In colorectal cancer, the immune response is particularly pronounced against tumors displaying the high microsatellite instability (MSI-H) phenotype. MSI-H tumors accumulate mutations affecting microsatellites located within protein encoding regions (coding microsatellites, cMS), which lead to translational shifts of the respective reading frames. Consequently, novel tumor-specific frameshift-derived neopeptides (FSP) are generated and presented by MSI-H tumor cells, thus eliciting effective cytotoxic immune responses. To analyze whether the immunoselective pressure was reflected by the phenotype of MSI-H colorectal cancer cells, we compared here the expression of antigen processing machinery (APM) components and human leukocyte antigen (HLA) class I antigen subunits in 20 MSI-H and 20 microsatellite-stable (MSS) colorectal cancer using a panel of newly developed APM component-specific monoclonal antibodies. In addition, we did a systematic analysis of mutations at cMS located within APM genes and β2-microglobulin (β2m). Total HLA class I antigen loss was observed in 12 (60.0%) of the 20 MSI-H lesions compared with only 6 (30.0%) of the 20 MSS colorectal cancer lesions. Moreover, total loss of membraneous HLA-A staining was significantly more frequent in MSI-H colorectal cancer (P = 0.0024). Mutations at cMS of β2m and genes encoding APM components (TAP1 and TAP2) were detected in at least 7 (35.0%) of 20 MSI-H colorectal cancers but in none of the MSS colorectal cancers (P = 0.0002). These data show that defects of HLA class I antigen processing and presentation seem to be significantly more frequent in MSI-H than in MSS colorectal cancer, suggesting that in MSI-H colorectal cancer the immunoselective pressure leads to the outgrowth of cells with defects of antigen presentation.

Antigen presentation of the human leukocyte antigen (HLA) class I pathway is mediated by a network of several proteins. Briefly, peptide fragments that are generated from cellular proteins by the proteasome complex are transported to the endoplasmatic reticulum (ER) by TAP1 and TAP2 proteins. There, peptides are loaded on β2-microglobulin (β2m)–associated HLA class I heavy chain molecules, a process that is supported by several chaperone molecules (for a detailed review, see ref. 1).

Malignant tumors may generate a variety of antigens that may be presented to and recognized by the host's immune system and may eventually lead to destruction of the respective tumor cells by CTLs (reviewed in refs. 2, 3). Abnormalities of antigen presentation have been found in various malignancies and may reflect the immune selective pressure imposed by the continuous antitumoral immune response of the host (reviewed in ref. 4). The immunoselection hypothesis has been supported by the increased frequency of HLA class I antigen presentation defects observed in tumors upon application of T cell–based immune therapies (5, 6). Different mechanisms may eventually lead to impairments of HLA class I antigen presentation (7). Inactivation of both alleles of the gene encoding β2m, most frequently because of mutation in one allele in combination with loss of heterozygosity of the β2m gene locus seems to be the most frequent mechanism leading to complete loss of HLA class I expression on the cell surface (8). Alterations of proteasome subunits (LMP proteins) and the transporter proteins TAP1 and TAP2 also play an important role, and have been observed in different tumor entities (911).

In colorectal cancer, there is growing evidence that the extent of the antitumoral immune response correlates with the type of genetic instability. Colorectal cancer may emerge through two principally different carcinogenic pathways. The majority of colorectal cancers are characterized by numerical and structural chromosomal alterations (chromosomal instability). In about 15% of colorectal cancers, defects of the DNA mismatch repair system result in rapid accumulation of deletions or insertions of single nucleotides particularly in repetitive DNA sequences (microsatellites), a process commonly referred to as high microsatellite instability (MSI-H). There are several characteristics of MSI-H colorectal cancer which indicate a particularly high immunogenicity of these tumors, for example, a dense lymphocytic infiltration (12, 13) and the comparably low frequency of distant metastases (14). Recent studies suggest that the high immunogenicity of MSI-H colorectal cancer is induced by the abundant expression of frame-shift neopeptides (FSP) that are generated as a consequence of insertion/deletion mutations at coding microsatellites (cMS; refs. 15, 16). The immunogenicity of several MSI-associated FSPs has been shown in vitro (1518).

Taken together, there is strong evidence for a tight immunologic surveillance particularly of MSI-H colorectal cancer. MSI-H colorectal cancer cells with defects of the antigen processing machinery (APM) might therefore be insistently selected for during tumorigenesis. Indeed, β2m mutations leading to total HLA class I loss have been found to be closely associated with the MSI-H phenotype in colorectal cancer (19, 20).

In this study, we analyzed in detail whether the different level of the immunoselective pressure on colorectal cancer of the MSI-H or the microsatellite-stable (MSS) phenotype was reflected by a differential pattern of defects in HLA class I antigen processing and presentation. Therefore, we systematically examined the APM profile of 20 MSI-H and 20 MSS colorectal cancers by immunohistochemistry using a panel of newly developed monoclonal antibodies (mAb). To obtain additional information, mutation analysis was done, focusing on cMS located in APM component-encoding genes. Our data suggest, that MSI-H colorectal cancers are significantly more often affected by alterations of HLA class I antigen processing and presentation. In particular, total HLA class I and selective HLA-A antigen loss on tumor cell surface was closely linked to the MSI-H phenotype. Furthermore, we found that transporter subunits TAP1 and TAP2 may be directly inactivated by MSI-related frame-shift mutations at cMS.

Tissue samples and preparation. Forty formalin-fixed, paraffin-embedded colorectal cancer lesions (MSS, n = 20; MSI-H, n = 20) were randomly selected from tumors previously tested for their microsatellite status according to the recommendations of the National Cancer Institute/ICG-HNPCC (21). No patient included in this study had received neoadjuvant therapy. Blocks were cut into 2-μm sections for immunohistochemical staining with mAb and into 5-μm sections for the preparation of genomic DNA. Informed consent was obtained from all patients included in this study.

