Mechanisms of Immune Suppression in Patients with Head and … · 96 CD34 Cells in Head and Neck Cancers ies with the munine LLC model have shown that several immune suppressive populations

Vol. 1, 95-103, January 1995 Clinical Cancer Research 95

Mechanisms of Immune Suppression in Patients with Head and

Neck Cancer: Presence of CD34� Cells Which Suppress Immune

Functions within Cancers That Secrete Granulocyte-Macrophage

Colony-Stimulating Factor’

Annette Schmidt Pak, Mark A. Wright,

John P. Matthews, Sharon L. Collins,

Guy J. Petruzzelli, and M. Rita I. Young2

Departments of Research Service [M. R. I. Y., M. A. W.] and

Otolamyngology [S. L. Cl, Hines VA Hospital, Hines, Illinois 60141,and the Departments of Pathology [M. R. I. Y., A. S. P., i. P. M.1

and Otolaryngology [G. J. P.], Loyola University Stritch School ofMedicine, Maywood, Illinois 60153

ABSTRACT

Production of granulocyte-macrophage colony-stimu-

lating factor (GM-CSF) by murine tumors has been shown

to induce immune suppressive cells having homology with

GM progenitor cells. The purpose of this study was to

determine if human head and neck cancers secrete GM-

CSF, if this is associated with an intratumoral presence of

similar cells expressing the hematopoietic progenitor cell

antigen CD34, and if such CD34� cells suppress functions of

intratumoral T cells. This was evaluated with fresh head and

neck cancers, and in some instances regional lymph nodes

and control tissue. Ten of the 14 squamous cell carcinomas(SCCs) studied secreted greater than 5 ng GM-CSF/g tissue.

GM-CSF was not secreted in significant levels by either the

other cancer types or by control normal muscle. Each of the

high GM-CSF-secreting SCCs, but none of the cancers that

did not secrete GM-CSF, contained cells expressing the

hematopoietic progenitor cell antigen CD34 that had the

capacity to grow into colonies in soft agar. Available re-

gional lymph nodes from patients with high GM-CSF-pro-

ducing cancers also contained CD34� cells. Depletion of

CD34’� cells from dissociated cancers increased interleukin2 secretion by the intratumoral lymphocytes while addition

of the CD34� cells to dissociated cancers reduced interleu-

kin 2 production, indicating that the presence of CD34� cells

within GM-CSF-producing head and neck SCCs results in

suppressed functional competence oflymphocytes within the

SCCs. These results show that GM-CSF-secreting SCCs

contain cells expressing the hematopoietic antigen CD34

which are inhibitory to the capacity of lymphocytes within

the SCCs to secrete interleukin 2.

INTRODUCTION

Clinically, patients with head and neck SCCs3 have been

recognized to have profound defects in cellular immune respon-

siveness (1). Risk factors for development of this type of cancer

include the initiating carcinogens in tobacco and possibly pro-

moting factors in alcohol. Causes of the diminished immune

responsiveness include ‘ ‘extrinsic’ ‘ factors such as the malnour-

ished state of the patients due to interference of the cancer with

swallowing (2). In addition, patients with recurrent cancers have

often had prior radiation therapy which, although localized to

the head and neck, is well known to cause profound long-term

alterations in the immune response. Attempts to define the

cellular basis of the ‘ ‘intrinsic’ ‘ factors in patients with head and

neck SCCs that subvert immune responsiveness have, until the

past decade, been limited to relatively nonspecific assessments

of systemic immune capabilities.

More recent basic research studies have shown that cancers,

including head and neck cancers, are vulnerable to immune effeetom

cells such as natural killer cells, lymphokine-activated killer cells,

and tumor-specific eytotoxic T lymphocytes (1-4). That head and

neck cancers can evoke immune responses in vivo is suggested by

the selective homing of CD8� T cells into the cancer (5). Studies

with other cancers have suggested that immune responses to can-

cems may also be important in maintaining cancer dormancy (6).

Although cancers are susceptible to immune-mediated de-

struetion, patients with head and neck cancers and other cancer

types have defects in their cellular immune functions, including

the tumor-infiltrating T cells and natural killer cells (5, 7, 8).

Some of this immune deficiency may be directly induced by

soluble suppressive factors produced by the cancers (9, 10).

Prostaglandins are among the suppressor factors that have been

shown to be produced by some, although not all, head and neck

cancers (9, 10). Cancers have also been shown to indirectly

suppress immune functions by inducing immune suppressor

cells. The immune suppressor cells that may be induced by head

and neck cancers have not been extensively studied or de-

senibed, but have been well documented in many other cancer

types in both humans and animals. Some human cancers induce

suppressive T cells that are inhibitory to antitumor T-eell meac-

tivity (11). Other reports have shown suppression of T-eell

function in cancer patients to instead be due to PGE2 that is

secreted by mature monocytes/macnophages (12, 13). Our stud-

Received 7/18/94; accepted 9/16/94.

I Supported by grants from the Medical Research Service of the De-pamtment of Veterans Affairs, by Grants CA-45080 and CA-48080 from

NIH, and by the American Cancer Society, Illinois Division.2 To whom requests for reprints should be addressed, at PathologyResearch (151-Z2), Hines VA Hospital, Hines, IL 60141.

3 The abbreviations used are: SCCs, squamous cell carcinomas; PGE,,

prostaglandin E2; LLC, Lewis lung carcinoma; GM-CSF, gmanulocyte-

macmophage colony-stimulating factor; TGF�3, transforming growth fac-

tom �3; FACS, fluorescence-activated cell sorting; ELISA, enzyme-linked

immunosombent assay; IL-2, intemleukin 2.

Research. on February 17, 2021. © 1995 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

96 CD34� Cells in Head and Neck Cancers

ies with the munine LLC model have shown that several immune

suppressive populations can appear during the course of tumor

growth, including PGE,-pmodueing macrophages during early

phases of tumor growth (14). The significance of suppressor

cells in limiting immunological destruction of tumor is under-

scored by the tumor regression, reduced metastasis, and in-

creased host survival that occurs when suppressor T cells are

reduced in tumor-bearing animals with low-dose chemotherapy,

on when macnophage suppressive activity is blocked with pros-

taglandin synthesis inhibitors (15, 16).

