PD1 negative and PD1 positive CD4+ T regulatory cells in mild cognitive impairment and Alzheimer's disease

Journal of Alzheimer’s Disease 21 (2010) 927–938 927DOI 10.3233/JAD-2010-091696IOS Press

PD1 Negative and PD1 Positive CD4+ TRegulatory Cells in Mild CognitiveImpairment and Alzheimer’s Disease

Marina Saresellaa, Elena Calabreseb, Ivana Marventanoa, Federica Pianconea, Andrea Gattia,Maria Gaetana Calvoa, Raffaello Nemnib and Mario Clericia,c,∗

aLaboratory of Molecular Medicine and Biotechnology, Don C.Gnocchi ONLUS Foundation IRCCS, Milano, ItalybDepartment of Neurorehabilitation, Don C. Gnocchi ONLUS Foundation IRCCS, Milano, ItalycDepartment of Biomedical Sciences and Technologies, University of Milano, Milano, Italy

Handling Associate Editor: Daniela Galimberti

Accepted 22 April 2010

Abstract. Regulatory T lymphocytes (Treg) play a fundamental importance in modulating the relative balance between inflam-mation and immune tolerance, and alterations of these cellsare observed in inflammatory diseases. To better characterize theneuroinflammatory processes suggested to be associated with Alzheimer’s disease (AD) and to clarify the possible role of Treg

cells in this process, we extensively analyzed these cells (CD4+ CD25highFoxp3+) in patients with either severe AD (n = 25)or mild cognitive impairment (MCI) (n = 25), comparing the results with those of two groups of healthy controls (HC) (n =

55). Because the intra- or extracellular expression of programmed death receptor 1 (PD1) identifies functionally diverse subsetsof Treg we also analyzed such subpopulations. Results showed that,whereas both Treg and PD1pos Treg are increased in MCIand AD patients compared to HC, PD1neg Treg, the subpopulation of Treg cells endowed with the strongest suppressive ability,are significantly augmented in MCI patients alone. In these patients amyloid-β-stimulated-T cells proliferation was reducedand Treg-mediated suppression was more efficient compared to both ADand HC. The observation that PD1neg Treg, cells areincreased in MCI patients reinforces the inflammatory origin of AD and supports a possible beneficial role of these cells in MCIthat is lost in patients with full-blown AD.

Keywords: Alzheimer’s disease, immunology, mild cognitive impairment, PD1, T regulatory cells

INTRODUCTION

A chronic inflammatory process mediated by acti-vated glial cells is suggested to play a pivotal role inthe neurodegeneration that is observed in age-related

∗Correspondence to: Professor Mario Clerici, M.D., Chair ofIm-munology, Department of Biomedical Sciences and Technologies,University of Milano, Via Fratelli Cervi 93, 20090 Segrate (Mi-lano), Italy. Tel.: +39 0250330412; Fax: +39 02 50330414; E-mail:[email protected].

diseases characterized by the presence of amyloid-β

(Aβ) plaques in the brain. Inflammatory mediators,such as cytokines, are probably finalized at destroyingsenile plaques in these pathologies, but are also lethalto surrounding neurons [1]. Alzheimer’s disease (AD)epitomizes such diseases.

Various studies indicate the presence of increasedconcentrations of proinflammatory cytokines andchanges in lymphocyte subsets in AD. Thus, elevatedlevels of interleukin (IL)-6 in plasma [2], augmentedproduction of interferon-γ, and tumor necrosis factor-

ISSN 1387-2877/10/$27.50 2010 – IOS Press and the authors. All rights reserved

928 M. Saresella et al. / Treg in AD-Associated Neuroinflammation

α by natural killer cells [3], and an increase in IL-1β

associated with a concomitant decrease in IL-10 havebeen observed in AD [4]. Notably, an increase in IL-10 has been described in non-demented healthy elderlyindividuals [5], and a particular genotype resulting inhigher levels of IL-10 was found to be associated withlongevity [6]. A significant decrease of CD8+ T lym-phocytes, [7] as well as increases of CD4+, CD25+,CD28+ and CD8+CD38+ (activated) T cells have al-so been described in AD [8]. Finally, a significant re-duction of CD45RA+ lymphocytes and of circulatingB and CD8+CD28− cells was also observed in thesepatients [9,10].

Some authors hypothesized that autoimmune re-sponses could be beneficial in preventing disease de-velopment within the central nervous system (CNS)in healthy individuals; as a consequence, an imbal-ance of immune regulation could lead to increased sup-pressive function or to a pathology mediated by pro-tein aggregation [11]. Thus, the reciprocal balance be-tween immune tolerance and inflammatory responsesseems to be important in AD-associated neuroinflam-mation. Regulatory T cells (Treg) are pivotal agentsin the regulation of tolerance by dampening harmfulautoimmune T cells and harnessing inflammation [12–14]. As a consequence, loss of Treg function appears tobe a fundamental factor in autoimmunity [15]. Quan-titative and qualitative analyses of Treg have been per-formed in autoimmune diseases in the attempt to shedlight on the pathologic impairments associated withthese conditions. Results showed that Treg lympho-cytes are defective in most of these diseases, includ-ing rheumatoid arthritis, lupus erythematosus systemi-cus, type 1 diabetes, and multiple sclerosis [16–18].Treg are characterized by a number of markers, includ-ing CD4, high CD25 expression and the transcriptionfactor Foxp3 [19]; other surface markers expressed byTreg include CD127, CD39, and CTLA-4 [20–24]. Auniversal consensus on the phenotypic classification ofthese cells nevertheless has not yet been reached, rather,it is emerging that the cells defined as Treg include dif-ferent functional and phenotypic subpopulations. Oneof such populations was recently described [25], show-ing that CD4+/CD25high/Foxp3+ Treg cells can besub-classified based on the surface expression of pro-grammed death receptor 1 (PD1). Briefly, PD1 can ei-ther be retained within the intracellular compartment,or it can be expressed on cell surface, possibly uponactivation. Treg lymphocytes that retain PD1 in thecytoplasm, and are endowed with stronger suppressiveproperties, are identified as PD1neg. The mechanisms

used by PD1neg Treg to suppress immune responses arenot fully clarified, but likely involve both cell-cell in-teraction and suppression mediated by the productionof IL-10 and TGFβ, as is the case with Treg as a whole.