Antibodies. The mAb HC-10 which recognizes a determinant expressed on β2m-free HLA-A10, HLA-A28, HLA-A29, HLA-A30, HLA-A31, HLA-A32, and HLA-A33 heavy chains and on all β2m-free HLA-B and HLA-C heavy chains (22, 23); the mAb HC-A2 which recognizes a determinant expressed on β2m-free HLA-A (excluding HLA-A24), HLA-B7301, and HLA-G heavy chains (22, 24); the anti-β2m mAb L368 (25); the anti-calnexin mAb TO-5; the anti-ERp57 mAb TO-2; the anti-calreticulin mAb TO-11; and the anti-tapasin mAb TO-3 were developed and characterized as described (26). The anti-LMP2 mAb SY-1 and the anti-TAP1 mAb TO-1 were developed and characterized using the strategy described elsewhere (26). Briefly, the mAb-secreting hybridomas were derived from BALB/c mice immunized with synthetic peptides derived from the amino acid sequence of the native protein and with recombinant proteins. Antibodies of the desired specificity were identified by their specific binding to the immunizing peptides in ELISA. The specificity of the selected mAb was proven by their reactivity with the corresponding antigens when tested with lymphoid cell lysates with the appropriate phenotype in Western blotting. The specificity of anti-LMP2 mAb SY-1 and anti-TAP1 mAb TO-1 was corroborated further by their lack of reactivity with a lysate of the T2 cell line, which does not express these molecules (27). Second antibodies (biotinylated anti-mouse IgG) were purchased from Vector (Burlingame, CA).

Immunohistochemistry. Tissue sections were stained using the Vectastain elite avidin-biotin complex detection system (Vector) according to the manufacturer's instructions. Briefly, 2-μm sections were deparaffinized with xylene and passaged through decreasing concentrations of ethanol. Subsequently, antigens were retrieved by heating the slides in a microwave oven (thrice, 5 minutes, 560 W) at pH 6.0. Tissue sections were incubated with an optimal amount of first antibody at 4°C overnight (5 μg/mL of mAb HC-10, TO-1, TO-5, and SY-3; 6.7 μg/mL of mAb HC-A2, L368, SY-1, SY-2, SY-4, SJJ-3, TO-3, TO-11, and NB1). After washing and incubation with an optimal amount of biotinylated anti-mouse IgG antibodies as the secondary antibody (Vector) for 1 hour at 37°C, tissue sections were stained using 3-amino-9-ethylcarbazole (DakoCytomation, Glostrup, Denmark) or diaminobenzidine (DakoCytomation) as substrates.

Tissue sections were scored as positive, heterogeneous, and negative when the percentage of stained tumor cells was >75, 25 to 75, and <25, respectively. All slides were scored by two observers independently (C.B. and M.K.). Staining of normal cells in each tissue section was used as an internal positive control; slides that exhibited no positive staining of normal cells were not included in the evaluation.

Identification and mutation analysis of coding microsatellites. A systematic European Molecular Biology Laboratory (EMBL) database search for translated (coding) microsatellites in human was done (ref. 28; update based on EMBL Rel. 71, June 2002). Mononucleotide repeat tracts consisting of six or more nucleotides were considered for further analysis.

For mutation analysis, tumor and normal tissue DNA was isolated from microdissected tissue sections according to the manufacturer's instructions (DNeasy Tissue Kit, Qiagen, Hilden, Germany). Oligonucleotide primers used for APM gene fragment analysis are listed in Table 1. Fragment length analysis was done on an ABI 3100 genetic analyzer (Applied Biosystems, Darmstadt, Germany). Genescan Analysis Software (Applied Biosystems) was used for data evaluation. Exonwise sequencing of β2m gene was carried out as described (29).

Table 1.

Oligonucleotide primers used for mutation analysis of cMS in APM component-encoding genes

GeneAccession no.RepeatPosition*Primer sensePrimer antisenseProduct (bp)Ta
Calnexin BC003552 T8 1409 TATTGGTTTGGAGCTGTGGTC ATCAGCAGCTTTCTTCAGGC 124 59 
Delta X61971 G6 273 CGACTGGTGACTCCTCTCCT TCCCGGTAGGTAGCATCAAC 136 59 
ERp57 a U42068 T6 552 ACGTTTATGGTTTGGAATGTCC CGGTAGTTATCCCTCAAGTTGC 148 59 
ERp57 b U42068 A6 752 GTTTGAGGACAAGACTGTGGC CTTTTTAATTTTGCCACTGGTC 61 59 
ERp57 c U42068 C6 1536 GATCTTTCTGTTTTCAGGGTGG AAAGTGGTGTTTGGCTACTGC 153 59 
LMP7 U17496 C6 54 GAGAGCGGACAGATCTCTGG GAAACTGTAGTGTCCTGGGTCC 138 59 
TAP1 X57522 G6 2376 GTACTCCCGCTCAGTGCTTC AGAAGGCTTTCATTCTGGAGC 184 59 
TAP2 AB073779 C6 218 TACTGTGGCTGCTTCAGGG AGGGAGACAGTCAGGGGG 181 59 
Tapasin a AB010639 C6 569 GCTCTGCTGGACTTGAGCTT CCAGGAGCAGATGTCCCTTA 130 58 
Tapasin b AB010639 C6 884 GCAGGCAAACTGAGGGTCT GTTGCTGGCATCAGGGAC 107 59 
GeneAccession no.RepeatPosition*Primer sensePrimer antisenseProduct (bp)Ta
Calnexin BC003552 T8 1409 TATTGGTTTGGAGCTGTGGTC ATCAGCAGCTTTCTTCAGGC 124 59 
Delta X61971 G6 273 CGACTGGTGACTCCTCTCCT TCCCGGTAGGTAGCATCAAC 136 59 
ERp57 a U42068 T6 552 ACGTTTATGGTTTGGAATGTCC CGGTAGTTATCCCTCAAGTTGC 148 59 
ERp57 b U42068 A6 752 GTTTGAGGACAAGACTGTGGC CTTTTTAATTTTGCCACTGGTC 61 59 
ERp57 c U42068 C6 1536 GATCTTTCTGTTTTCAGGGTGG AAAGTGGTGTTTGGCTACTGC 153 59 
LMP7 U17496 C6 54 GAGAGCGGACAGATCTCTGG GAAACTGTAGTGTCCTGGGTCC 138 59 
TAP1 X57522 G6 2376 GTACTCCCGCTCAGTGCTTC AGAAGGCTTTCATTCTGGAGC 184 59 
TAP2 AB073779 C6 218 TACTGTGGCTGCTTCAGGG AGGGAGACAGTCAGGGGG 181 59 
Tapasin a AB010639 C6 569 GCTCTGCTGGACTTGAGCTT CCAGGAGCAGATGTCCCTTA 130 58 
Tapasin b AB010639 C6 884 GCAGGCAAACTGAGGGTCT GTTGCTGGCATCAGGGAC 107 59 
*