Contrasting with some tumor types, head and neck cancers

have been suggested not to induce lymphoid immune suppres-

sive cells (7). However, studies with experimental animal to-

moms have indicated that some tumors induce immune suppmes-

son cells that are neither suppressor T cells nor macrophages,

and whose presence is induced by tumor production of GM-CSF

(17-20). These suppressor cells are an immature population of

bone marrow-derived cells which appear as a result of the

myelopoietie stimulation that is induced by tumor-produced

GM-CSF. We have shown that in LLC tumor bearers, this

myelopoiesis-assoeiated suppressor population becomes the

dominant immune suppressor cell as LLC tumor growth

progresses, appearing first in the bone marrow and then also

becoming visible within the tumor (18, 19, 21). Extensive phe-

notypic and functional characterizations have shown that these

GM-CSF-indueed suppressor cells resemble GM progenitor

cells and mediate their suppressive effects in part by producing

TGF�3 (18, 19, 22, 23). The in vito functional significance of

such suppressor cells in influencing the course of tumor devel-

opment was suggested when LLC-indueed suppressor bone

marrow cells enhanced growth and metastasis of coinoculated

LLC tumor cells (18).

Many tumor types, such as human lung cancer, breast

cancers and hepatocellular carcinomas, and an equally broad

spectrum of munine tumors, express GM-CSF mRNA and pro-

tein (17, 24-27). Despite the demonstration in animals that

tumor production of GM-CSF results in the induction of sup-

pressom cells, no studies have evaluated whether human head

and neck cancers secrete GM-CSF or if this induces an intratu-

moral influx of cells that are homologous to the immune sup-

pressive GM progenitor cells that have been described for

GM-CSF-pmodueing animal tumors. The objective of this study

was to determine if human head and neck cancers secrete

GM-CSF and if this is associated with the presence of immune

suppressive cells within the cancer that resemble GM progenitor

cells. Our results show that a high frequency of human head and

neck SCCs secrete GM-CSF and that these cancers contain cells

resembling GM progenitor cells whose presence results in me-

duced function of intratumomal T lymphocytes.

MATERIALS AND METHODS

Human Head and Neck Cancers. Fresh nonneerotie

human head and neck cancer specimens, involved regional

lymph nodes, and adjacent normal sternoeleidomastoid muscle

biopsies were obtained immediately after surgical removal.

Most of these cancers, whose histologies are described in Table

1 , were SCCs. The head and neck cancer specimens were

prepared for analysis by separating tumor tissue from necrotic

tissue and then mincing the tissue into 2-4-mm fragments. The

fragments were either directly placed into culture for measure-

ment of GM-CSF secretion, or were dissociated into single-cell

suspensions. For analyses requiring single-cell suspensions of

cancer specimens, the fragments were enzymatically dissociated

by incubation for 2 h in a mixture of 1 mg/mI collagenase type

IV-0.5 mg/mI hyaluronidase type V-0.1 mg/ml DNase type 1

(Sigma, St. Louis, MO). Single-cell suspensions of the dissoci-

ated tumor were centrifuged in a Ficoll-Hypaque gradient (1.077

g/liter), recovered from the interface, washed, and either FACS

analyzed or seeded into culture. The culture medium used for all

studies was RPMI 1640 containing 10% endotoxin-free defined

fetal bovine serum, 100 units/mI penicillin, 100 mg/ml strepto-

mycin, 0.02 M 4-(2-hydmoxyethyl)-1-pipemazineethanesulfonie

acid buffer, S X i0� M 2-mercaptoethanol and 2 msi L-glu-

tamine (Sigma). For two of the head and neck SCC specimens

studied, cancer lines were established by culturing fragments

and selectively trypsinizing explants to enrich for cancer cells.

Once cancer cultures became established, they were maintained

in culture medium and passaged twice weekly.

FACS Analysis. Dissociated tumor samples were stained

by direct immunofluorescence using fluomescein isothiocyanate-

conjugated CD34 antibody (QBEND/10 clone; BioSource, Ca-

mamillo, CA) or a fluorescein isothiocyanate-conjugated isotype

control antibody. After 30 mm of staining on ice in the dank, the

cells were washed and the percentage of positive staining cells

was measured using a FACS 420 (Beeton Dickinson, Sunny-

vale, CA). Positive cells were quantified using marker settings

established to exclude 95% of the cells that were incubated with

the isotype control antibody. Data shown are FACS histograms

and the percentage of cells identified by these analyses to be

positive.

GM-CSF Secretion. To measure GM-CSF secretion,

minced fragments (2-4 mm) of cancer specimens were individ-

ually cultured with 1 ml culture medium for 24 h. The super-

natants were then collected and the tumor pieces were weighed.

The amount of GM-CSF in the supemnatant was quantitated by

ELISA, having a sensitivity of 20 pg/ml using a combination of

capture and secondary antibodies that was paired for this appli-

cation (PhamMingen, San Diego, CA). Results were expressed as

ng GM-CSF secreted per g tumor tissue (mean ± SD).

Immunohistology. Multiple areas throughout the tissue

samples were immunohistologically analyzed to obtain a col-

leetive representation of the cells expressing CD34 or producing

GM-CSF. Specimens were eryoseetioned, fixed, and incubated

with antibody to GM-CSF (PharMingen), CD34 antibody, or

with isotype control antibody (Biosounee). To visualize cells

producing GM-CSF or expressing CD34, sections were stained

with secondary antibody and the Vectastain ABC immunoper-

oxidase technique, and then eountemstained with ethyl green.

