www.elsevier.com/locate/ycimm

Cellular Immunology 235 (2005) 109–116

Alternative splice variants of the human PD-1 gene

Christian Nielsen a, Line Ohm-Laursen a, Torben Barington a,Steffen Husby b, Søren T. Lillevang a,*

a Department of Clinical Immunology, Odense University Hospital, Odense, Denmarkb Department of Pediatrics, Odense University Hospital, Odense, Denmark

Received 1 December 2004; accepted 28 July 2005Available online 19 September 2005

Abstract

PD-1 is an immunoregulatory receptor expressed on the surface of activated T cells, B cells, and monocytes. We describe fouralternatively spliced PD-1 mRNA transcripts (PD-1Dex2, PD-1Dex3, PD-1Dex2,3, and PD-1Dex2,3,4) in addition to the full lengthisoform. PD-1Dex2 and PD-1Dex3 are generated by alternative splicing where exon 2 (extracellular IgV-like domain) and exon 3(transmembrane domain) respectively are spliced out. PD-1Dex3 is therefore likely to encode a soluble form of PD-1. PD-1Dex2,3 lacks exon 2 and 3. These three variants have unaffected open reading frames. PD-1Dex2,3,4 lacks exon 2, 3, and 4 (intra-cellular domain) and contains a premature stop codon in exon 5. Activation of human PBMCs with anti-CD3 + anti-CD28monoclonal antibodies induces an increased level of each PD-1 transcript. A parallel increase in the expression of PD-1Dex3 andflPD-1 upon activation suggests an important interplay between the putative soluble PD-1 and flPD-1 possibly involved in main-tenance of peripheral self-tolerance and prevention of autoimmunity.� 2005 Elsevier Inc. All rights reserved.

Keywords: PD-1; Alternative splicing; PMBCs; Co-stimulation; Soluble

1. Introduction

Programmed death 1 (PD-1) is a type I transmem-brane protein, structurally belonging to the CD28/CTLA-4 subfamily of the Ig superfamily [1]. Engage-ment of PD-1 with its two ligands, PD-L1 (B7-H1)[2,3] or PD-L2 (B7-DC) [4,5], opposes lymphocyte acti-vation and attenuates cytokine production by recruit-ment of SHP-2 to the phosphorylated tyrosine residuein the cytoplasmatic region [6]. In contrast to the Tcell-restricted expression of other CD28 superfamilymembers, PD-1 is induced on T cells, B cells, and mono-cytes after activation [6], suggesting that PD-1 is acostimulatory receptor family member with uniquefunctions.

0008-8749/$ - see front matter � 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.cellimm.2005.07.007

* Corresponding author. Fax: +45 66127975.E-mail address: lillevang@dadlnet.dk (S.T. Lillevang).

PD-1 is considered to play an important inhibitoryrole in immune responses as PD-1-deficient animalsdevelop autoimmune conditions such as lupus-like glo-merulonephritis and proliferative arthritis in mice witha C57BL/6 background [7] and fatal autoantibody-med-iated dilated cardiomyopathy in BALB/c mice [8]. Be-cause these phenotypes differ markedly compared fromthe phenotypes of CTLA-4-deficient mice, this suggestsdifferential regulatory roles for PD-1 and CTLA-4[9,10].

An important function of PD-1 in maintaining theperipheral self-tolerance and prevention of autoimmu-nity is further supported by the reported association be-tween single nucleotide polymorphisms (SNPs) in thehuman PD-1 gene with susceptibility to systemic lupuserythematosus (SLE) [11], type 1 diabetes [12], rheuma-toid arthritis [13], and the presence of nephropathy inSLE patients [14].

110 C. Nielsen et al. / Cellular Immunology 235 (2005) 109–116

Recently, an alternativeCTLA-4mRNAsplice variantlacking the entire transmembrane-encoding exon hasbeen identified. This alternative mRNA transcript is ex-pressed in non-stimulated T cells, and encodes a solubleform of CTLA-4 (sCTLA-4). sCTLA-4 is translated,secreted and present in human serum, and represents anintact functional receptor for the B7.1 ligands with down-regulatory function like the membrane bound CTLA-4molecule in vitro [15–17]. Both serum protein andmRNAlevels of sCTLA-4 have been associated with the develop-ment of several human autoimmune diseases [16,18,19].

The aim of this work was to search for alternativesplice variants of PD-1 using cDNA from both non-stimulated and stimulated human peripheral bloodmononuclear cells (PBMCs). In the present study, we re-port the identification of four PD-1 splicing variants,one of which is likely to encode a soluble form of thePD-1 receptor, and show that the mRNA expressionof these is increased during activation of PBMCs.

