Alternative translation initiation site usage results in two structurally ...

Communication

THE JOURNAL OP BIOLOGICAL CHEMISTRY Vol. 266, No. 20, hue of July 15, pp. 12832-12635, 1991 0 1991 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

Alternative Translation Initiation Site Usage Results in Two Structurally Distinct Forms of Pit-l* (Received Jeffrey Michael

for publication,

W. Voss$.$, Tso-Pang G. Rosenfeld$3IJJ

Yao*,

February

EXPERIMENTAL

1, 1991)

and

From

the (Howard Hughes Medical Institute, jEukaryotic Biology Program, and 11Center for Molecular Genetics School of Medicine, University of California, San Diego, La Jolla, California 92093-0648 Regulatory

Pit- 1 is a pituitary-specific transcription factor that plays a critical role in the normal development of the anterior pituitary gland. Previous analyses have shown that this protein exists in the rat pituitary gland and in rat pituitary-derived cell lines as two forms of relative molecular mass 33 and 31 kDa. This aspect of Pit-l expression has been conserved throughout the evolution from rodents to humans. Here, we determine the origin of these structurally distinct forms of Pit-l protein and find that these arise as a consequence of the alternative usage of translation initiation sites present in Pit-l mRNA.

Pit-l, a pituitary-specific transcription factor, plays an essential role in the normal development of the anterior pituitary gland (1). A member of the POU domain family of transcriptional activators (2-ll), this protein has been shown by several means to tram-activate expression of the growth hormone and prolactin genes (3,12, 13). Pit-l is first detected on day 15 of rat embryonic development (14), although Pit-l gene transcripts are detected as early as embryonic day 11 in the rat neural tube (9). Western blot analyses of proteins, isolated from rat pituitary-derived GC3 cells or from developing and adult rat pituitary glands, reveal the presence of two immunologically related species of 31 and 33 kDa (12). Both forms of Pit-l have also been detected, using the same technique, in murine and human pituitary tissue.’ Moreover, labeled Pit-l DNA recognition elements bind to species of 31 and 33 kDa on protein transfers of extracts from rat pituitaryderived cell lines (2, 14). Because the alteration that leads to the formation of these forms of Pit-l could affect a domain required for a function such as DNA binding or transactivation, it was of interest to distinction. Here, we describe Pit-l species and demonstrate Pit-l result from alternative

characterize the nature of this the relationship between these that the alternative forms of translation initiation site usage.

* These studies were supported by a National Institutes of Health grant. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § American Cancer Society Fellow PF-3407. ’ J. Voss, unpublished observations.

PROCEDURES

Antisera, Immunoprecipitations, and Western Blot Analysis132ruPit antisera was generated by injecting a female New Zealand White rabbit with purified bacterially expressed Pit-l protein. 5lOotPit serum was generated by injecting a female New Zealand White rabbit with a synthetic peptide (MSCQPFTSADTFIPLNSDASAALPPR) conjugated to bovine serum albumin. For immunoprecipitation, cells were lysed with a buffer that consisted of 0.1% SDS,’ 20 mM Tris. DH 7.8. 2 mM EDTA. 300 mM NaCl. 5 UP/ ml leupeptin, and 5 rg/mipepstatin A. The lysates were imme&ate& boiled, sonicated, and then cleared by centrifugation at 13,000 X g. Antiserum was added to a final dilution of 1:2000 for l-2 h, and the antibody-antigen complexes were precipitated using protein A-Sepharose, followed by several washes with lysis buffer. Western blot analyses were performed as described (14). Transient Co-transfection Analysis-African green monkey kidney cells (CV-1) were transfected as described (14. 16). The transfected cells were harvested 48 h after transfection, and the cytoplasmic fraction was assayed for luciferase activity as previously described (17).

