Robust Allo-Anti-D with Subsequent Anti-K Production After Transfusion of D-Positive Rbcs to a Patient with Weak D Type 1

Robust Allo-Anti-D with Subsequent Anti-K Production after Transfusion of D-Positive RBCs to a Patient with Weak D Type 1

Jennifer S. Woo, Anastasia Gikas, Morvarid Moayeri, Sara Bakhtary, Karen Rodberg, Sunitha Vege, Christine Lomas-Francis, Andrew Rossin, Ashok Nambiar and Connie M. Westhoff

10/13/2018 Introduction

• D antigen (RH1) is the most immunogenic of the blood group antigens

• Weak D types 1, 2, and 3 are generally considered not at risk for allo-anti-D

• We present a case of robust allo-anti-D production in a patient with weak D type 1 after transfusion with D-positive RBCs

www.aabb.org 2 Clinical Presentation

• 45-year-old Caucasian woman with metastatic ovarian carcinoma on Carboplatin/Paclitaxel therapy • No history of pregnancy (G0) or prior transfusion

Rh Serologic Typing Results Anti-D (Gel) Anti-D (IS Tube) 2+ 0 Rh Phenotype: C+c+E-e+

Serologic Weak D phenotype Reflex to RHD genotyping

www.aabb.org 3 RHD Genotyping

• Automated RHD BeadChip Prototype • PCR-multiplex analysis for RHD exons 4, 7, the inactivating RHD pseudogene and C/c genotyping • PCR-RFLP for RHD exon 8 (1136C>T) • Zygosity determination by hybrid box detection

RHD Genotyping Results RHD hemizygote: RHD*weak D type 1 c.809T>G (p.Val270Gly) Predicted RhD phenotype: D+ (weak)

www.aabb.org 4 Clinical Presentation Days after # RBC Hgb (g/dL) Antibody Screen D-positive transfused transfusion (units) 0 1 O-pos 7.2 Negative 7 2 O-pos 6.4/6.2 Negative 14 3 O-pos 4.5 Negative 18 1 O-pos 7.3 Negative 21 1 O-neg 7.4 Negative

www.aabb.org 5 Clinical Presentation Days after # RBC Hgb (g/dL) Antibody Screen D-positive transfused transfusion (units) 0 1 O-pos 7.2 Negative 7 2 O-pos 6.4/6.2 Negative 14 3 O-pos 4.5 Negative 18 1 O-pos 7.3 Negative 21 1 O-neg 7.4 Negative 49 6.8 Positive (anti-D)

No history of IVIG or RhIG to suggest passive anti-D www.aabb.org 6 Methods (Allo- or Auto-anti-D?)

• Serologic testing by standard methods using tube methodology

• RHD cDNA amplification and sequencing – Confirmed weak D type 1 allele with no additional nucleotide changes

• RHAG cDNA amplification and sequencing – No changes from conventional

www.aabb.org 7 Immunohematology

4 weeks post last D-positive transfusion Antibody Screen and Identification Anti-D (variable reactivity up to 3+ PEG IAT) Auto-control Negative DAT Negative Elution Anti-D (w+)

www.aabb.org 8 Immunohematology (Reference Lab)

4 weeks post last D-positive transfusion Antibody Identification All D-positive cells were reactive (2-3+) DAT Negative (elution study not performed) DTT-treated RBCs Ruled out anti-LW Auto-adsorption Could not be performed due to history of recent transfusion with Rh+ RBCs Allo-adsorption with DTT-ficin(ZZAP) rr RBCs Anti-D in adsorbed plasma

Har D variant RBCs (DIII, DIV, DV, DVI, DVII, R0 ) Reactive Weak D type 1 RBCs (n=3) Non-reactive

Findings all together suggestive of allo-anti-D

www.aabb.org 9 Drug-induced hemolytic anemia? Anti-D

18 Carboplatin and Taxol

) 16

dL Carboplatin and Taxol 14 12 10 8 6 4 Hemoglobin Hemoglobin (g/ Elevated total bilirubin 2 Reticulocytosis 0 Undetectable Haptoglobin