2. Materials and methods

Preliminary cDNA PCR analyses with primers (for-ward 5 0-GCGGCCAGGATGGTTCTTA-3 0 and reverse5 0-TACTCCGTCTGCTCAGGGA-3 0 corresponding topositions 125–143 and 793–811, respectively, of thePD-1 mRNA, GenBank Accession No. NM-005018)designed to amplify the entire coding sequence of thehuman PD-1 mRNA revealed the expression of five dif-ferent transcripts in stimulated human T cells (Fig. 1).Amplification was carried out in a Peltier Thermal Cy-

Fig. 1. Gel electrophoresis of PD-1 cDNA amplification productsfrom human anti-CD3 and anti-CD28 stimulated (48 h) PBMCs.With the primers used in the present study, the sizes of the respectivePCR products were 687 bp for the full-length PD-1 transcript, 531 bpfor PD-1Dex3, 327 bp for PD-1Dex2, 171 bp for PD-1Dex2,3, whilethe band at 136 bp is the amplification of the PD-1Dex2,3,4transcript.

cler using the following specific temperature cycling pro-file: an initial hold at 95 �C for 15 min, 40 cycles of 94 �Cfor 1 min, 59 �C for 1 min, and 72 �C for 1 min; and afinal extension step of 72 �C for 10 min. PCR productswere separated by 2% agarose gel electrophoresis andvisualized by ethidium bromide. The PCR fragmentswere subcloned in the pCR2.1-TOPO (Invitrogen) clon-ing vector (using the manufacturer�s instructions) andsequenced. Sequencing was performed using the ABIPRISM BigDye Terminators v3.1 Cycle SequencingKit (Applied Biosystems) and an ABI Prism 3100Genetic Analyzer (Applied Biosystems).

2.1. Isolation and stimulation of PBMCs

Using Ficoll-Hypaque gradient centrifugation,PBMCs were isolated from freshly made buffy-coatsfrom 11 healthy blood donors. PBMCs were culturedin complete medium (1 · 106 cells/ml) consisting ofRPMI 1640 medium supplemented with 10% heat-inac-tivated AB-serum, 200 lM L-glutamine, 100 U/ml peni-cillin G, and 100 lg/ml streptomycin sulphate(Invitrogen Corporation). Cells were either non-stimu-lated or, to assess the effects of T cell activation, stimu-lated with 1 lg/ml anti-CD3 mAb (clone OKT-3; OrthoBiotech, New Jersey, USA) and 0.05 lg/ml anti-CD28mAb (clone CD28.1; DAKO A/S, Glostrup, Denmark)in 96 wells (Nunc) microtiter plates incubated at 37 �C inan atmosphere with 5% CO2. Anti-CD3 and anti-CD28mimic the binding of the peptide:MHC complex to the Tcell receptor and the binding of the co-stimulatory mol-ecule B7 to CD28, respectively. A fixed number(800,000) of cells were harvested following 0, 5, 24, 48,72, and 96 h of stimulation, respectively, lysed, andstored at �80 �C until mRNA extraction. Control cul-tures received culture medium alone and were harvestedfollowing 96 h of incubation. In addition, cells from fourpersons were stimulated with phytohemagglutinin(PHA).

2.2. Reverse transcription

mRNA extraction was performed using KingFishermL (Thermo Labsystems, Helsinki, Finland). mRNAwas reverse transcribed with the Applied Biosystems re-verse transcriptase using random hexamers (AppliedBiosystems) and the following temperature profile:22 �C for 12 min, 42 �C for 22 min, and 99 �C for 7 min.

2.3. qPCR primer design

To secure a specific amplification of the different splicevariants and to avoid amplification of chromosomalDNA, the different primer sets were designed to annealat the junction of two exons spliced together in the partic-ular splice variant in question. The following primers

C. Nielsen et al. / Cellular Immunology 235 (2005) 109–116 111

were used to specifically amplify the different splice vari-ants: full-length forward 5 0-CTCAGGGTGA CAGA-GAGAAG-3 0 (corresponding to positions 492–511 ofthe PD-1 mRNA, GenBank Accession No. NM-005018) and reverse 5 0-GACACCAACCACCAGGGTTT-3 0 (positions 568–587) primers; the PD-1Dex2forward 5 0-GGTTCTTAGAGAGAAGGGCA-3 0 (posi-tions 136–144 and 505–515) and reverse 5 0-GACACCAACCACCAGGGTTT-3 0 (positions 568–587) primers;the PD-1Dex3 forward 5 0-AGG GT GACAGGGACAATAGG-3 0 (positions 495–504 and 661–670) andreverse 5 0-CCATAGTCCACAGAGAACAC-3 0 (posi-tions 720–739) primers; the PD-1Dex2,3 forward 5 0-TGGTTCTTAGGGACAATAGG-3 0 (positions 135–144and 661–670) and reverse 5 0-TCTTCTCTCGCCACTGGAAA-3 0 (positions 749–768) and the PD-1Dex2,3,4 for-ward 5 0-TGGTTCTTAGAAGGAGGACC-3 0 (posi-tions 135–144 and 696–705) and reverse 5 0-TCTTCTCTCGCCACTGGAAA-3 0 (positions 749–768). Further,to minimize differences in the amplification efficienciesof the amplicons, the length of these were kept between83 and 96 bp, and the theoretical melting temperatureof the different primers between 60 and 62 �C. Due to aprimer dimer problem with the initial primers used forthe amplification of the PCR product representing thePD-1Dex2,3 transcript, the reverse primer had to be locat-ed further downstream. As a result, the PCR product rep-resenting the PD-1Dex2,3 transcript is 118 bp in length.

