ARTHRITIS & RHEUMATISMVol. 44, No. 10, October 2001, pp 2358–2366© 2001, American College of RheumatologyPublished by Wiley-Liss, Inc.

Characterization of a Prolactin Gene Polymorphism andIts Associations With Systemic Lupus Erythematosus

Adam Stevens, David Ray, Aliya Alansari, Ali Hajeer, Wendy Thomson, Rachelle Donn,William E. R. Ollier, Jane Worthington, and Julian R. E. Davis

Objective. Hyperprolactinemia is associated withsystemic lupus erythematosus (SLE), but the mecha-nism is unknown. Prolactin is expressed in T lympho-cytes and is under the control of an alternative promoterregion. We characterized a G/T single-nucleotide poly-morphism (SNP) at position �1149 of this promoterand assessed its prevalence in patients with SLE.

Methods. Electrophoretic mobility shift assays(EMSAs) were performed to determine DNA proteincomplex formation in the prolactin promoter. Transienttransfection of reporter gene constructs containing theG/T promoter alleles into the Jurkat T cell line wereused to determine transcription activity. Peripheralblood lymphocytes (PBLs) were treated in vitro withphytohemagglutinin (PHA) to determine levels of pro-lactin messenger RNA (mRNA).

Results. EMSAs indicated that binding of aGATA-related transcription factor was altered by theG/T SNP at position �1149. Transient transfectionstudies in Jurkat cells showed that the G allele consis-tently produced higher promoter activity. PHA treat-ment of PBLs in vitro induced a greater increment ofprolactin mRNA from patients with the GG�1149 geno-type than from those with the TT�1149 genotype. Diseaseassociation studies in a cohort of SLE patients demon-strated an increased frequency of the prolactin �1149 Gallele compared with control subjects.

Conclusion. We found a functionally significantpolymorphism that alters prolactin promoter activity

and mRNA levels in the lymphocytes. Altered localprolactin production by immune cells may contribute todisease progression by affecting T cell function.

Prolactin is a 23-kd protein hormone that isproduced by the anterior pituitary gland as well as atvarious extrapituitary sites, including T lymphocytes. Itinduces lactation and has a wide variety of actions onimmune cell types (1,2). Since the prolactin receptor is amember of the cytokine receptor superfamily, it ispossible that prolactin plays a multifunctional cytokinerole in addition to its endocrine actions (3).

Prolactin is synthesized and secreted by humanperipheral blood mononuclear cells (PBMCs) and hasbeen proposed to have autocrine effects, for example asa growth factor for lymphoproliferation that is inhibitedby antiprolactin antisera (4). Prolactin has been shownto augment the release of interleukin-2 (IL-2), increaseIL-2 receptor expression and interferon-� levels in lym-phocytes, and potentiate IL-2–induced proliferation ofnatural killer cells (5,6). Recently, prolactin has alsobeen shown to have an effect on dendritic cell matura-tion in synergy with granulocyte–macrophage colony-stimulating factor (7). Prolactin is also associated withincreased antibody production by human lymphocytes(8), and in mice, hyperprolactinemia increases circulat-ing levels of immunoglobulins (9). These data have ledto the hypothesis that prolactin may be implicated in thepathogenesis or progression of a number of autoimmunediseases, notably, rheumatoid arthritis (10–13) and sys-temic lupus erythematosus (SLE) (9,14–18).

The effects of prolactin and bromocriptine onSLE have been tested in the NZB/NZW mouse model ofthe disease. Female mice made chronically hyperpro-lactinemic by transplantation of syngeneic pituitaryglands under the renal capsule had elevated serum IgGand premature mortality from autoimmune renal dam-age. Mice that received bromocriptine, which was ad-

Supported by the Arthritis Research Council. Dr. Ray is asenior Glaxo Wellcome Clinical Research Fellow.

Adam Stevens, PhD, David Ray, MB, PhD, Aliya Alansari,PhD, Ali Hajeer, PhD, Wendy Thomson, PhD, Rachelle Donn, MB,PhD, William E. R. Ollier, PhD, Jane Worthington, PhD, Julian R. E.Davis, PhD: University of Manchester, Manchester, UK.

Address correspondence and reprint requests to Julian R. E.Davis, MD, Endocrine Sciences Research Group, Stopford Building,University of Manchester, Oxford Road, Manchester M13 9PT, UK.

Submitted for publication February 9, 2001; accepted inrevised form May 10, 2001.

2358

ministered to suppress pituitary prolactin production,had delayed elevation of anti-DNA antibodies and se-rum IgG and an increased lifespan (9,15,16). It has alsobeen demonstrated in these mice that treatment withrecombinant prolactin can exacerbate SLE symptoms,and that treatment with a recombinant prolactin antag-onist can ameliorate the syndrome and prevent renaldamage (19).

In humans, prolactin concentrations have beenfound to be elevated in some patients with SLE. Studiescomparing serum prolactin levels in patients with SLEwith those in control subjects have consistently shownthat prolactin is elevated in the SLE patients (14,20–22).Studies controlled for the presence of other identifiablecauses of hyperprolactinemia (e.g., pregnancy) haveshown increased disease activity (23), and studies thathave failed to demonstrate an association between SLEand prolactin levels have been shown to have insufficientstatistical power (17). In a small open-label trial in whichbromocriptine was given to suppress circulating prolac-tin levels, significant lowering of disease activity wasachieved: after bromocriptine therapy was stopped, 5 ofthe 7 patients became hyperprolactinemic, and manifes-tations of disease increased in all patients (24).

