Transforming DNA Sequences Present in Human ProlactinSecreting Pituitary Tumors
Rivkah Gonsky, Vivien Herman, Shlomo Melmed, and James Fagin Department of Medicine Cedars-Sinai Medical Center-University of California School of Medicine Los Angeles, California 90048
Five human PRL-secreting pituitary tumors were tested for the presence of DNA-transforming sequences. After calcium phosphate transfection to NIH-3T3 mouse fibroblast cells, DNA samples derived from two prolactinomas induced foci of morphologically transformed cells which subsequently grew in soft agar. After retransfection of transformant DNA, resulting secondary transformants elicited rapidly growing solid tumors in nude mice. Southern analysis of transformant DNA revealed the integration of Alu-positive human DNA sequences into the mouse fibroblast NIH-3T3 cells, as judged by hybridization to a Blur-8 probe. The Alu signal became increasingly more difficult to detect with the multiple passaging (>20) of transformant cells in culture. Alu polymerase chain reaction (PCR) was, therefore, used to selectively amplify human DNA sequences from the NIH-3T3 rodent background. PCR using a human Alu-specific primer resulted in amplification of an Alu-containing DNA region within these transformants. The transformant DNA did not hybridize to human genomic probes for genes known to evoke focus formation in this assay, including H-ras, K-ras, N-ras, trk, ret, ros, or met. Further identification of the Alu-containing region revealed that it contained sequences from the human hst gene, a member of the fibroblast growth factor family. The presence of human hst was demonstrated by strong hybridization to a 40-mer oligonucleotide probe to the second exon of hst, by amplification of this region with human hst-nested amplimers within the first and second introns, and finally by direct sequencing. Northern analysis showed hst mRNA transcripts (~3 kilobases) in nude mouse tumors generated from secondary transformants. Using reverse PCR, a 200-basepair cDNA was amplified from prolactinoma tumor RNA, with amplimers bracketing the first and third exons and shown to hybridize to an hst oligonucleotide probe for the second hst exon. A highly sensitive ribonuclease protection assay also confirmed the presence of hst 0888-8809/91/1687-1695S03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
transcripts in a prolactinoma. These results demonstrate that DNA sequences from human prolactinomas containing at least part of the coding region for the hst gene are transforming in the NIH-3T3 cell focus assay. Expression of hst mRNA in prolactinomas suggests that this growth factor gene may be associated with pituitary tumorigenesis. (Molecular Endocrinology 5: 1687-1695, 1991)
INTRODUCTION
Human pituitary adenomas are commonly occurring benign neoplasms, which may arise due to an intrinsic pituitary cellular defect or as a result of hypothalamic dysregulation (1). The most compelling evidence in favor of clonal expansion of genomically altered pituitary cells leading to adenoma formation is the fact that tissue surrounding the pituitary adenoma is not hyperplastic, implying no exogenous hypothalamic hyperstimulation (2). This argument has been strengthened by recent observations using X-chromosome inactivation analysis showing that pituitary adenomas are monoclonal in origin (3, 4). Evidence showing that these adenomas arise from a single cell supports the notion that somatic mutations may be involved in their formation. Mutations in the Gs a-gene gsp have recently been reported in a subset of GH-secreting pituitary tumors with constitutive activation of adenylyl cyclase and GH hypersecretion (5, 6). Little is known regarding the nature of gene mutations that may lead to initiation or progression of PRLsecreting adenomas, the most frequent type of pituitary neoplasia. Bysrom ef a/. (7) reported that two of three prolactinoma tumors tested had loss of heterozygosity in the 11q13 chromosomal locus, the site believed to contain the tumor suppressor gene for multiple endocrine neoplasia type I (MEN-I), suggesting that this putative tumor suppressor gene may also play a role in sporadic pituitary tumors. However, activating mutations of dominantly acting oncogenes have not been reported for PRL-secreting adenomas. To elucidate the possible role of a genetic mutation
1687
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 14:56 For personal use only. No other uses without permission. . All rights reserved.
Vol 5 No. 11
MOL ENDO-1991 1688
in pituitary tumorigenesis, DNA transfection analysis was performed in NIH-3T3 cells to determine the presence of putative transforming DNA sequences in human PRL-secreting adenomas.