2.4. Real-time quantitative PCR

Real-time PCR product accumulation was monitoredin the ABI Prism 7700 Sequence Detector (Applied Bio-systems, Foster City, CA, USA) using the intercalatingdye SYBR Green I. The reaction was carried out withthe 1· SYBR Green PCR Master Mix, 3 ll of cDNA,and 5 pmol of each specific primer in a total volume of15 ll. Thermocycling program was 50 cycles of 95 �Cfor 10 min, 95 �C for 15 s, and 60 �C for 1 min with aninitial cycle of 50 �C for 2 min. All amplifications anddetections were carried out in capped 96-well opticalplates (Applied Biosystems).

The quantitative real-time PCR method used in thisstudy is based on determining the threshold cycle (CT)at which the amplified target increases above a setthreshold level. The CT value therefore correlates nega-tively to the amount of target mRNA, i.e., the higher theamount of mRNA, the sooner the threshold is reachedand the lower the CT value.

Each sample was tested in triplicate to check forreproducibility. To control for genomic DNA amplifica-tion, no-reverse-transcription controls (NRT) were runin duplicates for each sample.

To check for any significant levels of contaminants andto exclude false positives, two non-template controls wereincluded for each primer pair with every amplification.

The ‘‘mRNA Induction index’’ of the different splicevariants was calculated using the formula 2�DCT , whereDCT is the CT value at the stimulation time in questionminus the CT value at 0 h (non-stimulated cells).

2.5. PCR amplification efficiency and reproducibility

To evaluate whether the five PCR products represent-ing the differently spliced transcripts were amplified withthe same efficiency during the real time PCR, the rela-tion between the number of input cells and the CT valuewas tested for each PCR product in the range from25,000 to 6,400,000 cells originating from the samecell-culture stimulated for 72 h. The slope is a measureof the PCR amplification efficiency of the PCR productin question.

To assess the reproducibility the procedure of mRNAextraction, reverse transcription and real-time PCR wererepeated in 5 replicates of 800,000 cells originating fromthe same cell-culture stimulated for 72 h. Means andstandard deviations were obtained to calculate the coef-ficient of variation (CV).

2.6. SDS–PAGE, native PAGE, and Western blot

analysis

The production of putative soluble PD-1 was assessedby Western blotting in the cell-free supernatants of bothstimulated and non-stimulated cells. Three differentcommercial antibodies against PD-1: goat anti-humanPD-1 (extracellular domain, 0.15 lg/ml, R&D Systems),goat anti-human PD-1 (peptide mapping at the aminoterminus of human PD-1, diluted 100-fold, Santa CruzBiotechnology), and goat anti-human PD-1 (peptidemapping near the carboxy terminus of human PD-1,diluted 100-fold, Santa Cruz Biotechnology), respective-ly, were used. No positive controls were available to beincluded. The detection limit for the used PD-1 antibod-ies is reportedly approximately 2 ng/ll.

2.7. Statistical analysis

Data are presented as means ± SD. Multiple compar-isons among slopes using Tukey�s HSD test was used tocompare the amplification efficiencies (slopes) of the fivePCR products, representing the differently splicedtranscripts.

ThemRNAexpression of thePD-1 full-length (flPD-1)and PD-1Dex3 transcripts was evaluated using two-wayANOVA followed by Bonferroni adjusted Fisher�sLSD-test as the multiple comparison procedure (Systat7_0, 1997 by SPSS, Chicago, IL, USA).

The mRNA Induction index of the flPD-1 andPD-1Dex3 transcripts at 5, 24, 48, 72, and 96 h ofstimulation, respectively, was evaluated using two-wayANOVA followed by Bonferroni adjusted Fisher�s

112 C. Nielsen et al. / Cellular Immunology 235 (2005) 109–116

LSD-test as the multiple comparison procedure. Datawere (log + 1) transformed to satisfy the ANOVAassumptions of normality and homogeneity of varianc-es. One sample t tests were used to compare the mRNAInduction index of the flPD-1 and PD-1Dex3 transcriptsat time 0 h with the levels at 5, 24, 48, 72, and 96 h ofstimulation, respectively, and the levels in non-stimu-lated cells (96 h). A corrected value of P < 0.05 was con-sidered statistically significant.

3. Results

3.1. Identification of alternatively spliced variants of PD-1mRNA

PCR amplification of the human PD-1 coding se-quence revealed the expression by human PBMCs of fivesplice variants: flPD-1, PD-1Dex2, PD-1Dex3, PD-1Dex2,3, and PD-1Dex2,3,4 (Fig. 2). The flPD-1 tran-script showed complete homology with the publishedmembrane PD-1 sequence (GenBank Accession No.NM_005018), while the PD-1Dex2 and PD-1Dex3 tran-scripts, according to the genomic organization of the hu-man PD-1 gene, are generated by alternative splicing ofthe PD-1 mRNA, where exon 2 and exon 3 are splicedout, respectively. The open reading frames in thesetwo variants are unaffected. The splicing of PD-1Dex2results in a codon coding for glutamic acid at position

exon 1 exon 2 exon 3 ex

exon 1 exon 4 exon exon 3

exon 1 exon 2 exon 4 ex

exon 1

exon 5

STOP (TGA)

E → G

D → E

exon 1

exon 5

E → G

exon 4

Fig. 2. Schematic representation of exon organization in the different PD-1introduced by alternative splicing are marked with arrows.

