Tumor marker-encoding genes: a bunch of mysterious diamonds

in the pile of evolutionary compost

A.V. Baranova, George Mason University

TUMOR MARKERSTumor markers are proteins or mRNAs

that are detected in higher-than-normal amounts in the blood, urine or body tissues

of some patients with certain types of cancer.

Most tumor markers are produced by the tumor itself,

but not by normal tissue

Tumor markers could be immunogenic;

they could serve as targets for anti-tumor immune response

Changes in tumor marker concentration during the course of disease:

No response to treatment

Good response

Remission

Relapse

No treatemnt

Secondremission

Known examples of tumor markers PSA and PAP

Prostate Specific Antigen (PSA) and Prostatic Acid Phosphatase (PAP)

Prostatic carcinoma, but also in prostatitis and benign prostatic hyperplasia (BPH)

Carcinoembryonic Antigen (CEA)Used for monitoring of colorectal and lung cancer

Alpha-fetoprotein (AFP) AFP is normally produced by a developing fetus.

An elevated level of AFP strongly suggests the presence of either primary liver cancer or germ cell cancer

CA125Ovarian carcinoma.

Very good for monitoring of response and recurrence

Healthy

Benign inflammatory conditions

“Grey diagnostic area”

CANCER

?

Level of single tumor marker often is not definitive enough

Combining of known tumor markersinto marker panels

will help to increase sensitivity and specificity of diagnostics

MULTIPLE TUMOR MARKERS NEED TO BE OBTAINED FOR EACH COMMON TYPE OF CANCER

preferably by high-throughput methods

Protein based methods of tumor marker discovery

• Occasional findings in course of other research

• Various modifications of SEREX (serological expression of cDNA expression libraries

• Proteomics–based methods, including:2D –PAGE analysis

SELDI-TOF mass-spectroscopy analysisReverse proteomics arrays

Remove background

True screening

Aim: to reveal the repertoire of antigens

eliciting an antibody response in cancer patients.

Cancer Immunome Database [www2.licr.org/CancerImmunomeDB]

Near 100 of good TMs are revealed by this method

From: Wellcome Trust website

Two-dimensional polyacrylamide gel electrophoresis (2D-PAGE)

is a technique for separating complex mixtures of proteinsby their mass and/or charge .

From: Wellcome Trust website

APPLY WOLTAGE TOP TO BOTTOM

From: Wellcome Trust website

APPLY WOLTAGE SIDE TO SIDE

2D –PAGE analysisNormal serum Tumor serum

Potential tumor marker

Example: from University of Kent website

MORE TYPICAL 2D-PAGE GEL

Surface-enhanced laser desorption/ionization (SELDI-TOF)

mass-spectrometry

proteins of interest are selectively absorbed toa chemically modified surface.

Mass/charge

From Virginia Prostate Center website

TUMOR

TUMOR

Hyperplasia

Hyperplasia

Normal

Normal

DNA-based Differential DisplayA common term for group of methods

for a search of the transcripts which expression in the sample of interest

differs from that in so-called “control sample”.

High throughput methodsLow throughput methods

High budget:SAGE,microarrays

Low budget:Computational approaches

It’s time forcomputational genomics

In UniGene, individual human ESTs are already clustered

by homology.

Each EST is derived from described tissue source

including tumor and normal samples.

GenBank contains 5,120,207 of human ESTs

on May 30, 2005

Example of ESTs that belong

to the same gene/UNIGENE cluster

Manual verification of cDNA libraries sources

cDNA library type

Number of libraries

Number of clusters

number of ESTs

Tumor 2 681 56 x 103 9 x 105 Normal 1 087 53 x 103 8 x 105 Undefined 227 - 5 x 105 Total 3 995 88,4 x 103 2,2 x 106

cDNA libraries in GeneBank

TUMOR

NORMAL

UNDEFINED

A sources for cross-verification: UNIGENE, CGAP, TIGR, STRATAGENE, PubMed

Figure of year 2001

HSAnalyst and ClustOut software

HSAnalyst and ClustOut software developed in Functional Genomics Group,

IGG RAS

Software is able to sort EST clusters according to their size, tissue origin or

other categories describing individual ESTs

First round of work (year 2000). Cluster-sorting software HSAnalyst

To list all human clusters containing 100% of ESTs from tumor-derived cDNA libraries with more than 16 ESTs in the cluster -