RESULTS Pituitary tumor samples were collected from five patients at transsphenoidal surgery and immediately frozen in liquid N2. All five patients were females, aged 21-27 yr, who harbored macroadenomas, which were well differentiated, sparsely granulated PRL cell adenomas. High mol wt DNA was isolated from these human PRL-secreting pituitary tumor specimens. These DNA samples were assayed for their ability to induce morphological transformation after transfection into NIH-3T3 mouse fibroblasts. Foci consisting of highly refractile node-forming cells were scored in a double blind manner. The efficiency of transformation was monitored by including pSM-FCSV DNA, a potent transforming viral oncogene, as a positive control in each assay. Transformed foci (33-40 FFU/50 ^g DNA) were induced by DNA derived from two different prolactinoma DNA samples. The positive foci were scattered throughout all dishes rather than clustered in one or two plates. Transforming efficiency was markedly increased (3- to 6-fold) after retransfection of transformant DNA into NIH-3T3 cells yielding secondary transformants (Table 1). High mol wt DNA extracted from normal pituitary tissue, normal lymphocyte DNA, and nontransfected NIH-3T3 cell DNA did not elicit cell transformation. The rate of spontaneous appearance of transformed foci in transfected NIH-3T3 cells was extremely low (50 cells) in semisolid agar (0.5%; Fig. 1D). Nontransfected NIH-3T3 cells (Fig. 1C) did not form colonies in soft agar. Cells from 10 secondary transformants were injected sc into athymic mice (106 cells/mouse). One secondary transformant, clone 307, was highly tumorigenic. All six mice injected with clone 307 developed a solid tumor at the site of inoculation after 4-5 weeks. During this time, control NIH-3T3 cells were not tumorigenic. When tumor cells were dispersed and reinoculated to nude mice, tumors developed in less than 3 weeks (Table 1). Detection of Human DNA Sequences in Prolactinoma-Derived Transformants To determine the presence of human DNA sequences in the transformants, DNA was extracted from subcloned transformants and screened by slot blot hybridization with an Alu-specific Blur-8 probe for detection of human specific repetitive Alu sequences (Fig. 2). The Alu-specific Blur-8 probe did not hybridize with nontransformed 3T3 cell DNA (negative controls). The arrows in Fig. 2 indicate six positive clones that were further characterized by Southern analysis (Fig. 3). After EcoRI digestion, DNA derived from primary and secondary transformants revealed the presence of high mol wt Alu-positive DNA sequences (lanes 6 and 10). Enzymatic restriction of DNA with Alu\, which identifies a restriction site within Alu repeats, resulted in the dissipation of the Alu-positive bands (lanes 5 and 9). Southern blots of DNA derived from secondary transformants or from nude mouse tumors were screened for the presence of human protooncogene sequences known to evoke focus formation in the NIH-3T3 cell assay (8-13). Genomic probes for H-ras, K-ras, N-ras,
Table 1. In Vitro and in Vivo Transforming Activity of Pituitary Adenoma DNA In Vitro Transfection (FFU/50 nQ DNA) Donor DNA
Prolactinoma8 3 4 Negative Controls Normal pituitary Normal lymphocyte NIH-3T3 Positive Control pSM-FCSV DNA
Primary transfection
Secondary transfection
33 40
180 130
0 0 0
25 x 103
In Vivo Tumorigenicity (nude mice)
6/6"; 10/10" ND 0 0 0
ND
5/5
FFU, Focus-forming units; ND, not done. a Two of five prolactinoma DNAs transfected were transforming in the assay. b Injected with NIH-3T3 cells derived from secondary transfection (clone 307).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 14:56 For personal use only. No other uses without permission. . All rights reserved.