Growth of Immunomagnetically Isolated CD34� Cells

in Soft Agar. The capacity of CD34� cells to grow into

granulocytie and/or monocytie colonies in soft agar was mea-

sured. CD34� cells were immunomagnetically isolated from

freshly dissociated head and neck cancer specimens, using a

procedure previously shown to result in a high degree of pro-

genitom cell purity without diminishing their capacity for growth

(28). Briefly, dissociated tumors were adjusted to 10”/ml and

mixed with CD34 antibody-coated magnetic beads (Advanced



I 2 3 4 5 6 7 8 9 1011 12131415161718

Patients

Clinical Cancer Research 97

Magnetics, Cambridge, MA) at 4#{176}Cfor 15-20 mm. The cells

that complexed with the beads were separated with a magnetic

plate at a perpendicular angle to gravity. After 3-S mm, the cells

remaining in suspension were removed by pipetting and dis-

carded. The cells that remained in the flask attracted to the

magnetic plate (CD34” population) were released into suspen-

sion, washed, and again magnetically isolated from the cells in

suspension. Various numbers of cells in the CD34� fraction

were added to 35- x 10-mm tissue culture dishes with 1 X 10”

irradiated (25 Gy) spleen cells as feeder cells in 1 ml semisolid

RPMI 1640 supplemented medium. The medium was supple-

mented with 20% fetal bovine serum, 0.3% agam (Bacto-Agam;

Difco, Detroit, MI), and a mixture of 20% (v/v) GCT superna-

tant (Sigma) plus 10% supemnatant from GM-CSF-secreting

head and neck cancer lines as sources of CSFs. Cultures were

incubated at 37#{176}Cfor 6 days, after which colonies (>50 cells)

were counted. The cells which composed the colonies were

identified histologically after fixing colonies with pama-

formaldehyde and methanol, and staining with hematoxylin and

eosin.

IL-2 Production by Intratumoral Lymphocytes. The

effect of the presence of CD34� cells on IL-2 production by

lymphocytes of autologous cancers was measured after immu-

nomagnetic separation of CD34� cells from freshly dissociated

head and neck cancer specimens, as described above. The pos-

itively selected CD34� cell population was added to aliquots of

unfractionated dissociated tumor to also establish CD34�-en-

niched tumor cultures. Thus, the following dissociated cancer

cultures were prepared to assess the effects of CD34” cells on

IL-2 production by intratumoral T cells: (a) CD34-depleted

cultures; (b) CD34-enmiched cultures; and (c) unfractionated

cultures, when sufficient cells were available. FACS analysis of

a fraction of each population showed the absence of CD34’

cells in the CD34-depleted cultures and a proportionate increase

(by 5-10%) of CD34� cells in the CD34-enriehed cultures. The

remaining cells of each of these three populations were incu-

bated on CD3 antibody- (Biodesign, Kennebunkpont, ME)

coated tissue culture plates to induce IL-2 production (29). After

2 days, culture supemnatants were collected and the IL-2 content

was measured at least in quadmuplicate using ELISA (Endogen,

Boston, MA, or Inestar, Stillwatem, MN). The significance of the

difference between values was determined using Student’s t test.

RESULTSGM-CSF Secretion by Head and Neck Cancers.

Freshly excised specimens of human head and neck cancers

were obtained, with specimens from 18 patients being of suffi-

cient size to measure GM-CSF secretion and the frequency of

intratumomal cells expressing the hematopoietic progenitor anti-

gen CD34 (Fig. 1). Fourteen of these 18 cancers were SCCs, as

summarized in Table 1. Using ELISA to measure levels of

GM-CSF secreted by fragments of these cancer tissues, 10

(56%) of these 18 cancers were shown to secrete high levels

of GM-CSF, with an overall mean ± SE of 15.0 ± 2.1 ng

GM-CSF/g tumor tissue. The remaining 8 samples secreted low

levels of GM-CSF, averaging 2.6 ± 0.7 ng GM-CSF/g tumor

tissue. All of the high GM-CSF-producing cancers were SCCs

and none of the non-SCCs secreted significant levels of GM-

Fig. I Human head and neck cancer secretion of GM-CSF and intra-tumomal CD34� cell content. Fragments of cancer specimens were

individually cultured for 24 h, after which supemnatants were removed

and their GM-CSF content was measured by ELISA. Top, Data shown

is the mean of tmiplicates (ng GM-CSF/g tissue) ± SD. Enzymaticallydissociated cancer fragments were centrifuged through a discontinuous

density gradient, recovered from the interface, and stained by direct

immunofluomescence using fluomescein isothiocyanate-conjugated

CD34. Bottom, The proportion of CD34� staining cells was measuredby FACS and shown as percentage of positive-staining cells.

CSF. GM-CSF production by normal muscle biopsies (sterno-

cleidomastoid muscle from the operative field) from 9 of the

studied patients was also measured and was found to be insig-

nificant (values for several patients in Table 2).

Whether the cancer cells were capable of secreting GM-

CSF was determined by establishing two head and neck cancer

cell lines from GM-CSF-secneting cancers used in the studies

described in Fig. 1. The amount of GM-CSF secreted by these

cells was measured after oven 3 months of passage to assume loss

of infiltrating cells. Both of these cancer lines secreted GM-

CSF, with cells from patients HN13 and HN1S secreting 451 pg

and 598 pg GM-CSF, respectively, pen 10” cells during 18 h of

culture. In addition, head and neck 5CC tissues were analyzed

for cancer cell production of GM-CSF by immunostaining eryo-

sections with anti-GM-CSF antibodies. Shown in Fig. 2 is a high

frequency of cells staining strongly and moderately for GM-

CSF in sections from a cancer that was shown by ELISA to

secrete GM-CSF. Also shown is negatively staining tissue from

a cancer shown by ELISA not to secrete GM-CSF. These data

support the results of ELISA analyses showing that head and

neck cancers can secrete GM-CSF.




Table I Head and neck cancers

Prior

radiation Multiple 1#{176}

Patient Age Sex TNM Histology Site Smoking Alcohol therapy tumors Outcome

HN1 62 M T,N) 5CC, poorly dif.” Larynx (supraglottic) No, 25 yr None Yes No NED

HN2 49 M T3N SCC, moderately dif. Larynx (supraglottic;

false and true cord)Heavy Heavy No No NED

HN3 46 M T2N 5CC, well dif. Right floor mouth Heavy Heavy Yes Yes Recurrent lO

HN4 40 M T3N Adenocamcin., moderatelydif.