In Vitro Transcription and in Vitro Translation-in vitro transcription was performed using T7 RNA polymerase (Promega Biotec), and in vitro translation was performed using a rabbit reticulocyte lysate system and [%]methionine (Promega Biotec). RESULTS

Two Tissues

Forms Derived

of Pit-l

Protein the Anterior

Are

Expressed

in

Cells

and

Pituitary-Shown in Fig. 1 are “Western” blot analyses of a variety of cell and tissue extracts using Pit-l antiserum (132aPit) as a probe. Epitope mapping experiments using truncated Pit-l molecules have established that the Pit-l reactive antibodies in this serum are directed primarily at an epitope that lies between amino acids 50 and 125 of the Pit-l primary sequence.’ Panel A contains the extracts of a variety of pituitary and nonpituitary-derived cell lines, and panel B contains extracts of pituitary glands isolated on various days postfertilization from embryonic and neonatal rats. Common to all samples that contained detectable levels of Pit-l protein were species of 33 and 31 kDa that were present in what appeared to be a consistent ratio of about 2:l. Thus, the presence of two forms of Pit-l appeared to be a common if not ubiquitous feature of cells that express Pit-l protein. Pulse and Pulse-Chase Analysis of Pit-l Protein SynthesisTo determine if a precursor-product relationship existed between the two forms of Pit-l protein, pulse-chase labeling experiments were performed. Based on the results of these experiments shown in Fig. 2, we concluded that a precursorproduct relationship between the two forms of Pit-l was not likely since the rates of synthesis and degradation of the two forms of Pit-l appeared to be identical. Furthermore, we estimated the half-life of Pit-l protein in the pituitary-derived 235 cells to be approximately 4 h. 31- and 33-kDa Forms of Pit-l Protein Result from Alternative

Transhtion

from

Initiation

Site

Usage-The

results

of the

pulse-chase experiments described above raised the possibility that the two forms of Pit-l were distinct primary translation products. Examination of the Pit-l cDNA sequence (2, 18) revealed the presence of a number of potential initiator methionine codons (Fig. 3 A). The first two AUG triplets of the Pit-l leader sequence are followed almost immediately by

12832

* The

abbreviation

used

is: SDS,

sodium

dodecyl

sulfate.

Alternative Translation Site Initiation

Usage

12833

B.

A.

.-In "

a

Pit-1 Plt-1

--

a

I +pit-1 -Pits1

+33 Kd3 '1 KdCI

1

2

3

4

5

6

7

8

91011

FIG.1. Western blot analysis of Pit-1 derived from pituitary and pituitary-derived cell lines. Panel A, total cellular proteins derivedfrom 1 X IO4 cells of each of theindicated cell lines were transferredto nitrocellulose after fractionation on an SDS-polyacrylamide gel. The transfer was reacted with 132nPit antibody and visualized by horseradish peroxidase staining. Panel R, lunes 1-9 contain total proteins (40 pgllane) derived from developing rat pituitary glands dissected on the day (post-fertilization) indicatedat thetop of the panel that were transferred to nitrocellulose after fractionation on a 12% SDS-polyacrylamide gel. Also included as controls are proteins derived from fetal liver (lane IO) and pituitary-derived 235 cells (lane 11). The transfer was reacted with 132nPit antiserum and visualized by horseradish peroxidase staining.

A

B

PulsePulse Minutes: 0 10 20 30 100 Pit-1Pit-1-

t

- Chase

Hours: 0

-- i 5

6 3 3 Kd"31 Kd-

1

4

22 Pit-1 *Pit-l

t

1 1 2 2 3 3 4 4 5

FIG.2. Pulse and pulse-chase analysis of Pit-1 protein expressed in 235 cells. Panel A, Lanes 1-5 contain '"S-labeled Pit-1 proteinimmunoprecipitated from 235 cell lysatesusing132aPit serum. Cells were pulsed and then harvested. PanelR, 235 cells were labeledwith[:',%]methioninefor4h andthenchasedwith cold methionine. After the chase, interval cells were lysed and immunoprecipitated with 132nPit serum.