Time  Legend: Rh-negative RBC transfusion

Rh-positive RBC transfusion

NO antibodies to carboplatin or paclitaxel (Taxol)

www.aabb.org 10 Immunohematology

>3 months (16.5 wks) post D-positive transfusion

Antibody Identification Anti-D (3+ PEG IAT) One cell was 3+ (AUS?) Autoadsorption x1 Anti-D (3+ to 4+) Autocontrol Positive (1+) DAT Positive (2+ IgG, w+ C3) Elution D-positive: 3+ to 4+ D-negative: 0 to 1+s (no specificity)

www.aabb.org 11 Immunohematology

8 months post D-positive transfusion

Antibody Identification Anti-D (3+ PEG IAT) and anti-K* Auto-control Positive (w+) DAT/Elution Not performed

*Phenotype of prior transfused RBCs: at least 2 of 7 D+ units were also K+

www.aabb.org 12 Immunohematology (Reference Lab) 9.5 months post D-positive transfusion

Antibody Identification Anti-D (1+ PEG IAT) and anti-K DAT Positive (microscopic IgG only) Elution Very weak (microscopic) anti-D Weak D type 1 RBCs Non-reactive Auto-adsorption (2x) with Anti-D showed marginal weakening in papain-treated cells and titration adsorbed plasma Alba Advanced Partial D Typing Kit No Partial D pattern detected Matched the pattern of reactivity

Consistent with allo-anti-D www.aabb.org 13 Discussion

• Weak D type 1 individuals are not considered to be at risk for allo-anti-D based on observational studies primarily from Europe

• Majority of anti-D detected in weak D type 1, 2, or 3 individuals have characteristics consistent with auto-reactivity

• However, rare cases of allo-anti-D have been reported in weak D type 1, 2, and 3 individuals

www.aabb.org 14 Literature Review: Auto* and/or Allo-Anti-D in Weak D Type 1, 2, or 3 Individuals

Author Year Journal Article Type Weak D Type Beckers 2004 Vox Sanguinis Abstract Type 1

Roxby 2004 Vox Sanguinis Abstract Type 1

Daniels 2007 Transfusion Letter to the Editor Type 1 Medicine *unpublished observations

Pelc-Klopotowska 2010 Vox Sanguinis Abstract Type 2

Vege 2011 Transfusion Abstract Type 2 (n=7)

Nixon 2016 Transfusion Abstract Type 3

These reported cases do not speak to clinically significant RBC destruction from allo-anti-D *Some antibodies when tested had characteristics of both allo- and autoantibody www.aabb.org 15 Summary of Course and Immunohematology

Anti-D in Anti-D in plasma, Anti-D in plasma plasma Anti-D first identified eluate not and eluate and eluate tested

18 Carboplatin and Taxol ) 16 Taxol dL Carboplatin and Taxol 14 12 10 8 6

Hemoglobin Hemoglobin (g/ 4 2 0 Legend: Time  Rh-negative RBC transfusion

Rh-positive RBC transfusion Discussion

• It is unclear why this weak D type 1 patient made robust allo-anti-D • Patient factors at time of transfusion may have contributed – Inflammation or other unknown factors • As this patient also formed anti-K, transfusion of at least 2 D+K+ units and sensitization to both antigens raises the possibility of the two antigens complementing alloimmunization – Alloantibody “responders” often form multiple alloantibodies, usually at the same/time exposure

www.aabb.org 17 Conclusion

• We report robust production of allo-anti-D in a patient with weak D type 1

• Due to underlying anemia in the setting of chemotherapy treatment, it is not possible to definitively conclude that allo-anti-D caused RBC destruction and the use of D-negative blood was indicated

