes readily penetrate tissues in frog embryos. A potential limitation to the technique is also illustrated in Fig. 2C. In addition to the cement gland, some non-specific staining is seen in the dorsal structures. From the analysis of RNA in dissected animals (Sive et al. 1989), it is unlikely that this signal could be due to XAG-1 RNA. Various sense and antisense probes show variation in the amount of such background, in a probe-dependent way; it may therefore be necessary to optimize the fragment of DNA used to generate the probe (as is the case with RNAse protection) in order to minimize background.

The distribution of XA-1 in the head ectoderm of a stage 36 embryo is shown in Fig. 2F. As development proceeds the continuous lines of cells that expressed the gene at stage 23 (Fig. 2D,E) break up into patches. The ridge of cells in the dorsal midline is still continuous and extends to a point behind the otic vesicle (not shown), but the facial expression has broken up into patches of a few cells. Expression is excluded from the nasal pits.

The distribution of XA-1 RNA in the head ectoderm partially overlaps the distribution of UVS.2 protein (Sato and Sargent, 1989) and tyrosine hydroxylase (Drysdale and Elinson, 1989). Both these markers appear to be expressed in the hatching gland. However, XA-1 and UVS.2 are clearly distinct genes, with different temporal distribution and RNA size.

The representation of XA-1 cDNA clones in libraries indicates that the RNA is rare (less than 1 in 105 embryo mRNAs; Sive et al. 1989). However, since the expression is clearly highly localized it is difficult to assess the number of mRNA copies per cell, and hence the absolute sensitivity of the method. It is however, encouraging to us that rare transcripts such as XA-1 and engrailed (Xenopus En-2, Hemmati-Brivanlou et al. manuscript in preparation) can be detected with this method.

In summary, we have developed a sensitive and reliable method for in situ hybridization in whole Xenopus embryos. The successful detection of the localized RNAs, muscle actin and α-globin, shows that the method is specific and reproducible, and a comparison of our results to in situ detection of the muscle actin gene with a radioactive probe shows that the non-radioactive method is as sensitive. As with other methods, there are problems with background staining, particularly at early stages. Though the background currently limits the sensitivity of the technique, further efforts will improve the specificity of staining. The solubility of the stain in the currently available mounting media also limits the sensitivity, but this difficulty also should be overcome with further experimentation. The use of a non-radioactive assay reduces the time required to develop the signal from weeks to minutes; in addition the visualization of RNAs in whole embryos greatly facilitates the spatial characterization of gene expression. Even where detailed examination of sectioned material is required, sectioning is more easily carried out after the hybridization procedure than before it. Finally, the hybridization and staining can be carried out on batches of embryos permitting the generation of many pieces of information in one experiment. Thus, the sensitivity, specificity and convenience of this new method make it a powerful and highly useful tool for the detailed study of the expression of individual RNAs throughout development. The same approach should also prove successful in other vertebrate embryos.

A.H.B. would like to thank Nipam Patel for his encouragement and helpful comments during the early phase of the work. D.F. is a Rothschild postdoctoral fellow. M.E.B. is an N.S.F. predoctoral fellow. H.L.S. is indebted to Pei Feng Cheng for expert sequencing, and to Bruce Draper for help with isolation of a full-length XA-1 cDNA clone. Supported by an American Cancer Society postdoctoral fellowship and a NIH Molecular Training Program in Cancer Research grant to H.L.S. who also thanks Hal Weintraub for support. This work was supported by a grant from the NIH (GM 42341).

