Immunobiology (1999/2000) 201, 631-644© 2000 Urban & Fischer Verlaghttp://www.urbanfischer.de/journals/immunobiol

Department of Dermatology, Medical Faculty (Charite), Humboldt University, Berlin, Germany

Clonal Evolution in a Primary Cutaneous Follicle Center BCell Lymphoma Revealed by Single Cell Analysis inSequential Biopsies

SVEN GOLEMBOWSKI, SYLKE GELLRICH, MALGORSZATA VON ZIMMERMANN,SASCHA RUTZ, STEFFEN LIPPERT, HElKE AUDRING, PAMELA LORENZ, WOLFRAMSTERRY and SIGBERT JAHN

Received December 9, 1999 . Accepted january 11,2000

Abstract

B cell neoplasias descending from germinal center cells harbor the hallmark of intraclonal diver­sity resulting from ongoing mutation in the variable parts of their immunoglobulin-encodinggenes. To characterize a primary cutaneous follicle center B cell lymphoma in more detail, weanalyzed the respective VH and VL genes in single cells mobilized from four sequential biopsies,three taken from the skin and one obtained after internal dissemination from a retrobulbar infil­trate. The lymphoma cells were found to contain V5-51/D6-12/jH5b (heavy chain) andA27/jkappa2 (light chain) gene rearrangements detected on both the genomic and the transcrip­tional level. To provide an accurate mutation analysis, the specific VH gene counterpart(VS-S1 UK) was cloned from the patient's germline. Analyzing 226 single cells, we found: (i)complete nucleotide identity when VH and VL genes of lymphoma cells from one particularbiopsy were compared among each other; (ii) intraclonal diversity due to ongoing mutationcomparing the sequences obtained from sequential biopsies; (iii) both VH and VL genes to behighly mutated. Deducing from the sequence data, we propose a scenario of the clonal evolutionof the B cell tumor in this patient. From the molecular-biological point of view, this primarycutaneous follicle center B cell lymphoma shows the features of a germinal center cell lym­phoma. To draw this conclusion from single cell PCR data, however, a sample of sequentialbiopsies had to be analyzed.

Introduction

The investigation of the unique immunoglobulin genes of B cells has been used asa diagnostic tool to prove monoclonality in B cell lymphomas. The availability ofthe peR and sequencing techniques, allowing the analysis of mutation rate andpattern of the variable regions in Ig genes, resulted in a molecular-biologicallybased classification of the B cell malignancies: Unmutated pre-germinal center Bcells, found in chronic lymphocytic B cell leukemias (1) and mantle cell lym­phomas (2) are clearly distinguishable from mutated post-germinal center B cells

0171-2985/99-00/201105-631 $ 12.0010

632 . S. GOLEMBOWSKI et al.

found in multiple myelomas (3) and monocytoid B cell lymphomas (4). Germinalcenter B cells show the hallmark of an active germinal center reaction, namelyintraclonal diversity due to ongoing mutation, which has been demonstrated totake place in nodal follicle center cell lymphomas (5, 6). Not finding such muta­tion events does not automatically imply that the analyzed lymphoma cells are ofpost-germinal center B cell origin since intraclonal diversity can easily be missedby studying a limited part of one particular biopsy (shown in this paper below),and sequential biopsies are not always available. Furthermore, for a post-germinalcenter B cell it is possible to re-enter the germinal center reaction (7). Thus, theterm «germinal center B cells or their descendants» has been introduced for B cellswith mutated immunoglobulin genes (8, 9). On the other hand, it is not possibleto reliably interpret intraclonal diversity by comparing different clones aftercloning of only one PCR product because of the likelihood of PCR errors, i.e.,Taq errors and the formation of PCR hybrid artifacts (10). Therefore, we appliedthe micromanipulation technique followed by single cell PCR amplification of theIg-genes for the analysis of primary cutaneous lymphomas (11, 12).

The EORTC cutaneous lymphoma study group has proposed a distinct classi­fication for these tumors since cutaneous lymphomas differ from their systemiccounterparts by their clinical course and therapeutic approaches (13). Newmolecular-biological data will provide further input into this classification. Here,we present the clonal evolution of a primary cutaneous follicle center B celllym­phoma from which we were able to analyze four different biopsies.

Material and Methods

Patient

A 54-year-old male presented in 1992 with a skin-associated tumor on his head which was par­tially exulcerated. After the excision of the tumor, a local relapse developed two years later fromwhich we got the first biopsy «A» 11/94. Despite a repeated excision accompanied by systemicIFN-a-2b and local radiotherapy, a local tumor relapse occurred in 1995 (biopsies «B» 12/95and «C» 07/96). Then, a unilateral exophtalmus occurred as a sign of internal involvement fromwhich we analyzed a biopsy taken from the retrobulbar space (<<D» 03/97). At this time, no evi­dence was found for tumor dissemination into visceral organs, blood or bone marrow.

Histology and immunohistochemistry

Histology showed a diffuse infiltrate of the dermis and subcutis mainly consisting of cells withpale nuclear chromatin and a paracentrally situated large nucleus which were immunohisto­chemically characterized as CD20+, CD5-, CDI0-, bcl-2- and CD35-. About 800/0 of the patho­logic cell population was in proliferation, as indicated by positive staining for the proliferationmarker MIBl (Ki-67). Along the vessels there was clustering of small lymphocytes withoutnuclear polymorphism which were CD3 positive and hence most likely represented reactive Tcells. Follicular structures and mantle cells were not seen.

Micromanipulation

Frozen skin sections (10 pm thick) were stained with the anti-CD20 antibody (Dako Diagnos­tika, Hamburg, Germany) using biotinylated Fab anti mouse and streptavidine-alkaline phos-

Single cell analysis in a primary cutaneous follicle center B cell lymphoma · 633

phatase complexes (Dako) as developing reagents. Control sections were stained with anti-CD3antibodies. After washing, bound alkaline phosphatase was visualized by staining with NewFuchsin. The slides were counterstained with haematoxylin.