21 cluster

To list all human clusters containing less than 10% of ESTs from normal cDNA libraries with more than 16 ESTs in the cluster –

83 clusters

StatisticsEST number Number of tumour related clusters at threshold, %*

> 90 % 100%

Range Entries clusters

Tumour

specific EST, % observed calculated observed calculated

1-2 59111 44373 42% 18342 23073 18342 23073

3-4 45400 13401 35% 1880 1884 1880 1884

5-8 53569 8742 37% 567 279 567 172

9-16 63421 5407 39% 168 5 99 4

17-32 83968 3607 41% 45 0 17 0

33-64 176845 3762 43% 16 0 2 0

65-128 349008 3790 45% 10 0 2 0

129-256 460493 2588 47% 8 0 0 0

257-512 339482 975 50% 3 0 0 0

513-1024 208171 303 53% 1 0 0 0

1025-2048 130524 96 57% 0 0 0 0

2049-4096 95180 36 60% 0 0 0 0

4097-8192 49804 10 66% 0 0 0 0

8193-16384 14725 1 67% 0 0 0 0

FEBS Letters, 508 (2001), 143-148

RT-PCR of Brachiury genein human tumor

and normal tissues

Cytokine production and cytotoxic activity of CTLs specific for three Brachyury-derived peptides.

A, CD8+ Tcells generated from PBMC of a healthy donor against peptidesT-p2 and T-p3 were stimulated for 24 h in the presence of

Brachyury (T)-specific peptides or irrelevant peptide-pulsed autologous DCs. IFN-g was evaluated in the supernatants by ELISA.

B. Cytotoxic activity (6-h assay) of CTLs generated with peptideT-p2 against peptide-pulsed C1R-A2 targets. Two effector-to-target ratios (E:T) were used as

indicated.

Cytotoxic activity of Brachyury-specific CTLs against tumor targets. CTLs established from the blood of (C) a colorectal cancer patient

(patient1) and (D) an ovarian cancer patient (patient 2) were used after three IVS for cytotoxic killing of H441 and AsPC-1 tumor

cells.

Tumor marker properties of Brachiury were predicted computationally

• It was the first study to show that (a) a T-box transcription factor and (b) a molecule implicated in mesodermal development, i.e., EMT, can be a potential target for human T-cell–mediated cancer immunotherapy.

• up-regulation of Brachyury occurs in certain tumor tissues and cancer cell lines

• Brachyury-specific CTLs can be generated from the blood of cancer patients and normal donors, which, in turn, can effectively lyse Brachyury-expressing tumor cells.

Recent round of work (year 2006). Cluster-sorting software ClustOut

Clusters containing 100% ESTs from human libraries derived from tumor samples

- 328 clusters

Clusters containing less than 10% ESTs from human libraries derived from normal samples

- 501 clusters

Tumor prevalent clusters

52 are genes with Gene IDs (10%)

501

10 known tumor markers

Insulin 268/3113

Parathyroid hormone 1/83

Unknown human genes with unstudied ORFs

13

Non-coding mRNAs

466 (87%)