1689
DNA Sequences in PRL-Secreting Pituitary Tumor
Fig. 1. Photomicrograph of NIH-3T3 Cells Transformed with Feline Sarcoma Virus DNA (A) or Prolactinoma DNA (B) Derived from Human PRL-Secreting Pituitary Tumor Five micrograms of DNA were transfected to NIH-3T3 cells. Photographs of foci were taken 14 days after transfection. Soft agar colonies formed with prolactinoma-transformed NIH-3T3 cells (D), but not with untransformed NIH-3T3 cells (C). Cells (1 x 10") were plated in 0.5% agar onto 1.5% agar underlayer. Clusters of cells forming colonies visible to the naked eye were photographed after 14 days.
met, trk, ros, and ret did not hybridize with the transformant or tumor-derived DNA. PCR Amplification of Alu sequences DNA extracted from nude mouse tumors as well as from secondary transformants that had been passaged in culture for more than 20 passages showed decreased intensity of hybridization to Blur-8. To facilitate characterization of the transforming sequence presumed to be in close proximity to an Alu region, an Aluspecific polymerase chain reaction (PCR) was performed. The primer used, 517 (see Materials and Methods), amplifies regions between two Alu repeats if they are in an inverted orientation relative to each other (14). No amplification of mouse 3T3-cell DNA (negative control) was observed. After amplification, DNA was sizeseparated and probed with Blur-8. Figure 4 shows that discrete bands of DNA ranging from 0.5-1 kilobases (kb) were present after Alu-specific PCR amplification and hybridization to Blur-8. To further investigate the nature of the Alu-containing products generated by PCR and to determine whether known growth factor or protooncogene regions were present (some of which are also capable of inducing cell transformation), Southern blots were hybridized to an /isM oligonucleotide probe (a 40-base probe within the second exon of the hst gene). The hst^ is a member
of the fibroblast growth factor family (29), and its gene is located in close proximity to the MEN-I locus on chromosome 11q13 (7,15). Figure 5 shows that amplified PCR products of 0.5-1 kb hybridized to the radiolabeled hst-~\ oligonucleotide probe, indicating that sequences containing one or more exons of hst-A were present in these transformants. PCR amplification with nested primers (Fig. 6A) specific for the introns flanking the second exon of human hst confirmed the presence of an appropriately sized human hst product (Fig. 6B). When the primary PCR product was reamplified with nested amplimers, DNA fragments of the expected size of 0.5 kb were obtained, which hybridized to the hst oligonucleotide probe. Nontransformed NIH-3T3 cell DNA was not amplified with either set of hst amplimers (Fig. 6B, lanes 7 and 8). About 100 basepairs (bp) of the 0.7-kb transformant hst fragment was sequenced and found to be identical to the normal human sequence within this intron (Fig. 6C). As the hst gene does not contain Alu-like sequences within the coding or intervening regions, this may explain the loss of hybridization with Alu probes after repeated retransfection. The hst Expression in Nude Mouse Tumors Total cellular RNA was extracted from nude mouse tumors generated from secondary transformants, gel separated, and probed with an /isf-specific oligonucle-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 14:56 For personal use only. No other uses without permission. . All rights reserved.
Vol 5 No. 11
MOL ENDO-1991 1690
B
C
f!
I
1 2 3 4 5 6 7 8
Fig. 2. Presence of Human Alu Sequence in Transformed NIH3T3 Cell DNA: Slot Blot Hybridization Screen with Blur-8, an Alu-Specific 32P-Labeled Probe Lanes A, B, and C: +, HeLa cell DNA (positive control); - , NIH-3T3 mouse DNA (negative control); 1-8, 10 ^g DNA from transformant. Arrows indicate transformants exhibiting positive Alu hybridization that were further analyzed. 4A, 7A, 3B, 8B, and 1C are primary transformant DNA derived from two separate prolactinomas; 2B is secondary transformant DNA (clone 307) originally derived from same prolactinoma DNA as 8B.