Single cells were mobilized under the microscope (Nikon, Dusseldorf, Germany) with thehelp of a hydraulic micromanipulator (Narishige, Tokyo, Japan) using 600x magnification.Then, single cells were aspirated into a micropipette (Narishige), put into 20 pI PCR buffersupplemented by 1 ng/pl rRNA (both from Boehringer, Mannheim, Germany) and stored at-20°C. Photographs were taken before and after the micromanipulation of each cell.

Amplification and sequencing of the rearranged immunoglobulin heavy chain (lgH) andimmunoglobulin light chain (lgL) genes of single cells

A set of oligonucleotide primers according to KUPPERS et al. was used for PCR amplification ofrearranged VHgenes. Six VHgene family specific primers (14) were used together with JH specificoligonucleotides (15). A semi-nested PCR approach was chosen. In the first round of amplifica­tion, 6 VH gene primers together with the outer (3') JH primer mix were used simultaneously inone tube. For the second round of amplification, aliquots of the first round were reamplifiedusing the same VH primers but nested JH primer mixes in separate reactions for each VH genefamily. Before PCR, the single cells were incubated with 0.25 mg/ml proteinase K (BoehringerMannheim, Mannheim, Germany) for 55 min at 50°C and additionally at 95°C for 10 min(enzyme inactivation). The reaction mix (50 pI) for the first round of amplification contained:PCR buffer, 2.5 pM MgClb 200 pM of each nucleotide (dATP, dCTG, dGTP, dTTP; AGS),7 nM of each primer and 3.5 U Expand™ High Fidelity Taq-polymerase (all chemicals fromBoehringer Mannheim). PCR amplification was carried out on a Personal Cycler (Biometra,Gottingen, Germany): one cycle at 95°C for 2 min, 65°C for 1 min (when Taq-polymerase wasadded), 72°C for 1 min, followed by 35 cycles at 95 °C for 30 sec, 59°C for 30 sec, 72°C for1 min, and final extension at 72 °C for 5 min. The second round of the PCR was carried out on aTRIO-Thermoblock (Biometra) in separate reactions for each of the six VH family-specificprimers using 1 pI of the amplification product from the first round. Reaction mixture: PCRbuffer (Roche Molecular Systems, Branchburg, USA for Perkin Elmer), 1.5 pM MgCl2 (PerkinElmer), 200 pM of each nucleotide, 7 nM of each primer and 5U Taq-polymerase (PerkinElmer). The cycle program consisted of one cycle at 95°C for 2 min, 68°C for 5 min (additionof Taq-polymerase), 72°C for 1 min, 45 cycles at 95°C for 1 min, 61°C (VH1, VH2, VH5, VH6specific primers) or 65°C (VH3, V H4) for 30 sec, and a final extension at 72°C for 5 min.

For the simultaneous amplification of the rearranged IgL genes (here kappa) an additional setof nine oligonucleotide primers, six VK family specific and 3 JK segment specific primers accord­ing to KUPPERS et al. (16), was used together with the nine heavy chain gene-specific primers inthe first round of amplification, each 7 nM in the final volume of 50 pI. For the second round ofamplification, aliquots of the first round were reamplified using only one of the same VK primersplus four more internally located JK primers in separate reactions for each VK gene family. Thereaction mix (50 pI) for the second round of amplification contained 1 pI of the amplificationproduct from the first round, PCR buffer, 2.5 pM MgCI2, 200 pM of each dNTP, 50 nM of eachprimer and 1.25 U Taq-polymerase (all Perkin Elmer). The second round cycle program con­sisted of one cycle at 95°C for 2 min, 68°C for 5 min (addition of Taq-polymerase), 72°C for 1min, 45 cycles at 95°C for 1 min, 61°C for 30 sec, 72°C for 1 min and a final extension at 72 °Cfor 5 min. A 5 pI aliquot of the reaction mixture was analyzed on a 2% agarose gel.

Strong attention was paid to avoid any contamination with desoxyribonucleic acid (DNA).The micromanipulation, as well as the first and second round of PCR were carried out sepa­rately in different rooms. As negative controls, T cells were picked from CD3-stained adjacentsections during the same experiment.

The PCR products were purified using the QIAquick PCR Purification Kit (Qiagen, Hilden,Germany), followed by direct sequencing with the Dye Terminator Cycle Sequencing ReadyReaction Kit (Perkin Elmer) on a Sequencing System 373 (Applied Biosystems, Weiterstadt,Germany) using between 0.5 and 10 pI of purified PCR product and 2.5 pM of the respective

634 · S. GOLEMBOWSKI et aI.

primers as used in the second round PCR. The sequences were online analyzed using the Soft­ware «DNAPLOT» (created by H.-H. ALTHAUS, University of Cologne, Germany) and com­pared with the gene bank «V BASE Sequence Directory» (TOMLINSON et aI., MRC Center forProtein Engineering, Cambridge, UK) for determining the most similar germline VH/Vl( genes.

Amplification of the germline VH gene rearranged in the lymphoma Bcells

In order to exclude an allelic polymorphism for the mutation analysis, the patient's specificgermline VH gene rearranged in the lymphoma B cells had to be sequenced. Therefore genomicDNA was needed. We used paraffin embedded skin tissue which was first deparaffinized andthen proteinazed. Primers were designed to amplify the proposed germline gene (VS-Sl [17]),one annealing to the leader sequence of the VH gene (S'-CCCCTGAAATTCAAATTTTGT­GTTTCTCC-3'), and one annealing to the recombination signal sequence (S'-CTCGGGGCTGGTTTCTCTCACTGTG-3'). The PCR product was purified with the QIAquick GelExtraction Kit (Qiagen), then cloned using the TA-Cloning Kit (Invitrogen, Leek, Netherlands),and finally sequenced.