To clarify the role of neuroinflammation in AD weperformed phenotypic and functional analysis of Treg inpatients affected by either full-blown AD or mild cog-nitive impairment (MCI), comparing the results withthose obtained in two groups of healthy controls.

MATERIALS AND METHODS

Patients and controls

Eighty elderly and twenty-five middle age individ-uals were enrolled in the study.Twenty-five subjectswere diagnosed as being affected by AD and 25 indi-viduals had a diagnosis of MCI; these patients were fol-lowed by the Neurology Department of the Don Gnoc-chi Foundation in Milano. Clinical diagnosis of ADwas performed according to NINCDS-ADRDA [26]and DMS III-R [27] criteria. Epidemiological and clin-ical characterization of patients and controls is present-ed in Table 1. All AD patients underwent completemedical, neurological and neuropsychological evalua-tion, as well as laboratory analysis to exclude reversiblecauses of dementia. Nineteen patients were late and 6early AD onsets; all cases were sporadic. The degreeof cognitive impairment was assessed by neuropsycho-logical tests including the Mini-Mental State Examina-tion (MMSE) [28], the Digit Span Forward and DigitSpan Backward, the Logical Activated Test, the PairedAssociated Words Test, the Token Test, the Supra SpanCorsi Block Tapping Test, the Verbal Fluency Task,the Raven Colored Matrices, the Wisconsin Card Sort-ing Test, the Rey Complex Figure, the Global Deteri-oration Scale [29], and the Hachinski Ischemic Scale.MMSE and CDR scales were used to assess the severityof dementia. AD patients were divided into three sub-groups: mild dementia, moderate dementia, and severedementia (Table 1). Blood count, urine analysis, bloodchemistry screen, serum folate, B12 levels, and thyroidfunctions tests were normal in all patients and none ofthem suffered from malnutrition or vitamin deficiencysyndromes.

The diagnosis of MCI was based on the followingunanimously adopted criteria: 1) reported cognitivedecline; 2) impaired cognitive function; 3) essentiallynormal functional activities; and 4) exclusion of de-mentia [30,31]. Cognitive impairment was defined by

M. Saresella et al. / Treg in AD-Associated Neuroinflammation 929

Table 1Characterization of patients and controls

Mild AD Moderate AD Severe AD Total AD MCI HC MAC

n 10 10 5 25 25 30 25Median Age 83 74 86 81 79 83 63i.q. range 80–85 73–77 84–88 75–85 73–83 75–85 58–70Gender 5F 5M 4F 6M 3F 2M 13F 12M 13F 12M 18F 12M 15F 10MMMSE 23.5± 1.5 16.6± 1.3 9.5± 0.3(Mean± S.D.) (range) 20–25 15–19 0–14

F: females; M: males; S.D.: Standard Deviation; i.q.: interquartile.

performance below 1 standard deviation (SD) of theage-, gender-, and education-adjusted mean in any ofthe tested cognitive domains above indicated. Activi-ties of daily living (ADL) were assessed with the Bay-er ADL Scale. A score< 4 points was consideredto reflect grossly intact ADL. In all cases, an infor-mant (wife, husband, or relative) corroborated the cog-nitive impairment of the patient. All subjects under-went thorough clinical history, neurological examina-tion, laboratory test, and MRI. Patients with structuralabnormalities that could impair cognitive function, oth-er than cerebrovascular lesions, were excluded. Noneof the individuals enrolled in the study was undergoingcholinesterase inhibitor or anticholinergic treatment.

Patients with MCI were further classified into oneof the two MCI subtypes: aMCI, if the impairment in-cluded the memory domain, or naMCI, if the impair-ment was in one or more non-memory domains withrelative preservation of memory. All but two of theMCI patients included in the study were diagnosed asbeing affected by aMCI, therefore the data presentedin the paper substantially relate to aMCI patients, whoare considered at high risk to specifically develop AD.

Thirty healthy elderly subjects, age- and gender-matched with the patients (healthy controls; HC) and25 middle age healthy controls (MAC) were also en-rolled in the study. These individuals were either un-related healthy spouses of AD and MCI patients orhealthy volunteers, and they had no family history ofdementia or evidence of acute or chronic diseases atthe time of enrollment. Healthy elderly human groupswere selected in according to the SENIEUR protocolfor immuno-gerontological studies of European Com-munity’s Control Action Programme on Aging [9,32].The cognitive status was assessed by administration ofthe MMSE (score for inclusion as normal controls sub-jects> 28). All the individuals enrolled in the study, ortheir relatives when appropriate, provided written in-formed consent according to a protocol approvedby thelocal ethics committee of the Don Gnocchi Foundationbefore admission to the study.

Blood sample collection and cell separation

Whole blood was collected by venopuncture in vacu-tainer tubes containing ethylenediaminetetraaceticacid(Becton Dickinson & Co., Rutherford, NJ, USA). Pe-ripheral blood mononuclear cells (PBMC) were sep-arated on lymphocyte separation medium (OrganonTeknika Corp., Durham, NC, USA) and washed twicein phosphate buffered saline (PBS); viable leukocyteswere determined by trypan blue exclusion.

Flow staining for PD1 and Foxp3 of CD4+ CD25high

T reg

Freshly isolated, PBMC were washed and incubat-ed with CD4-, CD25- (Beckman-Coulter, Fullerton,CA, USA), and PD1 (eBioscience, San Diego, CA,USA)-specific monoclonal antibodies for 30 min at4◦C. PBMC were then washed and the intracellularcostaining of PD1 and Foxp3 was conducted using theFoxp3 staining protocol (eBioscience). Intracellularor surface costaining of PD1 and intracellular Foxp3was performed on CD4+CD25high (CD4+CD25++)gated T cell by flow cytometry.Treg PD-1 neg lym-phocytes are cells with a positive intracellular stainingfor both PD1 (Mab anti-PD-1 FITC-conjugated) andFoxp-3 (Mab anti-Foxp-3 PC-5-conjugated); Treg PD-1 pos lymphocytes are cells in which surface stainingof PD-1(Mab anti-PD-1 PE-conjugated) is associatedwith an intracellular positivity for Foxp-3 (Mab anti-Foxp-3 PC-5-conjugated); both populations were cal-culated on CD4+CD25high (CD4+CD25++) gated Tcell.