26 compared to aspartic acid in fl-PD-1, while the splic-ing of PD-1Dex3 introduces a glycine codon at position136 instead of glutamic acid in fl-PD-1. The PD-1Dex2,3transcript lacks exon 2 and 3 has an unaffected openreading frame, and introduces a glycine codon at posi-tion 26 compared to aspartic acid in fl-PD-1. The PD-1Dex2,3,4 transcript lacks exons 2, 3, and 4. Thisalternative splicing variant introduces a frame shiftresulting in a premature stop codon (corresponding topositions 797–799 of the PD-1 mRNA, GenBank Acces-sion No. NM-005018) in exon 5.

3.2. PCR amplification efficiency and reproducibility

The detection and quantification of the five PCRproducts, representing the different splice transcripts,were logarithmic (Fig. 3) over the examined range withthe following equations: flPD-1: CT = �3.28 · log 10input cell number + 48.56, r2 = 0.98; PD-1Dex2:CT = �3.09 · log 10 input cell number + 57.96, r2 =0.89; PD-1Dex3: CT = �3.03 · log 10 input cell num-ber + 46.21, r2 = 0.98; PD-1Dex2,3: CT = �2.58 · log10 input cell number + 45.07, r2 = 0.99; andPD-1Dex2,3,4: CT = �2.46 · log 10 input cell num-ber + 45.38, r2 = 0.95. Further, the slopes were parallelto the ‘‘ideal’’ slope (�3.32) based on the existing rela-tion between CT and template concentration where thethreshold cycle decreases by 1 cycle as the concentrationof template doubles. The slopes for PD-1Dex2,3 and

on 4 exon 5

5

on 5

flPD-1

PD-1∆ex2

PD-1∆ex3

PD-1∆ex2,3,4

PD-1∆ex2,3

splice variants. The amino acid shifts and the premature stop codon

Time (h)

0 20 40 60 80 100 120

Thr

esho

ld c

ycle

(C

)

24

26

28

30

32

34

36

38

40

42

44

46

Time (h)

0 20 40 60 80 100 120

Thr

esh

old

cycl

e (C

T)

24

26

28

30

32

34

36

38

a

b

cd d cd

bc

*

A

B

C C CBC

A

a

A

B

T

Fig. 4. Changes in mRNA expression of flPD-1 (closed circles), PD-1Dex3 (open circles), PD-1Dex2 (closed squares), PD-1Dex2,3 (closeddiamonds), and the PD-1Dex2,3,4 (open squares) transcript in PBMCsstimulated with anti-CD3 and anti-CD28 over time (A), and the flPD-1and PD-1Dex3 transcript alone (B). Closed triangle up, open triangleup, closed triangle down, open diamond, and open triangle downrepresent the mRNA expression of the flPD-1, PD-1Dex3, PD-1Dex2,PD-1Dex2,3, and PD-1D ex2,3,4, respectively, in non-stimulatedPBMCs following 96 h of incubation. Values are means ± SE of 11healthy individuals. Values with shared letters are not significantlydifferent (P > 0.05). Upper and lower case letters are used for the PD-1Dex3 and flPD-1 transcript, respectively; asterisks indicate significantdifferences between mRNA expression of PD-1Dex3 and flPD-1transcript at the sample time in question.

aa

b

ab

a

ab

b

a

ab

b

aa

Log input cell number

4,0 4,5 5,0 5,5 6,0 6,5 7,0

Thr

esho

ld c

ycle

(C

)

20

25

30

35

40

45

50

ab

ba

a

b

T

Fig. 3. Relationship between the input cell number from a cell cultureof PBMCs stimulated with anti-CD3 and anti-CD28 for 72 h, and thethreshold cycle (CT) of the five different PCR products representing thedifferent PD-1 splice variants. flPD-1 (closed circles), PD-1Dex3 (opencircles), PD-1Dex2 (closed squares), PD-1Dex2,3 (open diamonds), andthe PD-1Dex2,3,4 (open squares), together with an ‘‘ideal’’ slope witharbitrary units (open triangle up), based on the existing relationbetween CT and template concentration where the threshold cycledecreases by 1 cycle as the concentration of template doubles. Theequations of the fitted linear regressions, where the slope is a measureof the PCR amplification efficiency of the PCR product in question,are given in Section 3. Data are means ± SD of triplicate amplifica-tions of the same sample. Slopes with shared letters are notsignificantly different (P > 0.05). The arrow marks the input cellnumber used in the present experiment.