Amplification of the IgG gene of the tumor cells in the retrobulbar infiltrate

Since we did not find single tumor cells in the retrobulbar infiltrate, we applied a semi-nestedPCR approach on crude DNA using a complementarity determining region 3 (CDR3)-specificprimer (S'-CCCTGGCCCMAAGGACTRTGA-3') deduced from the known tumor sequencefor the second round PCR.

mRNA isolation and cDNA synthesis

We also checked whether the tumor B cells transcribe their Ig genes. mRNA was isolated withthe Oligotex Direct mRNA-Kit (Qiagen), and eDNA was prepared using the SuperscriptPreamplification System (Gibco BRL, Eggenstein, Germany). For the PCR, the family specific5'-primer was paired with a primer specific for the constant pl-region (S'-CCAAGCTTAGAC­GAGGGGGAAAAGGGTT-3') in the first round. The second semi-nested PCR round wascarried out with the above mentioned CDR3-specific primer.

Results

Altogether, we analyzed 226 single cells isolated from four different biopsies.Table 1 gives a summary of the experiments. From the first biopsy taken inNovember 1994 (<<A») we mobilized 66 single CD20-positive B cells. In 19 cellswe found identical VHDJH rearrangements (AHSC)' in four of them we couldamplify the l( light chain simultaneously (AKSC). In another two cells, we foundthis light chain sequence only, but no heavy chain Ig gene rearrangements couldbe amplified. From the second biopsy «B» taken in December 1995 we only hadparaffin sections in hand. Here, we were unable to carry out single cell analysis.From our own experience, a more or less efficient single cell analysis requirescryomaterial (own unpublished observations). Therefore, using biopsy material«B» (12/95) we had to analyze crude DNA. To validate this sequence data, twoindependent PCRs were performed and their products sequenced. The heavyand the light chain sequences (BHCR' BKCR) were clonally related to the respective«A» sequences since they contained the same CDR3 stretches. Biopsy «C» could

Single cell analysis in a primary cutaneous follicle center B cell lymphoma . 635

Table 1. Summary of the experiments performed

biopsya) analyzed findings sequence name

A 11/94 66 single cells ~ 19 identical VHDJHb) AHsc~ 6 identical VJKb) BHCR and BKCR

B 12/95 crude DNA ~ same sequence found in BHSCseparate PCRs

C 07/96 63 single cells ~ 13 identical VHDJHb) C HSC~ 5 identical VJKb) C HCR and C KCR

D 03/97 97 single cells ~ no clonally related sequences foundcrude DNA ~ same sequence found in D HcRcDNA separate PCRsc) DHcDNA

a) Biopsies A-C taken from lesions of the head, biopsy D from a retrobulbar infiltrate, date ofbiopsies givenb) In some single cells both the rearranged heavy and the light chain were amplified simultane­ously in one PCRc) A CDR3 tumor-specific primer was used

be used for single cell analysis. Out of 53 investigated B cells, 13 gave identicalVHDJH (CHSC)' and five harbored identical VJ K(CKSC) rearrangements, clonallyrelated to the sequences found in «A» and «B». Applying our single cell PCRapproach, we did not detect the tumor-specific sequence found in «A», «B» or«C» when analyzing 97 single cells in biopsy «D» from the retrobulbar infiltrate.However, using crude DNA prepared from this biopsy material for a seminestedCDR3 specific PCR approach, we detected sequence D HCR which is clonallyrelated to the above described tumor-specific Ig gene rearrangements. We didnot find, however, a light chain rearrangement related to the tumor-specifickappa chain sequence. Taking RNA from this biopsy, we could demonstrate thatthe VH-D-JH genes of the tumor cells are transcribed (DHcDNA).

Whenever we compared the respective Ig gene rearrangements of single tumorB cells obtained from one particular biopsy (biopsy «A» or «C») we observedcomplete nucleotide identity. However, a sequence comparison between Ig genesfrom different biopsies revealed some distinctions at certain positions due tonucleotide exchanges. Figure 1 and Figure 2 show the sequences of the rear­ranged immunoglobulin heavy and light chain genes which we suggest to be spe­cific for the tumor B cells.

To avoid the misinterpretation of an allelic polymorphism as a somatic muta­tion, we sequenced the patient's specific germline VH gene. Interestingly, wefound an allelic polymorphism at position -1 when the patient's specificgermline gene V5-51 UK was compared to the published corresponding germ­line gene V5-51 (17). Therefore, it was possible to prove that the tumor cellsindeed had rearranged this germline gene: they all harbor the guanine as the firstnucleotide at position -1. In contrast to the heavy chain, the patient's specificgermline gene of the kappa chain had not to be cloned since only a few allelic

30

31

40

50

52

aG

GA

TA

CA

GC

TT

TA

CC

AG

CT

AC

TGG

ATC

GG

CTG

GG

TGC

GC

CA

GA

TGC

CC

GG

GA

AA

GG

CC

TGG

AG

TGG

ATG

GG

GA

TC

ATC

TA

TC

CT

GG

TG

AC

TC

T

Intr

on

><

Ex

on

-1+

11

02

0C

CC

AC

AG

GA

GT

CT

GT

GC

CG

AG

GTG

CA

GC

TGG

TGC

AG

TC

TG

GA

GC

AG

AG

GTG

AA

AA

AG

CC

CG

GG

GA

GT

CT

CTG

AA

GA

TCT

CC

TG

TA

AG

GG

TT

CT

T--

--G

---

---

-A-

-T-

---

---

---

---

---

---

---

---

---

---

G--

G--

--G

---

--a

-A-

-T-

---

---

---

---

-AC

---

---

G--

--G

---

--a

-A-

-T-

---

---

---

---

---

---

---

---

---

---

-AC

---

---

G--

--G

---

---

-A-

-T-

---

---

---

---

---

---

---

---

---

---

---

---

---

---

G--

G--

--G

---

---

-A-

-T-

---

---

---

---

---

---

---

---

---

---

---

G--

G--

Hi

CD

R1

_

0'

(,,;.