CFDA-SE labeling

PBMC resuspended at 107 ml in PBS were added toan equal volume of 5µM of 5–6 carboxyfluorescein di-acetate succimidyl ester (CFSE) (Molecular Probe, Eu-gene, OR) in PBS and mixed for 10 min at room temper-ature. This procedure has a labeling efficiency exceed-

930 M. Saresella et al. / Treg in AD-Associated Neuroinflammation

ing 99%, and cells remain labeled for at least 10 daysduring tissue culture [33]. CFSE reacts with secondaryamino of intracellular proteins providing a uniform flu-orescent label. Upon cell division a CFSEhighcell (highfluorescence intensity) will lose half of its CFSE fluo-rescence intensity (CFSElow) resulting in populationsof daughter cells (proliferating cells), which can bevisualized by flow cytometry.

Proliferation assay

CFSE-labeled cells were washed twice and resus-pended at 3× 106 cells/ml in polystyrene tissue cul-ture tubes containing 1 ml RPMI-1640 medium supple-mented with 10% AB serum. Cells were then stimulat-ed with either non-immunogenic peptides [34] or witha pool of three different fragments of Aβ-protein: frag-ment 1-40, fragment 1-16 and fragment 1-35 (10µg/ml)(Sigma, St. Louis, MO, USA) at 37◦C in a humidified5% CO2 atmosphere for 5 days. These fragments arelargely present in cerebral deposits of Aβ -protein insenile plaques, induce inflammatory mediators [35,36],and circulate in plasma and cerebrospinal fluid [37,38].Cells were subsequently harvested and washed twicein PBS. Surface staining for CD3, CD4, and CD8 wasperformed for 30 min at 4◦C, finally cells were washedin PBS and fixed in 1% paraformaldehyde. Prolifera-tion in non-immunogenic peptides- stimulated sampleswas considered to be the backgroundvalue. The∆ Pro-liferating Fraction (PF) was calculated by subtractingbackgroundproliferation from the antigen-specific pro-liferation. Stimulation indexes (SI) were calculated bydividing antigen-induced proliferation by backgroundproliferation. Both a∆ PF> 1% and an SI> 2.0 wererequired to classify a response as positive [39].

Cell purification

CD4+ T cells were isolated using a negative CD4+

T cell isolation kit (EasySep, StemCell Technolo-gies, Grenoble, France). CD4+CD25high (Treg) andCD4+CD25neg (Tresp) cells were separated using theALTRA EPICS cell sorter (Beckman-Coulter). Fresh-ly isolated CD4+ T cells were stained for 40 min at4◦C with human CD25-specific PC-5 labeled antibod-ies. Sort gates were restricted to the CD4+ CD25high

(CD4+CD25++) and CD4+CD25−.

Suppression assays

CD4+ and CD4+CD25neg cells were labeled with1 µM of CFSE. 1x105 autologous mytomicin–C treat-

ed (25µg/ml) PBMC plus 2x104 CD4+CD25neg inabsence/presence of 2× 104 CD4+CD25high Treg

were resuspended in 200µl RPMI 1640 medium andstimulated with coated anti-CD3 (10µg/ml) or withthe Aβ peptide pool (10µg/ml). Cultures were setup as duplicates in U-bottom wells (COSTAR, Cam-bridge, MA, USA) and incubated at 37◦C in a hu-midified atmosphere with 5% CO2. After 5 days cellswere harvested and the CFSE signal of gated lympho-cytes was analyzed by flow cytometry. The suppres-sive capacity of Tregtowards responder cells in cocul-ture (Tresp-Treg ratio 1:1) was expressed as the relativeinhibition of the percentage of CFSElow cells (prolif-erating [100× (1 − %CFSElow CD4+CD25− in co-culture/ %CFSElowCD4+CD25− T cells alone)] forCFSE based measurement of proliferation [40].

Monoclonal antibodies

The monoclonal antibodies used for PBMC stimula-tion or labeling are shown in Table 2. Flow-cytometricdata (100.000 non-gated events) were acquired on aBeckman-Coulter Cytomics FC-500 flow cytometer.For analysis the CXP Analysis software (Beckman-Coulter) was used.

Data analysis

Quantitative data were not normally distributed(Shapiro-Wilk test) and are thus summarized as medi-an and Interquartile Range (25◦ and 75◦ percentile).Comparisons between groups were analyzed to evalu-ate immunological differences. Kruskal-Wallis analy-sis of variance was performed for each variable; Bon-ferroni correction was applied to the results. p valuesand all tests are two-sided.

RESULTS

Treg lymphocytes in peripheral blood

The intracytoplasmatic or surface expression ofthe PD1 protein identifies two different subpopu-lations of Treg within CD4+/CD25high/Foxp3+ Tlymphocytes. We examined these three populations:CD4+/CD25high/Foxp3+ (Treg), CD4+/CD25high/Foxp3+/PD1pos,and CD4+/CD25high/Foxp3+/PD1neg

in the peripheral blood of all MCI and AD patients,comparing the results with those of 30 gender- andage-matched healthy controls (HC) and 25 MAC. Rep-

M. Saresella et al. / Treg in AD-Associated Neuroinflammation 931

Table 2Monoclonal antibodies used in the study

mAbs Clone Isotype Source Company Fluorochrome Concentration

CD4 SFCI12T4D11 IgG1 mouse Beckman-Coulter Phycoerythrin-Cyanin-7 (PC7) 0.2µgCD3 UCHT1 IgG1 mouse Beckman-Coulter 10µgCD8 SFCI21Thy2D3 IgG1 mouse Beckman-Coulter Phycoerythrin-Cyanin-5(PC5) 0.2µgCD25 B1.49.9 IgG2a mouse Beckman-Coulter Phycoerythrin-Texas Red (ECD) 0.2µgFoxp3 PCH101 IgG2a rat e- Bioscience Phycoerythrin-Cyanin-5(PC5) 0.5µgPD1 MIH4 IgG1 mouse e- Bioscience Fluorescein Isothiocyanate (FITC) 0.5µgPD1 MIH4 IgG1 mouse e- Bioscience Phycoerythrin (PE) 0.5µg

Fig. 1. Data observed in a representative MCI patient are shown. A) Expression of CD25high (CD4+CD25++)T lymphocytes on CD4+ Tcells. B) Intracellular Foxp3; cells gated on CD4+CD25high. C) Surface expression of PD1 (PD-1pos cells); cells gated on CD4+CD25high.D) Intracellular PD1 (PD-1neg cells); cells gated on CD4+CD25high. All populations of Treg shown in the figure are stained for Foxp3. Inpanels B-D the percentage of positive cells is indicated above the longitudinal bars. Cell counts (y-axis) and intracellular Foxp3 (B), surface PD1(C), and intracellular PD1 (D) staining are also shown on thex-axis. Dotted lines in panels B-D represent the isotype control.

resentative results showing the three cell populationsanalyzed are presented in Fig. 1.