Given that a link has been established betweenprolactin and SLE, it remains unclear whether this linkrelates to pituitary-derived prolactin, as is usually as-sumed, or to prolactin that is produced locally in peri-pheral tissues by activated T lymphocytes. Synthesis ofauthentic prolactin by human circulating T cells, to-gether with expression of the prolactin receptor in avariety of lymphoid cell types, has suggested an impor-tant autocrine or paracrine role for the hormone(5,25,26). Indeed, PBMCs from SLE patients have beenreported to secrete higher amounts of a prolactin-likeprotein than PBMCs from control subjects (18). In thepituitary gland, a 5,000-bp region of 5�-flanking DNAdirects human prolactin gene expression, with multiplebinding sites for the pituitary-specific transcription fac-tor Pit-1 (27–29). In contrast, prolactin production inhuman lymphocytes is regulated by an alternative pro-moter region that lies 5,840 bp upstream of the pituitary-specific start site of transcription. This region is inde-pendent of Pit-1, binds different transcription factors, isunresponsive to dopamine, and is regulated by differentagents (30–32). This alternative promoter also directsprolactin transcription in other extrapituitary tissues,such as the endometrium (31).

The human prolactin gene is situated on the shortarm of chromosome 6, close to the major histocompat-ibility complex (MHC) (33). Previous studies have found

linkage disequilibrium between MHC genes known to beassociated with rheumatoid arthritis and SLE (34),raising the possibility that polymorphisms in the up-stream regulatory regions of the human prolactin gene,which perhaps determine levels of lymphocyte-specifictranscription, might be associated with disease. Thephenomenon of functional promoter region polymor-phisms has been observed in other cytokine genes. In theIL-10 gene, for example, the interactions of 3 differentsingle-nucleotide polymorphisms (SNPs) within the pro-moter region have been examined. In this case, highlytranscriptionally active haplotypes have been definedand have been found to be associated with SLE (35).

In the present study, we carried out a compre-hensive functional characterization of a previously de-fined prolactin promoter region polymorphism (Gen-Bank accession no. AF068856) (36). We report thepresence of a functionally significant T3G�1149 poly-morphism in the prolactin gene upstream promoterregion that is overrepresented in patients with SLE andthat increases prolactin gene transcription activity in Tlymphocytes. The polymorphism alters the pattern ofnuclear protein binding, and the homozygous GG geno-type is associated with higher levels of prolactin messen-ger RNA (mRNA) in circulating leukocytes. The dataare consistent with the hypothesis that increased produc-tion of prolactin by T lymphocytes affects disease activityin patients with SLE.

PATIENTS AND METHODS

Potential transcription factor binding sites. To analyzeDNA sequences for potential transcription factor binding sites,the TFSEARCH program was used (available online at http://molsun1.cbrc.aist.go.jp/research/db/TFSEARCH.html). Thisprogram relies on the TRANSFAC database of transcriptionfactors (37) to create a profile of all potential DNA bindingfactors that may bind to a specific sequence.

Electrophoretic mobility shift assays (EMSAs).Double-stranded 27-mer oligonucleotides representing con-sensus GATA-3 binding sequence and mutant sequence ascompetitors were obtained from Santa Cruz Biotechnology(Santa Cruz, CA). Oligonucleotides were also synthesized(Operon Technologies, Newcastle, UK) to represent a 27-bpregion of the human prolactin gene extrapituitary promoter(�1135/�1161) containing the polymorphic site at position�1149; these oligonucleotides were annealed to form double-stranded probes (G allele 5�-AAAGGAGGAAAGAGAATT-TTATGGAGT-3� and T allele 5�-AAAGGAGGAAAGAT-AATTTTATGGAGT-3�). Radiolabeling was performed usingT4 polynucleotide kinase (Gibco BRL, Paisley, UK) and�32P-ATP (�10 mCi/ml; Amersham Pharmacia Biotech, LittleChalfont, UK). Labeled oligonucleotides were column-purified (“Nick” spin columns, Amersham Pharmacia Bio-tech).

PROLACTIN GENE POLYMORPHISM IN SLE 2359

EMSA reactions were carried out using �32P-labeledoligonucleotide probe as described previously by Lee et al (38).Briefly, 1.0 �l of probe (100,000 counts per minute/�l) wasused in 25 �l of buffer (10 mM HEPES, pH 7.9, 1 mM EDTA,100 mM KCl, 10% glycerol, 10 mM dithiothreitol, and 1 �g/�lpoly[dI-dC]) with or without Jurkat cell nuclear protein (5 �g)and competitor DNA (7 pmoles). Electrophoresis was per-formed using a 4% acrylamide gel (37.5:1 ratio of acrylamide:bis-acrylamide) in 0.5� Tris–borate–EDTA buffer.

Generation of luciferase reporter gene constructs.Specific primers were designed to generate a human prolactingene extrapituitary promoter fragment with restriction enzymerecognition sites added (Nhe I forward primer and Bam HIreverse primer; forward primer 5�-GCTAGCCTTAGAGACT-GAAATGGC-3� for the �1824 to �1842 extrapituitary pro-moter region and reverse primer 5�-GGATCCTCTGTCTT-TGAGGGTACT-3� for the �18 to �35 exon 1a). The poly-merase chain reaction (PCR) product was generated fromcontrol DNA samples of known genotype (36) using theAdvantage HF-2 Taq polymerase kit (Clontech, Palo Alto,CA). The PCR product was initially inserted into the TOPO2.1 TA cloning vector (Invitrogen, Groningen, The Nether-lands) according to the manufacturer’s instructions and theninto the Nhe I and Bgl II sites of the pGL3 enhancer reporterplasmid (Promega, Madison, WI). All constructs were se-quenced to confirm the presence of the polymorphism.