)

0'

~ CJ o t"'-4

t"11~ tiJ o ~ (J

') g:; (D M ~

--a

--a

_H

2_

__

__

__

__

CD

R2

_

--g

--g

C-­

C--

--a

--a

-G-

-G-

V5-

51U

KV

S-S:

LU

sc(1

1/9

4)

Bll

ea

(12

/95

)C

Hsc

(07

/96

)D

lle

a(0

3/9

7)

Dll

eD

_

V5-

51U

KV

S-S:

LU

sc(1

1/9

4)

Bll

ea

(12

/95

)C

Hsc

(07

/96

)D

lle

a(0

3/9

7)

Dlle

DR

A

70

80

82

ab

cC

CG

TC

CT

TC

CA

AG

GC

CAG

GTC

AC

CA

TCT

CA

GC

CG

AC

AA

GT

CC

ATC

AG

CA

CC

GC

CTA

CC

TG

CA

GTG

GA

GC

AG

CC

TGA

AG

CD

R2

60V

5-51

UK

GA

TA

CC

AG

AT

AC

AG

CV

S-S:

LU

sc(1

1/9

4)

AC

G--

a-A

---t

Bll

ea

(12

/95

)-C

G---

---

--t

--t

CH

sc(0

7/9

6)

-CG

---

--t

--t

Dll

ea

(03

/97

)A

TG--

a-A

---t

Dll

eD

_A

TG--a

-A-

--t

-T-

--t

-T-

--t

-T-

--t

-T-

--t

-T-

--t

GA

A--t

-CA

--t

-CA

--t

GA

A--t

GA

A--t

-T-

-T-

-T-

-A-

-A-

-A-

-A-

-A-

_H

3_

__

__

__

__

_C

DR

3_

V5-

51U

KV

S-S:

LU

sc(1

1/9

4)

Bll

ea

(12

/95

)C

Hsc

(07

/96

)D

lle

a(0

3/9

7)

Dll

eD

_

90

94G

Ce

rCG

GA

CA

CC

GC

CA

TGT

AT

TAC

TG

TG

CG

AG

AC

A

-Tt

---

-T-

-Tt

---

-T-

-Tt

---

-T-

--g

--

-Tt

---

-T-

-Tt

---

-T-

D6-

:L3

10

0T

ATA

GC

AG

CA

GC

TG

GT

-G

---T

t-T

t-T

t-

G--

-Tt

-G

---T

t

aN

CA

TA

GT

CA

CA

GT

CA

CA

GT

103

J.S

bC

CC

TGG

GG

CC

AG

GG

A

--t

--t

--t

Fig

ure

1.N

ucle

otid

ese

quen

ces

of

the

VH

DJH

rear

rang

emen

tfr

omtu

mo

rce

llso

fa

prim

ary

cuta

neou

sfo

llic

lece

nter

Bce

llly

mph

oma

inch

rono

logi

cal

orde

r.N

ucle

otid

edi

ffer

ence

sin

com

pari

son

toth

epa

tien

t'ssp

ecif

icge

rmli

nege

ne(V

5-51

UK

)ar

ein

dica

ted

by

capi

tal

lett

ers

(rep

lace

men

t)o

rlo

wer

case

lett

ers

(sil

ent

mut

atio

ns).

V5-

51is

the

corr

espo

ndin

gpu

blis

hed

germ

line

gene

(17)

tode

mon

stra

teal

lelic

poly

­m

orph

ism

(not

epo

siti

on-1

).M

onth

sw

hen

biop

sies

(A-D

)w

ere

take

nar

egi

ven

inpa

rent

hesi

s.F

ootn

otes

indi

cate

type

of

anal

yzed

mat

eria

l:S

C-

sing

lece

ll;C

R-

crud

eD

NA

;H

-he

avy

chai

n.D

iffe

rent

leng

ths

for

the

sequ

ence

sar

edu

eto

diff

eren

tpri

mer

sus

ed(s

eeM

&M

).E

MB

L-a

ssoc

iati

onnu

mbe

rs:

V5

-51

UK

-Y

1281

7;A

H-

Yl1

556

(<<D

KV

H5a

»);

BH

-Y

l155

8(<

<UK

VH

5c»)

;C

H-

Yl1

557

(<<D

KV

H5b

»);

DH

­A

J223

759

(<<D

KV

H5d

»).

Fig

ure

2.N

ucle

otid

ese

quen

ces

of

the

VJK

rear

rang

emen

tfr

omtu

mo

rce

llso

fa

pri

mar

ycu

tane

ous

foll

icle

cent

erB

cell

lym

phom

a.N

ucle

otid

edi

ffer

ence

sin

com

pari

son

toth

eco

rres

pond

ing

publ

ishe

dge

rmli

nege

neA

27(3

7)ar

ein

dica

ted

by

capi

tal

lett

ers

(rep

lace

men

t)o

rlo

wer

case

lett

ers

(sil

ent

mut

atio

ns).

Foo

tnot

esgi

veth

ety

peo

fan

alyz

edm

ater

ial:

SC

-si

ngle

cell;

CR

-cr

ude

DN

A;

K-

kapp

ach

ain.

EM

BL

-ass

ocia

tion

num

bers

:A

K-

AJ2

2376

0(<

<DK

VK

3a»)

;B

K-

AJ2

2376

2(<

<DK

VK

3c»)

;C

K-

AJ2

2376

1(<

<DK

VK

3b»)

.