Known tumor markers in cancer-specific subtraction

Gene Function Normal/cancerMYEOV myeloma overexpressed gene 2/113MAGEA6

melanoma antigens, family A

8/103MAGEA3 7/76MAGEA12 0/27

MAGEA1 1/13MAGEA9 1/13CEACAM8 carcinoembryonic antigen-related cell

adhesion molecule 82/30

CTAG2 Cancer/testis antigen 2 2/20TRAG3 taxol resistance associated gene 3 0/10APOBEC1 apolipoprotein B mRNA editing enzyme,

catalytic polypeptide0/59

10/501

Prospective marker proteins in cancer-specific subtraction

Gene Function Norm/cancerASCL1 achaete-scute complex-like 1 (Drosophila) 33/394

WNT7B wingless-type MMTV integration site family, member 7B

22/285

DLL3 delta-like 3 (Drosophila) 18/193

NOX1 NADPH oxidase 1 4/76

KREMEN2 kringle containing transmembrane protein 2 5/74

SPRR2A small proline-rich protein 2A 1/69

MS4A3 membrane-spanning 4-domains, subfamily A, member 3

5/65

GPR35 G protein-coupled receptor 35 3/46

SPRR1A small proline-rich protein 1A 4/46

OTP orthopedia homolog (Drosophila) 1/41

Known human genes represented by only tumor-derived EST

Published in FEBS Letters, Nov 2001

№ UniGene

cluster ID

Gene name

and function

Tumor-

derived ESTs/

total ESTs Tumor types

1 Hs.560 APOBEC1 mRNA editing enzyme

catalytic subunit 55/55 Colon carcinoma, CLL

2 Hs.1085 Guanylate cyclase 2C (heat-stable

enterotoxin receptor) 17/17 Colon carcinoma

6 Hs.152531 Heart and neural crest derivatives

expressed 1 18/18

Germ cell tumor, sarcoma, schwannoma, neuroblastoma

12 Hs.239891 G protein-coupled receptor 35 33/33 Colon carcinoma

16 Hs.285026 HERV-H LTR-associating 1 22/22 Colon carcinoma, germ cell tumor, HNSCC, AML

501 clusters with less than 10% of ESTs from normal tussues

Reverse subtraction: genes prevalent in normal samples

Normal prevalent clusters :

1048 clusters with less than 10% of ESTs from tumor tussues

560 with 100%

Tumor prevalent clusters:

328 with 100%

Normal prevalent clusters

612 are genes with Gene IDs (56,6%)

1048

10 crystallins

7 keratins

Tear prealbumin, lens intrinsic membrane protein 2

Unknown human genes with unstudied ORFs

37

Non-coding mRNAs

399 (36,9%)

16 (!) different solute carrier proteins (2,6%)

Non-coding RNAs in tumor-specific and normal–specific subtractions

In normal – prevalent clusters:

399 out of 1048 (36,9%)

In tumor prevalent clusters:

466 out of 501 (87%)

Possible explanation:

• Tumors express more non-coding mRNA than normal tissue due to:

-- overall weakness of gene expression control

(probably due to demethylation of genome)

Awakening of dormant promoters

Expression of sequences adjacent to dormant promoters with splicing according to cryptic splice sites

Experimental study

of Tumor Specificity:

Whether software based predictions are true or not???

Kozlov et al., 2003

Krukovskaja et al., 2004

Hs.133107- Non-coding mRNA- Located on Chromosome 12p13- Represented by 20 EST from the 5 tumor- derived libraries from Brain Tumors, Lung Carcinomas, Ovarian Carcinomas

20/20

Amplification of Hs.133107 Fragment on Human Tumor MTC Panel (Clontech)

1 2 3 4 5 6 7 8 9 NC PC

344 bp

1 – 100bp DNA ladder2 – Breast Carcinoma3 – Lung Carcinoma4 – Colon Adenocarcinoma5 – Lung Carcinoma

6 – Prostatic Adenocarcinoma7 – Colon Adenocarcinoma8 – Ovarian Carcinoma9 – Pancreatic AdenocarcinomaPC - Human DNA

Amplification of Hs.133107 Fragment onHuman NORMAL MTC Panel 1 (Clontech)