Transient transfection of Jurkat cells. Jurkat E6.1(ECACC 88042803), a T lymphoblastoid cell line (39), wasmaintained in RPMI 1640 supplemented with 10% fetal calfserum, 300 mg/liter of L-glutamine, 1% of a 0.1M solution ofsodium pyruvate, and 1% nonessential amino acids (all fromGibco BRL).

Jurkat cells were transiently transfected as describedby Berwaer et al (30). Briefly, 5 � 107 cells in 1 ml ofserum-free medium were placed in a 0.4-cm path-lengthcuvette (Bio-Rad, Hercules, CA) with 10 �g of reporterplasmid and 2 �g of CMV-�-gal plasmid. Cells were electro-porated at 325V, 1,800 microfarads (Easyject One electropo-rator; EquiBio, Liege, Belgium). The transfected cells wereresuspended in growth medium, incubated overnight, platedinto 6-well plates (2.5 � 106 cells/ml of growth medium), andthen incubated with various treatments (as indicated below).

Luciferase activity in cell lysates was detected using aluminometer (LB 9501; Berthold, Wildbad, Germany), and�-galactosidase activity (as a control for transfection effi-ciency) was measured as previously described (40), using aSpectraMAX 250 96-well plate reader (Molecular Devices,Sunnyvale, CA).

Primary culture of peripheral blood leukocytes. Sam-ples of peripheral blood were diluted 1:2 with RPMI 1640(with L-glutamine and 10 units/ml of heparin; Gibco BRL).PBMCs were isolated using Lymphoprep (Amersham Pharma-cia Biotech) and then resuspended in growth medium (RPMI1640 with 300 mg/liter of L-glutamine supplemented with 10%fetal calf serum, 1% of a 0.1M solution of sodium pyruvate, 1%nonessential amino acids, and 1% penicillin/streptomycin mix[10,000 units of each/ml]).

PBMCs were grown in 6-well plates at a density of3.0 � 106 cells in 2 ml of growth medium per well. Three wellswere unstimulated, and 3 were stimulated by the addition ofphytohemagglutinin (PHA; final concentration 5 �g/ml). The

plates were then incubated at 37°C in a CO2 incubator for 72hours, and total RNA was extracted using a commercial kit(Qiagen, Hilden, Germany) according to the manufacturer’sinstructions.

Reverse transcription–PCR (RT-PCR). To quantifyprolactin mRNA in cultured peripheral blood lymphocytes(PBLs), a competitive RT-PCR protocol was used with aninternal standard. The DNA competitor was coamplified withthe same primers as the endogenous target sequence and wasslightly larger than the target fragment to allow accuratecomparison. Specific prolactin primers were located in exons 2and 4 to amplify complementary DNA (cDNA; forward primer5�-CTCTCCTCAGAAATGTTCAGC-3�[5F] and reverseprimer 5�-GGTTTGCTCCTCAATCTCTAC-3�[5R]).

The competitor DNA fragment was designed with theLacZ gene as the “backbone” of the fragment. Primers con-sisting of LacZ-specific oligonucleotides contiguous with theprolactin-specific primers were used (forward 5�-CTC-TCCTCAGAAATGTTCAGCCATAAAGAAAGGCCCGG-CGC-3� and reverse 5�-GGTTTGCTCCTCAATCTCTACC-GAAACACCACGGTAGGCTGCG-3�). The amplification ofthis backbone was performed using AmpliTaq Gold (PerkinElmer, Cheshire, UK) under the following conditions: 95°C for20 minutes, then 36 cycles of 95°C for 45 seconds, 60°C for 30seconds, and 72°C for 1 minute, followed by a final extensionstage of 72°C for 5 minutes (PTC-225 thermal cycler; MJResearch, Waltham, MA).

RT-PCR was performed as a 2-stage procedure. First,400 ng of the total RNA was reverse transcribed using theOmniscript Reverse Transcriptase kit (Qiagen) and poly(dT)primers. Five microliters of the resulting reaction was used asthe template for specific PCR with the specific primers and aknown amount of competitor fragment. RNase inhibitor(Gibco BRL) was added to prevent RNA degradation.

The second-stage specific RT-PCR reactions were setup using the 5F/5R primers described above and BioTaq PCRreagents (Bioline, London, UK) using 200 ng of DNA in a50-�l reaction containing 1� PCR buffer (10� buffer: 160 mMammonium sulfate, 670 mM Tris HCl, pH 8.8, 0.1% Tween 20)and final concentrations of 0.4 �M of each primer, 2 mM ofeach dNTP, 1.5 mM magnesium chloride, 2 units of Taqpolymerase, 1M betaine, and 4 ng of cDNA competitor frag-ment. The PCR conditions were as follows; 95°C for 5 minutes,then 36 cycles of 95°C for 45 seconds, 58°C for 45 seconds, and72°C for 45 seconds.

The PCR reactions were visualized on an ethidiumbromide–stained 2% agarose gel. Using spot densitometry(Alpha Imager 200; Alpha Innotech, San Leandro, CA), theintensity of bands was calculated relative to the local gelbackground, and the ratio of the intensity of specific andcompetitor fragments could be calculated. This ratio has beenpreviously demonstrated to be in direct proportion to therelative amount of starting mRNA (41).