1U

7

AK

sc(1

1/9

4)

BK

c.(1

2/9

5)

CK

sc(0

7/9

6)

1U7

AK

sc(1

1/9

4)

BK

c.(1

2/9

5)

CK

sc(0

7/9

6)

1U

7

AK

sc(1

1/9

4)

BK

c.(1

2/9

5)

CK

sc(0

7/9

6)

__

__

__

__

__

CD

R1

_

10

20

30

31

31

a3

2A

CC

CT

GT

CT

TT

GT

CT

CC

AG

GG

GA

AA

GA

GCC

AC

CC

TC

TC

CTG

CA

GG

GC

CA

GT

CA

GA

GT

GT

TA

GC

AG

CA

GC

TAC

TT

AG

CC

TGG

TA

CC

AG

CA

GA

AA

--c

A--

---

---

---

--t

---

---

---

---

---

--t

--a

-G-

---

A--

---

---

---

---

---

---

---

---

--t

--a

---

-G-

A--

---

---

---

---

---

---

---

---

---

---

A--

---

---

---

---

---

---

---

--t

--a

---

-G-

__

__

__

CD

R2

_

40

50

60

70

CC

TG

GC

CA

GG

CT

CC

CA

GG

CT

CC

TC

ATC

TAT

GG

TG

CA

TC

CA

GC

AG

GG

CC

AC

TG

GC

ATC

CC

AG

AC

AG

GT

TC

AG

TG

GC

AG

TG

GG

TC

TG

GG

AC

AG

AC

---

A--

---

---

-AA

--­

A--

---

---

-AA

---

---

A--

---

-AA

---

__

__

__

__

CD

R3

_

809

0N

JIa

TT

CA

CT

CT

CA

CC

ATC

AG

CA

GA

CTG

GA

GC

CT

GA

AG

AT

TT

TG

CA

GTG

TA

TTA

CT

GT

CA

GC

AG

TA

TG

GT

AG

CT

CA

CC

TC

CTG

TAC

AC

TT

TT

G--t

---

---

---

---

---

---

---

--c

---

--t

---

---

---

-A-

-A-

GA

--

-G-

--t

---

---

-A-

CA

----

A--

GA

--

-G-

--t

---

---

-A-

CA

----

GA

--

-G-

V'J S· cr

q (£"""

n 2= ~ ~ ~ ~ ~.

en S· ~ ~""1 s· ~ ""1 "< n ~ M ~ ~ (D o ~ en ~ (S. (£"

""n (D ~ M (D ""1 to n 2= ~ s ~ p- o S ~ 0

'V

J'-

J

638 · S. GOLEMBOWSKI et al.

variants are described for the kappa genes and the whole gene locus has alreadybeen sequenced (18).

Sequence information for BHCR and CHCR starts within the leader region sincewe got PCR products with the leader primer designed initially for the germlinegene analysis. Starting to count at position 20, twenty-one (sequence BH) ortwenty-two mutations (sequences AH, CH, and DH) are scattered over the VHgene with a predominance in the CDR2 and FR3. For the light chains somaticmutations occurred less frequently (12 and 14).

It became obvious that the VHD]H and VJK sequences detected either by thesingle cell technique or the PCR analysis of crude DNA/RNA were clonallyrelated, having rearranged the V5-51/D6-13/]H5b germline genes for the heavychain and the A27/]kappa2 germline genes for the light chain. They contain aunique CDR3 region and share some common mutations. However, the chrono­logical order of the biopsies does not correspond to the chronological order ofthe tumor progression: sequences BHCR/BKCR (biopsy 12/95) are not, as onewould expect, further mutated variants of the AHsc/AKsc sequences (biopsy11/94). Although some new mutations were introduced during the 13 monthsperiod (see positions 30 and 47 in the respective VH genes), there are 5 nucleo­tides still in germline configuration in sequence BHCR at positions were the AHSCsequence had already been mutated (for example see positions 51-53). Similarobservations are true for the light chain sequences (positions 74 and 85). There­fore, we postulate a common precursor cell sequence X (sequence not shown)harboring all those mutations which are observed in both its descendants. Forthis precursor cell sequence we performed an R/S-ratio analysis which is shownin Table 2. The R/S ratio of the observed mutations within the CDR2 is slightlyhigher, those of the mutations within the FRs somewhat lower than the intrinsicR/S ratio (for evaluation see discussion).

After we had shown initially that the clonally expanded B cells contain aV5-51/D6-13/]H5b (heavy chain) and a A27/]kappa2 (light chain) gene rear­rangement, we focused on the respective V gene families during the subsequentcourse of the experiments. Sequencing was performed only for PCR amplificatesobtained using our VH5 or VK3-specific primers. Therefore, we can not draw anyconclusions concerning the nature of other infiltrating B cells within the lym­phoma lesions, their Ig gene usage or rate of somatic mutation.

Table 2. R/S analysis for the putative precursor cell«X»

Region Relative size Observed mutations Intrinsic

R S R/S R/Sa)

CDR 66/234 = 9,282 5 1 5 3.6FR 168/234 = 0.718 6 3 2 3.4

ph)

0.13

a) Following the calculation by CHANG & CASALI (35)h) Probability for the observed R-mutations in the CDRs occurring by chance. Binomial proba­bility model by SHLOMCHIK et al. (36)

Single cell analysis in a primary cutaneous follicle center B cell lymphoma . 639

In the T cells used as negative controls throughout the experiments, no PCRproduct has ever been obtained.

The number of light chain gene rearrangements amplified was generally foundto be somewhat lower than the number of heavy chain peR products. Thisobservation argues for a less efficient PCR amplification in case of the lightchain.

Discussion

Primary cutaneous B cell lymphomas are defined as non-Hodgkin lymphomaspresenting in the skin with no evidence of extracutaneous disease at the time ofdiagnosis and within the first six months after diagnosis (13). Molecularbiologi­cal analysis may provide further details to understand the pathogenetic mecha­nisms of non-Hodgkin lymphomas (NHL) occurring primarily in the skin envi­ronment in comparison to their nodal counterparts.

Sequence analyses on the single cell level give reliable information (11).Sequence errors can almost fully be excluded when the danger of contaminationis properly addressed (16). Exact sequence information, if possible from both theheavy and the light chain, is crucial to perform mutation and genealogical analy­ses. Sequences obtained from multiple colonies after cloning of only one PCRproduct are not reliable because of Taq erorrs, cloning errors and the possibilityof hybrid formation. The disadvantage of the single cell approach is the fact, thatonly a minute part of a biopsy can be analyzed. Even if numerous cells of differ­ent sections are micromanipulated, the data obtained may not necessarily be rep­resentative of the ~hole tumor. In our case, when we were micromanipulatingthe «A» biopsy, we missed the cells harboring sequence B which had to be pre­sent at that time (see below). Only the circumstance that we had the biopsies«B» and «C» in hand, taken several months later, revealed ongoing mutation or,more exactly, intraclonal diversity.