Ethmoid sinus None Moderate No No NED

HN5 63 M T4N,� SCC Pharyngeal wall Heavy Heavy No No NED

HN6 67 F T4N0 5CC, poorly dif. Larynx (pymifomm sinus) Heavy Small Yes Yes Recurrent I”

HN7 45 M T4N� 5CC, moderately dif. Trachea (2#{176}occurrence) Heavy Small Yes No Recurrent

10, DWD

HN8 62 M T4N2� 5CC, moderately dif. Larynx (supmaglottic) Heavy Heavy No No Recurrent lO

HN9 70 F T,N(, SCC Tongue None None No No NED

HNIO 48 F T4N1 Carcinoma ex-pleomomphic

adenoma (salivary gland)

Pamaphamyngeal space Moderate Moderate No No NED

HN1 1 73 M T3N 5CC Larynx (false cord) Heavy Heavy No No NED

HNI2 48 M T,N� High grade intestinal-type

adenocamcinoma

Left ethmoid maxillary

sinus

Heavy Small No No NED

HN13 54 F T4N2h SCC, poorly dif. Larynx Heavy Heavy No Yes NEDHN14 62 M T3N0 5CC, moderately dif. Larynx (true cord) Heavy Moderate No No NED

HN1S 59 F T4N1 SCC, moderately dif. Tongue Heavy Heavy No No NED

HN16 62 M T,N,� 5CC, moderately dif. Larynx (supraglottic) Heavy Heavy No No NED

HN17 61 M T4N0 High grade osteogenic

sarcoma

Mandible Heavy Heavy No No NED

HN18

a di

66

f., dif

M

fementi

T3N1

ated; N

SCC, moderately dif.

ED, no evidence of disease;

Larynx (pymiform sinus)

DWD, dead with disease.

No, 23 yr Small No No NED

Table 2 GM-CSF production and CD34� cell content of cancers andregional lymph nodes

Tumor-��---- Specimen

GM-CSFsecreted”

(ng/g tissue)

CD34�

cells”(%)

HNS Larynx, SCC 8.9 ± 0.3 13.8

Jugulocarotid node 20.2 ± 2.0 13.5

Jugulodigastmic node 10.9 ± 4.6 14.2

Lower jugular node 0.0 ± 0.4 13.0HNIO Salivary gland, came. expliomomphic 0.1 ± 0.1 2.1

adenomaCervical node 2.0 ± 3.8 0.7

Normal muscle 0.3 ± 0.3

HNI3 Larynx, 5CC 10.2 ± 1.2 10.1

Cervical node 12.6 ± 1.8 14.8

Tongue, 5CC 5.1 ± 0.9 10.4


HN1S Tongue, 5CC 25.9 ± 7.5 10.6

Left upper jugular node 7.0 ± 3.3 6.2


HN18 Larynx, pymiform sinus, 5CC 16.9 ± 3.0 9.2

Jugulodigastnic node 0.0 ± 0.2 7.4

�1 GM-CSF secreted by tissue fragments is expressed as ng GM-

CSF/g tissue (mean ± SD)./‘ The proportion of CD34� cells in enzymatically dissociated tis-

sue is expressed as percentage of positive cells from flow cytometmie

analysis of each sample.

CD34� Cells within GM-CSF-Producing Human Headand Neck SCCs. Since our prior studies with mumine tumors

showed GM-CSF-producing tumors induce an intratumoral in-

flux of immune suppressive cells resembling GM progenitor

cells, the human head and neck tumors were analyzed for the

presence of cells that express the human progenitor cell antigen

CD34. This was accomplished both by FACS and immunohis-

tochemical analysis of enzymatically dissociated tumor samples.

Examples of FACS analyses for 4 of these specimens that

contained CD34” cells and 1 that did not contain CD34� cells

(from patient HN 1 1) are shown in Fig. 3. Summarized in Fig. 1

is the percentage ofCD34’ cells within the 18 tested dissociated

head and neck cancers. Overall, the cancer samples could be

divided into two groups on the basis of the CD34� cell content.

One group consisting of 8 cancers contained only a minimal

number of CD34�-staining cells (1.4 ± 0.6%), which appmox-

imated the level of staining with the isotype control antibody.

Although the number of non-SCC samples was small (n = 4),

all of them were in this group of cancers that did not contain

CD34� cells. The second group of cancers contained a small,

but distinct, population of CD34� cells which composed ap-

proximately 10 ± 1% ofthe cells recovered from the dissociated

cancers. The group of cancer specimens that contained CD34’

cells was the same group of SCCs that secreted the high levels

of GM-CSF. Linear regression analysis did not show a highly

significant correlation between the absolute amount of GM-CSF

the tumors secreted and the number of intratumomal CD34� cells

(,2 0.65), but all 10 of the SCCs that secreted more than S ng

GM-CSF/g tissue contained the higher content of CD34” cells.

To determine if the presence of CD34 cells may have been

simply a reflection of the extent of immune infiltration in the

cancer, and not necessarily associated with the amount of GM-

CSF secreted by the cancer, the frequency of CD4” plus CD8�

T cells within the cancer was measured. The results of these

FACS analyses showed that the CD34” cell content was not

associated with the number of infiltrating T cells and, in fact, the




-S �

HNI6

HNI6

k’�t .. -.-

Fig. 2 Immunostaining of cancers for GM-CSF. Cryosections of SCCs were stained with immunopemoxidase for GM-CSF, and then counterstained

with ethyl green. Shown are cells staining positive for GM-CSF production in sections from patient HN13 (left) and a lack of staining in a sectionfrom patient HN16 (right). Top row, X 400; bottom row, X 1000.

high GM-CSF-secreting cancers contained fewer T cells (6.8 ±

3.3%) than did the low GM-CSF-secmeting cancers (12.9 ±

3.4%).

Regional lymph node specimens containing metastatic can-

cen were also available in sufficient size from four patients with

high GM-CSF-pmoducing SCCs and from one patient with a low

GM-CSF-producing adenoma to measure both GM-CSF pro-

duction and CD34’ cell content (Table 2). Node samples from

three of the four patients with high GM-CSF-producing primary

cancers also secreted high levels of GM-CSF. A node from one

of these three patients and the node sample from the fourth

patient with a GM-CSF-pmodueing primary cancer did not se-

crete GM-CSF. Evaluation of these nodes for presence of

CD34” cells showed that the nodes that secreted high levels of

GM-CSF also had a significant number of CD34� cells. In

addition, the nodes that did not secrete GM-CSF but which were

obtained from patients with GM-CSF-seereting SCCs contained

CD34� cells. The node from the patient with a low GM-CSF-

producing cancer did not secrete significant levels of GM-CSF

and did not have significant levels of CD34” cells. Although

these results are from a relatively small number of samples, they

showed that CD34” cells could be found in GM-CSF-secmeting

SCCs and in nodes involved with metastatic cancer.