C. Nielsen et al. / Cellular Immunology 235 (2005) 109–116 113

PD-1Dex2,3,4 were significantly different from the slopesfor flPD-1 and PD-1Dex3, while the slope for PD-1Dex2did not differ significantly from any of the other slopes.There was a relatively poor logarithmic relation betweenCT and template concentration of the PD-1Dex2transcript, and the expression of this transcript was verylow and near the detection limit (CT above 40). Due tothese observations and the fact that the PD-1Dex3transcript is likely to encode a soluble form of PD-1 sim-ilar to sCTLA-4 [17], we decided to only statisticallycompare the mRNA expression of the flPD-1 andPD-1Dex3 transcripts (Figs. 4B and 5B).

The variability of the mRNA extraction, reverse tran-scription, and real-time PCR was very low (mean CT,SDV, CV (%): 28.34, 0.08, 0.27) in 5 replicates of800,000 stimulated cells (72 h) originating from the samecell culture.

The level of contaminants was low as non-templatecontrols always, except for the PD-1Dex2 transcriptdue to the high CT values, resulted in at least 10–15 cy-cles higher CT values compared to the templates con-taining samples.

3.3. Regulation of PD-1 mRNA in human PBMC

The five PD-1 mRNA splice variants were constitu-tively expressed by non-stimulated PBMCs, with the

flPD-1, PD-1Dex3, PD-1Dex2,3, and PD-1Dex2,3,4apparently being expressed at higher levels than thePD-1Dex2 (Fig. 4A). The expression of the latter tran-script was very low and near the detection limit of theABI Prism 7700 Sequence Detector used in the presentstudy. Stimulation with immobilized anti-CD3 andanti-CD28 resulted in an increased expression of all fivesplice variants in PBMCs.

The PD-1Dex3 mRNA transcript was constitutivelyexpressed by non-stimulated PBMCs at a significantlyhigher level compared with the flPD-1 transcript(Fig. 3B). The expression pattern of the PD-1Dex3

Time (h)

0 20 40 60 80 100 120

mR

NA

Indu

ctio

n in

dex

0

20

40

60

80

100

120

140

160

180

Time (h)

0 20 40 60 80 100 120

mR

NA

Indu

ctio

n in

dex

0

20

40

60

80

100

120

140

160

180A AB

C

D

BC

a

b

** *

*

* *

*

A

B

Fig. 5. Changes in the mRNA Induction index of flPD-1 (closedcircles), PD-1Dex3 (open circles), PD-1Dex2 (closed squares), PD-1Dex2,3 (closed diamonds), and PD-1Dex2,3,4 (open squares) inPBMCs stimulated with anti-CD3 and anti-CD28 over time (A) andthe flPD-1 and PD-1Dex3 transcript alone (B). Closed triangle up,open triangle up, closed triangle down, open diamond, and opentriangle down represent the mRNA Induction index of flPD-1, PD-1Dex3, PD-1Dex2, PD-1Dex2,3, and PD-1Dex2,3,4 in non-stimulatedPBMCs, respectively, following 96 h of incubation. Values are mean-s ± SE of 11 healthy individuals. Upper case letters are used for boththe PD-1Dex3 and flPD-1 transcript. Values with shared letters are notsignificantly different (P > 0.05). The different lower case lettersindicate the significant difference between the mRNA Induction indexof the PD-1Dex3 and flPD-1 transcripts at all time. Asterisks indicatesignificant difference between the mRNA expression in non-stimulatedPBMCs (0 h) and the PD-1Dex3 and flPD-1 expression at the time inquestion.

114 C. Nielsen et al. / Cellular Immunology 235 (2005) 109–116

and flPD-1 transcripts in stimulated PBMCs followedthe same pattern: 5 h of stimulation resulted in a sig-nificantly increased mRNA expression, reaching a pla-teau expression at 24 h lasting to 72 h of stimulation,followed by a decrease in the expression at 96 h ofstimulation. At no time did the expression of thePD-1Dex3 and flPD-1 transcripts differ significantlyin the stimulated PBMCs. There were no significantdifferences in the expression of these transcripts in

non-stimulated cells at 0 h and non-stimulated cellsincubated for 96 h.

The mRNA Induction index of all splice variants in-creased following stimulation and reached a plateau le-vel after 48–72 h of stimulation, with the flPD-1mRNA expression reaching a higher level comparedwith the splice variants (Fig. 5A). At 96 h of stimulation,the mRNA expression of all splice variants wasdecreased.

The mRNA Induction index of the PD-1Dex3 andflPD-1 transcripts in stimulated PBMCs followed thesame pattern with the mRNA Induction index ofthe flPD-1 transcript being significantly higher thanthe mRNA Induction index of PD-1Dex3 at all time(Fig. 5B). The mRNA Induction index of both tran-scripts were significantly elevated after 5 h of stimula-tion and peaked at 48–72 h of stimulation, where themRNA level of the flPD-1 transcript was upregulatedapproximately 106–120 times, while the PD-1Dex3mRNA level was upregulated approximately 25–30times. After 96 h of stimulation, the mRNA Inductionindex of both transcripts significantly decreased butwere still significantly elevated compared to the non-stimulated cells.

3.4. Soluble PD-1 protein

Using Western blotting with three different commer-cial antibodies against human PD-1, we were unableto detect the presence of soluble PD-1 in the cell-freesupernatants of both the stimulated and the non-stimu-lated PBMCs cell cultures (data not shown). Since a po-sitive control was not available, the detecting power ofthe commercial antibodies used could not bedemonstrated.