The criticism that the so-called primary cutaneous B cell lymphomas do notreally exist and that they are disseminated undetected nodal/systemic malignan­cies cannot formally be refuted yet. But at least for our case, we can be sure thatthe retrobulbar infiltrate is composed of tumor cells descended from the lym­phoma cells of the capillitium and not vice versa. A scheme for the tumor pro­gression is given in Figure 3. In order for the retrobulbar space to be the startingpoint for tumorigenesis, it must have produced the putative precursor cells, aswell, for explaining the Band C sequences. This precursor cell must havemigrated into the capillitium, and must there have given rise to the cells with theA, Band C sequences. Cells with the sequence A had then to migrate back intothe eye. Adding the fact that we did not find any other tumor related sequencesin the retrobulbar infiltrate, this scenario is very unlikely. A possible accompa­nying nodal/systemic process can still not be ruled out.

The predominance of a most likely polyclonal infiltrate, which made it diffi­cult to find single tumor cells in the retrobulbar space, is in contrast to the skininfiltrate where we found tumor cells almost exclusively. This points strongly

640 · S. GOLEMBOWSKI et al.

capillitium retrobulbar infiltrate

Figure 3. Possible scenario for the tumor progression. Tumor cells A-D correspond to the fourbiopsies (see Table 1 for biopsy date), whereas X and Xl represent the putative precursor cells.For sequence X the number of mutations observed in both its descendants is given. Only thevariable gene sequences (VHand VxJ starting from position 20 or 11, respectively are considered.Mutations including their positions are indicated.

towards a different microenvironment, especially a different cytokine profileresponsible for the accumulation of non-tumor cells in the retrobulbar space.Whether the VH mutation at position 56 in sequence D took place in the skin orin the eye can not be distinguished.

Although it is unknown where and when the transforming event took place, itis a fact that the hypermutation process was still switched on after the B cellbecame malignant. The presence of differently mutated lymphoma subclonescould not be explained otherwise. A scenario, in which different physiologicalsubclones transform independently is very unlikely. Cutaneous follicle centercell lymphoma B cells are thus still able to mutate their immunoglobulin genes.This has also been described for nodal follicular center cell lymphomas (5, 6) andmaltomas (19). It raises the question for an antigen responsible for the processcomparable to the low-grade maltomas (20). The mutation frequency is with 9%(22 mutations within the VH gene) significantly higher than the 4% given for thephysiological class-switched memory B cells (21). Since no follicle center-inde-

Single cell analysis in a primary cutaneous follicle center B cell lymphoma . 641

pendent hypermutation has been described in humans, as it is the case in sheep(22), the tumor cells must have taken part in the germinal center reaction longerthan usual. This has also been shown for the diffuse larger B cell lymphomas(23, 24).

It is common practice to assess antigenic selection for antibody expression byanalyzing the mutational pattern. The observed ratio of the replacement muta­tions to the silent mutations (R/S ratio) is compared to the intrinsic R/S ratioregarding the complementarity determining regions (CDR) and frameworkregions (FR). This has been done for the putative precursor cell in Table 2.Although the observed R/S ratio for the CDRs of 5 is higher than the intrinsicratio of 3.6 and the observed R/S ratio for the FRs of 2 is lower than the intrinsicratio of 3.4, one can not necessarily conclude an antigen-dependent selection asthis kind of analysis should be applied with caution. Firstly, it is unfavorable touse the method on single sequences from a statistical point of view. Only a singlesilent mutation (in the denominator) alters the ratio massively. Secondly, not allresidues of the CDRs defined by KABAT et al. (25) take part in the antigen bind­ing. Recently, MCCALLUM et al. (26) presented their contact definition, supple­menting the data by CHOTIA et al. (27). Thirdly, only a few or even only one ofthe numerous observed mutations could playa role for the affinity to the anti­gen (28, 29). All the others would be bystanders influencing only the ratio.Fourthly, there is some evidence, that residues of the FRs do influence the affin­ity, too (30, 31). Therefore, we plan to investigate the specificity of the antibodyproduced by the tumor B cells by binding studies on the protein level. On theDNA level, the sequences appear to be functional (i.e. no stop codon) and tran­scripts on the mRNA level could be detected.

Another interesting observation is worth mentioning. The routine molecular­biological investigation of the crude DNA did not detect a clonal PCR productfor the heavy chain of the immunoglobulin using a widely applied FR3 consen­sus primer (5'-ACACGGCYSTGTATTACTGT-3') with a JH primer (5'­GTGACCAGGGTNCCTTGGCCCCAG-3'). After the successful amplifica­tion of the tumor cell DNA with the oligonucleotides used for the single cellPCR, the reason became obvious: the residues of the immunoglobulin genes cor­responding to the third and fourth least position of the FR3 primer (amino acidposition 91) and the residue corresponding to the last 3' position of the JHprimer (position 101) were mutated. And just the hybridization of the 3' part ofthe primer at its matrix is crucial for a successful PCR (32). It has been knownfor a while that the framework region 3 is subject to hypermutation, too (33).The use of FR3 consensus primers for the routine diagnosis of B cell malignan­cies with a high load of somatic mutations must be questioned.

In conclusion, the analyzed primary cutaneous follicle center B cell lym­phoma is a germinal center cell lymphoma with the hallmark of intraclonaldiversity and ongoing mutation. We demonstrate the progression of the cuta­neous infiltrate and the dissemination into the retrobulbar space summarized inthe genealogical tree in Figure 3. From a molecular-biological point of view,there is no striking difference between the nodal and the primary cutaneousentity of follicle center cell lymphomas.

642 . S. GOLEMBOWSKI et al.

Sequence analysis on a single cell level gives reliable data for mutation studiesand allows to simultaneously obtain the heavy chain and the light chainsequences from single cells for further specificity studies on the protein level. Tothis end, we are planning to construct Fab-expression plasmids for expression inE. coli (34).