1 2 3 4 5 6 7 8 9 NC PC

344 bp

1 – 100 bp DNA ladder2 – Brain3 – Heart4 – Kidney5 - Liver

6 – Lung7 – Pancreas8 – Placenta9 – Skeletal MusclePC - Human DNA

Amplification of Hs.133107 Fragment onHuman NORMAL MTC Panel 2 (Clontech)

1 2 3 4 5 6 7 8 9 NC PC

344 bp

1 – 100bp DNA ladder2 – Colon3 – Ovary4 – Peripheral Blood Leukocyte5 - Prostate

6 – Small Intestine7 – Spleen8 – Testis9 – ThymusPC - Human DNA

Amplification of Hs.133107 Fragment on Human Immune System MTC Panel, (Clontech)

1 2 3 4 5 6 7 8 NC PC

344 bp

1 – 100 bp DNA ladder2 – Bone Marrow3 – Fetal Liver4 – Lymph Node5 – Peripheral Blood Leukocyte

6 – Spleen7 – Thymus8 – TonsilPC - Human DNA

Amplification of Hs.133107 Fragment onHuman Fetal MTC Panel (Clontech)

1 2 3 4 5 6 7 8 9 NC PC

344 bp

1 – 100 bp DNA ladder2 – Brain3 – Heart4 – Kidney5 - Liver

6 – Lung7 – Skeletal Muscle8 – Spleen9 – ThymusPC - Human DNA

Conclusion:

Hs.133107 is a Tumor-Specific Sequence

Hs.154173

- Non coding mRNA - Located in the intergenic spacer region within rRNA

encoding unit - characteristic for lung carcinoma, testicular

teratocarcinoma

23/23

Amplification of Hs.154173 Fragment on Human Tumor MTC Panel (Clontech)

1 2 3 4 5 6 7 8 9 NC PC

443 bp

1 – 100bp DNA ladder2 – Breast Carcinoma3 – Lung Carcinoma4 – Colon Adenocarcinoma5 – Lung Carcinoma

6 – Prostatic Adenocarcinoma7 – Colon Adenocarcinoma8 – Ovarian Carcinoma9 – Pancreatic AdenocarcinomaPC - Human DNA

Amplification of Hs.154173 Fragment on Human Normal MTC Panel 1 (Clontech)

1 2 3 4 5 6 7 8 9 NC PC

443 bp

1 – 100 bp DNA ladder2 – Brain3 – Heart4 – Kidney5 - Liver

6 – Lung7 – Pancreas8 – Placenta9 – Skeletal MusclePC - Human DNA

Three other normal panels looks same as previous;

Conclusion:

Hs.154173 is a Tumor-Specific Sequence

Hs.133294- Protein encoding mRNA- Located on Chromosome 1q21

IQGAP3 gene

- Weakly similar to IQGAP (human RAS GTPase-activating-like protein IQGAP1)

- Represented in: prostate tumor, HNSCC, breast carcinoma, oligodendroglioma, colon carcinoma, CML, lung carcinoma, ovarian carcinoma, uterus carcinoma, adrenal adenoma

- Minor occurrences in normal testis and germinal B-cells

58/62

5`

polyA 5380-5407

ATG 76 TGA 4971

3`

ATG 5014

IQGAP coding region

5`

polyA 6052-6079

ATG 76 TGA 4971

5235 GUAGGA

5905 GAUUCU

3`intron

ATG 5014 ATG 5242

IQGAP coding region

5336

5905

mRNA isoforms of IQGAP3 gene

4924

412 bp

34924

1084bp

PCR products from spliced and unspliced RNA forms of Hs.133294

Only 3’ underspliced mRNA variant of IQGAP3 gene is tumor –specific; (this part of the gene is non-coding either)

Conclusion:

One of the mRNA isoforms of the IQGAP3 gene

represented by Hs.133294

is a Tumor-Specific Sequence

Back to fundamental aspect of the tumor-specific mRNAs

What is a degree of conservativeness between human tumor-specific genes

and mouse genes?