Characterization of Intratumoral CD34� Cells. The

results described above showing that cancer production of GM-

CSF was associated with an intmatumonal presence of cells that

express the human progenitor cell antigen CD34 prompted

studies to evaluate the identity of these CD34’ cells. Immuno-

histological analyses were first performed to confirm the pres-

ence of CD34’ cells and to visualize their distribution within

the cancers. An example of CD34� cells within a GM-CSF-

secreting cancer is shown in Fig. 4. These CD34� cells stained

intensely and typically appeared in small clusters, contrasting

with the occasionally observed endothelial cells which stain

diffusely with CD34 (30). To evaluate the possibility that the

CD34� cells could be myeloid progenitor cells, the capacity of

CD34� cells isolated from head and neck SCCs to grow into

colonies in soft agan containing CSFs was determined. In the

presence of CSFs, the isolated CD34’ cells grew into colonies

with an efficiency of 46 ± 8% (mean ± SE of results from two

cancer samples). Fig. 5 shows the appearance of colonies that

formed from one of these head and neck SCCs. Most of the

colonies that formed (>70%) consisted of both monocytes and

gmanulocytes. The remaining colonies contained either granulo-

cytes on monocytes. In the absence of CSFs, isolated CD34�

cells failed to form colonies, supporting the possibility that a



“HN2” “HN8”

00

100 CE)34 #{149}Cells in Head and Neck Cancers

I-

ci��0E:�

C

a)0

�‘I 10 100 lOOm 10 100 1.0001 10 100 1.0001 10 100 1.uOf� 10 100 1,000

l�’ig. 3 Examples of FACS analyses for CD34 � cells within dissociated human head and neck cancer specimens. Fragments of GM-CSF-producing

cancers ( HN2. HN8. HN I 3. and HN I 5 ) and a non-GM-CSF-producing cancer ( HN I I ) were enzyniatically dissociated and centrifuged through a

discontinuotis density gradient. Cells recovered from the interface were then stained by direct imniunotluorescence with fluorescein isothiocyanate-

labeled isotype control or CD34 antibody. Histograms from flow cytometmic analyses show that the GM-CSF-prodttcing cancers have an intratumomal

(‘1)34 ‘ cell suhpopulation while the cancer that does not secrete GM-CSF is devoid of CD34 � cells.

large proportion of the intratunioral CD34 � cells are GM pro-

genitor cells.

IL-2 Production by T Lymphocytes from Dissociated

Cancer after Depletion of CD34� Cells. Our prior studies in

a munine tumor model showed that GM-CSF-producing tumors

induced intratumoral infiltration of cells that resembled GM

progenitor cells and were immune suppressive (18. 21. 22).

Since human head and neck cancers that produce GM-CSF

contained cells expressing the progenitor cell antigen CD34

(Fig. I ). the possibility was examined that these cells niight be

immune suppressive to the activity of the intratumoral T cells.

Primary 5CC specimens from 5 of the I 8 above-described head

and neck cancer patients were large enough for these analyses.

This was accomplished by deterniining if depletion of CD34

cells from dissociated cancer would result in an increased ca-

pacity of the intraturnoral T cells to he induced by immobilized

CD3 to produce lL-2. In the absence of stimulation with immo-

hilized CD3. lL-2 was not secreted by the dissociated cancer

(not shown). Afler incubation for 2 days on immobilized CD3.

lL-2 secretion was induced in each of the unfractionated disso-

ciated cancer cultures (Table 3). In each of the 4 analyzed

GM-CSF-producing samples that contained CD34 � cells. IL-2

production was approximately doubled when CD34 ‘ cells were

depleted as compared to the cultures to which CD34 � cells were

added. Addition of the CD34� cells to unfractionated dissoci-

ated tumor cultures further diminished by almost one half their

capacity to be induced to produce IL-2. By contrast. similar

treatment with CD34-coated magnetic beads had no effect on

lL-2 secretion by an analyzed tumor sample that did not contain

CD34 � cells. These studies suggest that in the presence of

CD34 � cells the capacity of the intratumoral T cells to be

induced to secrete IL-2 is reduced. and that removal of these

CD34 � cells enhances IL-2 production.

DISCUSSION

Many human and animal tumor types have been shown to

produce GM-CSF (24, 25. 31 ) which can modulate antitumor

immune responses both positively and negatively (17, 26, 32).

The demonstration that GM-CSF can have potent immune stim-

ulatory effects has led to the inclusion of GM-CSF in strategies

to develop tumor vaccines (32. 33). For example. vaccines

composed of irradiated tumor cells engineered to secrete GM-

CSF have been shown to stimulate long-lasting antitumor im-

munity in murine tumor models (32). Negative immune regula-

tory effects of GM-CSF have also been reported. but are

typically indirect and are mediated by the populations of im-

mune suppressive cells that can be induced by GM-CSF, includ-

ing mature suppressor macrophages ( I 7). A second population

of immune suppressor cells is induced as a result of GM-CSF

stimulation of myelopoiesis and appears to resemble GM pro-

genitor cells (17. 18, 20). However, human head and neck

cancers have not been previously studied for whether they

secrete GM-CSF or whether this might be associated with the

presence of immune suppressor cells that resemble the GM

progenitor cells.

The results of the present study have shown that approxi-

mately two thirds of the evaluated head and neck SCCs. but

none of the non-SCCs. secrete GM-CSF and contain cells ex-

pressing the hematopoietic progenitor cell antigen CD34. The

SCCs that did not secrete GM-CSF did not contain significant

numbers of CD34� cells. Cancer-containing lymph nodes that

were regional to GM-CSF-secreting primary SCCs also con-

tamed CD34 cells and most were shown to produce GM-CSF.