Acknowledgement

Authors wish to thank Dr. R. KUPPERS (Institute for Genetics, University Cologne, Germany)for methodical advice. This work was supported by a grant from the Deutsche Forschungsge­meinschaft (Ste 366/7-1, Clinical Research Group «Cutaneous Tumors»).

Referenceds

1. KUPPERS, R., A. GAUSE, and K. RAjEWSKY. 1991. B cells of chronic lymphatic leukemiaexpress V genes in unmutated form. Leuk. Res. 15: 487-496.

2. HUMMEL, M., J. TAMNARU, B. KALVELAGE, and H. STEIN. 1994. Mantle cell (previously cen­trocytic) lymphomas express VH genes with no or very little somatic mutations like thephysiologic cells of the follicle mantle. Blood 84: 403-407.

3. BAKKUS, M.H., C. HEIRMAN, 1. VAN RIET, B. VAN CAMP, and K. THIELEMANS. 1992. Evi­dence that multiple myeloma Ig heavy chain VDJ genes contain somatic mutations but showno intraclonal variation. Blood 80: 2326-2335.

4. KUPPERS, R., M. HAjADI, L. PLANK, K. RAjEWSKY, and M.L. HANSMANN. 1996. MolecularIg gene analysis reveals that monocytoid B cell lymphoma is a malignancy of mature B cellscarrying somatically mutated V region genes and suggests that rearrangement of tpe kappa­deleting element (resulting in deletion of the Ig kappa enhancers) abolishes somatic hyper­mutation in the human. Eur. J. Immunol. 26: 1794-1800.

5. BAHLER, D.W and R. LEVY. 1992. Clonal evolution of a follicular lymphoma: evidence forantigen selection. Proc. Natl. Acad. Sci. U. S. A. 89: 6770-6774.

6. ZHU, D., R.E. HAWKINS, T.J. HAMBLIN, and F.K. STEVENSON. 1994. Clonal history of ahuman follicular lymphoma as revealed in the immunoglobulin variable region genes. Br. J.Haematol. 86: 505-512.

7. BEREK, C. and C. MILSTEIN. 1988. The dynamic nature of the antibody repertoire.Immunol. Rev. 105: 5-26.

8. KLEIN, U., G. KLEIN, B. EHLIN HENRIKSSON, K. RAjEWSKY, and R. KUPPERS. 1995. Burkit­t's lymphoma is a malignancy of mature B cells expressing somatically mutated V regiongenes. Mol. Med. 1:495-505.

9. KANZLER, H ., R. KUPPERS, M.L. HANSMANN, and K. RAjEWSKY. 1996. Hodgkin and Reed­Sternberg cells in Hodgkin's disease represent the outgrowth of a dominant tumor clonederived from (crippled) germinal center B cells. J. Exp. Med. 184: 1495-1505.

10. FORD, J.E., M.G. McHEYZER WILLIAMS, and M.R. LIEBER. 1994. Chimeric molecules cre­ated by gene amplification interfere with the analysis of somatic hypermutation of murineimmunoglobulin genes. Gene 142: 279-283.

11. GELLRICH, S., S. GOLEMBOWSKI, H. AUDRING, S. JAHN, and W STERRY. 1997. Molecularanalysis of the immunoglobulin VH gene arrangement in a primary cutaneous immunoblas­tic B-cell lymphoma by micromanipulation and single-cell PCR. J. Invest. Dermatol. 109:541-545.

12. FORSTE, N., S. GELLRICH, S. GOLEMBOWSKI, S. RUTZ, H. AUDRING, W. STERRY, and S. JAHN.1997. Analysis of V(H) genes rearranged by individual B cells in dermal infiltrates ofpatients with mycosis fungoides. Clin. Exp. Immunol. 110: 464-471.

Single cell analysis in a primary cutaneous follicle center B cell lymphoma · 643

13. WILLEMZE, R., H. KERL, W. STERRY, E. BERTI, L. CERRONI, S. CHIMENTI, J.L. DIAZ-PEREZ,M.L. GEERTS, M. Goos, R. KNOBLER, E. RALFKIAER, M. SANTUCCI, N. SMITH, J. WECH­SLER, W.A. VAN VLOTEN, and C.J. MEIJER. 1997. EORTC classification for primary cuta­neous lymphomas: a porposal from the Cutaneous Lymphoma Study Group of the Euro­pean Organization for Research and Treatment of Cancer. Blood 90: 354-371.

14. KUPPERS, R., K. RAjEWSKY, M. ZHAO, G. SIMONS, R. LAUMANN, R. FISCHER, and M.L.HANSMANN. 1994. Hodgkin disease: Hodgkin and Reed-Sternberg cells picked from histo­logical sections show clonal immunoglobulin gene rearrangements and appear to be derivedfrom B cells at various stages of development. Proc. Natl. Acad. Sci. U.S.A. 91: 10962-10966.

15. KUPPERS, R., M. ZHAO, M.L. HANSMANN, and K. RAjEWSKY. 1993. Tracing B cell develop­ment in human germinal centres by molecular analysis of single cells picked from histologi­cal sections. EMBO J. 12: 4955-4967.

16. KUPPERS, R., M.L. HANSMANN, and K. RAjEWSKY. 1997. Micromanipulation and PCR anal­ysis of single cells from tissue sections. In: WEIR, D.M., C. BLACKWELL, and L.A. HERZEN­BERG (eds.). Handbook of Experimental Immunology. Blackwell Scientific, Oxford, pp.206.1-206.4.

17. MATSUDA, F., E.K. SHIN, H. NAGAOKA, R. MATSUMURA, M. HAINO, Y. FUKITA, S. TAKAISHI, T. IMAI, J.H. RILEY, R. ANAND, and et al. 1993. Structure and physical map of 64 vari­able segments in the 3'O.8-megabase region of the human immunoglobulin heavy-chainlocus. Nat. Genet. 3: 88-94.