Position of first nucleotide in cMS repeat in sequence of given accession number.

Primer annealing temperature.

Statistical analysis. For the comparison of immunohistochemical staining results and frequencies of mutational events at cMS, Fisher's exact test was applied. In addition, a logistic regression model (30) was applied for the comparison of MSI-H and MSS colorectal cancer.

Human leukocyte antigen class I heavy chains and β2m expression in colorectal cancer lesions. MSI-H and MSS colorectal cancer lesions were examined for HLA class I subunit expression. The β2m-specific mAb L368 and the mAb HC-10 and HC-A2 that recognize different epitopes expressed on HLA class I heavy chains were used as probes. Because HLA class I complexes are denatured during formalin fixation, total and membraneous localization of staining signals were documented separately to distinguish between free heavy chains or β2m molecules and assembled HLA class I heavy chains/β2m complexes transferred to the cell surface.

Staining results are summarized in Table 2. A loss of membraneous heavy chain signals was observed in all tumors exhibiting total loss of β2m staining (except two cases with regionally retained membraneous HC-A2 signals; data not shown).

Table 2.

HLA class I antigen expression in MSI-H, MSS, and total colon carcinoma specimens

HC-10HC-10 membrane*HC-A2HC-A2 membrane*β2mβ2m membrane*
MSI-H       
    Positive (%) 7 (35.0) 3 (15.0) 3 (15.8) 0 (0.0) 9 (45.0) 6 (30.0) 
    Heterogeneous (%) 4 (20.0) 3 (15.0) 8 (42.1) 2 (10.5) 3 (15.0) 2 (10.0) 
    Negative (%) 9 (45.0) 14 (70.0) 8 (42.1) 17 (89.5) 8 (40.0) 12 (60.0) 
    Samples analyzed 20 20 19 19 20 20 
MSS       
    Positive (%) 12 (60.0) 5 (25.0) 8 (44.4) 5 (27.8) 10 (50.0) 7 (35.0) 
    Heterogeneous (%) 6 (30.0) 8 (40.0) 7 (38.9) 7 (38.9) 6 (30.0) 7 (35.0) 
    Negative (%) 2 (10.0) 7 (35.0) 3 (16.7) 6 (33.3) 4 (20.0) 6 (30.0) 
    Samples analyzed 20 20 18 18 20 20 
Total       
    Positive (%) 19 (47.5) 8 (20.0) 11 (29.7) 5 (13.5) 18 (45.0) 13 (32.5) 
    Heterogeneous (%) 10 (25.0) 11 (27.5) 15 (40.5) 9 (24.3) 10 (25.0) 9 (22.5) 
    Negative (%) 11 (27.5) 21 (52.5) 11 (29.7) 23 (62.2) 12 (30.0) 18 (45.0) 
    Samples analyzed 40 40 37 37 40 40 
    P 0.05 0.11 0.10 0.0024 0.24 0.09 
HC-10HC-10 membrane*HC-A2HC-A2 membrane*β2mβ2m membrane*
MSI-H       
    Positive (%) 7 (35.0) 3 (15.0) 3 (15.8) 0 (0.0) 9 (45.0) 6 (30.0) 
    Heterogeneous (%) 4 (20.0) 3 (15.0) 8 (42.1) 2 (10.5) 3 (15.0) 2 (10.0) 
    Negative (%) 9 (45.0) 14 (70.0) 8 (42.1) 17 (89.5) 8 (40.0) 12 (60.0) 
    Samples analyzed 20 20 19 19 20 20 
MSS       
    Positive (%) 12 (60.0) 5 (25.0) 8 (44.4) 5 (27.8) 10 (50.0) 7 (35.0) 
    Heterogeneous (%) 6 (30.0) 8 (40.0) 7 (38.9) 7 (38.9) 6 (30.0) 7 (35.0) 
    Negative (%) 2 (10.0) 7 (35.0) 3 (16.7) 6 (33.3) 4 (20.0) 6 (30.0) 
    Samples analyzed 20 20 18 18 20 20 
Total       
    Positive (%) 19 (47.5) 8 (20.0) 11 (29.7) 5 (13.5) 18 (45.0) 13 (32.5) 
    Heterogeneous (%) 10 (25.0) 11 (27.5) 15 (40.5) 9 (24.3) 10 (25.0) 9 (22.5) 
    Negative (%) 11 (27.5) 21 (52.5) 11 (29.7) 23 (62.2) 12 (30.0) 18 (45.0) 
    Samples analyzed 40 40 37 37 40 40 
    P 0.05 0.11 0.10 0.0024 0.24 0.09 
*

General staining and membraneous staining are documented separately.

Loss of β2m expression accompanied by total HLA class I antigen loss on the cell surface (Fig. 1A) was observed in 12 (60.0%) of the 20 MSI-H colorectal cancer lesions and in only 6 (30.0%) of the 20 MSS colorectal cancer lesions (Table 2). In addition, MSI-H colorectal cancers frequently displayed a specific loss of certain HLA class I heavy chains detected by the mAb HC-10 (9 of 20, 45.0%) and HC-A2 (8 of 20, 42.1%). In detail, the frequency of reduced staining with mAb HC-10 was significantly higher in MSI-H tumors than in MSS tumors (9 of 20, 45.0% versus 2 of 20, 10%; P = 0.05). Loss of membrane staining with mAb HC-A2 was significantly more often observed in MSI-H than in MSS colorectal cancer lesions (17 of 19, 89.5% versus 6 of 18, 33.3%; P = 0.0024).