4. Discussion

To compare mRNA levels (obtained by RT-PCR),the expression of a reference gene is commonly assessedin parallel with the expression of the gene of interest.Fundamental to a reference gene is that its expressionshould be at a constant level among different tissues,at all stages of development and should be unaffectedby the experimental treatment [20]. Commonly, a house-keeping gene is selected, but numerous studies haveshown that these do not necessarily behave as usablereference genes in certain experimental situations andthe most appropriate normalizer should be tailored byindividual experimental conditions [20].

Using the same experimental conditions as in thepresent study, previous experiments in our laboratoryhave revealed that the mRNA expressions of eightfrequently used reference genes including GAPDHand b-actin, fluctuate during stimulation of PBMCs

C. Nielsen et al. / Cellular Immunology 235 (2005) 109–116 115

(unpublished data). The expression level of referencegenes is therefore unreliable for normalizing the quan-titative RT-PCR assay in the present study and hencewe decided to normalize the different splice variantsto the input cell number as suggested by [21]. Further,the mRNA extraction, reverse transcription, and real-time PCR proved to be highly reproducible and thePCR products, representing the flPD-1 and PD-1Dex3splice variants, were amplified with the same efficiency.This justifies the present statistical comparison of theabsolute CT values and the mRNA Induction index ob-tained for these two transcripts.

We have identified four alternative spliced PD-1mRNA transcripts of PD-1 (PD-1Dex2, PD-1Dex3,PD-1Dex2,3, and PD-1Dex2,3,4) in addition to the fulllength isoform encoded by exon 1 (leader peptide), exon2 (extracellular IgV-like domain), exon 3 (transmem-brane domain), exon 4 and 5 (intracellular domain)[22]. The different splice variants are unlikely to repre-sent artifacts amplified during the PCR reactions asthe cellular activation appears to regulate the relativelevels of each PD-1 transcript. Further, the expressionof the different transcripts was induced by both anti-CD3/anti-CD28 and PHA stimulation of PBMCs, withthe full length isoform exhibiting a higher degree ofinduction than the variants.

The PD-1Dex2 mRNA transcript is generated froman alternative spliced transcript of the PD-1 gene wherethe sequence encoded by exon 2 is deleted. This suggeststhat if the PD-1Dex2 mRNA transcript was translated,the putative protein product would be expressed as amembrane molecule lacking its binding properties toPD-L1/2. The splicing does not affect the reading framebut replaces aspartic acid at position 26 with glutamicacid. A biological significance of this mRNA transcriptis unlikely and the level of this transcript was very lowcompared with the other transcripts and near the detec-tion limit, suggesting that this transcript could be the re-sult of a splicing error occurring in parallel with theincreased mRNA expression during the extensive cellproliferation.

The change of reading frame caused by alternativesplicing of exons 2, 3, and 4 in the PD-1Dex2,3,4 splicevariant generates a premature translation–terminationcodon in exon 5. If translated, this putative truncatedprotein product would lack the ligand-binding domain,the transmembrane region, and the cytoplasmatic tailresponsible for the binding, signalling, and scaffoldingmolecules [23]. Even with no obvious biologicalfunction, the constant high expression of this mRNAtranscript in stimulated PBMCs proposes an immuno-logical function. As the introduced premature transla-tion–termination codon in PD-1Dex2,3,4 is located inthe 3 0-terminal exon, this transcript is not candidate tobe subjected to nonsense-mediated mRNA decay(NMD). NMD represents a conserved surveillance

pathway by which mRNAs that carry translation–termi-nation codons specifically are eliminated [24]. Only tran-scripts that contain premature stop codons >50nucleotides 5 0 of the final exon are candidates forNMD [25].

Likewise, despite the elevated expression of the PD-1Dex2,3 mRNA transcript in stimulated PBMCs, noobvious biological function of a putative protein encodedby this transcript is apparent, as this putative truncatedprotein would lack the binding properties to PD-L1/2.The PD-1Dex2,3 transcript has an unaffected open read-ing frame and, if translated, introduces glycine at posi-tion 26 compared to aspartic acid in fl-PD-1.

Of the four alternative splice variants, PD-1Dex3 isparticularly interesting. The reading frame of PD-1Dex3 is unaffected by the alternative splicing, but glu-tamic acid at position 136 is replaced with glycine. Thistranscript lacks the entire membrane-spanning domainof the PD-1 molecule, suggesting that the putative trans-lation product is expressed as a soluble form of the PD-1molecule analogous to sCTLA-4, known to possess adownregulatory function like membrane boundCTLA-4 [17]. In a putative soluble PD-1 encoded byPD-1Dex3, the extracellular region critical to PDL1/2binding would be intact. Theoretically, by bindingPDL1/2 expressed on peripheral APCs, the candidatesoluble PD-1 could interfere with the PD-L1/2:flPD-1interactions, thereby blocking the negative signalimparted via the transmembrane form of PD-1.

As the mRNA expression of the PD-1Dex3 transcriptwas significantly higher compared to the flPD-1 tran-script in non-stimulated PBMCs, this may suggest thatthe putative translated soluble PD-1 protein could playan important role in the early stage of an immuneresponse.

