Advertisement

Computational modeling and confirmation of leukemia-associated minor histocompatibility antigens

Jefferson L. Lansford, Udara Dharmasiri, Shengjie Chai, Sally A. Hunsucker, Dante S. Bortone, James E. Keating, Ian M. Schlup, Gary L. Glish, Edward J. Collins, Gheath Alatrash, Jeffrey J. Molldrem, Paul M. Armistead and Benjamin G. Vincent

Data supplements

Article Figures & Data

Figures

  • Figure 1.

    mHA prediction algorithm schema. (A) The alleles at cSNPs for all 101 DRPs in the patient cohort were determined using the Illumina NS-12 array platform (n = 13 917 cSNPs). (B) Each allele for each cSNP was mapped to the reference coding transcriptome. (C) The peptide sequences for all 8-, 9-, 10-, 11-, 16-, 20-, and 24-mer amino acid sequences that contained the allele of interest were computationally generated (n = 2 189 712 amino acid sequences covering both alleles for all 11 172 cSNPs). (D) All peptides that could be mHAs for a given DRP (ie, one allele was present in the recipient, but not the donor) were tested for predicted binding affinity to the DRP’s HLA types. (E) All predicted mHAs were classified as GVL or GVHD associated based upon the tissue expression of the source protein for the mHA.

  • Figure 2.

    Patient cohort survival outcomes. (A) Two-year OS was 55%, and there was no significant difference in OS based upon disease type (AML, MDS, or CML). (B) There was a statistically significant (P = .0010) difference in OS based upon disease status at time of SCT (25% vs 65%, not in CR/CP vs CR/CP). (C-D) There were no significant differences in OS based upon conditioning type (myeloablative conditioning [MAC] or RIC) or donor source (MRD or MUD).

  • Figure 3.

    Characteristics of mHAs by HLA type and donor relation. (A) There is a large difference in the number of peptides that can bind with high affinity to class I HLAs (0.09 to 1.32 per gMM). (B) MUD SCT is associated with roughly twice as many class I mHAs compared with MRD SCT. (C) Because of variable frequencies of peptide presentation by different HLA molecules, there is a greater variability in the mHAs compared with gMMs among both MRD and MUD patients. (D) There is a linear association between the number of GVL mHAs and GVHD mHAs. (E) There is no association between the number of GVL mHAs and relapse for MRD or MUD. (F) There is similarly no association between the number of GVH mHAs and GVHD.

  • Figure 4.

    Confirmation of previously discovered mHAs. (A) Eighteen previously described HLA class I and class II mHAs were potentially identifiable in the patient genotyping data set by having the relevant cSNP contained on the NS-12 array and the appropriate HLA type contained in the patient population. For each pie chart, the total number represents the number of DRPs in the 101-patient cohort that expressed the appropriate HLA for the examined mHA. The red wedges represent the number of DRPs where the actual mHA could be presented (ie, the appropriate gMM existed in the DRP, and the actual peptide was contained in the total peptide pool). Of the 18 mHAs, 14 were successfully identified by the prediction algorithm. Two mHAs (LB-NUP133 and HA-3) were represented on rare HLAs, and no patients in the cohort had the appropriate gMMs to predict the mHA. For the other 2 mHAs not predicted, the peptide epitope from the mHA derived from HEATR has a predicted binding affinity of >500 nM, and the mHA LB-ADIR is derived from an alternative reading frame. (B) Twelve cSNPs from the SNP array data mapped to known class I mHAs that could be contained in the generated peptide data set. The HLA binding affinity for all 8- to 11-mer DRP peptide pairs to all class I HLA types contained in the patient data set for the 12 cSNPs is shown. A total of 20 926 DRP peptide pairs are shown with increasing shades of gray corresponding to the frequency of the peptide pair in the patient cohort. The 12 cSNPs have been shown to yield 17 mHA peptides that are represented by the red dots. Only 2 of the 17 mHAs fall out of our threshold Kd of 500 nM: the HEATR mHA and 1 confirmed mHA derived from LB-APOBEC3B. The UNC-GRK4-V peptide pair (see Figures 5 and 6) is also mapped (yellow dot).

  • Figure 5.

    UNC-GRK4-V is expressed in AML. (A) GRK4 has limited tissue expression, with transcripts only detectable by RT-PCR in human AML and testis, but not in PBMCs, liver, colon, or skin. (B) Western blot confirms GRK4 protein expression in testis and 3 of 4 human AML samples. (C) Extracted ion chromograms using DIMS identifies the EC to allow the maximum signal from UNC-GRK4-V (m/z = 1049.5) into the MS using the pure standard (red trace), which was 86 V/cm. Extracted ion chromogram of m/z = 1049.5 in the U937.A2 cell epitope pool (blue trace) shows that other species with m/z = 1049.5 can be detected across a range of EC. (D) The fragmentation pattern for pure UNC-GRK4-V peptide was determined for comparison with results from the peptide pool. (E) Targeted MS was performed on the epitope pool by setting EC = 86 V/cm and fragmenting the parent ion with an m/z = 1049.5. The resulting MS/MS spectrum is virtually identical to that of the pure peptide.

  • Figure 6.

    UNC-GRK4-V–specific T cells are present post-SCT. UNC-GRK4-V–specific T-cell populations were tested in 12 HLA-A*02:01–expressing AML patients who had undergone allo-SCT. Negative tetramers were used for each sample to define tetramer-positive gates for each patient. (A-C) Three patients in DRPs with gMMs for UNC-GRK4-V were tested, with 1 patient (A, alive 29 months post-SCT with chronic GVHD) showing an expanded tetramer-positive population. (D-I) Six patients in DRPs where both the recipient and donor carried the UNC-GRK4-V allele were tested, and 3 patients (D, alive 75 months post-SCT with mild chronic GVHD; E, died of relapsed AML 21 months post-SCT; F, alive 84 months post-SCT without GVHD) showed evidence of a tetramer-positive population. (J-L) Three patients in DRPs in which the patient did not carry the UNC-GRK4-V allele were tested, and no samples showed a tetramer-positive population.

Tables

  • Table 1.

    Patient characteristics

    CharacteristicNo. of patients (N = 101)
    Age (range), y48 (21-72)
    Sex, n (%)
     Male64 (63)
     Female37 (37)
    Disease, n (%)
     AML61 (60)
     CML25 (25)
     MDS14 (14)
     MPN1 (1)
    Disease stage, n (%)
     CR/CP48 (48)
     Not in CR/CP53 (52)
    Conditioning intensity, n (%)
     Myeloablative71 (71)
     Reduced intensity30 (29)
    Donor type, n (%)
     HLA-matched related donor72 (72)
     HLA-matched unrelated donor29 (28)
    Graft source, n (%)
     Peripheral blood stem cells68 (67)
     Bone marrow33 (33)
    Sex mismatch, n (%)
     Female donor into male recipient28 (28)
     Male donor into female recipient19 (19)