• Exposure to D and K antigens on the same RBCs along with additional possible patient factors may have increased the risk of alloimmunization to both antigens

www.aabb.org 18 Acknowledgements

University of California San New York Blood Center Francisco (UCSF) • Connie M. Westhoff • Anastasia Gikas • Christine Lomas-Francis • Morvarid Moayeri • Sunita Vege • Ashok Nambiar • Andrew Rossin • Sara Bakhtary

American Red Cross, Pomona • Karen Rodberg

www.aabb.org 19 References

• Beckers EAM, Ligthart PC, Overbeeke MAM, Masskant PA, van Rhenen DJ. A patient with weakD type 1 and anti-D: auto or allo? Vox Sanguinis. 2004; 87(Suppl. 2), 75 (abstract). • Daniels G, Poole G, Poole J. Partial D and weak D: can they be distinguished? Transfus Med. 2007 Apr;17(2):145-6. • Nixon C, Ochoa-Garay G, Sweeney J. A Patient with Weak D Type 3 and Anti-D Alloimmunization. Transfusion. 2016;56(Suppl):33A (abstract) • Noiret L, Slater A, Higgins JM. Determinants of red blood cell alloantibody detection duration: analysis of multiply alloimmunized patients supports peritransfusion factors. Transfusion. 2017 Aug;57(8):1930-1937. • Pelc-Klopotowska M, Guz K, Walaszcyk A, Orzinska A, Michalewska B. Weak D type 2 patient with allo anti-D. Vox Sanguinis. 2010;99(Suppl. 1),367 (abstract). • Pham BN, Roussel M, Peyrard T, Beolet M, Jan-Lasserre V, Gien D, Ripaux M, Bourgouin S, Kappler-Gratias S, Rouger P, Le Pennec PY. Anti-D investigations in individuals expressing weak D Type 1 or weak D Type 2: allo- or autoantibodies? Transfusion. 2011 Dec;51(12):2679-85. • Roxby D, Coloma M, Flegel WA, Poole J, Martin P, Abbott R. Observation of anti-D after D-positive transfusion in an individual with weak D type-1 phenotype. Vox Sanguinis. 2004; 87(Suppl. 2), 77-78 (abstract). • Sandler SG, Flegel WA, Westhoff CM, Denomme GA, Delaney M, Keller MA, Johnson ST, Katz L, Queenan JT, Vassallo RR, Simon CD; College of American Pathologists Transfusion Medicine Resource Committee Work Group. It's time to phase in RHD genotyping for patients with a serologic weak D phenotype. College of American Pathologists Transfusion Medicine Resource Committee Work Group. Transfusion. 2015 Mar;55(3):680-9. • Tormey CA, Stack G. The characterization and classification of concurrent blood group antibodies. Transfusion. 2009 Dec;49(12):2709- 18. • Vege S, Fong C, Lomas-Francis C, Nickle P, Horn T, Delaney M, Westhoff CM. Weak D Type 2 and Production of Anti-D. Transfusion. 2011;51(Suppl.),41A (abstract).

www.aabb.org 20 Differences in monocyte/macrophage phagocytic activity due to different Fc regions of human IgG3 isoallotypes

Selena Cen MSc. BSc. Research assistant/Dr. Donald Branch Lab Oct 13th, 2018 AABB, Boston, USA Human Immunoglobulins and their isoallotypes

IgG1 IgG2 IgG 29 Isoallotypes IgG3

IgG4 Fab IgA

IgM Fc IgE IgD Alloantibodies and their detection

Polyclonal anti-IgG Monoclonal anti-IgG - Serum of animals - Serum of murine (rabbit) sources (blends) - Multiple specificities - Narrow range of reactivity - Suffers from - Stable and consistent variations from animal to animal Generation of IgG isoallotypes Generation of 15 IgG3 isoallotypes Serological detection of IgG3 isoallotypes