The source of the GM-CSF was most likely the cancer within

the nodes, although this has not yet been rigorously studied. In

a few instances, there were differences in GM-CSF production



I

4z� �

.-*‘� C.

C


#{149}

;“: -%�

‘�-

Fig. 4 Immunostaining for CD34 � cells within SCCs. Cryosections of

a GM-CSF-secmeting 5CC (HN2) were stained by immunoperoxidase

for CD34� cells, and then counterstained with ethyl green. Shown are

positive staining cells in a section stained for CD34� cells (top) and a

control negative stain of the same cancer specimen (bottom). X 1000.

between the cancer samples and involved node. This may be due

to the variability in the extent or distribution of cancer within the

node.

That the head and neck cancer cells within the surgical

specimens were a significant source of the secreted GM-CSF

was supported by the demonstration that the head and neck

cancer lines that we established from GM-CSF-pmoducing spec-

imens were also capable of secreting GM-CSF. In addition,

immunohistochemical analysis of cryosections from several of

the cancers shown by ELISA to secrete GM-CSF also showed a

high number of cancer-like cells expressing GM-CSF.

As to whether the CD34� cells are in fact GM progenitor

cells was supported by the capacity of isolated CD34 ‘ cells to

grow into colonies composed of granulocytes and monocytes in

the presence of CSFs but not in the absence of a CSF source. Of

principal significance is our demonstration that in the presence

of these CD34’ cells, the capacity of the SCC-denived lympho-

cytes to be activated to secrete IL-2 is reduced. Additionally,

removal of the CD34� cells from the tumor enhances the

capacity of the intratumoral T cells to be induced to secrete IL-2.

Although immune suppressive cells that are homologous to GM

progenitor cells have been shown to be induced by GM-CSF-

Fig. 5 Growth of CD344 cells isolated from human head and neckSCCs into colonies in soft agar. Immunomagnetically isolated CD34’

cells were seeded in soft agam with irradiated mouse feeder spleen cellsand CSF-containing supemnatants. Shown are colonies that formed after

6 days. Top. X 40; bottom, X 100.

secreting animal tumors (17, 18, 20), this has not been pnevi-

ously suggested for human cancers. However, studies with

normal human bone marrow have shown the presence of an

immune suppressive cell population that copumifies with my-

eloid progenitor cells (34). Our speculation is that this is the

same population of cells as we have shown to be present within

the GM-CSF-secmeting cancers.

The patients with GM-CSF-seemeting SCCs were not eval-

uated for whether the tumor-derived GM-CSF might have in-

creased the frequency of GM progenitor cells in the marrow,

although other studies have indicated myeloid stimulation in

patients with other types of GM-CSF-secmeting cancers (27, 31).

Interestingly, head and neck cancers have also been shown to

secrete high levels of PGE�, which can function as a negative

regulator of both immune competence and myelopoiesis (9, 35).

Thus, it is possible that production of PGE2 by the cancers might

function to restrict the frequency of CD344 cells which might

otherwise be further increased if PGE, production were not

increased on if prostaglandin synthesis inhibitors were being

administered.

Presence of immune suppressor cells that interfere in host

anti-tumor immune competence has been suggested to facilitate



Tumor

code

IL-2

secreted

(IU/ml)

Effect offractionation

(%)


Table 3 Increased IL-2 secretion after depletion of CD34 from the

tumor infiltrate”

Enrichment

vs.

depletion

(%)Cell treatment

HN1 None

CD34 depletionCD34 enrichment

4.5 ± 0.7

9.4 ± 0.7

2.5 ± 0.2

+ 1 10�....73**

HN7 None

CD34 depletionCD34 enrichment

3.5 ± 0.5

6.7 ± 1.5

2.4 ± .02

+88*

HN9 CD34 depletion

CD34 enrichment

4.5 ± 0.2

1.9 ± 0.1

HN16 CD34 depletion

CD34 enrichment

4.2 ± 0.7

5.1 ± 0.4 +21

HNI8 CD34 depletion

CD34 enrichment

5.9 ± 0.3

0.5 ± 0.1

‘, Dissociated cancers that had been shown to secrete high levels of

GM-CSF (HN1, HN7, HN9, and HNI8) or low levels of GM-CSF(HN16) were adjusted to 10” cells/mI and then either immunomagneti-

cally depleted of CD34 ‘ cells or enriched with immunomagneticallyisolated CD34� cells. IL-2 secreted by cell populations after incubation

on CD3-coated dishes is expressed as IU IL-2 secreted per ml ± SD ofat least 4 determinations. *, P < 0.05; **, P < 0.001.

the growth and metastasis ofeancem (15, 16, 36, 37). Expression

of GM-CSF by progressively growing murine tumors has been

associated with increased metastasis (22, 25). The studies pres-

ently being described did not demonstrate any clear correlation

between tumor stage and either the capacity of the tumor to

produce GM-CSF on to induce the intratumoral presence of

CD34’ cells, but they suggest an importance of these CD34�

cells in subverting immune competence. We are not, however,

suggesting that CD34� cells are necessarily the sole mediators

of immune suppression in head and neck cancer patients, whose

immune functions have previously been shown to be impaired

(5, 7). Cancers have been shown to directly suppress immune

competence by secreting immune suppressive mediators (9, 10),

or indirectly induce immune suppression by inducing immune

suppressor cells, including the more frequently studied suppres-

son macmophages on T cells (1 1, 13). Although we do not know

if these other suppressor cascades are present in GM-CSF-

secreting cancers that contain CD34’ cells, it is unlikely that

they are all mutually exclusive and most likely combinations of

suppressive mechanisms contribute to immune suppression in

patients with head and neck SCCs.