Figure 1.

A, immunohistochemical stainings with mAbs L368 (β2m, left), HC-A2 (middle), and HC-10 (right). Tumor U13 (MSI-H, top row): tubular tumor crypts on the right are β2m positive and retain membrane staining with mAbs specific for HLA class I heavy chains. Carcinoma cells (left) present complete loss of β2m expression. In the β2m-negative region, reactivity with mAbs HC-10 (reduced in carcinoma compared with surrounding stroma cells) and HC-A2 is exclusively cytoplasmatic. Tumor U18 (MSI-H, bottom row): Complete loss of β2m expression in tumor cells; note positive staining in tumor-infiltrating lymphocytes. Weak cytoplasmatic staining with mAb HC-10 indicates reduction of HLA class I heavy chain expression; HC-A2 reactivity is lost completely. Surrounding stromal cells serve as internal positive control. B, immunohistochemical staining of tumors that harbor mutations at coding microsatellites of APM genes. Tumor U5 (TAP1 mutation positive, top row) and U10 (TAP2 mutation positive, bottom row) with mAbs TO-1 and SY-2, respectively. Two regions of each tumor. Expression of TAP1 and TAP2 is regionally retained in both tumors (right) but lost or markedly reduced in regions near the infiltration front (left). Infiltrating lymphocytes serve as internal positive controls.

Figure 1.

A, immunohistochemical stainings with mAbs L368 (β2m, left), HC-A2 (middle), and HC-10 (right). Tumor U13 (MSI-H, top row): tubular tumor crypts on the right are β2m positive and retain membrane staining with mAbs specific for HLA class I heavy chains. Carcinoma cells (left) present complete loss of β2m expression. In the β2m-negative region, reactivity with mAbs HC-10 (reduced in carcinoma compared with surrounding stroma cells) and HC-A2 is exclusively cytoplasmatic. Tumor U18 (MSI-H, bottom row): Complete loss of β2m expression in tumor cells; note positive staining in tumor-infiltrating lymphocytes. Weak cytoplasmatic staining with mAb HC-10 indicates reduction of HLA class I heavy chain expression; HC-A2 reactivity is lost completely. Surrounding stromal cells serve as internal positive control. B, immunohistochemical staining of tumors that harbor mutations at coding microsatellites of APM genes. Tumor U5 (TAP1 mutation positive, top row) and U10 (TAP2 mutation positive, bottom row) with mAbs TO-1 and SY-2, respectively. Two regions of each tumor. Expression of TAP1 and TAP2 is regionally retained in both tumors (right) but lost or markedly reduced in regions near the infiltration front (left). Infiltrating lymphocytes serve as internal positive controls.

Close modal

Mutations affecting the β2m gene. Exonwise sequencing of the β2m gene detected mutations in 5 (29.4%) of 17 MSI-H colorectal cancer samples (Table 3). In three cases, β2m sequencing was not possible due to poor DNA quality of the paraffin-embedded lesions. All tumors that displayed β2m mutations were not stained by β2m-specific mAb L368 in immunohistochemistry (Fig. 1A). All identified mutations were localized at coding repeats of the β2m gene, the (CT)4 repeat in exon 1 (three samples) and two A5 repeats in exon 2 (four samples; Table 3B). Two of the β2m-negative tumors had two mutations, one in exon 1 and one in exon 2 (U5 and U17). We could not determine whether these mutations were localized on the same or on different alleles of the β2m gene. No β2m mutations were detected in the 20 MSS tumor samples.

Table 3.

A. Mutations at cMS located within APM component encoding genes
cMSU1U2U3U4U5U6U7U8U9U10U11U12U13U14U15U16U17U18U19U20Total (%)
β2m NA mut1 wt wt mut2/mut4 wt NA wt wt wt wt wt mut5 wt NA wt mut3/mut5 mut5 wt wt 5/17 (29.4) 
Calnexin T8 wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/20 (0) 
Delta G6 NA wt wt wt wt wt wt wt wt wt wt wt wt wt NA wt wt wt NA wt 0/17(0) 
ERp57 T6 wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/20 (0) 
ERp57 A6 wt wt wt wt NA wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/19 (0) 
ERp57 C6 NA wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/19 (0) 
LMP7 C6 wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/20 (0) 
TAP1 G6 wt wt wt wt mut6 mut6 wt wt wt wt wt wt NA wt wt wt wt NA wt wt 2/18 (11.1) 
TAP2 C6 NA wt wt wt wt wt wt wt wt mut7 wt wt wt wt NA wt wt wt NA wt 1/17 (5.9) 
Tapasin a C6 wt NA wt wt wt wt wt NA wt NA wt NA wt wt wt NA wt NA NA NA 0/12 (0) 
Tapasin b C6 NA wt wt wt wt wt wt wt wt NA wt NA wt wt NA wt wt wt NA wt 0/15 (0) 
                       