Results showed that CD4+/CD25high/Foxp3+ weresignificantly increased in patients with a diagnosis ofeither MCI (median percentage= 1.5%) or AD (me-dian percentage= 1%) compared to HC (median per-centage= 0.01%; p < 0.001 in both cases), but nodifferences were observed between MCI and AD pa-tients. CD4+/CD25high/Foxp3+/PD1pos Treg lympho-cytes were increased in both MCI (median percent-age= 0.05%) and AD patients (median percentage=

0.2%) compared to HC (median percentage= 0.01%;p = 0.01 andp < 0.01 respectively) and in ADcompared to MCI patients (p = 0.02). In contrastwith these results, CD4+/CD25high/Foxp3+/PD1neg

Treg cells were augmented in MCI patients alone (me-dian percentage= 1.9%). This increase was high-ly significant compared to both HC individuals (me-dian percentage= 0.01%; p < 0.001) and patientswith full-blown AD (median percentage= 0.03%;p = 0.006) (Fig. 2). Notably, the three popula-tions of Treg cells were significantly increased (p <

932 M. Saresella et al. / Treg in AD-Associated Neuroinflammation

Fig. 2. Regulatory T cell (CD4+CD25high Foxp-3+) (A), PD-1pos Treg lymphocytes (CD4+25high Foxp3+ PD-1pos) (B), and PD1negTreg

lymphocytes (CD4+25high Foxp3+PD1neg) (C) in the peripheral blood of patients affected by MCI or AD; data obtained in age- and gen-der-matched healthy controls (HC) and in middle age healthycontrols (MAC) are also shown. The boxes stretch from the 25th to the 75th per-centile; the lines across the boxes indicate the median values; the lines stretching from the boxes indicate extreme values. Statistical significanceis shown:∗p < 0.05.∗∗p < 0.01, and∗∗∗p < 0.001.

0.001 in all cases) in MAC compared to HC (me-dian percentages: CD4+/CD25high/Foxp3+Treg =

2.4%;CD4+/CD25high/Foxp3+/PD1posTreg = 1.3%;CD4+/CD25high/Foxp3+/PD1neg Treg =3.7%); where-as CD4+/CD25high/Foxp3+/PD1neg Treg cells alonewere augmented in MAC compared to AD (p <

0.001).Theseresults indicate that an increase of PD1neg

Treg cells, the regulatory T lymphocytes endowedwith the strongest suppressive abilities, characterizesMCI individuals, distinguishing them from patientswith full-blown AD. In additional analyses, we veri-fied whether the tree populations of Treg lymphocyteswould differ in AD patients affected by mild, moderate,or severe disease. Results showed that the highest per-centages of all three cell populations were observed inpatients with mild disease, whereas the lowest percent-

ages characterized those with severe AD. These trendsdid not reach statistical significance (data not shown).

Aβ pool-stimulated proliferation

Aβ aggregates in the brain are suspected to drivemicroglia- and astrocytes-mediated inflammation, aprocess suggested to be at the basis of AD [35,36,41]. To measure Aβ-stimulated proliferation, periph-eral blood cells of AD, MCI, and HC individuals werestimulated with either a pool of Aβ immunogenic pep-tides or with a pool of non-immunogenic (control) pep-tides, cell division was subsequently evaluated usingCFSE staining. Results indicated that the Aβ pool-stimulated proliferation of CD4+ T lymphocytes, ex-pressed as a stimulation index (S.I.), was significantlyaugmented in AD (S.I. median value= 2.5) compared

M. Saresella et al. / Treg in AD-Associated Neuroinflammation 933

to MCI patients (S.I.= 0.02; p = 0.007), with thelowest values being observed in HC (S.I.= 0.01;p <

0.001vs. AD).Aβ pool-stimulated S.I. of CD8+ T lymphocytes

were similarly greatly increased in AD compared toMCI patients and HC (AD= 2.0; MCI = 0.01; HC=

0.01). This difference was statistically significant whenAD patients were compared to HC (p < 0.001). Noproliferation was observed by cells of any of the groupsenrolled in the study upon stimulation with the pool ofnon-immunogenic control peptides. These results areshown in Fig. 3. The Aβ-stimulated proliferation wasperformed also in PBMC isolated from MAC and noproliferation was observed (data not shown).

Suppression of Aβ pool and anti CD3-stimulatedproliferation by Treg

The ability of Treg to suppress the proliferationof antigen- and anti CD3-stimulated proliferation wasanalyzed in both groups of patients enrolled in thestudy; results were compared to those obtained in HC.CD4+CD25high Treg were isolated from freshly drawnperipheral blood, (PD1-based sorting is not technical-ly possible) and Treg-depleted cells were stimulatedwith either the pool of immunogenic Aβ peptides orwith anti CD3. Treg were then added back in a 1:1ratio of Tresp/Treg lymphocytes. Results obtained up-on stimulating cells with the pool of immunogenicAβ peptides indicated that the suppressive ability ofCD4+CD25high Treg cells on activated cells was sig-nificantly higher in MCI (median percentage= 40%)compared to AD (21%;p = 0.04). These values weresignificantly higher also compared to those observed inHC, in whom the stimulation with immunogenic Aβpeptides did not elicit the generation of Aβ-specific Teffector cells (p < 0.001 in both cases). Finally, resultsobtained in anti CD3-stimulated cell cultures showedan augmented suppression of CD4+CD25high Treg onactivated cells in AD (median percentage= 50%) com-pared to MCI (19%) patients and HC (25%); these dif-ferences did not reach statistical significance. Theseresults are shown in Fig. 4. The suppression of – Aβ

pool and anti CD3-stimulated proliferation by Treg wasachieved also in PBMC isolated from MAC and theresults were similar to those obtained in HC (data notshown).