SLE association studies. A cohort of Caucasoid SLEpatients (n � 143) from the north west of England whosatisfied the 1982 American College of Rheumatology criteriafor SLE (42) were studied. The patients were consecutiveattendees at Hope Hospital (Salford, UK) and ManchesterRoyal Infirmary (Manchester, UK). Extensive clinical andimmunologic details were recorded for this patient group,including dermatologic features, renal involvement, and anti-

2360 STEVENS ET AL

nuclear antibody status (43). This disease group has a male-to-female ratio of 1:9.6, a range of ages at disease onset of10–73 years, and a mean � SD age of 37.2 � 15.1 years.

Genotyping for the �1149 extrapituitary promoterregion polymorphism was performed using PCR–restrictionfragment length polymorphism with Apo I digestion. Diseaseassociation was assessed by comparison of the findings with apanel of healthy UK Caucasoid control subjects (n � 394), asdescribed previously (36,44).

Statistical analysis. All reporter gene and RT-PCRdata points represent the mean � SEM of 3 independentexperiments. Statistical significance was analyzed usingANOVA and t-tests as appropriate. Analysis of genotypefrequencies was carried out using a chi-square test with 2degrees of freedom. Analysis of the presence of the G allelewas performed using a chi-square test with 1 degree offreedom and odds ratios (OR) with 95% confidence intervals(95% CI).

RESULTS

Potential transcription factor binding sites. Thepolymorphic region of the prolactin extrapituitary pro-moter was examined for homologies with the consensussequences of binding sites for known transcription fac-tors. A G/T polymorphism at position �1149 in theextrapituitary promoter was previously identified (36), inwhich the T allele displayed complete homology to aconsensus GATA transcription factor binding site. SinceGATA factors are linked with differentiation in thehematopoietic cell lineage, this polymorphism was agood candidate for functional analysis (Figure 1).

Results of EMSAs. EMSAs demonstrated a num-ber of DNA protein complexes forming at the �1135/�1161 region of the prolactin promoter. The �1149 Tallele was demonstrated to generate an additional low-mobility band (complex C in Figure 2) compared with

the G allele. The formation of this band, along withothers in both alleles, was reduced by the addition of theunlabeled specific GATA consensus oligonucleotide butnot by the mutant oligonucleotide.

This result implies that GATA-related proteinsare involved in binding to each allele, but the T allele iscapable of binding an additional GATA-related proteincomplex (Figure 2).

Results of transient transfection experiments.Transient transfection studies in Jurkat cells consistentlyshowed increased basal and maximal alterations in tran-scription activity. The �1149 G allele had significantlyincreased promoter activity compared with the �1149 Tallele (P � 0.002) (Figure 3A).

The �1149 G allele also showed consistentlyhigher promoter activity in response to increasingamounts of chlorophenylthio (cpt) cAMP (P � 0.042)(Figure 3B). In time course studies, the �1149 G allelealso had a greater response to 250 �M cpt cAMPcompared with the T allele (P � 0.017) (Figure 3C).

Results of semiquantitative RT-PCR. For each�1149 extrapituitary promoter region polymorphism,we identified 3 individuals who were homozygous for

Figure 1. DNA sequence of the region surrounding the polymor-phism at position �1149 of the human prolactin extrapituitary pro-moter region (GenBank accession no. AF068856). A, The G allele isnot part of any lymphoid-specific consensus transcription factor bind-ing sites. B, The T allele forms a consensus GATA transcription factorbinding site (boxed area).

Figure 2. Electrophoretic mobility shift assay for prolactin variant.Unstimulated or phorol myristate acetate (PMA)–stimulated Jurkatnuclear extract was incubated with labeled G or T variant prolactinoligonucleotides, with or without the unlabeled competitor oligonucleo-tide shown. Five DNA protein complexes were resolved (A–E), as wellas the unbound probe (F). Complexes A and B are reduced byconsensus GATA oligonucleotides with both alleles. Complex C ispresent only in the T allele, where it is competed. Complexes D and Eare present in both alleles, but are not competed.

PROLACTIN GENE POLYMORPHISM IN SLE 2361

each allelic variant. The PBLs from these subjects wereseparated and treated as described above. Three repli-cates of each treatment were performed for each subjecton different days. The ratio of specific-to-competitorbands was analyzed upon completion of RT-PCR, and amean value was calculated for each subject. The datafrom the 3 individuals of a single genotype were com-bined, and the mean � SEM was calculated. ProlactinmRNA levels in PBLs were increased 3.1-fold by PHAstimulation in samples from the GG homozygotes, but

only 1.2-fold in the samples from the TT homozygotes (P� 0.03) (Figure 4).

Genetic association with autoimmunity. In theUK cohort of SLE patients, there was a significantassociation with the �1149 extrapituitary promoterpolymorphism genotype compared with the controlsubjects (�2 � 6.11 with 2 degrees of freedom, P �0.047). Specifically, the presence of the G allele wasstrongly associated with SLE (�2 � 2.51, P � 0.016,OR 2.51 [95% 1.14–6.28]). The numbers of patients

Figure 3. Luciferase reporter gene analysis of allelic variants in Jurkatcells. A, Basal difference between the G and T alleles. Jurkat cells weretransfected with G or T reporter constructs and harvested after 6 hourswithout treatment (P � 0.002). B, Dose response to chlorophenylthio(cpt) cAMP. Jurkat cells were transfected with G or T reporterconstructs and harvested after 6 hours of treatment. The dose re-sponses were significantly different (P � 0.042). C, Time course ofresponse to 250 �M cpt cAMP. Jurkat cells were transfected with G orT reporter constructs and harvested after the incubation times shown.The time courses were significantly different (P � 0.017). RLU �relative light units (corrected).