termination codons, but the third and fourthAUG triplets of Pit-1 mRNA are in-frame to initiate the translation of the long open reading frame that encodes Pit-1 protein. Although the sequences surrounding these AUG triplets exhibit little overall homology to the proposed Kozak consensus sequence (19) (Fig. 3A), initiation from these sites would be predicted to yield polypeptides of 33 and 31 kDa, respectively. To test the possibility that eachof these AUG triplets was competent to direct the synthesisof Pit-1 protein, we used site-directed in vitro mutagenesis to construct templatesin which the first (MET-l,,,) or second methionine (MET-2,,,,) codon of the Pit-1 coding sequence was replaced by a n isoleucine codon (Fig. 3A).The mutagenized Pit-1 genes were then cloned into a eucaryotic expression vector, and each was tested by transienttransfectionanalysis followed by Westernblotting (using 132aPit serum) for their ability to give rise to the respective 33- or 31-kDa proteins predicted from the coding sequence. The resultsof the Western blot analysis of extracts of CV-1 cells transfected with MET-l,,,, MET-P,,,,, or wild type expression plasmids is shown in Fig. 3, Panel B. These experiments indicated that, although transfection of wild type Pit-1 expression plasmid yielded the characteristic 33- and 31-kDa forms of Pit-1 (lane 3 ) , the mutant expression plasmids were each restricted in their expression toa single form of Pit-1. Interestingly, the ratio of 33- to 31-kDa proteins appeared to be considerably higher in CV-1 cells transfected with the wild type expression plasmidthan in pituitaryderived cells (see Fig. 1)that naturally express Pit-1 protein. Because the expression plasmid MET-2,,,, gives rise to the 33-kDa form of Pit-1 exclusively, we were able to exclude the

possibility that the 31-kDa form of Pit-1 arises from proteolytic cleavage of the 33-kDa form during theisolation procedure and to conclude that expression of the 33- and 31-kDa forms of Pit-1 is dependent upon translation initiation events that takeplace a t methionine codons 1 and 2 of the predicted Pit-1 coding sequence. 31- and 33-kDa Formsof Pit-1 Protein AreDerived from a Single R N A Transcript and Are Modified during Translation in Vitro-The results described above predict that both 33and 31-kDa forms of Pit-1 arisefrom a singleRNA transcript. To test this, capped Pit-1 mRNA was first synthesized i n vitro from a cloned Pit-1 cDNA template and then translated i n vitro using [:%]methionine in a rabbit reticulocyte system. The translation products were then fractionated on a 10% SDS-polyacrylamide gel, as shown in Fig. 3C. These results demonstrated that the 33- and 31-kDa forms of Pit-1 could arise during in vitro translation of a single, unprocessed Pit1 mRNA transcript. MutantPit-1mRNAs derived fromthe MET-l,,,, and MET-2,,, mutants described above were used to generate""Slabeled Pit-1 polypeptides by i n vitro translation usinga rabbit reticulocyte system.Aftertranslation,thereaction products were fractionated a t high resolution on a 16% SDSpolyacrylamide gel (Fig. 3 0 , lanes 1-3) and yielded results analogous tothoseobtainedinthetransienttransfection analysis described above. Specifically, synthesis of the 33and 31-kDa forms of Pit-1 were again dependentonthe presence of methionine codons at positions 1 and 27 of the predicted Pit-1 coding sequence. However, resolution of the Pit-1 i n vitro translationproductsundertheseconditions revealed that the33- and 31-kDa Pit-1species each consisted of two species that could be resolved into doublets differing in relative molecular weight by less than a kilodalton. Because it was likely that these "doublets" resulted from a post-translational modification of Pit-1, such asphosphorylation, we treated i n vitro translated Pit-1 proteins with calf intestine alkaline phosphatase. As shown in Fig. 3, Panel D, lanes 5 and 6 , treatment of the in vitro translation products with alkaline phosphatase converted each pair of doublets into single bands, indicating that the source of the heterogeneity within each33- and 31-kDaspecies was likely to be the presenceorabsence of phosphate residues. Theseresults suggested that Pit-1 serves as an efficient substrate fora protein kinase found in rabbitreticulocyte lysate and indicate that Pit-1 may be efficiently phosphorylated in vivo. Prelim-

Alternative Translation Initiation Site Usage

12834

A. MET met ter TER CTCAGAGCCGCCCTGATGTATATATGCAATAGGGAGCCGTGAATCGGCCCTTTGATACAGTAATATAATAAAAGCGGACTGGCAAGCGGTGGCTCTTAGTTC