DISCUSSION

Our working hypothesis was that if inflammation isa negative factor for AD, and if Treg play a beneficial

anti-inflammatory role in the attempt to control such in-flammation, it could be expected that such cells wouldbe quantitatively and qualitatively impaired in AD com-pared to MCI patients. We thus examined Treg cells inpatients with a diagnosis of either MCI or AD; recentdata indicating that a subset of Treg characterized by theintracellular expression of PD1 is endowed with par-ticularly potent anti-inflammatory properties promptedus to concentrate on such cells. Results herein indi-cate that development of AD is associated with lowerquantities of circulating Treg lymphocytes and, in par-ticular, with reduced percentages of PD1neg Treg cells.These quantitative alterations are associated with qual-itative changes, summarized as an increased Aβ- spe-cific proliferation and a reduced ability of Treg to sup-press such proliferation. These results, together withthe preliminary observation that the lowest percentagesof all subpopulations of Treg cells are seen in patientswith severe AD, lend support to the inflammatory ori-gin of AD and suggest that alterations in Treg lympho-cytes play a pivotal role in the inflammation associatedwith AD. The fact that data herein stem from analysesperformed in peripheral blood lymphocytes and not incells circulating in the cerebrospinal fluid could appar-ently weaken our results. The ethical committee of ourcenter defined the lumbar puncture procedure as non-ethical and not necessary for AD diagnosis. Neverthe-less, the observations that: 1) immune cells continu-ously re-circulate throughout the body; and 2) immunecells migrate across the blood-brain barrier (BBB), andthe BBB is permeable to cytokines, seem to by-passsuch possible criticism, lending support to our results.

Deposit of fibrillar Aβ [42] stimulates microgliaand astrocytes to act as immunological mediators [43,44]. Thus, Aβ-stimulated microglia releases cytotox-ic molecules such as NO, oxygen radicals, proteas-es, proinflammatory cytokines, and expresses MHCI,MHCII, CD40, and adhesion molecules [45]. No-tably, the up-regulation of IL-1β enhances BBB per-meability facilitating leukocytes infiltration [46]. Im-mune responses thus do occur in CNS and can be driv-en by endogenous (glial) and/or exogenous (peripheralleukocytes) sources and can serve either productive orpathological roles [47]. The recruitment of peripheralmonocytes/macrophages in the CNS could restrict Aβ

plaques [47–49]; activated endothelium or microgliacan also present Aβ peptides to T cells, thus induc-ing immune responses [50–52]. Whether the arrivalof T-cells is beneficial or detrimental is ill-understood.However in the case of massive T-cell response (forinstance in Aβ vaccine-related meningoencephalitis),the effects seem to overwhelmingly negative.

934 M. Saresella et al. / Treg in AD-Associated Neuroinflammation

Fig. 3. Aβ-peptide pool-stimulated proliferation (CFSE-staining). Representative dot plots obtained in AD patient are shown.CFSE staining isshown on x-axis and CD4+ or CD8+ staining on the y-axis. Non-dividing CFSEhigh labeled cells (high fluorescence intensity) are shown inthe upper and lower right quadrants, whereas daughter cell populations which have lost half of their CFSE fluorescence signal with each divisionround (CFSElow) are shown in upper and lower left quadrants.The percentage next to the populations represents the proliferating fraction ofCD4+ T cells and CD8+ T cells. Summary results of Stimulation Index of CD4+ (panel A) and CD8+ (panel B) T lymphocytes stimulatedwith Aβ peptide pools in patients affected by MCI, AD and in age-and gender-matched healthy controls (HC) are also shown. The boxes stretchfrom the 25th to the 75th percentile; the lines across the boxes indicate the median values; the lines stretching from theboxes indicate extremevalues. Statistical significance is shown:∗p < 0.05,∗∗p < 0.01, and∗∗∗p < 0.001.

Treg cells have convincingly been shown to modu-late immune reactivity and inflammation, and quanti-tative/qualitative alterations of such cells result in thehoning of inflammation, favoring autoimmuneprocess-es. Treg are still relatively unexplored in AD. A re-cent paper showed an increase of such cells (identifiedas Foxp3-expressing lymphocytes) in AD and in olderhealthy controls; the authors also reported an increasedability of Treg cells of AD patients to suppress mitogen-stimulated proliferationin vitro [11]. These data, thus,suggest that the frequency of Treg increases with ageand is accompanied by a stronger suppressive activity.Our results confirm that an overall increase of Treg cellsis seen in AD, and add the novel information that thisincrease does not involve PD1neg Treg lymphocytes,

the cells with the highest functional activity, that arediminished in AD. In contrast, data herein do not con-firm that Treg cells are increased in older healthy indi-viduals, as we detected a decrease of such cells in suchindividuals. These results were further strengthenedby data obtained in MAC subjects, in whom all theexamined populations of Treg cells were significant-ly increased compared to the values observed in olderpersons. Further studies will be needed to clarify thisissue. Notably, other authors [53] reported a decreasedpercentage of CD4+CD25high cells in AD comparedto older controls and an increase of these cells was ob-served in healthy elderly groups [54,55]. The apparentdiscrepancy between these data and the ones herein isexplained by the fact that we identified Treg cells by

M. Saresella et al. / Treg in AD-Associated Neuroinflammation 935

Fig. 4. Median percent of suppressive capacity mediated by CD4+CD25high (Treg) in patients affected by MCI or AD; data obtained inage-and-gender-matched healthy controls (HC) are also shown. A) Aβ peptide pools-stimulated proliferation. B) plate bound anti-CD3 +

anti-CD28 -stimulated proliferation. In all experiments a1:1 Treg:Tresponders ratio was used. The boxes stretch from the 25th to the 75thpercentile; the lines across the boxes indicate the median values; the lines stretching from the boxes indicate extremevalues. Statistical significanceis shown:∗p < 0.05,∗∗p < 0.01, and∗∗∗p < 0.001.

different and more specific markers. Additionally, theage range of our healthy control was higher than thatof the individuals studied in [53].

Functional analyses of Treg cells activity showedthat cells of AD patients suppress mitogen-stimulatedproliferation. Whereas these data confirm older re-sults [11] the observation of a significantly increasedTreg-mediated suppression of Aβ-specific proliferationin MCI and AD is novel. This result could be seen ei-ther as protective or as a harmful response. Thus, Aβ-specific Treg could suppress an inflammatory and po-tentially harmful response, or rather, they could con-tribute to disease progression by impairing an immuneactivity trying to destroy Aβ plaques. The latter hy-pothesis is less credible because Treg-mediated sup-pression is decreased in AD compared to MCI, sug-gesting an association between disease worsening anddecrease of Treg suppressive capacity. The previous ob-servation that CD8+CD28-lymphocytes, cells knownto have a suppressor/regulatory function resulting inT helper unresponsiveness [56,57], are also signifi-cantly reduced in AD patients, further supports theconcept that a generalized impairment of the suppres-sor/regulatory ability of T lymphocytes is present inAD.