B. Description of mutations on nucleotide and amino acid level                       
Gene  Mutation   Description on nucleotide level      Description on amino acid level       Consequence     
β2m  mut1   c.43_44delCT      L15fsX55       Frameshift, stop 41 aa downstream     
  mut2   c.47_48delCT      S16fsX55       Frameshift, stop 40 aa downstream     
  mut3   c.45_48delTTCT      L15fsX42       Frameshift, stop 28 aa downstream     
  mut4   c.204_205insA      V68fsX88       Frameshift, stop 21 aa downstream     
  mut5   c.204delA      K67fsX102       Frameshift, stop 36 aa downstream     
TAP1  mut6   c.2351delG      G784fsX878       Frameshift, nonsense stretch of 95 aa     
TAP2  mut7   c.223delC      L75X       Stop codon     
A. Mutations at cMS located within APM component encoding genes
cMSU1U2U3U4U5U6U7U8U9U10U11U12U13U14U15U16U17U18U19U20Total (%)
β2m NA mut1 wt wt mut2/mut4 wt NA wt wt wt wt wt mut5 wt NA wt mut3/mut5 mut5 wt wt 5/17 (29.4) 
Calnexin T8 wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/20 (0) 
Delta G6 NA wt wt wt wt wt wt wt wt wt wt wt wt wt NA wt wt wt NA wt 0/17(0) 
ERp57 T6 wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/20 (0) 
ERp57 A6 wt wt wt wt NA wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/19 (0) 
ERp57 C6 NA wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/19 (0) 
LMP7 C6 wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt 0/20 (0) 
TAP1 G6 wt wt wt wt mut6 mut6 wt wt wt wt wt wt NA wt wt wt wt NA wt wt 2/18 (11.1) 
TAP2 C6 NA wt wt wt wt wt wt wt wt mut7 wt wt wt wt NA wt wt wt NA wt 1/17 (5.9) 
Tapasin a C6 wt NA wt wt wt wt wt NA wt NA wt NA wt wt wt NA wt NA NA NA 0/12 (0) 
Tapasin b C6 NA wt wt wt wt wt wt wt wt NA wt NA wt wt NA wt wt wt NA wt 0/15 (0) 
                       
B. Description of mutations on nucleotide and amino acid level                       
Gene  Mutation   Description on nucleotide level      Description on amino acid level       Consequence     
β2m  mut1   c.43_44delCT      L15fsX55       Frameshift, stop 41 aa downstream     
  mut2   c.47_48delCT      S16fsX55       Frameshift, stop 40 aa downstream     
  mut3   c.45_48delTTCT      L15fsX42       Frameshift, stop 28 aa downstream     
  mut4   c.204_205insA      V68fsX88       Frameshift, stop 21 aa downstream     
  mut5   c.204delA      K67fsX102       Frameshift, stop 36 aa downstream     
TAP1  mut6   c.2351delG      G784fsX878       Frameshift, nonsense stretch of 95 aa     
TAP2  mut7   c.223delC      L75X       Stop codon     

Abbreviations: aa, amino acid; mut, mutant; wt, wild type.

*

(CT)4 dinucleotide repeat and three mononucleotide repeats are summarized.

Expression and mutation analysis of antigen processing machinery components. To identify APM genes that harbor coding mononucleotide repeats and thus may be targeted by mutations during MSI-H tumorigenesis, a database search was done. One T8 mononucleotide repeat was found in the calnexin gene, nine additional repeats composed of six nucleotides were detected in the ER chaperone ERp57, tapasin, the proteasome subunit Delta, the immunoproteasome subunit LMP7, and the TAP subunits TAP1 and TAP2 (Table 1).

Fragment length analysis of these loci revealed frame-shift mutations in the transporter genes TAP1 and TAP2 in 2 of 18 (11.1%) and 1 of 17 (5.9%) of the lesions, respectively. Coding repeats of proteasome subunits Delta and LMP7 and of chaperone molecules calnexin, ERp57, and tapasin were not found to be mutated in the present collective (Table 3).

In addition, the expression of APM components was evaluated by immunohistochemistry. No significant differences were observed between MSI-H and MSS colorectal cancer (Table 4). Interestingly, the three carcinoma lesions that encompassed mutant TAP1 and TAP2 alleles all retained regional immunoreactivity with the respective antibodies and were scored heterogeneous (Fig. 1B).

Table 4.