Condie
,
B. G.
and
Harland
,
R. M.
(
1987
).
Posterior expression of a homeobox gene in early Xenopus embryos
.
Development
101
,
93
105
.
Dent
,
J. A.
,
Polson
,
A. G.
and
Klymkowsky
,
M. W.
(
1989
).
A whole-mount immunocytochemical analysis of the expression of the intermediate filament protein vimentin in Xenopus
.
Development
105
,
61
74
.
Drysdale
,
T. A.
and
Elinson
,
R. P.
(
1989
).
Localization of tyrosine hydroxylase to hatching gland cells of early Xenopus laevis embryos
.
J. Cell Biol, (abstracts)
109
,
63a
.
Dworkin-Rastl
,
E.
,
Kelly
,
D. B.
and
Dworkin
,
M. B.
(
1986
).
Localization of specific mRNA sequences in Xenopus laevis by in situ hybridization
.
J. Embryol. exp. Morph
.
91
,
153
168
.
Harland
,
R. M.
and
Weintraub
,
H.
(
1985
).
Translation of mRNA injected into Xenopus oocytes is specifically inhibited by antisense RNA
.
J. Cell Biol
101
,
1094
1099
.
Hemmati-Brivanlou
,
A.
and
Harland
,
R. M.
(
1989
).
Expression of an engrai/ed-related protein is induced in the anterior neural ectoderm of early Xenopus embryos
.
Development
106
,
611
617
.
Henikoff
,
S.
(
1987
).
Unidirectional digestion with exonuclease III in sequence analysis
.
Meth. Enzymol
.
155
,
156
165
.
Hopwood
,
N. D.
,
Pluck
,
A.
and
Gurdon
,
J. B.
(
1989a
).
MyoD expression in the forming somites is an early response to mesoderm induction in Xenopus embryos
.
EMBO J
.
8
,
3409
3417
.
Hopwood
,
N. D.
,
Pluck
,
A.
and
Gurdon
,
J. B.
(
1989b
).
A Xenopus mRNA related to Drosophila twist is expressed in response to induction in the mesoderm and the neural crest
.
Cell
59
,
893
903
.
Jamrich
,
M.
and
Sato
,
S.
(
1989
).
Differential gene expression in the anterior neural plate during gastrulation of Xenopus laevis
.
Development
105
,
779
786
.
Kintner
,
C. R.
and
Melton
,
D. A.
(
1987
).
Expression of Xenopus N-CAM RNA in ectoderm is an early response to neural induction
.
Development
99
,
311
325
.
Lynn
,
D. A.
,
Angerer
,
L. M.
,
Bruskin
,
A. M.
,
Klein
,
W. H.
and
Angerer
,
R. C.
(
1983
).
Localization of a family of mRNAs in a single cell type and its precursors in sea urchin embryos
.
Proc. natn. Acad. Sci. U.S.A
.
80
,
2656
2660
.
Melton
,
D. A.
,
Kreig
,
P. A.
,
Rebagliati
,
M. R.
,
Maniatis
,
T.
,
Zinn
,
K.
and
Green
,
M. R.
(
1985
).
Efficient in vitro synthesis of biologically active RNA and RNA hybridization probes from plasmids containing a bacteriophage SP6 promoter
.
Nuc. Acids Res
.
12
,
7035
7056
.
Mohun
,
T. J.
,
Brennan
,
S.
,
Dathan
,
N.
,
Fairman
,
S.
and
Gurdon
,
J. B.
(
1984
).
Cell type activation of actin genes in the early amphibian embryo
.
Nature
311
,
716
721
.
Nieuwkoop
,
P. D.
and
Faber
,
J.
(
1967
).
Normal table of Xenopus laevis (Daudin). North Holland, Amsterdam
.
Ruiz
,
I
Ataba
,
A.
and
Melton
,
D. A.
(
1989
).
Bimodal and graded expression of the Xenopus homeobox gene Xhox3 during embryonic development
.
Development
106
,
173
183
.
Sato
,
S. M.
and
Sargent
,
T. D.
(
1990
).
Molecular approach to dorsoanterior development in Xenopus laevis
.
Devi Biol
.
137
,
135
41
.
Sive
,
H. L.
,
Draper
,
B. W.
,
Harland
,
R. M.
and
Weintraub
,
H.
(
1990
).
Identification of a retinoic acid sensitive period during primary axis formation in Xenopus laevis
.
Genes Dev. (in press)
.
Sive
,
H. L.
,
Hattori
,
K.
and
Weintraub
,
H.
(
1989
).
Progressive determination during formation of anteroposterior axes in Xenopus laevis
.
Cell
58
,
171
180
.
Tautz
,
D.
and
Pfeifle
,
C.
(
1989
).
A non-radioactive in situ hybridization method for the localization of specific RNAs in Drosophila embryos reveals translational control of the segmentation gene hunchback
.
Chromosoma
98
,
81
85
.
Weeks
,
D. L.
and
Melton
,
D. A.
(
1987
).
A maternal mRNA localized to the vegetal hemisphere of Xenopus eggs codes for a growth factor related to TGF-β Cell
51
,
861
867
.
Widmer
,
H. J.
,
Andres
,
A. C.
,
Niessing
,
J.
,
Hosbach
,
H. A.
and
Weber
,
R.
(
1981
).
Comparative analysis of cloned larval and adult globin cDNA sequences of Xenopus laevis
.
Devi Biol
.
88
,
325
332
.
Wilkinson
,
D. G.
,
Bailes
,
J. A.
and
Mcmahon
,
A. P.
(
1987
).
Expression of the proto-oncogene int-1 is restricted to specific neural cells in the developing mouse embryo
.
Cell
50
,
79
88
.