otide. As shown in Fig. 7, hst mRNA transcripts (—3.3 kb) were present in the nude, mouse tumors. The F9 teratocarcinoma cell line, known to express hst in its undifferentiated state, served as a positive control and indeed expressed hst, whereas normal mouse liver tissue did not. The histological appearance of the tumors showed them to be highly vascularized, compatible with the presence of an angiogenic growth factor in these tumors (data not shown). The hst Expression in Human Prolactinomas Expression of hst in PRL-secreting tumors was studied in surgical specimens obtained at transsphenoidal adenomectomy. Since the size of the tumor tissue samples was very small (
Rivkah Gonsky, Vivien Herman, Shlomo Melmed, and James Fagin Department of Medicine Cedars-Sinai Medical Center-University of California School of Medicine Los Angeles, California 90048
Five human PRL-secreting pituitary tumors were tested for the presence of DNA-transforming sequences. After calcium phosphate transfection to NIH-3T3 mouse fibroblast cells, DNA samples derived from two prolactinomas induced foci of morphologically transformed cells which subsequently grew in soft agar. After retransfection of transformant DNA, resulting secondary transformants elicited rapidly growing solid tumors in nude mice. Southern analysis of transformant DNA revealed the integration of Alu-positive human DNA sequences into the mouse fibroblast NIH-3T3 cells, as judged by hybridization to a Blur-8 probe. The Alu signal became increasingly more difficult to detect with the multiple passaging (>20) of transformant cells in culture. Alu polymerase chain reaction (PCR) was, therefore, used to selectively amplify human DNA sequences from the NIH-3T3 rodent background. PCR using a human Alu-specific primer resulted in amplification of an Alu-containing DNA region within these transformants. The transformant DNA did not hybridize to human genomic probes for genes known to evoke focus formation in this assay, including H-ras, K-ras, N-ras, trk, ret, ros, or met. Further identification of the Alu-containing region revealed that it contained sequences from the human hst gene, a member of the fibroblast growth factor family. The presence of human hst was demonstrated by strong hybridization to a 40-mer oligonucleotide probe to the second exon of hst, by amplification of this region with human hst-nested amplimers within the first and second introns, and finally by direct sequencing. Northern analysis showed hst mRNA transcripts (~3 kilobases) in nude mouse tumors generated from secondary transformants. Using reverse PCR, a 200-basepair cDNA was amplified from prolactinoma tumor RNA, with amplimers bracketing the first and third exons and shown to hybridize to an hst oligonucleotide probe for the second hst exon. A highly sensitive ribonuclease protection assay also confirmed the presence of hst 0888-8809/91/1687-1695S03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society
transcripts in a prolactinoma. These results demonstrate that DNA sequences from human prolactinomas containing at least part of the coding region for the hst gene are transforming in the NIH-3T3 cell focus assay. Expression of hst mRNA in prolactinomas suggests that this growth factor gene may be associated with pituitary tumorigenesis. (Molecular Endocrinology 5: 1687-1695, 1991)
INTRODUCTION
Human pituitary adenomas are commonly occurring benign neoplasms, which may arise due to an intrinsic pituitary cellular defect or as a result of hypothalamic dysregulation (1). The most compelling evidence in favor of clonal expansion of genomically altered pituitary cells leading to adenoma formation is the fact that tissue surrounding the pituitary adenoma is not hyperplastic, implying no exogenous hypothalamic hyperstimulation (2). This argument has been strengthened by recent observations using X-chromosome inactivation analysis showing that pituitary adenomas are monoclonal in origin (3, 4). Evidence showing that these adenomas arise from a single cell supports the notion that somatic mutations may be involved in their formation. Mutations in the Gs a-gene gsp have recently been reported in a subset of GH-secreting pituitary tumors with constitutive activation of adenylyl cyclase and GH hypersecretion (5, 6). Little is known regarding the nature of gene mutations that may lead to initiation or progression of PRLsecreting adenomas, the most frequent type of pituitary neoplasia. Bysrom ef a/. (7) reported that two of three prolactinoma tumors tested had loss of heterozygosity in the 11q13 chromosomal locus, the site believed to contain the tumor suppressor gene for multiple endocrine neoplasia type I (MEN-I), suggesting that this putative tumor suppressor gene may also play a role in sporadic pituitary tumors. However, activating mutations of dominantly acting oncogenes have not been reported for PRL-secreting adenomas. To elucidate the possible role of a genetic mutation
1687
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 14:56 For personal use only. No other uses without permission. . All rights reserved.
Vol 5 No. 11
MOL ENDO-1991 1688
in pituitary tumorigenesis, DNA transfection analysis was performed in NIH-3T3 cells to determine the presence of putative transforming DNA sequences in human PRL-secreting adenomas.