Expression of APM components in MSI-H and MSS colorectal cancer lesions

TAP 1TAP2LMP 7MB1LMP 2DeltaLMP 10ZErp57TapasinCalnexinCalreticulin
MSI-H             
    Positive (%) 6 (30.0) 15 (75.0) 15 (78.9) 12 (60.0) 13 (65.0) 13 (65.0) 10 (58.8) 5 (26.3) 16 (80.0) 9 (50.0) 9 (45.0) 14 (70.0) 
    Heterogeneous (%) 7 (35.0) 4 (20.0) 3 (15.8) 5 (25.0) 3 (15.0) 4 (20.0) 6 (35.3) 5 (26.3) 4 (20.0) 3 (16.7) 7 (35.0) 5 (25.0) 
    Negative (%) 7 (35.0) 1 (5.0) 1 (5.3) 3 (15.0) 4 (20.0) 3 (15.0) 1 (5.9) 9 (47.4) 0 (0) 6 (33.3) 4 (20.0) 1 (5.0) 
    Samples analyzed 20 20 19 20 20 20 17 19 20 18 20 20 
MSS             
    Positive (%) 4 (21.1) 9 (45.0) 12 (60.0) 11 (57.9) 13 (65.0) 14 (70.0) 9 (50.0) 7 (36.8) 11 (55.0) 10 (50.0) 15 (75.0) 10 (52.6) 
    Heterogeneous (%) 6 (31.6) 6 (30.0) 7 (35.0) 3 (15.8) 3 (15.0) 5 (25.0) 4 (22.2) 3 (15.8) 6 (30.0) 5 (25.0) 2 (10.0) 6 (31.6) 
    Negative 9 (47.4) 5 (25.0) 1 (5.0) 5 (26.3) 4 (20.0) 1 (5.0) 5 (27.8) 9 (47.4) 3 (15.0) 5 (25.0) 3 (15.0) 3 (15.8) 
    Samples analyzed 19 20 20 19 20 20 18 19 20 20 20 19 
Total             
    Positive (%) 10 (25.6) 24 (60.0) 27 (69.2) 23 (59.0) 26 (65.0) 27 (67.5) 19 (54.3) 12 (31.6) 27 (67.5) 19 (50.0) 24 (60.0) 24 (61.5) 
    Heterogeneous (%) 13 (33.3) 10 (25.0) 10 (25.6) 8 (20.5) 6 (15.0) 9 (22.5) 10 (28.6) 8 (21.1) 10 (25.0) 8 (21.1) 9 (22.5) 11 (28.2) 
    Negative (%) 16 (41.0) 6 (15.0) 2 (5.1) 8 (20.5) 8 (20.0) 4 (10.0) 6 (17.1) 18 (47.4) 3 (7.5) 11 (28.9) 7 (17.5) 4 (10.3) 
    Samples analyzed 39 40 39 39 40 40 35 38 40 38 40 39 
    P 0.85 0.11 0.37 0.62 1.00 0.78 0.27 0.66 0.15 0.83 0.12 0.52 
TAP 1TAP2LMP 7MB1LMP 2DeltaLMP 10ZErp57TapasinCalnexinCalreticulin
MSI-H             
    Positive (%) 6 (30.0) 15 (75.0) 15 (78.9) 12 (60.0) 13 (65.0) 13 (65.0) 10 (58.8) 5 (26.3) 16 (80.0) 9 (50.0) 9 (45.0) 14 (70.0) 
    Heterogeneous (%) 7 (35.0) 4 (20.0) 3 (15.8) 5 (25.0) 3 (15.0) 4 (20.0) 6 (35.3) 5 (26.3) 4 (20.0) 3 (16.7) 7 (35.0) 5 (25.0) 
    Negative (%) 7 (35.0) 1 (5.0) 1 (5.3) 3 (15.0) 4 (20.0) 3 (15.0) 1 (5.9) 9 (47.4) 0 (0) 6 (33.3) 4 (20.0) 1 (5.0) 
    Samples analyzed 20 20 19 20 20 20 17 19 20 18 20 20 
MSS             
    Positive (%) 4 (21.1) 9 (45.0) 12 (60.0) 11 (57.9) 13 (65.0) 14 (70.0) 9 (50.0) 7 (36.8) 11 (55.0) 10 (50.0) 15 (75.0) 10 (52.6) 
    Heterogeneous (%) 6 (31.6) 6 (30.0) 7 (35.0) 3 (15.8) 3 (15.0) 5 (25.0) 4 (22.2) 3 (15.8) 6 (30.0) 5 (25.0) 2 (10.0) 6 (31.6) 
    Negative 9 (47.4) 5 (25.0) 1 (5.0) 5 (26.3) 4 (20.0) 1 (5.0) 5 (27.8) 9 (47.4) 3 (15.0) 5 (25.0) 3 (15.0) 3 (15.8) 
    Samples analyzed 19 20 20 19 20 20 18 19 20 20 20 19 
Total             
    Positive (%) 10 (25.6) 24 (60.0) 27 (69.2) 23 (59.0) 26 (65.0) 27 (67.5) 19 (54.3) 12 (31.6) 27 (67.5) 19 (50.0) 24 (60.0) 24 (61.5) 
    Heterogeneous (%) 13 (33.3) 10 (25.0) 10 (25.6) 8 (20.5) 6 (15.0) 9 (22.5) 10 (28.6) 8 (21.1) 10 (25.0) 8 (21.1) 9 (22.5) 11 (28.2) 
    Negative (%) 16 (41.0) 6 (15.0) 2 (5.1) 8 (20.5) 8 (20.0) 4 (10.0) 6 (17.1) 18 (47.4) 3 (7.5) 11 (28.9) 7 (17.5) 4 (10.3) 
    Samples analyzed 39 40 39 39 40 40 35 38 40 38 40 39 
    P 0.85 0.11 0.37 0.62 1.00 0.78 0.27 0.66 0.15 0.83 0.12 0.52 

Human cancer cells accumulate various structural alterations that may give rise to translation of modified or novel peptide sequences that may be presented to the immune system. Immune cells may recognize these peptides as foreign and attack the respective tumor cells (4). Colorectal cancer cells with deficient DNA mismatch repair that display the MSI-H phenotype are characterized by a pronounced local infiltration with CTL, a phenomenon that has been proposed as a histopathologic criterion to identify MSI-H colorectal cancer (13). This suggested that MSI-H cancer cells may present particularly antigenic peptides. In line with this hypothesis, recent studies indeed revealed a novel class of antigenic peptides generated by frame-shift mutations of genes encompassing coding repetitive sequences. These peptides are apparently abundantly expressed and presented by MSI-H colorectal cancer cells (1518). This might well explain the selective pressure favoring the outgrowth of β2m-negative MSI-H tumor cells which are incapable of HLA class I antigen-mediated presentation.

In the present study, we therefore examined in detail the type and frequency of APM component alterations that may be found in MSI-H colorectal cancer cells as a consequence of the immunoselective pressure and compared them with MSS colorectal cancer. Previous studies that examined alterations of antigen presentation in colorectal cancer either did not distinguish between MSI-H and MSS samples (e.g., refs. 31, 32), or relied upon a collection of colorectal cancer specimens containing only a very small proportion of MSI-H tumors (e.g., refs. 19, 20). In this study, we examined for the first time a cohort of colorectal carcinomas that had been typed for their microsatellite status a priori. Thus, equal numbers of MSI-H and MSS samples could be included in the analysis.

Due to the denaturation of the HLA class I complex during the formalin fixation process, the immunohistochemical detection of functionally active HLA heavy chain/β2m complexes was not feasible. Therefore, β2m positivity and membraneous localization of anti-HLA heavy chains antibody stainings were used as the criterion indicative for the expression of intact HLA class I complexes on tumor cell surface. Using this approach, a complete loss of membraneous β2m staining accompanied by loss of membraneous HLA class I heavy chain staining was observed in 60.0% of the MSI-H carcinoma lesions compared with only 30.0% of the MSS carcinomas. Mutations of the β2m gene were detected in five of these β2m-negative MSI-H colorectal cancers, in the remaining seven cases, no mutations were detected, suggesting, for example, large deletions which could not be detected by gene sequencing or loss of β2m expression not related to mutational events. No β2m mutations were detected in any of the MSI-H tumors displaying positive β2m staining in immunohistochemistry or any of the MSS tumors. Thus, in a larger collection of MSI-H cases, our study confirms previous studies assuming a close correlation of β2m mutations with the MSI-H phenotype in colorectal cancer (19, 20). The elevated mutation frequency of the β2m gene most probably indicates that β2m represents a relevant target gene during MSI carcinogenesis. In addition, we observed that selective HLA-A down-regulation was significantly more frequent in MSI-H colorectal cancers. Menon et al. (33) have previously reported an association of HLA-A negativity with a better prognosis in colorectal cancer patients; however, MSI typing had only been done in six HLA-negative tumors, with MSI present in three of six (50%) samples analyzed. The authors therefore speculated that there might be a relation of HLA-A negativity with MSI. Our data prove that down-regulation or loss of HLA-A heavy chains is closely associated with the MSI-H phenotype, thus validating the assumption of Menon et al. (33) in a collective that has been MSI-typed a priori.