The increase of PD1neg Treg we observed in MCIindividuals could be an attempt to down regulate in-flammatory response to Aβ. To this end, we measuredAβ-specific proliferation in all the groups of individu-als enrolled in the study. Results clearly indicated that

Aβ-specific proliferation is significantly higher in ADcompared to MCI patients; thein vitro Aβ peptidesstimulated generation of effector T cells was also great-ly increased in AD individuals. Interestingly, the ob-servation that no Aβ-specific effector cells were elicit-ed in HC suggests that normal aging is clearly distinctfrom MCI, underlining the hypothesis that individualswith MCI are much more similar to AD patients thanthey are to HC.

The suppressive ability of Treg on Aβ peptide-stimulated proliferation was also significantly reducedin AD compared to MCI individuals. Notably, as sort-ing of PD1neg cells is not technically possible, suppres-sion was measured using CD4+CD25high cells isolatedfrom freshly drawn blood; as a consequence we did notformally measure the suppressive capacity of PD1neg

Treg cells. Nevertheless, the observation that the per-centages of PD1neg Treg cells are significantly differentin MCI, AD, and HC individuals allows the specula-tion that the different suppressive abilities we observedare secondary to the diverse percentages of PD1negTreg

cells detected in each one of these subpopulations.Treg cells mediate their effect via two complemen-

tary mechanisms: IL-10-mediated functional impair-ment of immune cells and induction of apoptosis ofsuch cells. In this light it is interesting to notice that: 1)Aβ- stimulated IL-10 production has repeatedly beenshown to be reduced in AD compared to healthy indi-viduals [4,9] and 2) particular SNP of the IL-10 geneassociated with low IL-10 production are significantly

936 M. Saresella et al. / Treg in AD-Associated Neuroinflammation

over expressed in AD [58]. It will be interesting toevaluate PD1-mediated apoptosis of Aβ-specific cellsin patients with a diagnosis of either MCI or AD.

In conclusion, the data herein indicate that Treg cells,and in particular, PD1negTreg cells might play an im-portant role in the pathogenesis of AD: loss of suchcells and of their functional ability is associated withdevelopment of AD. These results will need to be sup-ported by wider analyses and by longitudinal studiesof patients with MCI followed over time; further ef-forts, comparing peripheral and cerebrospinal fluid cellanalysis, will also be needed to validate these resultsin wider cohorts of patients with different patterns ofdisease.

ACKNOWLEDGMENTS

This study was supported by grants from 2009 Ricer-ca Corrente [Italian Ministry of Health]; FondazioneCARIPLO; grants from the Istituto Superiore di Sani-ta “‘Programma Nazionale di Ricerca sull’ AIDS”;the nGIN EC WP7 Project; the Japan Health Sci-ence Foundation and Progetto FIRB RETI: Rete Ital-iana Chimica Farmaceutica CHEM-PROFARMA-NET[RBPR05NWWC].

Authors’ disclosures available online (http://www.j-alz.com/disclosures/view.php?id=437).

REFERENCES

[1] McGeer EG, McGeer PL (2003) Inflammatory processes inAlzheimer’s disease.Prog Neuropsychopharmacol Biol Psy-chiatry 27, 741-749.

[2] Singh VK, Guthikonda P (1997) Circulating cytokines inAlzheimer’s disease.J Psychiatr Res31, 657-660.

[3] Solerte SB, Cravello L, Ferrari E, Fioravanti M (2000) Over-production of IFN-gamma and TNF-β from natural killer (NK)cells is associated with abnormal NK reactivity and cognitivederangement in Alzheimer’s disease.Ann N Y Acad Sci917,331-340.

[4] Remarque EJ, Bollen EL, Weverling-Rijnsburger AW, Lat-erveer JC, Blauw GJ, Westendorp RG (2001) Patients withAlzheimer’s disease display a pro-inflammatory phenotype.Exp Gerontol36, 171-176.

[5] Castle SC, Uyemura K, Crawford W, Wong W, KlaustermeyerWB, Makinodan T (1999) Age-related impaired proliferationof peripheral blood mononuclear cells is associated with anincrease in both IL-10 and IL-12.Exp Gerontol34, 243-252.

[6] Lio D, Scola L, Crivello A, Colonna-Romano G, CandoreG, Bonafe M, Cavallone L, Franceschi C, Caruso C (2002)Gender-specific association between -1082 IL-10 promoterpolymorphism and longevity.Genes Immun3, 30-33.

[7] Shalit F, Sredni B, Brodie C, Kott E, Huberman M (1995)T lymphocyte subpopulations and activation markers corre-late with severity of Alzheimer’s disease.Clin Immunol Im-munopathol75, 246-250.

[8] Lombardi VR, Garcia M, Rey L, Cacabelos R (1999) Char-acterization of cytokine production, screening of lympho-cyte subset patterns andin vitro apoptosis in healthy andAlzheimer’s Disease (AD) individuals.J Neuroimmunol97,163-171.

[9] Speciale L, Calabrese E, Saresella M, Tinelli C, MarianiC,Sancito L, Longhi R, Ferrante P (2007) Lymphocyte subsetpatterns and cytokine production in Alzheimer’s disease pa-tients.Neurobiol Aging28, 1163-1169.

[10] Tan J, Town T, Abdullah L, Wu Y, Placzek A, Small B, KroegerJ, Crawford F, Richards D, Mullan M (2002) CD45 isoformalteration in CD4+ T cells as a potential diagnostic marker ofAlzheimer’s disease.J Neuroimmunol132, 164-172.

[11] Rosenkranz D, Weyer S, Tolosa E, Gaenslen A, Berg D, LeyheT, Gasser T, Stoltze L (2007) Higher frequency of regulatoryT cells in the elderly and increased suppressive activity inneurodegeneration.J Neuroimmunol188, 117-127.

[12] Fehervari Z, Sakaguchi S (2004) CD4+ Tregs and immunecontrol.J Clin Invest114, 1209-1217.