18. SCHABLE, K.F. and H.G. ZACHAU. 1993. The variable genes of the human immunoglobulinkappa locus. BioI. Chern. Hoppe Seyler 374: 1001-1022.

19. Du, M., T.C. DISS, C. Xu, H. PENG, P.G. ISAACSON, and L. PAN. 1996. Ongoing mutation inMALT lymphoma immunoglobulin gene suggests that antigen stimulation plays a role in theclonal expansion. Leukemia 10: 1190-1197.

20. WOTHERSPOON, A.C., C. ORTIZ HIDALGO, M.R. FALZON, and P.G. ISAACSON. 1991. Heli­cobacter pylori-associated gastritis and primary B-cell gastric lymphoma. Lancet 338:1175-1176.

21. KLEIN, U., R. KUPPERS, and K. RAjEWSKY. 1997. Evidence for a large compartment of IgM­expressing memory B cells in human. Blood 89: 1288-1298.

22. REYNAUD, C.A., C. GARCIA, WR. HEIN, and J.C. WEILL. 1995. Hypermutation generatingthe sheep immunoglobulin repertoire is an antigen-independent process. Cell 80: 115-125.

23. Hsu, F.J. and R. LEVY. 1995. Preferential use of the VH4 Ig gene family by diffuse large-celllymphoma. Blood 86: 3072-3082.

24. KUPPERS, R., K. RAJEWSKY, and M.L. HANSMANN. 1997. Diffuse large cell lymphomas arederived from mature B cells carrying V region genes with a high load of somatic mutationand evidence of selection for antibody expression. Eur. J. Immunol. 27: 1398-1405.

25. KABAT, E.A. and T.T. Wu. 1971. Attempts to locate complementarity-determining residuesin the variable positions of light and heavy chains. Ann. N. Y. Acad. Sci. 190: 382-393.

26. MACCALLUM, R.M., A.C.R: MARTIN, and J.M. THORNTON. 1996. Antibody-antigen inter­actions: contact analysis and binding site topography. J. Mol. BioI. 262: 732-745.

27. CHOTHIA, C., A.M. LESK, E. GHERARDI, LM. TOMLINSON, G. WALTER, J.D. MARKS, M.B.LLEWELYN, and G. Winter. 1992. Structural repertoire of the human VH segments. J. Mol.BioI. 227: 799-817.

28. RUDIKOFF, S., A.M. GIUSTI, W.D. COOK, and M.D. SCHARFF. 1982. Single amino acid substi­tution altering antigen-binding specificity. Proc. Natl. Acad. Sci. U.S. A. 79: 1979-1983.

29. BRUGGEMANN, M., H.J. MULLER, C. BURGER, and K. RAjEWSKY. 1986. Idiotypic selection ofan antibody mutant with changed hapten binding specificity, resulting from a point muta­tion in position 50 of the heavy chain. EMBO J. 5: 1561-1566.

30. HILLSON, J.L., N.S. KARR, LR. OPPLIGER, M. MANNIK, and E.H. SASSO. 1993. The struc­tural basis of germline-encoded VH3 immunoglobulin binding to staphylococcal protein A.J.Exp.Med.178:331-336.

31. FOOTE, J. and G. WINTER. 1992. Antibody framework residues affecting the conformationof the hypervariable loops. J. Mol. BioI. 224: 487-499.

644 . S. GOLEMBOWSKI et al.

32. ARNHEIM, N. and H. EHRLICH. 1992. Polymerase chain reaction strategy. Annu. Rev.Biochem. 61: 131-156.

33. TILGNER, ]., S. GOLEMBOWSKI, B. KERSTEN, W. STERRY, and S. ]AHN. 1997. VH genesexpressed in peripheral blood 19E-producing B cells from patients with atopic dermatitis.Clin. Exp. 1mmunol. 107: 528-535.

34. ]AHN, S., D. ROGGENBUCK, B. NIEMANN, and E.S. WARD. 1995. Expression of monovalentfragments derived from a human 19M autoantibody in E. coli. The input of the somaticallymutated CDRI/CDR2 and of the CDR3 into antigen binding specificity. 1mmunobiology193: 400-419.

35. CHANG, B. and P. CASALI. 1995. A sequence analysis of human germline 19 VH and VLgenes. The CDRls of a major proportion of VH, but not VL, genes display a high inherentsusceptibility to amine acid replacement. Ann. N. Y. Acad. Sci. 764: 170-179.

36. SHLOMCHIK, M.J., A.H. AUCOIN, D.S. PISETSKY, and M.G. WEIGERT. 1987. Structure andfunction of anti-DNA autoantibodies derived from a single autoimmune mouse. Proc. Natl.Acad. Sci. U. S. A. 84: 9150-9154.

37. STRAUBINGER, B., E. HUBER, W. LORENZ, E. OSTERHOLZER, W. PARGENT, M. PECH, H.D.POHLENZ, F.J. ZIMMER, and H.G. ZACHAU. 1988. The human VK locus. Characterization ofa duplicated region encoding 28 different immunoglobulin genes. ]. Mol. BioI. 199: 23-24.

SIGBERT ]AHN, M.D., Department of Dermatology, Medical Faculty (Charite), Schumannstrasse20/21, 10117 Berlin, Germany. Tel.: +49-30-2802 3383, Fax: +49-30-2802 3383, e-mail:sigbert.jahn@charite.de

Medicine Meets Millennium

World Congress on Medicine and Health, 21. 7.2000 - 20. 8.2000 at Con­gress Centrum Hannover, Germany under patronage of German ChancellorGerhard Schroder.

Main Topics Internal Medicine:Autoimmunity and Infectious DiseasesOncologyPharmaceuticalsHeart Diseases

3. 8.-5. 8. 20008.8.2000

10.8.200011.8.2000

Detailed program information:www.mh-hannover.de/mmm or Congress Partner GmbH, Birkenstr. 37,D - 28195 Bremen (Tel.: +49(0)421 303131; e-mail: MMM@cpb.de)