This report is the first demonstration of CD34� cells within

head and neck SCCs and of the reduced intmatumoral T-cell

function in the presence of these CD34’ cells. Other types of

cancers have not been previously evaluated for the presence of

CD34� GM progenitor cells having immune suppressive prop-

enties. We are currently continuing these studies to further

define whether the intnatumoral CD34� cells are present in

GM-CSF-seemeting cancers other than SCCs. Ongoing studies

are also aiming to delineate the mechanisms by which CD34’

cells mediate their suppression and whether they directly inhibit

functions of intratumonal T cells or if the suppressor cells are a

different population that is either induced by the CD34� cells or

is a progeny of the CD34� cells. These latter possibilities have

not been tested, but their likelihood is diminished by our dem-

onstmation that depletion of CD34 cells is sufficient to restore

the capability of intratumoral T cells to produce IL-2. Our

parallel studies with the mumine LLC tumor model have shown

that the GM progenitor cells mediate their suppressive activity

at least in part by producing TGF�3. The possibility that the

intratumoral CD34� cells of human head and neck SCCs also

mediate suppression by producing TGF�3 is currently being

evaluated. Since a variety of cancers have been shown to secrete

GM-CSF, it is possible that the presence of CD34’� cells that

inhibit intratumoral T-cell functions is not unique to head and

neck SCCs and may occur in other GM-CSF-secreting human

cancers.

ACKNOWLEDGMENTS

We are grateful to Dr. Wasim Raslan for his help in the patholog-

ical evaluation of the cancer specimens, Dms. iohn P. Leonetti and

Robert Bastian for providing some of the cancer specimens, and Dr.

Nirmala Bhoopalam for help in data analysis. We also thank Katie

O’Connor for hem help in preparation of this manuscript.

REFERENCES

1. Wanebo, H. J., Blackinton, D., Kouttab, N., and Mehta, S. Contmi-

bution of serum inhibitory factors and immune cellular defects to thedepressed cell-mediated immunity in patients with head and neck can-cem. Am. J. Sung. 166: 389-394, 1993.

2. Katz, A. E. Update on immunology of head and neck cancer. Med.Clin. North Am., 77: 625-631, 1993.

3. McGrath, P. C., Hamby, L. S., and Freeman, J. W. Phorbol dibutyrate

plus ionomycin improves the generation of cytotoxic T cells from

draining lymph nodes of patients with advanced head and neck cancer.

Am. J. Sung., 164: 610-614, 1992.

4. Zoechi, M. R., Ferramini, M., and Rugamli, C. Selective lysis of the

autologous tumor by d TSCI � wd� tumor-infiltrating lymphocytesfrom human lung carcinomas. Eur. J. Immunol., 20: 2685-2689, 1990.

5. Lakhdam, M., Ben Aribia, M. H., Maalej, M., and Ladgham, A.Selective homing of phenotypically lytic cells within nasophamyngealcarcinoma biopsies: numerous CD8- and CDI6-positive cells in thetumor. Int. J. Cancer, 48: 57-61, 1991.

6. Stewart, T. H., Hollinshead, A. C., and Raman, S. Tumour don-

mancy: initiation, maintenance and termination in animals and humans.Can. I. Sung., 34: 321-325, 1991.

7. Letessier, E. M., Sacehi, M., Johnson, J. T., Hembemman, R. B., and

Whiteside, T. L. The absence of lymphoid suppressor cells in tumor-

involved lymph nodes of patients with head and neck cancer. Cell.

Immunol., 130: 446-458, 1990.

8. Yoshino, I., Yano, T., Murata, M., Ishida, T., Sugimachi, K., and

Kimuma, G. Tumor-reactive T-cells accumulate in lung cancer tissues

but fail to respond due to tumor cell-derived factor. Cancer Res., 52:

775-781, 1992.

9. Lapointe, H., Lampe, H., and Banerjee, D. Head and neck squamous

cell carcinoma cell line-induced suppression of in vitro lymphocyte

proliferative responses. Otolamyngology, 106: 149-158, 1992.

10. Rice, S. Q., Crane, I. I., Scully, C., and Prime, S. S. Production ofa suppressor of lymphocyte proliferation by two human oral carcinoma

cell lines. Scand. I. Immunol., 36: 443-452, 1992.

11. Chakraborty, N. G., Twardzik, D. R., Sivanandham, M., Ergin,

M. T., Hellstmom, K. E., and Mukhemji, B. Autologous melanoma-induced activation of regulatory T cells that suppress cytotoxie me-

sponse. J. Immunol., 145: 2359-2364, 1990.

12. Wanebo, H. J., Riley, T., Katz, D., Pace, R. C., Johns, M. E., and

Cantrell, R. W. Indomethacin sensitive suppressor cell activity in headand neck cancer patients. The mole of the adherent mononuclear cell.Cancer, 61: 462-474, 1988.




13. Faminas, M. C., Rodniguez-Valvemde, V., Zamrabeitia, M. T., Pamma-

Blanco, J. A., and Sanz-Omtiz, J. Contribution of monocytes to thedecreased lymphoprolifemative response to phytohemagglutinin in pa-tients with lung cancer. Cancer, 68: 1279-1284, 1991.

14. Young, M. R., and Newby, M. Differential induction of suppressor

macmophages by cloned Lewis lung carcinoma variants. J. Natl. Cancer

Inst., 77: 1255-1260, 1986.

15. Inge, T. H., Hoover, S. K., Frank, I. L., Kawabata, T. T., Bethke,

K. P., and Bear, H. D. Enhancement of cytotoxic T lymphocyte growthfrom spleens of P815-tumor-beaming host mice with mafosfamide. Can-cem Immunol. Immunothem., 35: 119-126, 1992.

16. Khoo, N. K., Chan, F. P., Saarloos, M. N., and Lala, P. K. Immu-nothemapy of mammary adenocameinoma metastases in C3H/HeN micewith chronic administration of cyclo-oxygenase inhibitors alone or in

combination with IL-2. Clin. Exp. Metastasis, 10: 239-252, 1992.

17. Oghiso, Y., Yamada, Y., Ando, K., Ishihama, H., and Shibata, Y.

Differential induction of prostaglandin E2-dependent and -independentimmune suppressor cells by tumor-derived GM-CSF and M-CSF. J.Leuko. Biol., 53: 86-92, 1993.

18. Young, M. R. I., and Wright, M. A. Myelopoiesis-assoeiated im-mune suppressor cells in mice bearing metastatie Lewis lung carcinomatumors: �‘y intemfemon-g plus tumor necrosis factor a synergisticallyreduce immune suppressor and tumor growth-promoting activities of

bone marrow cells and diminishes tumor recurrence and metastasis.

Cancer. Res., 52: 6335-6340, 1992.