FDA-approved Anti-IgGs

Monoclonal

Polyclonal

Monoclonal

Polyclonal IgG3 and its clinical importance

1. IgG3 is typically considered a clinically significant IgG subtype that is often associated with hemolytic pathology 2. IgG3-03 and IgG3-13 are found at their highest frequencies in several ethnic groups of African origins or parts of the middle east. 3. The hemolytic potential of different IgG isoallotypes has not been assessed. 4. As a result, we would like to use an in-vitro assay (MMA) to examine the reactivity of these IgG3 isoallotypes, in particular IgG3-03 and IgG3-13. MMA (monocyte monolayer assay) Antibody mediated phagocytosis Aims

We aim to compare the activities of purified recombinant hIgG3 anti-K1 isoallotypes in an opsonized-erythrocyte phagocytosis assay. Results: Gel IAT – indirect antiglobulin test Results: Phagocytosis of IgG3 opsonized K+ RBCs Results: Phagocytosis of IgG3 opsonized K+ RBCs by M1 vs. M2 macrophages Results: Phagocytic activity partially corresponds to IgG3 concentration

T itra tio n o f a n ti-K 1 Ig G 3 u s in g M M A

5 0 Ig G 3 -0 1 * * * * x 4 0 Ig G 3 -0 3

e * * *

* * * * d

n Ig G 3 -0 8 I 3 0

c * * * i

t Ig G 3 -1 3 y

c * * * *

o 2 0 Ig G 3 -1 8 g

a * *

h Ig G 3 -1 9

P 1 0

0 0 0 0 0 0 5 5 0 2 5 C o n c e n tr a tio n o f h Ig G 3 [n g /m L ] Results: Differences in Fab binding affinity Summary

1. Certain amino acid variations in the Fc region of hIgG3 antibodies lead to an enhanced or reduced capacity for their ability to induce phagocytosis by monocyte- macrophages. 2. Results indicate that all anti-K1 hIgG3 isoallotypes can be clinically significant as measured by MMA. 3. Since anti-K1 is the second most common cause of severe hemolytic disease of the fetus and newborn, antiglobulin reagents that fail to detect hIgG3-03 or hIgG3-13 could present a problem. Ongoing experiments

• Apply surface plasmon resonance assay to determine Fab binding affinity.

• Apply surface plasmon resonance assay to determine Fc binding affinity of FcγRI, FcγRII, FcγRIII.

• Examine the induction of hemolytic anemia events of different IgG3 clones in mouse models. Acknowledgements

Dr. Donald R. Branch Dr. James Zimring Dr. Gregory Denomme Canadian Blood Services Bloodworks NW Bloodcenter of Wisconsin Lab: Lab: Beth Binnington Heather Howie Bonnie Lewis Jenna Lebedev Raymond Wong Anton Neschardim Cathy Branch Visit blood.ca Appendix Low Cost High Throughput Next Generation Sequencing Based Blood Group Typing Using Molecular Inversion Probes (MIPs) William Lane, Ozkan Aydemir, Patrick Marsh, Changxin Shen, Sunitha Vege, Nicholas Hathaway, Yong Zhao, Connie Westhoff, Jeffrey Bailey

Department of Pathology Disclosures

I have no relevant financial relationships to disclose for this presentation. Talk Outline

• Blood Group Typing Challenges • Molecular Inversion Probes (MIPs) • MIP based Next Generation Sequencing Blood Group Typing Extended Serologic Typing