Figure 4. Comparison of homozygote normal control subjects, usingsemiquantitative reverse transcription–polymerase chain reaction(RT-PCR) techniques. The RT-PCR competitor fragment is at 365 bp.The prolactin-specific fragment is at 276 bp. A, Peripheral bloodlymphocytes (PBLs) from a GG homozygote were incubated for 72hours with or without phytohemagglutinin (PHA). Total RNA wasthen separated from the samples and subjected to RT-PCR. B, PBLsfrom a TT homozygote were incubated and analyzed as in A andshowed little difference in prolactin mRNA before and after PHAstimulation. C, Analysis of 3 GG and 3 TT homozygous subjects, eachof whom was evaluated on 3 separate occasions. Induction of a specificprolactin PCR band was calculated using ethidium bromide fluores-cence and captured using a charge-coupled device with image analysissoftware. The GG homozygous PBLs showed a higher induction ofprolactin mRNA after PHA stimulation.

2362 STEVENS ET AL

and controls that were available were sufficient to allowa power of 94% to be assured using 1-tailed analysis(Table 1).

DISCUSSION

For many years, prolactin has been considered toplay a role in immune function, and numerous in vitroand in vivo studies have indicated that prolactin haseffects on lymphocyte proliferation and function. Al-though knockout studies of the mouse prolactin receptorand the prolactin gene have not demonstrated an obvi-ous immune phenotype (45–47), it is likely that consid-erable redundancy among diverse cytokines may compli-cate the interpretation of the results of theseexperiments. Indeed, recent clinical studies of recombi-nant prolactin as an immune stimulatory drug haveindicated many clear effects in humans that may haveclinical applications (48). There is therefore clear evi-dence to support further study of prolactin as a potentialproinflammatory factor in autoimmune disease.

In a clinical context, a variety of reports havesuggested that prolactin is involved in the pathogenesisor progression of SLE. This has become particularlyclear with the observations that recombinant prolactinantagonists can ameliorate SLE-like renal damage inNZB/NZW mice (19) and that PBLs from humans withSLE produce significantly more prolactin-like proteinthan PBLs from controls (18). Prolactin is likely to bemost significant here as a locally acting cytokine, andthat led us in the present experiments to address itsregulation in lymphocytes, rather than pituitary cells.The human prolactin gene is expressed in lymphocytesunder the control of an alternative upstream promoterthat is distinct from the proximal promoter that directs

transcription in pituitary lactotrophs (30,31). Only lim-ited information is available concerning the detail of itstranscriptional regulation in immune cells (33), but thefinding of a SNP at �1149 bp in the lymphoid-specificpromoter region was a strong candidate for further studybecause it altered a potential transcription factor bind-ing site.

Based on sequence alignments, the most likelytranscription factor to bind to this region of the prolactinextrapituitary promoter region is a member of theGATA family. In T lymphocytes, the most abundantGATA member is GATA-3 (49). This transcriptionfactor is involved with Th1/Th2 switching in the immunesystem, and GATA-3 expression has been linked withasthma (49,50). This may be because of its ability to bindto the IL-5, IL-10, and IL-8 cytokine gene promoters tostimulate gene transcription and, thus, act in a Th2-stimulatory manner (38,49). GATA-3 has also beenshown to bind to the promoter region of the interferon-�gene to inhibit its transcription, thus acting in a Th1-inhibitory manner (51). Any altered GATA-3 binding tothe prolactin promoter sequences may affect transcrip-tion of the gene in immune cells, with concomitantmodulation of immune function, particularly in relationto Th1/Th2 switching. Our data from the EMSA exper-iments confirm that a GATA-related factor binds to thisregion of the prolactin gene and indicate that thepolymorphism alters protein-binding characteristics. Weidentified multiple retarded complexes using EMSA,which suggests that additional factors may be involvedwith GATA binding in the 2 alleles. It remains to beestablished which members of the GATA family offactors are involved in binding to the �1149 element andwhich additional factors are also involved.

Studies of transient expression in the Jurkat cellline, using the prolactin promoter linked to a reportergene, indicated that the prolactin �1149 G variantgenerated higher basal and stimulated transcriptionactivity than the prolactin �1149 T variant. Thus, al-though the magnitude of activation by cAMP was notaffected, it is likely that the G variant stimulates overallhigher levels of prolactin gene expression in T lympho-cytes.

To investigate whether the promoter region ge-notype might alter the expression of the endogenousprolactin gene, in the context of native chromatin,mRNA levels were measured in normal circulatinghuman PBLs. Healthy control subjects with the homozy-gote GG genotype displayed a greater induction ofprolactin mRNA in response to T cell activation by PHAthan those with a TT genotype. The fact that studies of

Table 1. Genetic analysis of �1149 polymorphism in SLE patientsand control subjects

�1149 G/T genotype frequency

Control subjects SLE patients

GenotypeGG 0.388 0.448*GT 0.482 0.497*TT 0.129 0.056*

G allelePresent 0.871 0.945†Absent 0.129 0.056†

* P � 0.047 versus controls for comparison of genotypes.† P � 0.016, odds ratio 2.51, and 95% confidence interval 1.14–6.28 forcomparison of the presence (GG � GT) versus the absence (TT) ofthe G allele.