Notably, significant differences between MSI-H and MSS colorectal cancer were restricted to membraneous HLA-A staining, whereas general HLA-A staining differences did not reach statistical significance. These data may point to the functional relevance of HLA-A loss in antigen presentation breakdown, suggesting that β2m-positive tumors, potentially capable of HLA class I antigen presentation, are particularly affected by loss of HLA-A expression, thereby reflecting the particular immunoselective pressure exerted on these tumors. In contrast, HLA-B and HLA-C staining results showed significant differences only when regarding general but not membraneous staining. Hence, the significance of loss of HLA-B and HLA-C expression in MSI-H colorectal cancer awaits further clarification.

Previous studies suggested the involvement of proteasome subunits and transporters of antigen presentation in immune evasion of colorectal cancer cells. TAP1 has been reported to be lost in ≥14% of colorectal cancer (32, 34), whereas LMP7 and TAP2 down-regulation has been found to impair antigen presentation predominantly in MSS colorectal cancer (20). We were interested to define whether these APM components or ER chaperones might be targets of microsatellite instability and hit by mutations in cMS. Whereas no mutations were detected in proteasome subunits and ER chaperones, frame-shift mutations in transporter genes TAP1 and TAP2 were identified in 3 of 20 MSI-H carcinomas. These mutations were not detected in corresponding normal tissue samples, underlining that they were not polymorphisms but really somatic mutations occurring during MSI tumorigenesis. TAP1 mutations were detected in two cases; one case harbored a mutation at the C6 repeat of TAP2. Interestingly, none of the tumors with mutations at these loci presented a complete immunohistochemical loss of the corresponding protein. This observation may be explained either by retained protein translation from an intact second allele, or by a positive staining reaction due to the use of antibodies recognizing epitopes located upstream of the mutated repeat. In this case, immunohistochemical staining reactivity may be preserved, even if the corresponding proteins have lost their functional activity in the tumor cells.

Using immunohistochemistry, TAP1 and TAP2 were not detected in 41.0% and 15.0% of the lesions analyzed, respectively. Although not statistically significant, we observed a trend towards a higher frequency of TAP2 loss in MSS colorectal cancer. This observation is in agreement with the data of Cabrera et al. (20) who described dysregulation of TAP2 as one important factor contributing to APM impairment in MSS colorectal cancer.

In summary, our data show that the MSI-H phenotype in colorectal cancer is associated with a high frequency of defects in HLA class I antigen presentation. Most APM components did not reveal significant differences between MSI-H and MSS colorectal cancer, underlining that frame-shift mutations of the β2m gene represent the predominant cause of antigen presentation breakdown in MSI-H colorectal cancer. In addition, a specific loss of HLA-A heavy chains is significantly associated with the MSI-H phenotype. Moreover, we provide evidence that genes coding for TAP subunits TAP1 and TAP2 may be immediate mutational targets of MSI carcinogenesis. Further studies examining the prognostic effect of APM alterations on the survival of MSI-H colorectal cancer patients are currently in progress.

Note: M. Kloor and C. Becker contributed equally to this article.

Grant support: Deutsche Krebshilfe and National Cancer Institute, Department of Health and Human Services/USPHS grant CA67108.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank I. Voehringer and B. Kuchenbuch for excellent technical assistance and Y. Schwitalle for critical reading of the article and for helpful discussion.