A1; M+N+, U+, Vr−, Mt(a−), Ri(a−), Ny(a−), Or−, ERIK−, Os(a−), ENEP+, ENEH+, ENAV+, ENEV+, MNTD−; S+s+, He−, M(v−), s(D−), Mit−; Patient Cells P1+/P1−, pk+, NOR−; D+, Tar−; C−c+E+e−, CW−, CX−, EW−, V−, VS−, Rh26+LOCR− Be −, Crawford−CELO+, JAL−CEST+, STEM−, JAHK−; Lu(a−b+), LURC+, Lu4+, Lu5+, Lu6+, Lu7+, Lu8+, Lu13+, Lu16+, Lu17+, Au(a+b−), Lu20+, Lu21+; K−k+, Kp(a−b+c−); Js(a−b+), Ul(a−), K11+, K12+, K13+, K14+, K18+, K19+, K22+, K23−, VLAN−VONG−, TOU+, RAZ+, KALT+, KTIM+, KYO−, KUCI+, KANT+, KASH+, KELP+, KETI+, KHUL+; Le(a+b−); Fy(a+b+); Jk(a+b+); Di(a−b+), Wr(a−b+), Wd(a−), Rb(a−), WARR−, ELO−, Bp(a−), Mo(a−), Hg(a−), Vg(a−), Sw(a−), BOW−, NFLD−, Jn(a−), KREP−, Tr(a−), Fr(a−), SW1−, Wu−DISK+; Yt(a+b−); Sc1+Sc2−, Rd−, STAR+, SCER+, SCAN+; Do(a+b+), Jo(a+), DOYA+, Hy+, DOMR+, DOLG+; Co(a+b−), Co4+; LW(a+b−); Ch1+, Ch2+, Ch3+, Ch4+, Ch5+, Ch6+, Rg1−, Rg2−; Ch1−, Ch2−, Ch3−, Ch4−, Ch5−, Ch6−, Rg1+, Rg2+; H+; Kx+; Es(a+), Wb−, An(a−), Dh(a−), GEIS−, GELP+, Antibody Reagent GEAT+, GETI+; Cr(a+), Tc(a+b−c−), Dr(a+), Es(a+), WES(a−b+), UMC+, GUTI+, SERF+, ZENA+, CROV+, CRAM+, CROZ+; Kn(a+b−), McC(a+b−), Sla+Vil−, Yk(a+), Sl3+, KCAM+/KCAM−; In(a−b+), INFI+, INJA+; Ok(a+), OKGV+, OKVM+; MER2+; JMHK+, JMHL+, JMHG+, JMHM+, JMHQ+; I+; P+; GIL+; Duclos+, Ol(a−), DSLK+, RHAG4−; FORS+; Jra+; Lan+; Vel+; At(a+) Whole Genome Sequencing Typing

A1; M+N+, U+, Vr−, Mt(a−), Ri(a−), Ny(a−), Or−, ERIK−, Os(a−), ENEP+, ENEH+, ENAV+, ENEV+, MNTD−; S+s+, He−, M(v−), s(D−), Mit−; P1+/P1−, pk+, NOR−; D+, Tar−; C−c+E+e−, CW−, CX−, EW−, V−, VS−, Rh26+LOCR− Be −, Crawford−CELO+, JAL−CEST+, STEM−, JAHK−; Lu(a−b+), LURC+, Lu4+, Lu5+, Lu6+, Lu7+, Lu8+, Lu13+, Lu16+, Lu17+, Au(a+b−), Lu20+, Lu21+; K−k+, Kp(a−b+c−); Js(a−b+), Ul(a−), K11+, K12+, K13+, K14+, K18+, K19+, K22+, K23−, VLAN−VONG−, TOU+, RAZ+, KALT+, KTIM+, KYO−, KUCI+, KANT+, KASH+, KELP+, KETI+, image Courtesy of Illumina KHUL+; Le(a+b−); Fy(a+b+); Jk(a+b+); Di(a−b+), Wr(a−b+), Wd(a−), Rb(a−), WARR−, ELO−, Bp(a−), Mo(a−), Hg(a−), Vg(a−), Sw(a−), BOW−, NFLD−, Jn(a−), KREP−, Tr(a−), Fr(a−), SW1−, Wu−DISK+; Yt(a+b−); Sc1+Sc2−, Rd−, STAR+, SCER+, SCAN+; Do(a+b+), Jo(a+), DOYA+, Hy+, DOMR+, DOLG+; Co(a+b−), Co4+; LW(a+b−); Ch1+, Ch2+, Ch3+, Ch4+, Ch5+, Ch6+, Rg1−, Rg2−; Ch1−, Ch2−, Ch3−, Ch4−, Ch5−, Ch6−, Rg1+, Rg2+; H+; Kx+; Es(a+), Wb−, An(a−), Dh(a−), GEIS−, GELP+, GEAT+, GETI+; Cr(a+), Tc(a+b−c−), Dr(a+), Es(a+), WES(a−b+), UMC+, GUTI+, SERF+, ZENA+, CROV+, CRAM+, CROZ+; Kn(a+b−), McC(a+b−), Sla+Vil−, Yk(a+), Sl3+, KCAM+/KCAM−; In(a−b+), INFI+, INJA+; Ok(a+), OKGV+, OKVM+; MER2+; JMHK+, JMHL+, JMHG+, JMHM+, JMHQ+; I+; P+; GIL+; Duclos+, Ol(a−), DSLK+, RHAG4−; FORS+; Jra+; Lan+; Vel+; At(a+)