In remains to be demonstrated however, that a corre-sponding protein is secreted. We were not able to dem-onstrate this by Western blotting of culturesupernatants, but with an estimated detection limit ofour system of approximately 20 ng/ll it cannot be ruledout.

The parallel increase in the expression of thePD-1Dex3 and flPD-1 mRNA transcripts in mononucle-ar cells upon activation suggests an important interplaybetween the putative translated soluble PD-1 and flPD-1possibly involved in maintenance of peripheral self-tol-erance and prevention of autoimmunity.

In conclusion, our study shows that non-stimulatedand stimulated human PBMCs express at least fourdifferent PD-1 mRNA variants which could play differ-ent regulatory roles in immune activation. The identi-fication of a PD-1 splice transcript that possiblyencodes a soluble form of PD-1 should encourage fur-ther research considering the apparent important roleof sCTLA-4 in the development of human autoim-mune diseases.

116 C. Nielsen et al. / Cellular Immunology 235 (2005) 109–116

Acknowledgments

This work was supported by grants from a combinedgrant from The Danish Medical Research Council andthe counties of Southern Denmark. The experimentsperformed comply with the current laws of Denmark.

References

[1] A.H. Sharpe, G.J. Freeman, The B7-CD28 superfamily, Nat. Rev.Immunol. 2 (2002) 116–126.

[2] G.J. Freeman, A.J. Long, Y. Iwa, K. Bourque, T. Chernova, H.Nishimura, L.J. Fitz, N. Malenkovich, T. Okazaki, M.C. Byrne,H.F. Horton, L. Fouser, L. Carter, V. Ling, M.R. Bowman, B.M.Carreno, M. Collins, C.R. Wood, T. Honjo, Engagement of thePD-1 immunoinhibitory receptor by a novel B7 family memberleads to negative regulation of lymphocyte activation, J. Exp.Med. 192 (2000) 1027–1034.

[3] H. Dong, G. Zhu, K. Tamada, L. Chen, B7-H1, a third member ofthe B7 family, co-stimulates T-cell proliferation and interleukin-10secretion, Nat. Med. 5 (1999) 1365–1369.

[4] Y. Latchman, C.R. Wood, T. Chernova, D. Chaudhary, M.Borde, I. Chernova, Y. Iwai, A.J. Long, J.A. Brown, R. Nunes,E.A. Greenfield, K. Bourque, V.A. Boussiotis, L.L. Carter, B.M.Carreno, N. Malenkovich, H. Nishimura, T. Okazaki, T. Honjo,A.H. Sharpe, G.J. Freeman, PD-L2 is a second ligand for PD-1and inhibits T cell activation, Nat. Immunol. 2 (2001) 261–268.

[5] S.Y. Tseng, M. Otsuji, K. Gorski, X. Huang, J.E. Slansky, S.I.Pai, A. Shalabi, T. Shin, D.M. Pardoll, H. Tsuchiya, B7-DC, anew dendritic cell molecule with potent costimulatory propertiesfor T cells, J. Exp. Med. 193 (2001) 839–846.

[6] T. Okazaki, A. Maeda, H. Nishimura, T. Kurosaki, T. Honjo,PD-1 immunoreceptor inhibits B cell receptor-mediated signalingby recruiting src homology 2-domain-containing tyrosine phos-phatase 2 to phosphotyrosine, Proc. Natl. Acad. Sci. USA 98(2001) 13866–13871.

[7] H. Nishimura, M. Nose, H. Hiai, N. Minato, T. Honjo,Development of lupus-like autoimmune diseases by disruptionof the PD-1 gene encoding an ITIM motif-carrying immunore-ceptor, Immunity 11 (1999) 141–151.

[8] H. Nishimura, T. Okazaki, Y. Tanaka, K. Nakatani, M. Hara, A.Matsumori, S. Sasayama, A. Mizoguchi, H. Hiai, N. Minato, T.Honjo, Autoimmune dilated cardiomyopathy in PD-1 receptor-deficient mice, Science 291 (2001) 319–322.

[9] E.A. Tivol, F. Borriello, A.N. Schweitzer, W.P. Lynch, J.A.Bluestone, A.H. Sharpe, Loss of CTLA-4 leads to massivelymphoproliferation and fatal multiorgan tissue destruction,revealing a critical negative regulatory role of CTLA-4, Immunity3 (1995) 541–547.

[10] P. Waterhouse, J.M. Penninger, E. Timms, A. Wakeham, A.Shahinian, K.P. Lee, C.B. Thompson, H. Griesser, T.W. Mak,Lymphoproliferative disorders with early lethality in mice defi-cient in Ctla-4, Science 270 (1995) 985–988.

[11] L. Prokunina, C. Castillejo-Lopez, F. Oberg, I. Gunnarsson, L.Berg, V. Magnusson, A.J. Brookes, D. Tentler, H. Kristjansdottir,G. Grondal, A.I. Bolstad, E. Svenungsson, I. Lundberg, G.Sturfelt, A. Jonssen, L. Truedsson, G. Lima, J. Alcocer-Varela, R.Jonsson, U.B. Gyllensten, J.B. Harley, D. Alarcon-Segovia, K.Steinsson, M.E. Alarcon-Riquelme, A regulatory polymorphismin PDCD1 is associated with susceptibility to systemic lupuserythematosus in human, Nat. Genet. 32 (2002) 666–669.