RESULTS Pituitary tumor samples were collected from five patients at transsphenoidal surgery and immediately frozen in liquid N2. All five patients were females, aged 21-27 yr, who harbored macroadenomas, which were well differentiated, sparsely granulated PRL cell adenomas. High mol wt DNA was isolated from these human PRL-secreting pituitary tumor specimens. These DNA samples were assayed for their ability to induce morphological transformation after transfection into NIH-3T3 mouse fibroblasts. Foci consisting of highly refractile node-forming cells were scored in a double blind manner. The efficiency of transformation was monitored by including pSM-FCSV DNA, a potent transforming viral oncogene, as a positive control in each assay. Transformed foci (33-40 FFU/50 ^g DNA) were induced by DNA derived from two different prolactinoma DNA samples. The positive foci were scattered throughout all dishes rather than clustered in one or two plates. Transforming efficiency was markedly increased (3- to 6-fold) after retransfection of transformant DNA into NIH-3T3 cells yielding secondary transformants (Table 1). High mol wt DNA extracted from normal pituitary tissue, normal lymphocyte DNA, and nontransfected NIH-3T3 cell DNA did not elicit cell transformation. The rate of spontaneous appearance of transformed foci in transfected NIH-3T3 cells was extremely low (50 cells) in semisolid agar (0.5%; Fig. 1D). Nontransfected NIH-3T3 cells (Fig. 1C) did not form colonies in soft agar. Cells from 10 secondary transformants were injected sc into athymic mice (106 cells/mouse). One secondary transformant, clone 307, was highly tumorigenic. All six mice injected with clone 307 developed a solid tumor at the site of inoculation after 4-5 weeks. During this time, control NIH-3T3 cells were not tumorigenic. When tumor cells were dispersed and reinoculated to nude mice, tumors developed in less than 3 weeks (Table 1). Detection of Human DNA Sequences in Prolactinoma-Derived Transformants To determine the presence of human DNA sequences in the transformants, DNA was extracted from subcloned transformants and screened by slot blot hybridization with an Alu-specific Blur-8 probe for detection of human specific repetitive Alu sequences (Fig. 2). The Alu-specific Blur-8 probe did not hybridize with nontransformed 3T3 cell DNA (negative controls). The arrows in Fig. 2 indicate six positive clones that were further characterized by Southern analysis (Fig. 3). After EcoRI digestion, DNA derived from primary and secondary transformants revealed the presence of high mol wt Alu-positive DNA sequences (lanes 6 and 10). Enzymatic restriction of DNA with Alu\, which identifies a restriction site within Alu repeats, resulted in the dissipation of the Alu-positive bands (lanes 5 and 9). Southern blots of DNA derived from secondary transformants or from nude mouse tumors were screened for the presence of human protooncogene sequences known to evoke focus formation in the NIH-3T3 cell assay (8-13). Genomic probes for H-ras, K-ras, N-ras,
Table 1. In Vitro and in Vivo Transforming Activity of Pituitary Adenoma DNA In Vitro Transfection (FFU/50 nQ DNA) Donor DNA
Prolactinoma8 3 4 Negative Controls Normal pituitary Normal lymphocyte NIH-3T3 Positive Control pSM-FCSV DNA
Primary transfection
Secondary transfection
33 40
180 130
0 0 0
25 x 103
In Vivo Tumorigenicity (nude mice)
6/6"; 10/10" ND 0 0 0
ND
5/5
FFU, Focus-forming units; ND, not done. a Two of five prolactinoma DNAs transfected were transforming in the assay. b Injected with NIH-3T3 cells derived from secondary transfection (clone 307).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 14:56 For personal use only. No other uses without permission. . All rights reserved.