PROLACTIN GENE POLYMORPHISM IN SLE 2363

both promoter activity in a cell line model and endoge-nous prolactin mRNA levels yielded concordant resultssupports the hypothesis that prolactin genotype predictsprolactin mRNA levels in normal PBMCs. AlthoughmRNA production reflects a number of regulatoryevents, the fact that the inducibility of prolactin mRNAparallels promoter results in a cell line gives powerfulsupport to the presence of functional differences be-tween the prolactin �1149 polymorphic variants. Thegenotype-dependent regulation of prolactin mRNA byPHA is consistent with the hypothesis that GATA-3 isinvolved in down-regulating Th1 responses, as reportedby Ferber et al (51).

SLE is a complex disease that has a strong geneticcomponent (52,53). Genes within the HLA region arethe only susceptibility factors that have thus far beenconsistently identified, although others have been re-ported, and linkages have been suggested from wholegenome screening (e.g., IL-10, Fc�, and mannose-binding lectin) (35,54–56). In this study, we detected anassociation between a functional polymorphism in thelymphoid promoter of the human prolactin gene andSLE. The presence of the “high-producer” G allele wasmore frequent in SLE patients compared with thehealthy controls. The G allele is associated with in-creased prolactin transcription rates, perhaps as a resultof reduced binding of GATA-related protein complexes.Although the association is relatively weak, this can beexpected when multiple genes interact. In the presentstudy, power calculations confirmed that the availablecontrols and patients enabled a highly significant OR of2.5 to be found in SLE patients versus controls and with�90% power.

We propose that the �1149 extrapituitary pro-lactin polymorphism may contribute to the etiology ofSLE. PBLs have been demonstrated to possess theprolactin receptor (57), and there is some evidence tosuggest that B cells possess greater numbers of prolactinreceptors than do T cells (58). The putative GATA-related factor binding site formed by the T�1149 allele isassociated with lower levels of prolactin transcriptionupon lymphocyte activation. The higher levels of prolac-tin transcription that are associated with the G�1149

allele may well contribute to B cell activation andantibody production. The pathogenesis of SLE is medi-ated by immune complex formation, and therefore, anyfactor that may increase antibody production may alsoaugment the progression and pathology of the disease.Moreover, it has recently been demonstrated that highlevels of prolactin favor the breakdown of B cell toler-ance in mice, an observation that has significance for

SLE in humans (59). The origin of SLE is clearlymultifactorial, but we propose that the presence of thecommon G�1149 allele may contribute to diseaseseverity.

In summary, this is the first study to indicate thata frequent polymorphism in the human prolactin gene isfunctionally important in lymphocytes and is associatedwith a major autoimmune disease. The data provide apossible molecular mechanism for the previously ob-served relationship between prolactin and the develop-ment or progression of SLE in humans. The modulationof prolactin production by lymphocytes may be a thera-peutic target for SLE, and the present data support thecase for considering prolactin antagonists in its treat-ment.

ACKNOWLEDGMENTS

We are grateful to Jo Soden (ESRG) and DavidCarthy (ARC ERU) for excellent technical assistance.

REFERENCES

1. Sinha YN. Structural variants of prolactin: occurrence and physi-ological significance. Endocr Rev 1995;16:354–69.

2. Ben-Jonathan N, Mershon JL, Allen DL, Steinmetz RW. Extra-pituitary prolactin: distribution, regulation, functions and clinicalaspects. Endocr Rev 1996;17:639–69.

3. Velkeniers B, Dogusan Z, Naessens F, Hooghe R, Hooghe-PetersEL. Prolactin, growth hormone and the immune system in humansCell Mol Life Sci 1998;54:1102–8.

4. Sabharwal P, Glaser R, Lafuse W, Varma S, Lui Q, Arkins S, et al.Prolactin synthesized and secreted by human peripheral bloodmononuclear cells: an autocrine growth factor for lymphoprolif-eration. Proc Natl Acad Sci U S A 1992;89:7713–6.

5. Matera L. Action of pituitary and lymphocyte prolactin. Neuro-immunomodulation 1997;4:171–80.

6. Cesario TC, Yousefi S, Carandang G, Sadati N, Le J, Vaziri N.Enhanced yields of gamma interferon in prolactin treated humanperipheral blood mononuclear cells. Proc Soc Exp Biol Med1994;205:89–95.

7. Matera L, Galetto A, Geuna M, Vekemans K, Ricotti E, ContariniM, et al. Individual and combined effect of granulocyte-macrophagecolony-stimulating factor and prolactin on maturation of dendriticcells from blood monocytes under serum-free conditions. Immu-nology 2000;100:29–36.

8. Lahat N, Miller A, Shtiller R, Touby E. Differential effectsof prolactin upon activation and differentiation of human B-lymphocytes. J Neuroimmunol 1993;47:35–40.

9. McMurray R, Keisler D, Kanuckel K, Izui S. Prolactin influencesautoimmune activity in the female B/W mouse. J Immunol 1991;147:3780–7.

10. Berczi I, Nagy E, Asa SL, Kovacs K. The influence of pituitaryhormones on adjuvant arthritis. Arthritis Rheum 1984;27:682–8.

11. Nagy E, Chalmers IM, Barager FD, Friesen HG, Berczi I.Prolactin deficiency in rheumatoid arthritis. J Rheumatol 1991;18:1662–8.

12. Chikanza IC, Panayi GS. Excessive and dysregulated secretion ofprolactin in rheumatoid arthritis: immunopathogenic and thera-peutic implications. Br J Rheumatol 1993;32:445–8.

2364 STEVENS ET AL

13. Nagafuchi H, Suzuki N, Kaneko A, Asai T, Sakane T. Prolactinlocally produced by synovium infiltrating T lymphocytes inducesexcessive synovial cell functions in patients with rheumatoidarthritis. J Rheumatol 1999;26:1890–900.