[13] Kuniyasu Y, Takahashi T, Itoh M, Shimizu J, Toda G,Sakaguchi S (2000) Naturally anergic and suppressiveCD25+CD4+ T cells as functionally and phenotypically dis-tinct immunoregulatory T cell subpopulation.Int Immunol12,1145-1155.

[14] Sakaguchi S (2004) Naturally arising CD4+ regulatory t cellsfor immunologic self-tolerance and negative control of im-mune responses.Ann Rev Immunol22, 531-562.

[15] Hori S, Takahashi T, Sakaguchi S (2003) Control of autoim-munity by naturally arising regulatory CD4+ T cells. AdvImmunol81, 331-371.

[16] Crispin JC, Martınez A, Alcocer-Varela J (2003) Quantifica-tion of regulatory T cells in patients with systemic lupus ery-thematosus.J Autoimmun21, 273-276.

[17] De Kleer IM, Wedderburn LR, Taams LS, Patel A, Varsani H,Klein M, De Jager W, Pugayung G, Giannoni F, Rijkers G,Albani S, Kuis W, Prakken B (2004) CD4+CD25high regu-latory T cells actively regulate inflammation in the joints ofpatients with the remitting form of juvenile idiopathic arthritis.J Immunol172, 6435-6443.

[18] Kukreja A, Cost G, Marker J, Zhang C, Sun Z, Lin-Su K, TenS, Sanz M, Exley M, Wilson B, Porcelli S, Maclaren N (2002)Multiple immuno-regulatory defects in type-1 diabetes.J ClinInvest109, 131-140.

[19] Fontenot JD, Gavin MA, Rudensky AY (2003) Foxp3 pro-grams the development and function of CD4+CD25+ regu-latory T cells.Nat Immunol4, 330-336.

[20] Borsellino G, Kleinewietfeld M, Di Mitri D, Sternjak A,Dia-mantini A, Giometto R, Hopner S, Centonze D, Bernardi G,Dell’Acqua ML, Rossini PM, Battistini L, Rotzschke O, FalkK (2007) Expression of ectonucleotidase CD39 by Foxp3+

Treg cells: hydrolysis of extracellular ATP and immune sup-pression.Blood110, 1225-1232.

[21] Deaglio S, Dwyer KM, Gao W, Friedman D, Usheva A, EratA, Chen JF, Enjyoji K, Linden J, Oukka M, Kuchroo VK,Strom TB, Robson SC (2007) Adenosine generation catalyzedby CD39 and CD73 expressed on regulatory T cells mediatesimmune suppression.J Exp Med204, 1257-1265.

[22] Liu W, Putnam AL, Xu-Yu Z, Szot G, Lee MR, Zhu S, Got-tlieb PA, Kapranov P, Gingeras TR, Fazekas de St Groth B,Clayberger C, Soper DM, Ziegler SF, Bluestone JA (2006)

M. Saresella et al. / Treg in AD-Associated Neuroinflammation 937

CD127 expression inversely correlates with FoxP3 and sup-pressive function of human CD4+ T reg cells.J Exp Med203,1701-1711.

[23] Sansom DM, Walker LS (2006) The role of CD28 and cyto-toxic T-lymphocyte antigen-4 [CTLA-4] in regulatory T-cellbiology. Immunol Rev212, 131-148.

[24] Tang Q, Boden EK, Henriksen KJ, Bour-Jordan H, Bi M,Bluestone JA (2004) Distinct roles of CTLA-4 and TGF-betain CD4+CD25+ regulatory T cell function.Eur J Immunol34, 2996-3005.

[25] Raimondi G, Shufesky WJ, Tokita D, Morelli AE, ThomsonAW (2006) Regulated compartmentalization of programmedcell death-1 discriminates CD4+CD25+ resting regulatory Tcells from activated T cells.J Immunol176, 2808-2816.

[26] McKhann G, Drachman D, Folstein M, Katzman R, Price D,Stadlan EM (1984) Clinical diagnosis of Alzheimer’s disease:report of the NINCDS ADRDA Work Group under the aus-pices of Department of Health and Human Services Task Forceon Alzheimer’s Disease.Neurology34, 939–944.

[27] American Psychiatric Association (1987)Diagnostic and Sta-tistical Manual of Mental Disorders, DSM-III-R.

[28] Folstein MF, Folstein SE, McHugh PR (1975) “Mini-mentalstate” A practical method for grading the cognitive state ofpatients for the clinicians.J Psychiat Res12, 189-198.

[29] Hughes CP, Berg L, Danziger WL, Coben LA, Martin RL(1982) A new clinical scale for staging of dementia.Br JPsychiatry140, 566-572.

[30] Petersen RC (2004) Mild cognitive impairment as a diagnosticentity.J Intern Med256, 183–194.

[31] Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L,Wahlund LO, Nordberg A, Backman L, Albert M, AlmkvistO, Arai H, Basun H, Blennow K, de Leon M, DeCarli C,Erkinjuntti T, Giacobini E, Graff C, Hardy J, Jack C, JormA, Ritchie K, van Duijin C, Visser P, Peterson RC (2004)Mild cognitive impairment: beyond controversies, towardsaconsensus: report of the International Working Group on MildCognitive Impairment.J Intern Med256, 240–246.

[32] Ligthart GJ, Corberand JX, Fournier C, Galanaud P, HijmansW, Kennes B, Muller-Hermelink HK, Steinmann GG (1984)Admission criteria for immunogerontological studies in man:the SENIEUR protocol.Mech Ageing Dev28, 47-55.

[33] Lyons AB (2000) Analysing cell divisionin vivo andin vitrousing flow cytometric measurement of CFSE dye dilution.JImmunol Methods243, 147-154.

[34] Saresella M, Marventano I, Guerini FR, Zanzottera M, Del-bue S, Marchioni E, Macerati R, Longhi R, Ferrante P, Cleri-ci M (2008) Myelin Basic Protein-Specific T LymphocytesProliferation and Programmed Cell Death in DemyelinatingDiseases.Clin Imm129, 509-517.

[35] Gitter BD, Cox LM, Rydel LE, May PC (1995) Amyloidβ peptide potentiate cytokine secretion by interleukin 1-β-activated human astrocytoma cell.Proc Natl Acad Sci U S A92, 10738-10741.