1689
DNA Sequences in PRL-Secreting Pituitary Tumor
Fig. 1. Photomicrograph of NIH-3T3 Cells Transformed with Feline Sarcoma Virus DNA (A) or Prolactinoma DNA (B) Derived from Human PRL-Secreting Pituitary Tumor Five micrograms of DNA were transfected to NIH-3T3 cells. Photographs of foci were taken 14 days after transfection. Soft agar colonies formed with prolactinoma-transformed NIH-3T3 cells (D), but not with untransformed NIH-3T3 cells (C). Cells (1 x 10") were plated in 0.5% agar onto 1.5% agar underlayer. Clusters of cells forming colonies visible to the naked eye were photographed after 14 days.
met, trk, ros, and ret did not hybridize with the transformant or tumor-derived DNA. PCR Amplification of Alu sequences DNA extracted from nude mouse tumors as well as from secondary transformants that had been passaged in culture for more than 20 passages showed decreased intensity of hybridization to Blur-8. To facilitate characterization of the transforming sequence presumed to be in close proximity to an Alu region, an Aluspecific polymerase chain reaction (PCR) was performed. The primer used, 517 (see Materials and Methods), amplifies regions between two Alu repeats if they are in an inverted orientation relative to each other (14). No amplification of mouse 3T3-cell DNA (negative control) was observed. After amplification, DNA was sizeseparated and probed with Blur-8. Figure 4 shows that discrete bands of DNA ranging from 0.5-1 kilobases (kb) were present after Alu-specific PCR amplification and hybridization to Blur-8. To further investigate the nature of the Alu-containing products generated by PCR and to determine whether known growth factor or protooncogene regions were present (some of which are also capable of inducing cell transformation), Southern blots were hybridized to an /isM oligonucleotide probe (a 40-base probe within the second exon of the hst gene). The hst^ is a member
of the fibroblast growth factor family (29), and its gene is located in close proximity to the MEN-I locus on chromosome 11q13 (7,15). Figure 5 shows that amplified PCR products of 0.5-1 kb hybridized to the radiolabeled hst-~\ oligonucleotide probe, indicating that sequences containing one or more exons of hst-A were present in these transformants. PCR amplification with nested primers (Fig. 6A) specific for the introns flanking the second exon of human hst confirmed the presence of an appropriately sized human hst product (Fig. 6B). When the primary PCR product was reamplified with nested amplimers, DNA fragments of the expected size of 0.5 kb were obtained, which hybridized to the hst oligonucleotide probe. Nontransformed NIH-3T3 cell DNA was not amplified with either set of hst amplimers (Fig. 6B, lanes 7 and 8). About 100 basepairs (bp) of the 0.7-kb transformant hst fragment was sequenced and found to be identical to the normal human sequence within this intron (Fig. 6C). As the hst gene does not contain Alu-like sequences within the coding or intervening regions, this may explain the loss of hybridization with Alu probes after repeated retransfection. The hst Expression in Nude Mouse Tumors Total cellular RNA was extracted from nude mouse tumors generated from secondary transformants, gel separated, and probed with an /isf-specific oligonucle-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 November 2015. at 14:56 For personal use only. No other uses without permission. . All rights reserved.
Vol 5 No. 11
MOL ENDO-1991 1690
B
C
f!
I
1 2 3 4 5 6 7 8
Fig. 2. Presence of Human Alu Sequence in Transformed NIH3T3 Cell DNA: Slot Blot Hybridization Screen with Blur-8, an Alu-Specific 32P-Labeled Probe Lanes A, B, and C: +, HeLa cell DNA (positive control); - , NIH-3T3 mouse DNA (negative control); 1-8, 10 ^g DNA from transformant. Arrows indicate transformants exhibiting positive Alu hybridization that were further analyzed. 4A, 7A, 3B, 8B, and 1C are primary transformant DNA derived from two separate prolactinomas; 2B is secondary transformant DNA (clone 307) originally derived from same prolactinoma DNA as 8B.
otide. As shown in Fig. 7, hst mRNA transcripts (—3.3 kb) were present in the nude, mouse tumors. The F9 teratocarcinoma cell line, known to express hst in its undifferentiated state, served as a positive control and indeed expressed hst, whereas normal mouse liver tissue did not. The histological appearance of the tumors showed them to be highly vascularized, compatible with the presence of an angiogenic growth factor in these tumors (data not shown). The hst Expression in Human Prolactinomas Expression of hst in PRL-secreting tumors was studied in surgical specimens obtained at transsphenoidal adenomectomy. Since the size of the tumor tissue samples was very small (