Do they have orthologs in the mouse genome?

WHAT IS THEIR FUNCTION ???

Closer look into well-established tumor-markers

• 1) For most of the known tumor markers no clear function in the adult human cells is shown;

• 2) Looks like these genes not expressed in normal tissues just because they not needed there;

• 3) Looks like these genes are expressed in tumors just because tumors don’t care !!!

Prediction: no clear function low level of conservation

VISTA – special visualization tool for distant cross-species comparison

http://www-gsd.lbl.gov/vista/

Inna Dubchak and Eugene Rubin, 2002

Lawrence Berkeley National Laboratory (LBNL)

This tool allow graphical representation of the alignment plots

How normal gene looks in VISTA

CclbrAlignment 1Seqs: human/mouseCriteria: 75%, 50 bpRegions: 31X-axis: humanResolution: 15Window size: 50 bp

75%

75%

How tumor marker gene looks in VISTA (MAGE A4)

MGEA4 tumor marker

meningioma expressed antigen

75%

75%

You don’t believe me? Here is MAGE A9 – another one.

75%

75%

More detailed VISTA viewNRAS - oncogene

FUGU and CHICKEN ARE PRESENT !!!!

More detailed VISTA view

CEACAM6

Nothing here

Some genes are right in the middle: some exons are conservative, some

are not (non-coding ones) RFP2

Here it is

Graphical alignmentsNRAS

RFP2

CEACAM6

What about our own flock

of predicted tumor markers ???

170 cancer

strictly cancer specific EST

clusters

74 Clusters corresponding to mouse genes

>80% - nt

>60% -aa

83 Clusters not corresponding to mouse genes (evolutionary new)

< 70% nt

<40% - aa13 borderline clusters

“SIMPLE” BLAST analysis

Evolution often goes by duplications

What is the genomic source of these “evolutionary new” tumor markers????

22 of 170 cancer specific gene clusters are recently (only in primates) duplicated in human genome !!!

Including known tumor marker genes:

MAGE A4

MAGE A9

CTAG2 (cancer/testis specific antigen 2)

IDEA:

• In course of evolution new sequences become ready to be expressed

as they got joined to dormant promoter

• But unnecessary promotors are thoroughly suppressed in well-differentiated adult cells

• In tumors all these “yet-to-be” genes got their chance

TUMOR and NORMAL

TUMOR ONLY

Same is true for “parts of the genes” – separate exons,

especially non-coding exonscancer specific alternative splicing

Something to think about:

1. Can we use these non-coding tumor-specific mRNA as a tumor markers and targets for

immunotherapy?(protein-coding genes could be used more readily)

Yes. They always encode short ORFs linked with AUGs.

Such short ORFs are TRUE non-self antigens.

2. Why known tumor markers have carcino-embryonic or carcino-testicular profiles of expression?

-- embryos contain lots of less differentiated cells with relaxed expression control;

-- testes are covered by blood-testes barrier, so immune system is not watching for an expression of non-self ORFs

It explains

broadest pattern of expression observed in tested

Something to think about:

George Mason University, Fairfax, VA, USA

A.Baranova, M.Sikaroodi, G,Manyam, P. Gillevet

The Biomedical Center, St.Petersburg, Russia A.P. Kozlov,

L.L.Krukovskaja, D.Polev, I. Duhovlinov, Yu.Galachjants, N.Samusik

Vavilov Institute of General Genetics, Moscow, Russia

T.Tyazhelova

NCI, NIH

C.Palena, J.Schlom S.O’Brien

CEN TEL13q14.3

10 9 8 7 6 5 4 3 2 1A 1B

мРНК Dleu2мыши

мРНК DLEU2Bчеловека

1C

мРНК DLEU2Aчеловека

3B 3C

Experimental example:

DLEU2 gene in human and mouse (evol new gene)

Red = mouse-specific exons only; Green = human specific exons only