M S C O P F T S A D T F I P L N S D A S A A L P P R M H TCTACTCTCTTGiGGGAATGAGTTGCCAACCTTTCACCTCGGCTGATACCTTTATACCTCTGAATTCTGACGCTTCTGCTGCCCTGCCTCTGAGAATGCAC GCCACCATGG GCCACCATGG G G

MET-2

MET- 1

c. 2 VI

-0 C

Y

2? .&

w a 97.5

- 68Kd

’

NS 4 -. I

68

M3 -I

- 46Kd

4

46 33Kd 31 Kd - 29Kd

15

2

33Kd + *31Kd*

3

I-

M3 -+

29

1

2

33Kd +31 Kd

3

FIG. 3. Site-directed in vitro mutagenesis of Pit-I initiator methioninecodons. Panel A, DNA sequence of the 5’-untranslated leader sequence of Pit-1 mRNA (18). 5”Methionine codons that are separated from the (ler). Pit-1 long open reading frame are indicated in three-letter code with their corresponding termination codons The Pit-1 coding sequence is given in one-letter code with t.he Kozak (19) consensus translation initiation site sequence given below the MET-1 or MET-2 DNA sequence. Panel R, Western blot analysis of Pit-1 proteins of cells transfected with derived from CV-1 cells transfected with initiator methionine codon mutants. Lysates a 12% SDS-polyacrylamide each ofthe indicated wild t.ype or mutant Pit-1 expression plasmids were fractionatedon gel and transfected to nitrocellulose. The filter was then incubated with 132cuPit. antibody and visualized by horseradishperoxidasestaining. Panel C, ‘“S-labeled Pit-1 protein,translated in vitro fromcappedmRNA transcribed in vitro from a Pit-1 cDNA clone.1”ane.l I), high resolution analysisof in vitro translated Pit-1 proteins. Templates used in these consistedof wild type mRNA (lane I), MET-l,,,,,, mRNA(lane 2 ) . MET-2.,,,, mRNA (lane 3 ) , wild type Pit-1 mRNA (lane 5), and wild t.ype RNA treated with alkaline phosphat.ase (lane 6). M 3 , species that results from translation initiation at third methionine of Pit-1 coding sequence; N S , translation product not specific to Pit-1 mRNA.

terminal sequence, we generated antiserum (5lOaPit) to a synthetic polypeptide based on the sequence of the first 27 amino acids of the Pit-1 codingsequence (Fig. 3 A). This sequence would be confinedtothe larger (33-kDa)Pit-1 polypeptide if the smaller (31-kDa) polypeptide was generated as a result of translation initiation at thesecond methionine triplet of the codingsequence. As shown in Fig. 4C, the 5lOaPit serum reacted, on Western transfers of pituitaryderived cell extracts, specifically with the larger of the two Pit-1 polypeptides. These results demonstrated that amino acid sequences present in the first 26 residues of the Pit-1 coding sequenceare absentfrom the smaller Pit-1 polypeptide and areconfined exclusively to the largerspecies. 1 2 3 4 5 6 7 8 9 101112 31- and 33-kDa Forms Are Similar in Ability Their to transFIG. 4. Western blot analysis of pituitary- and nonpituitary-derived cell lines using 132aPit and 510aPit antisera. Activate the Prolactin Gene-Previous reports, using domain transfer and deletion mapping experiments, suggested that 1,ane.s 1-12 containproteinsderivedfrom t.he indicatedcelllines transferred to nitrocellulose after fractionation on an SDS-polyacrylthe amino terminusof Pit-1 may encode the trans-activation amide gel. Lanes 1-6, reaction with 132nPit antiserum. Lanes 7-12, domain of the protein (12, 21). It was, therefore, of interest reaction with the amino terminally directed5lOtuPit serum. All samto determine if the amino terminally truncated 31-kDaform ples were normalized bv cell number (1 X lo4 cells/lane). of Pit-1 was compromised in its ability to trans-activate a target gene such as the prolactin gene. Thus, theeffect of wild inary results, using pituitary-derived cell lines, indicate that type and mutant Pit-1 proteins on prolactin gene expression was measured by transient co-transfection analysis. The rePit-1 protein is phosphorylated in vivo (20). 33- and 31-kDa Forms of Pit-1 AreImmunologically Distinct porter gene construct (Fig. 5A)consisted of the distal enand Differ in PrimaryAmino Acid Sequence-To confirm that hancer and proximal promoter regions of the rat prolactin the two forms of Pit-1 synthesized in uiuo differ in amino- gene fused to the structuralsequences of the firefly luciferase 132 a Pit