14. Lavalle C, Loyo E, Paniagua R, Bermudez JA, Herrera J, Graef A,et al. Correlation study between prolactin and androgens in malepatients with systemic lupus erythematosus. J Rheumatol 1987;14:268–72.

15. McMurray R, Keisler D, Izui S, Walker SE. Hyperprolactinaemiain male NZB/NZW (B/W) F1 mice: accelerated autoimmunedisease with normal circulating testosterone. Clin Immunol Immu-nopathol 1994;71:338–43.

16. Walker SE, Allen SH, Hoffman RW, McMurray RW. Prolactin: astimulator of disease activity in systemic lupus erythematosus.Lupus 1995;4:3–9.

17. Blanco-Favella F, Quintal-Alvarez G, Leanos-Miranda A. Associ-ation between prolactin and disease activity in systemic lupuserythematosus: influence of statistical power. J Rheumatol 1999;26:55–9.

18. Larrea F, Martinez-Castillo A, Cabrera V, Alcocer-Varela J,Queipo G, Carino C, et al. A bioactive 60-kilodalton prolactinspecies is preferentially secreted in cultures of mitogen-stimulatedand nonstimulated peripheral blood mononuclear cells from sub-jects with systemic lupus erythematosus. J Clin Endocrinol Metab1997;82:3664–9.

19. Walker SE, Yang L, Kuo CB, Walker AM. A pilot study suggeststhat prolactin shortens, and a prolactin antagonist lengthens, thelifespan of NZB/NZW mice with spontaneous systemic lupuserythematosus [abstract]. Proceedings of the 81st Annual Meetingof the Endocrine Society; 1999 June; San Diego, CA. Bethesda(MD): The Endocrine Society; 1999. p. 2–171.

20. Folomeev M, Prokaeva T, Nassonova V, Nassonov E, Masenko E,Ovtraht N. Prolactin levels in men with SLE and RA. J Rheumatol1990;17:1569–70.

21. Jara-Quezada L, Graef A, Lavalle C. Prolactin and gonadalhormones during pregnancy in systemic lupus erythematosus.J Rheumatol 1992;18:349–53.

22. Jara LJ, Gomez-Sanchez C, Silveira LH, Martinez-Osuna P, VaseyFB, Espinosa LR. Hyperlactinaemia in systemic lupus erythema-tosus: association with disease activity. Am J Med Sci 1992;303:222–6.

23. Pauzner R, Urowitz MB, Gladman DD, Gough JM. Prolactin insystemic lupus erythematosus. J Rheumatol 1994;21:2064–7.

24. McMurray RW, Weidensaul D, Allen SH, Walker SE. Efficacy ofbromocriptine in an open label therapeutic trial for systemic lupuserythematosus. J Rheumatol 1995;22:2084–91.

25. Pellegrini I, Lebrun JJ, Ali S, Kelly PA. Expression of prolactinand its receptor in human lymphoid cells. Mol Endocrinol 1992;6:1023–31.

26. Montgomery DW, Shen GK, Ulrich ED, Steiner LL, Parrish PR,Zukoski CF. Human thymocytes express a prolactin like messen-ger ribonucleic acid and synthesize bioactive prolactin like pro-teins. Endocrinology 1992;131:3019–26.

27. Berwaer M, Monget P, Peers B, Mathy-Hartett M, Bellefroid E,Davis JRE, et al. Multihormonal regulation of the human prolac-tin gene expression from 5000 bp of its upstream sequence. MolCell Endocrinol 1991;80:53–64.

28. Peers B, Voz ML, Monget P, Mathy-Hartert M, Berwaer M,Belayew A, et al. Regulatory elements controlling pituitary-specific expression of the human prolactin gene. Mol Cell Biol1991;10:4690–700.

29. Takasuka N, White MR, Wood CD, Robertson WR, Davis JR.Dynamic changes in prolactin promoter activation in individualliving lactotrophic cells. Endocrinology 1998;139:1361–8.

30. Berwaer M, Martial JA, Davis JR. Characterization of an up-stream promoter directing extrapituitary expression of the humanprolactin gene. Mol Endocrinol 1994;8:635–42.

31. Gellersen B, Kempf R, Telgmann R, DiMattia G. Nonpituitaryhuman prolactin gene transcription is independent of PIT-1 anddifferentially controlled in lymphocytes and in endometrialstroma. Mol Endocrinol 1994;8:356–73.

32. Reem GH, Ray DW, Davis JRE. The human prolactin geneupstream promoter is regulated in lymphoid cells by activators ofT-cells and by cAMP. J Mol Endocrinol 1999;22:285–92.

33. Evans AM, Petersen JW, Sekhon GS, DeMars R. Mapping ofprolactin and tumor necrosis factor-� genes on human chromo-some 6p using lymphoblastoid cell deletion mutants. Somat CellMol Genet 1989;15:203–13.

34. Brennan P, Hajeer A, Ong KR, Worthington J, John S, ThomsonW, et al. Allelic markers close to prolactin are associated withHLA–DRB1 susceptibility alleles among women with rheumatoidarthritis and systemic lupus erythematosus. Arthritis Rheum 1997;40:1383–6.

35. Lazarus M, Hajeer AH, Turner D, Sinnott P, Worthington J,Ollier WE, et al. Genetic variation in the interleukin 10 genepromoter and systemic lupus erythematosus. J Rheumatol 1997;24:2314–7.