1
van Kaer L. Major histocompatibility complex class I-restricted antigen processing and presentation.
Tissue Antigens
2002
;
60
:
1
–9.
2
Jager D, Jager E, Knuth A. Immune responses to tumour antigens: implications for antigen specific immunotherapy of cancer.
J Clin Pathol
2001
;
54
:
669
–74.
3
Boon T, van den Eynde B. Tumour immunology.
Curr Opin Immunol
2003
;
15
:
129
–30.
4
Chang CC, Campoli M, Ferrone S. HLA class I defects in malignant lesions: what have we learned?
Keio J Med
2004
;
52
:
220
–9.
5
Restifo NP, Marincola FM, Kawakami Y, Taubenberger J, Yanelli JR, Rosenberg SA. Loss of functional β2-microglobulin in metastatic melanomas from five patients receiving immunotherapy.
J Natl Cancer Inst
1996
;
88
:
100
–8.
6
Garcia-Lora A, Algarra I, Gaforio JJ, Ruiz-Cabello F, Garrido F. Immunoselection by T lymphocytes generates repeated MHC class I-deficient metastatic tumor variants.
Int J Cancer
2001
;
91
:
109
–19.
7
Seliger B, Cabrera T, Garrido F, Ferrone S. HLA class I antigen abnormalities and immune escape by malignant cells.
Semin Cancer Biol
2002
;
12
:
3
–13.
8
Benitez R, Godelaine D, Lopez-Nevot MA, et al. Mutations of the β2-microglobulin gene result in a lack of HLA class I molecules on melanoma cells of two patients immunized with MAGE peptides.
Tissue Antigens
1998
;
52
:
520
–9.
9
Dissemond J, Goette P, Moers J, et al. Association of TAP1 downregulation in human primary melanoma lesions with lack of spontaneous regression.
Melanoma Res
2003
;
13
:
253
–8.
10
Dissemond J, Goette P, Moers J, et al. Immunoproteasome subunits LMP2 and LMP7 downregulation in primary malignant melanoma lesions: association with lack of spontaneous regression.
Melanoma Res
2003
;
13
:
371
–7.
11
Seliger B, Atkins D, Bock M, et al. Characterization of human lymphocyte antigen class I antigen-processing machinery defects in renal cell carcinoma lesions with special emphasis on transporter-associated with antigen-processing down-regulation.
Clin Cancer Res
2003
;
9
:
1721
–7.
12
Dolcetti R, Viel A, Doglioni C, et al. High prevalence of activated intraepithelial cytotoxic T lymphocytes and increased neoplastic cell apoptosis in colorectal carcinomas with microsatellite instability.
Am J Pathol
1999
;
154
:
1805
–13.
13
Smyrk TC, Watson P, Kaul K, Lynch HT. Tumor-infiltrating lymphocytes are a marker for microsatellite instability in colorectal carcinoma.
Cancer
2001
;
91
:
2417
–22.
14
Buckowitz A, Knaebel HP, Benner A, et al. Microsatellite instability in colorectal cancer is associated with local lymphocyte infiltration and low frequency of distant metastases.
Br J Cancer
2005
;
92
:
1746
–53.
15
Linnebacher M, Gebert J, Rudy W, et al. Frameshift peptide-derived T-cell epitopes: a source of novel tumor-specific antigens.
Int J Cancer
2001
;
93
:
6
–11.
16
Saeterdal I, Gjertsen MK, Straten P, Eriksen JA, Gaudernack G. TGF βRII frameshift-mutation-derived CTL epitope recognised by HLA-A2-restricted CD8+ T cells.
Cancer Immunol Immunother
2001
;
50
:
469
–76.
17
Ripberger E, Linnebacher M, Schwitalle Y, Gebert J, von Knebel Doeberitz M. Identification of an HLA-A0201-restricted CTL epitope generated by a tumor-specific frameshift mutation in a coding microsatellite of the OGT gene.
J Clin Immunol
2003
;
23
:
415
–23.
18
Schwitalle Y, Linnebacher M, Ripberger E, Gebert J, von Knebel Doeberitz M. Immunogenic peptides generated by frameshift mutations in DNA mismatch repair deficient cancer cells.
Cancer Immunity
2004
;
4
:
14
.
19
Bicknell DC, Kaklamanis L, Hampson R, Bodmer WF, Karran P. Selection for β2-microglobulin mutation in mismatch repair defective colorectal carcinomas.
Curr Biol
1996
;
6
:
1695
–7.
20
Cabrera CM, Jimenez P, Cabrera T, Esparza C, Ruiz-Cabello F, Garrido F. Total loss of MHC class I in colorectal tumors can be explained by two molecular pathways: b2-microglobulin inactivation in MSI-positive tumors and LMP7/TAP2 downregulation in MSI-negative tumors.
Tissue Antigens
2003
;
61
:
211
–9.
21
Boland CR, Thibodeau SN, Hamilton SR, et al. A National Cancer Institute workshop on microsatellite instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colon cancer.
Cancer Res
1998
;
58
:
5248
–57.
22
Stam NJ, Spits H, Ploegh HL. Monoclonal antibodies raised against denatured HLA-B locus heavy chains permit biochemical characterization of certain HLA-C locus products.
J Immunol
1986
;
137
:
2299
–306.
23
Perosa F, Luccarelli G, Prete M, Favoino E, Ferrone S, Dammacco F. β2-Microglobulin-free HLA class I heavy chain epitope mimicry by monoclonal antibody HC-10-specific peptide.
J Immunol
2003
;
171
:
1918
–26.
24
Sernee MF, Ploegh HL, Schust DJ. Why certain antibodies cross-react with HLA-A and HLA-G: epitope mapping of two common MHC class I reagents.
Mol Immunol
1998
;
35
:
177
–88.
25
Lampson LA, Fisher CA, Whelan JP. Striking paucity of HLA-A, B, C and β2-microglobulin on human neuroblastoma cell lines.
J Immunol
1983
;
130
:
2471
–8.
26
Ogino T, Wang X, Kato S, Miyokawa N, Harabuchi Y, Ferrone S. Endoplasmatic reticulum chaperone-specific monoclonal antibodies for flow cytometry and immunohistochemical staining.
Tissue Antigens
2004
;
62
:
385
–93.
27
Salcedo M, Momburg F, Hammerling GJ, Ljunggren HG. Resistance to natural killer cell lysis conferred by TAP1/2 genes in human antigen-processing mutant cells.
J Immunol
1994
;
152
:
1702
–8.
28
Woerner SM, Gebert J, Yuan YP, et al. Systematic identification of genes with coding microsatellites mutated in DNA mismatch repair-deficient cancer cells.
Int J Cancer
2001
;
93
:
12
–9.
29
Bicknell DC, Rowan A, Bodmer WF. β2-Microglobulin gene mutations: a study of established colorectal cell lines and fresh tumors.
Proc Natl Acad Sci U S A
1994
;
91
:
4751
–6.
30
Harrell FE. Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. New York: Springer; 2001.
31
Norazmi M, Hohmann AW, Skinner JM, Bradley J. Expression of MHC class I and class II antigens in colonic carcinomas.
Pathology
1989
;
21
:
248
–53.
32
Atkins D, Ferrone S, Schmahl GE, Storkel S, Seliger B. Down-regulation of HLA class I antigen processing molecules: an immune escape mechanism of renal cell carcinoma?
J Urol
2004
;
171
:
885
–9.
33
Menon AG, Morreau H, Tollenaar RA, et al. Down-regulation of HLA-A expression correlates with a better prognosis in colorectal cancer patients.
Lab Invest
2002
;
82
:
1725
–33.
34
Kaklamanis L, Townsend A, Doussis-Anagnostopoulou IA, Mortensen N, Harris AL, Gatter KC. Loss of major histocompatibility complex-encoded transporter associated with antigen presentation (TAP) in colorectal cancer.
Am J Pathol
1994
;
145
:
505
–9.