19. Young, M. R. I., Wright, M. A., and Young, M. E. Antibodies tocolony-stimulating factors block Lewis lung carcinoma cell-stimulationof immune suppressive bone marrow cells. Cancer Immunol. Immu-nothem., 33: 146-152, 1991.

20. Tsuehiya, Y., Igamashi, M., Suzuki, R., and Kumagai, K. Production

of colony-stimulating factor by tumor cells and the factor-mediatedinduction of suppressor cells. J. Immunol., 141: 699-708, 1988.

21. Young, M. R. I., MeCloskey, G., Wright, M. A., and Schmidt Pak,

A. Increasing infiltration and activation of CD8� tumor-infiltratinglymphocytes after eliminating immune suppressive gmanulocyte-mae-mophage progenitor cells with low doses of interfemon--y plus tumornecrosis factor-a. Cancer Immunol. Immunothem., 38: 9-15, 1994.

22. Young, M. R., Aquino, S., and Young, M. E. Differential induction

of hematopoiesis and immune suppressor cells in the bone marrowversus in the spleen by Lewis lung carcinoma variants. J. Leuko. Biol.,

45: 262-273, 1989.

23. Young, M. R. I., Wright, M. A., Coogan, M., Young, M. E., and

Bagash, J. Tumor-derived cytokines induce bone marrow suppressorcells that mediate immunosuppression through transforming growthfactor b. Cancer Immunol. Immunothem., 35: 14-18, 1992.

24. Tani, K., Ozawa, K., Ogura, H., Shimane, M., et al. Expression of

granulocyte and granulocyte-macmophage colony-stimulating factors by

human nonhematopoietic tumor cells. Growth Factors, 3: 325-331,

1990.

25. Takeda, K., Hatakeyama, K., Tsuchiya, Y., Rikiiski, H., and

Kumaga, K. A correlation between GM-CSF gene expression and me-tastases in mumine tumors. Int. J. Cancer, 144: 131 1-1316, 1991.

26. Sotomayom, E. M., Fu, Y.-X., Lopez-Cepemo, M., Herbert, L., Jime-

nez, J. J., Albarmacin, C., and Lopez, D. M. Role of tumor-derivedcytokines on the immune system of mice bearing a mammary adeno-carcinoma. II. Down-regulation of macrophage-mediated cytotoxicityby tumor-derived gmanulocyte-macrophage colony-stimulating factor. J.

Immunol., 147: 2816-2823, 1991.

27. Sawyers, C. L., Golde, D. W., Quan, S., and Nimem, S. D. Pmoduc-tion of granulocyte-macmophage colony-stimulating factor in two pa-

tients with lung cancer, leukocytosis, and eosinophilia. Cancer, 69:

1342-1346, 1992.

28. Silvestmi, F., Banavali, S., Yin, M., Gopal, V., Baceamani, M., and

Pmeislem, H. D. Comparison of two methods for concentrating CD34�cells from patients with acute non-lymphocytic leukemia. Leuk. Lym-phoma, 8: 389-396, 1992.

29. Schoof, D. D., Selleck, C. M., Massamo, A. F., Jung, S-E., and

Ebemlein, T. J. Activation of human tumor-infiltrating lymphocytes by

monoclonal antibodies directed to the CD3 complex. Cancer Res., 50:

1138-1143, 1990.

30. Anthony, P. P., Ramani, P. Endothelial markers in malignant vas-culam tumoums of the liven: superiority of QBEND/10 over von Will-ebmand factor and Ulex eumopaeus agglutinin 1. 1. Clin. Pathol., 44:

29-32, 1991.

31. Nakamuma, K., Takahashi, T., Tsuyuoka, R., Ueda, Y., Suzuki, A.,

Okuno, Y., Ihama, Y., Seko, S., Okada, T., and Kumagai, N. Identifica-tion of colony-stimulating factor activity in patients with malignant

tumor associated with excessive leukocytosis. Jpn. I. Clin. Oncol., 21:

395-399, 1991.

32. Dmanoff, G., Jaffee, E., Lazenby, A., Golumbek, P., Levitsky, H.,

Bmose, K., Jackson, V., Hamada, H., Pardoll, D., and Mulligan, R. C.Vaccination with irradiated tumor cells engineered to secrete muminegranulocyte-macrophage colony-stimulating factor stimulates potent,

specific, and long-lasting anti-tumor immunity. Proc. Natl. Acad. Sci.

USA, 90: 3539-3543, 1993.

33. Golumbek, P. T., Azhami, R., Jaffee, E. M., Levitsky, H. I., Lazenby,

A., Leong, K., and Pardoll, D. M. Controlled release, biodegradablecytokine depots: a new approach in cancer vaccine design. Cancer Res.,53:5841-5844, 1993.

34. Schmidt-Wolf, I. G., Dejbakhsh-Jones, S., Ginzton, N., Greenberg,

P., and Strobem, S. T-cell subsets and suppressor cells in human bonemarrow. Blood, 80: 3242-3250, 1992.

35. Bmoxmeyem, H. E. Suppressor cytokines and regulation of myelo-

poiesis. Biology and possible clinical uses. Am. J. Ped. Hematol. Oncol.,

14: 22-30, 1992.

36. Dunn, P. L., and North, R. J. Selective radiation resistance ofimmunologically induced T cells as the basis for irradiation-inducedT-cell-mediated regression of immunogenic tumor. J. Leuko. Biol., 49:

388-396, 1991.

37. Gelin, J., Andemsson, C., and Lundholm, K. Effects of indometha-

cm, cytokines, and cyclospomin A on tumor growth and the subsequent

development of cancer eachexia. Cancer Res., 51: 880-885, 1991.



1995;1:95-103. Clin Cancer Res A S Pak, M A Wright, J P Matthews, et al. colony-stimulating factor.functions within cancers that secrete granulocyte-macrophage

immuneneck cancer: presence of CD34(+) cells which suppress Mechanisms of immune suppression in patients with head and

Updated version

http://clincancerres.aacrjournals.org/content/1/1/95

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/1/1/95To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Mechanisms of Immune Suppression in Patients with Head and … · 96 CD34 Cells in Head and Neck Cancers ies with the munine LLC model have shown that several immune suppressive populations