510 a Pit

Alternative Translation Site Initiation

Usage

gene (22). The effector gene constructs used for Pit-1 expression in eucaryotic cells were identical with those described above. In these experiments, the transfected cells were harvested and separated into nuclear and cytoplasmic fractions that were assayed for Pit-1 content or luciferase activity, respectively. The values obtained for induced luciferase activity were then normalized to Pit-1 proteincontent in the transfected cells (as measured by scanning densitometry)and are presented in Fig. 5. The results indicate that the 33-kDa form of Pit-1, expressed by the MET-2,,, plasmid, is equivalently effective at activating prolactingene expression, as was the wild type mixture of 33- and 31-kDa proteins. However, the 31-kDa protein, expressed from the MET-l,,, plasmid, appears to be slightly more effective in its ability to transactivate the prolactin gene. Thus, it appears that, at least with respect to theprolactin gene, the absence of the 27 most amino-terminal amino acids of Pit-1 primary sequence do not compromise its transcriptional efficacy. DISCUSSION

Based on the results presented here, we conclude that two forms of Pit-1 arise as aconsequence of the alternative usage of translation initiation sites present in Pit-1 mRNA. Each of these forms is expressed in pituitary-derived cell lines, as well as in developing and adult pituitary tissue. They differ inthe presence or absence of a 27-amino acid sequence between thefirstand second methionine residues of the primary amino acid sequence. Alternative translation initiation appears the most likely mechanism leading to the formation of the two species of Pit-1 based on several lines of evidence. The synthesis of either form of Pit-1 could be eliminated in vivo using Pit-1 expression plasmids that contained mutations in the respective methionine codons. More-

12835

over, both species of Pit-1 could be derived from a single RNA in uitro, and either species could be eliminated by mutation of the respective AUG codon. This conclusion is reinforced by our observation, using the 51OaPit antiserum, that the amino-terminal amino acid sequences present in the larger (33-kDa) species are missing from the smaller (31-kDa) species. Although alternative translation initiation is an aspect of Pit-1 expression that has been preserved over the evolution of rodents and humans, we have, as of yet, no indication of how the process is, if at all, regulated. It is possible that the usage of initiation sites is a stochastic process because the relative abundance of the two forms of Pit-1 varied little in the pituitary-derived cell lines analyzed. Moreover, little relative variation was observed over the course of pituitary development. However, we are unable to exclude the possibility that, under some physiological circumstances, initiation site usage is regulated. For instance, there was a substantial difference in the relative abundance of the 33- and 31-kDa Pit-1 species between the transfected kidney-derived CV-1 cells and the panel of pituitary-derived cell lines. This would be consistent with the notion that there are certain specific conditions that modulate the relative usage of MET-1 and MET-2 to initiate Pit-1 translation. Currently, studies arein progress to determine if any of the three distinct cell types that express Pit-1 in the pituitary (13) are restricted in their expression to only one of the two forms of the protein and if the amino-terminal amino acid sequences influence the subcellular localization of Pit-1. Acknowledgments-We are grateful to Dan Drolet, Shannon Strybel, Dr. Alan Wells, and Dr. Ronald Emesonfor careful review of the manuscript and useful discussions. We also thank Chuck Nelson for maintenanceandaidintissuecultureexperiments. We arealso grateful to Dr.Holly Ingraham for providing reagents. REFERENCES 1. Li, S Crenshaw, E. B., 111, Rawson, B. J., Simmons, D. M., Swanson, L.