36. Stevens A, Hajeer A, John S, Thomson W, Worthington J, DavisJRE, et al. Identification of polymorphism in the prolactin generegion [abstract]. Eur J Immunogenet 1999;26:77.

37. Heinemeyer T, Wingender E, Reuter I, Hermjakob H, KelAE, Kel OV, et al. Databases on transcriptional regulation:TRANSFAC, TRRD, and COMPEL. Nucleic Acids Res 1998;26:364–70.

38. Lee HJ, O’Garra A, Arai K, Arai N. Characterization of cis-regulatory elements and nuclear factors conferring Th2-specificexpression of the IL-5 gene: a role for the GATA-binding protein.J Immunol 1998;160:2343–52.

39. Weiss A, Wiskocil RL, Stobo JD. The role of T3 surface moleculesin the activation of human T cells: a two-stimulus requirement forIL-1 production reflects events occurring at a pre-translationallevel. J Immunol 1984;133:123–8.

40. Fagart J, Wurtz J-M, Souque A, Hellal-Levy C, Moras D, Rafes-tin-Oblin M-E. Antagonism in the human mineralocorticoid re-ceptor. EMBO J 1998;17:3317–25.

41. Price T, Aitken J, Simpson ER. Relative expression of aromatasecytochrome P450 in human foetal tissues as determined by com-petitive polymerase chain reaction amplification. J Clin Endocri-nol Metab 1992;74:313–20.

42. Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, RothfieldNF, et al. The 1982 revised criteria for the classification of systemiclupus erythematosus. Arthritis Rheum 1982;25:1271–7.

43. Davies EJ, Hillarby MC, Cooper RG, Hay EM, Green JR, Shah S,et al. HLA-DQ, DR and complement C4 variants in systemic lupuserythematosus. Br J Rheumatol 1993;32:870–5.

44. Stevens FRA, Hajeer A, John S, Thomson W, Worthington J,Davis JRE, et al. The Bgl II polymorphism of the human prolactingene lies within intron C and can be detected by PCR/RFLP. EurJ Immunogenet 1999;26:261–3.

45. Horseman ND, Zhao W, Montecino-Rodriguez E, Tanaka M,Nakashima K, Engle SJ, et al. Defective mammopoiesis, butnormal hematopoiesis, in mice with a targeted disruption of theprolactin gene. EMBO J 1997;16:6926–35.

46. Bole-Feysot C, Goffin V, Edery M, Binart N, Kelly PA. Prolactin(PRL) and its receptor: actions, signal transduction pathways andphenotypes observed in PRL receptor knockout mice. Endocr Rev1998;19:225–68.

47. Bouchard B, Ormandy CJ, di Santo JP, Kelly PA. Immune systemdevelopment and function in prolactin receptor-deficient mice.J Immunol 1999;163:576–82.

48. Richards SM, Murphy WJ. Use of human prolactin as a therapeu-tic protein to potentiate immunohematopoietic function. J Neu-roimmunol 2000;109:56–62.

49. Zheng W, Flavell RA. The transcription factor GATA-3 is neces-

PROLACTIN GENE POLYMORPHISM IN SLE 2365

sary and sufficient for the Th2 cytokine gene expression in CD4 Tcells. Cell 1997;89:587–96.

50. Zhang DH, Yang L, Cohn L, Parkyn L, Homer R, Ray P, et al.Inhibition of allergic inflammation in a murine model of asthma byexpression of a dominant-negative mutant of GATA-3. Immunity1999;11:473–82.

51. Ferber IA, Lee HJ, Zonin F, Heath V, Mui A, Arai N, et al.GATA-3 significantly downregulates IFN-� production from de-veloping Th1 cells in addition to inducing IL-4 and IL-5 levels. ClinImmunol 1999;1:134–44.

52. Harley JB, Moser KL, Gaffney PM, Behrens TW. The genetics ofhuman systemic lupus erythematosus. Curr Opin Immunol 1998;10:690–6.

53. Wakeland EK, Wandstrat AE, Liu K, Morel L. Genetic dissection ofsystemic lupus erythematosus. Curr Opin Immunol 1999;11:701–7.

54. Manger K, Repp R, Spriewald BM, Rascu A, Geiger A, WassmuthR, et al. Fc� receptor IIa polymorphism in Caucasian patients withsystemic lupus erythematosus: association with clinical symptoms.Arthritis Rheum 1998;41:1181–9.

55. Garred P, Madsen HO, Halberg P, Petersen J, Kronborg G,Svejgaard A, et al. Mannose-binding lectin polymorphisms andsusceptibility to infection in systemic lupus erythematosus. Arthri-tis Rheum 1999;42:2145–52.

56. Ip WK, Chan SY, Lau CS, Lau YL. Association of systemic lupuserythematosus with promoter polymorphisms of the mannose-binding lectin gene. Arthritis Rheum 1998;41:1663–8.

57. Bresson JL, Jeay S, Gagnerault MC, Kayser C, Beressi N, Wu Z,et al. Growth hormone (GH) and prolactin receptors in humanperipheral blood mononuclear cells: relation with age and GH-binding protein. Endocrinology 1999;140:3203–9.

58. Dardenne M, de Moraes MD, Kelly PA, Gagnerault MC.Prolactin receptor expression in human hematopoietic tissuesanalyzed by flow cytofluorometry. Endocrinology 1994;134:2108–14.

59. Peeva E, Grimaldi C, Spatz L, Diamond B. Bromocriptinerestores tolerance in estrogen-treated mice. J Clin Invest 2000;106:1373–9.

2366 STEVENS ET AL