[12] C. Nielsen, D. Hansen, S. Husby, B.B. Jacobsen, S.T. Lillevang,Association of a putative regulatory polymorphism in the PD-1gene with susceptibility to type 1 diabetes, Tissue Antigens 62(2003) 492–497.

[13] S.C. Lin, J.H. Yen, J.J. Tsai, W.C. Tsai, T.T. Ou, H.W. Liu, C.J.Chen, Association of a programmed death 1 gene polymorphismwith the development of rheumatoid arthritis, but not systemiclupus erythematosus, Arthritis Rheum. 50 (2004) 770–775.

[14] C. Nielsen, H. Laustrup, A. Voss, P. Junker, S. Husby, S.T.Lillevang, A putative regulatory polymorphism in PD-1 isassociated with nephropathy in a population-based cohort ofsystemic lupus erythematosus patients, Lupus 13 (2004) 510–516.

[15] G. Magistrelli, P. Jeannin, N. Herbault, A. Benoit De Coignac,J.F. Gauchat, J.Y. Bonnefoy, Y. Delneste, A soluble form ofCTLA-4 generated by alternative splicing is expressed by nonsti-mulated human T cells, Eur. J. Immunol. 29 (1999) 3596–3602.

[16] M.K. Oaks, K.M. Hallett, Cutting edge: a soluble form of CTLA-4 in patients with autoimmune thyroid disease, J. Immunol. 164(2000) 5015–5018.

[17] M.K. Oaks, K.M. Hallett, R.T. Penwell, E.C. Stauber, S.J.Warren, A.J. Tector, A native soluble form of CTLA-4, Cell.Immunol. 201 (2000) 144–153.

[18] M.F. Liu, C.R. Wang, P.C. Chen, L.L. Fung, Increased expres-sion of soluble cytotoxic T-lymphocyte-associated antigen-4molecule in patients with systemic lupus erythematosus, Scand.J. Immunol. 57 (2003) 568–572.

[19] H. Ueda, J.M. Howson, L. Esposito, J. Heward, H. Snook, G.Chamberlain, D.B. Rainbow, K.M. Hunter, A.N. Smith, G. DiGenova, M.H. Herr, I. Dahlman, F. Payne, D. Smyth, C.Lowe, R.C. Twells, S. Howlett, B. Healy, S. Nutland, H.E.Rance, V. Everett, L.J. Smink, A.C. Lam, H.J. Cordell, N.M.Walker, C. Bordin, J. Hulme, C. Motzo, F. Cucca, J.F. Hess,M.L. Metzker, J. Rogers, S. Gregory, A. Allahabadia, R.Nithiyananthan, E. Tuomilehto-Wolf, J. Tuomilehto, P. Bing-ley, K.M. Gillespie, D.E. Undlien, K.S. Ronningen, C. Guja, C.Ionescu-Tirgoviste, D.A. Savage, A.P. Maxwell, D.J. Carson,C.C. Patterson, J.A. Franklyn, D.G. Clayton, L.B. Peterson,L.S. Wicker, J.A. Todd, S.C. Gough, Association of the T-cellregulatory gene CTLA4 with susceptibility to autoimmunedisease, Nature 423 (2003) 506–511.

[20] S.A. Bustin, Absolute quantification of mRNA using real-timereverse transcription polymerase chain reaction assays, J. Mol.Endocrinol. 25 (2000) 169–193.

[21] S.A. Bustin, Quantification of mRNA using real-time reversetranscription PCR (RT-PCR): trends and problems, J. Mol.Endocrinol. 29 (2002) 23–39.

[22] L.R. Finger, J. Pu, R. Wasserman, R. Vibhakar, E. Louie, R.R.Hardy, P.D. Burrows, L.G. Billips, The human PD-1 gene:complete cDNA, genomic organization, and developmentallyregulated expression in B cell progenitors, Gene 197 (1997) 177–187.

[23] X. Zhang, J.C. Schwartz, X. Guo, S. Bhatia, E. Cao, L. Chen,Z.Y. Zhang, M.A. Edidin, S.G. Nathenson, S.C. Almo, Structuraland functional analysis of the costimulatory receptor programmeddeath-1, Immunity 3 (2004) 337–347.

[24] T. Schell, T. Kocher, M. Wilm, B. Seraphin, A.E. Kulozik, M.W.Hentze, Complexes between the nonsense-mediated mRNA decaypathway factor human upf1 (up-frameshift protein 1) andessential nonsense-mediated mRNA decay factors in HeLa cells,Biochem. J. 373 (2003) 775–783.

[25] B.P. Lewis, R.E. Green, S.E. Brenner, Evidence for the wide-spread coupling of alternative splicing and nonsense-mediatedmRNA decay in humans, Proc. Natl. Acad. Sci. USA 100 (2003)189–192.