B.

r"--l MET-1 rmt

MET-2,,

( 1 1 1 1 1 1 1 1 1 1 0

20 40 60 80 100120

140 160 180200

% Wild Type Induction

FIG. 5. Transient expression analysis of Pit-1 translation initiation mutants. Panel A, schematic of prolactin gene reporter construct used in thesestudies. It consistsof the prolactingene distal enhancer proximal promoter fused to thefirefly luciferase gene (22). Also given are coordinates of Pit-1-binding sites within these sequences. Shown in the graph is a comparison of the average foldinduction of the prolactin-luciferase reporter construct in response to wild type ( W T ) or mutant Pit-1 proteins MET-I,,, and MET2,,,. Errorbars, S.E. See"ExperimentalProcedures" for details. CMV, regulatorysequences of human cytomegalovirus immediate early gene enhancer (12).

& Rosenfeld, M. G. (1990) Nature 347,528-533 2. Ingraham, H. A,, Chen, R., Mangalam, H. J., Elsholtz, H. P., Flynn, S. E., Lin, C. R., Simmons, D. M., Swanson, L. & Rosenfeld, M. G. (1988) Cell 55,519-529 3. Bodner, M., Castrillo, J. L., Theill, L. E., Deerinck, T., Ellisman, M. & Karin, M. (1988) Cell 50,267-275 4. KO, H.-S., Fast, P., McBride, W. & Staudt, L. M. (1988) Cell 56, 135-144 5. Clerc, R. G., Corcoran, L. M., LeBowitz, J. H., Baltimore, D. & Sharp, P. A. (1988) Genes & Deu. 2, 1570-1581 6. Scheidereit, C., Cromlish, J..A., Gerster, T., Kawakami, K., Balmaceda, C.G., Currie, R. A. & Roeder, R. G. (1988) Nature 336,552-557 7. Muller, M., Affolter, M., Leupin, W., Otting, G., Wiithrich, K. & Gehring, W. J. (1988) EMBO. J. 7.4299-4304 8. Sturm, R . A,, Das, G. & Herr, W. (1988) Genes & Deu. 2, 1582-1599 9. He, X., Treacy, M. M., Simmons, D. M., Ingraham, H. A,, Swanson, L. W. & Rosenfeld, M. G. (1989) Nature 340,945-948 10. B u r g h , T. R., Finney, M., Coulson, A. & Ruvkun, G. (1989) Nature 3 4 1 , 239-243 11. Johnson, W. A. & Hirsch, J. A. (1990) Nature 342,467-470 12. Ingraham, H. A,, Flynn, S. E., Voss, J. W., Albert, V. R., Kapiloff, M. S., Wilson, L. & Rosenfeld, M. G. (1990) Cell 61,1021-1033 13. Mangalam, H. J., Albert, V. R., Ingraham, H. A., Kapiloff, M., Wilson, L., Nelson, C., Elsholtz, H. & Rosenfeld, M. G. (1989) Genes & Deu. 3,946958 14. Simmons, D. M., Voss, J. W., Ingraham, H.A., Holloway J. M., Broide, R. S., Rosenfeld, M. G. & Swanson, L. W. (1990) Genes & Deu. 4, 695-711 15. Laemmli, U. K. (1970) Nature 227,680-685 16. Chen, C. & Okayama, H. (1987) Mol. Cell. Bid. 7,2745-2752 17. de Wet, J. R., Wood, K. V., de Luca, M., Helinski, D. R. & Subramani, S. (1987) Mol. Cell. Bid. 7,725-737 18. Chen, R., Ingraham, H. A,, Treacy, M. N., Albert, V. A,, Wilson, L. & Rosenfeld, M. G. (1990) Nature 346,583-586 19. Kozak, M. (1987) Nucleic Acids Res. 1 5 , 8125-8148 20. Kapiloff, M. S., Farkash, Y., Wegner, M. & Rosenfeld, M. G. (1991) Science. in press 21. Theill, L. E., Castrillo, J. L., Wu, D. & Karin, M. (1989) Nature 3 4 2 , 945948 22. Crenshaw, E. B., 111, Kalla, K., Simmons, D. W., Swanson, L. W., and Rosenfeld, M. G. (1989) Genes & Deu. 3,959-972