[36] Akiyama H, Barger S, Barnum S, Bradt B, Bauer J, Cole GM,Cooper NR, Eikelenboom P, Emmerling M, Fiebich BL, FinchCE, Frautschy S, Griffin WS, Hampel H, Hull M, Landreth G,Lue L, Mrak R, Mackenzie IR, McGeer PL, O’Banion MK,Pachter J, Pasinetti G, Plata-Salaman C, Rogers J, Rydel R,Shen Y, Streit W, Strohmeyer R, Tooyoma I, Van MuiswinkelFL, Veerhuis R, Walker D, Webster S, Wegrzyniak B, Wenk G,Wyss-Coray T (2000) Inflammation and Alzheimer’s disease.Neurobiol Aging21, 383-421.

[37] Shoji M, Kanai M (2001) Cerebrospinal fluid Abeta40 andAbeta42: Natural course and clinical usefulness.J Alzheimers

Dis 3, 313-321.[38] Strazielle N, Ghersi-Egea JF, Ghiso J, Dehouck MP, Fran-

gione B, Patlak C, Fenstermacher J, Gorevic P (2000)In vitroevidence that beta-amyloid peptide 1-40 diffuses across theblood-brain barrier and affects its permeability.J NeuropatholExp Neurol59, 29-38.

[39] Crawford MP, Yan SX, Ortega SB, Mehta RS, Hewitt RE, PriceDA, Stastny P, Douek DC, Koup RA, Racke MK, KarandikarNJ (2004) High prevalence of autoreactive, neuroantigen-specific CD8+ T cells in multiple sclerosis revealed by novelflow cytometric assay.Blood103, 4222-4231.

[40] Venken K, Thewissen M, Hellings N, Somers V, Hensen K,Rummens JL, Stinissen P (2007) A CFSE based assay formeasuring CD4+CD25+ regulatory T cell mediated suppres-sion of auto-antigen specific and polyclonal T cell responses.J Immunol Meth322, 1-11.

[41] Rojo LE, Fernandez JA, Maccioni AA, Jimenez JM, MaccioniRB (2008) Neuroinflammation: implications for the patho-genesis and molecular diagnosis of Alzheimer’s disease.ArchMed Res39, 1-16.

[42] Marins Okura Y, Kohyama K, Park IK, Matsumoto Y (2008)Nonviral DNA vaccination augments microglial phagocytosisof beta-amyloid deposits as a major clearance pathway in anAlzheimer disease mouse model.J Neuropathol Exp Neurol67, 1063–1071.

[43] Speciale L, Ruzzante S, Calabrese E, Saresella M, TaramelliD, Mariani C, Bava L, Longhi R, Ferrante P (2003) 1-40 Beta-amyloid protein fragment modulates the expression of CD44and CD71 on the astrocytoma cell line in the presence ofIL1beta and TNFalpha.J Cell Physiol 196, 190-195.

[44] Lynch MA (2009) The multifaceted profile of activated mi-croglia.Mol Neurobiol40, 139-56.

[45] Jana M, Palencia CA, Pahan K (2008) Fibrillar amyloid-beta peptides activate microglia via TLR2: implications forAlzheimer’s disease.J Immunol181, 7254-7262.

[46] Shaftel SS, Carlson TJ, Olschowka JA, Kyrkanides S, Ma-tousek SB, O’Banion MK (2007) Chronic interleukin-1betaexpression in mouse brain leads to leukocyte infiltration andneutrophil-independent blood brain barrier permeabilitywith-out overt neurodegeneration.J Neurosci27, 9301–9309.

[47] Rezai-Zadeh K, Gate D, Town T (2009) CNS infiltration of pe-ripheral immune cells: D-Day for neurodegenerative disease?J Neuroimmune Pharmacol4, 462-475.

[48] Town T, Laouar Y, Pittenger C, Mori T, Szekely CA, Tan J,Duman RS, Flavell RA (2008) Blocking TGF-beta-Smad2/3innate immune signaling mitigates Alzheimer-like pathology.Nat Med14, 681-687.

[49] Hawkes CA, McLaurin J (2009) Selective targeting of perivas-cular macrophages for clearance of beta-amyloid in cerebralamyloid angiopathy.Proc Natl Acad Sci U S A106, 1261-1266.

[50] Tan J, Town T, Suo Z, Wu Y, Song S, Kundtz A, Kroeger J,Humphrey J, Crawford F, Mullan M (1999) Induction of CD40on human endothelial cells by Alzheimer’s beta-amyloid pep-tides.Brain Res Bull50, 143-148.

[51] Tan J, Town T, Mullan M (2002) CD40-CD40L interaction inAlzheimer’s disease.Curr Opin Pharmacol2, 445-451.

[52] Town T, Nikolic V, Tan J (2005) The microglial activationcontinuum: from innate to adaptive responses.J Neuroinflam-mation2, 24.

[53] Larbi A, Pawelec G, Witkowski JM, Schipper HM, Derho-vanessian E, Goldeck D, Fulop T (2009) Dramatic shifts in cir-culating CD4 but not CD8 T cell subsets in mild Alzheimer’sdisease.J Alzheimers Dis17, 91-103.

938 M. Saresella et al. / Treg in AD-Associated Neuroinflammation

[54] Bryl E, Witkowski JM (2004) Decreased proliferative capabil-ity of CD4(+) cells of elderly people isassociated with fasterloss of activation-related antigens and accumulation of regu-latory T cells.Exp Gerontol39, 587-595.

[55] Trzonkowski P, Szmit E, Mysliwska J, Mysliwski A (2006)CD4+CD25+ T regulatory cells inhibit cytotoxic activity ofCTL and NK cells in humans-impact of immunosenescence.Clin Immunol119, 307-316.

[56] Cortesini R, LeMaoult J, Ciubotariu R, Suciu-Foca N (2001)CD8+CD28−T suppressor cells and the induction of antigen-

specific, antigen-presenting cell-mediated suppression of Threactivity. Immunol Rev182, 201-206.

[57] Suciu-Foca N, Manavalan JS, Scotto L, Kim-Schulze S, Gal-luzzo S, Naiyer AJ, Fan J, Vlad G, Cortesini R (2005) Molec-ular characterization of allospecific T suppressor and tolero-genic dendritic cells: review.Int Immunopharmacol1, 7-11.

[58] Ma SL, Tang NL, Lam LC, Chiu HF (2005) The associationbetween promoter polymorphism of the interleukin-10 geneand Alzheimer’s disease.Neurobiol Aging7, 1005-1010.