99.9% Accurate

Lane WJ, Westhoof CM, et al. Comprehensive Red Blood Cell and Platelet Antigen Prediction From Whole Genome Sequencing: Proof of Principle. Transfusion. 2016 Mar;56(3):743-54 Lane WJ, Westhoff CM, Gleadall NS, et al. Automated Typing of Red Blood Cell and Platelet Antigens from Whole Genome Sequencing. Lancet Haematology. 2018. Jun;5(6):e241-e251 Balancing Act

Whole Genome $$$ month Cost Whole Exome $$ weeks Speed Antigen Genes <$ week Accuracy Specific Antigen <$ days Molecular Inversion Probes (MIPs)

• Sequence multiple regions over many genes • Reliable • Scalable (> 2000 samples per week) • Minimal reagent and labor cost ($10/sample) • Since primer pairs are linked it is easier to optimize than traditional amplicon PCR enrichment • Sequence 240-285 bp regions with overlap • MIP design software MIPMaker (Aydemir, unpublished)

Aydemir JID 2018 MIP Workflow MIP Probes Target Specific Regions

Regions Amplified with sample barcodes and seq adapters MIP Blood Group Typing

260 MIPs 19 Genes 90 Antigens $10 /sample

image Courtesy of Illumina Typing Summary Typing Summary Exportable File Format

1 ABO 1 A - 2 MNS 5W S-,s-,U(w)+ 6 KEL 24 K24 - 14 DO 7 DOMR + 1 ABO 2 B - 2 MNS 6 S-,s-,U-,He - 6 KEL 25 VLAN - 14 DO 8 DOLG + 1 ABO 3 AB - 2 MNS 21 M(v) - 6 KEL 27 RAZ + 15 CO 1 Co(a) + 1 ABO 4 A1 - 2 MNS 23 s(D) - 6 KEL 28 VONG - 15 CO 2 Co(b) - 2 MNS 1 M + 2 MNS 24 Mit - 6 KEL 29 KALT + 15 CO 4 Co4 + 2 MNS 2 N - 4 RH 1 D + 6 KEL 31 KYO - 16 LW 5 LW(a) + 2 MNS 7 Mi(a) - 4 RH 2 C - 6 KEL 34 KASH + 16 LW 7 LW(b) - 2 MNS 9 - 4 RH 3 E - 6 KEL 37 KHUL + 21 CROM 1 Cr(a) + 2 MNS 12 Vr - 4 RH 4 c + 8 FY 1 Fy(a) - 21 CROM 11 GUTI + 2 MNS 14 Mt(a) - 4 RH 5 e + 8 FY 2 Fy(b) + 21 CROM 13 ZENA + 2 MNS 16 Ri(a) - 4 RH 8 C(W) - 9 JK 1 Jk(a) + 22 KN 1 Kn(a) + 2 MNS 18 Ny(a) - 4 RH 9 C(X) - 9 JK 2 Jk(b) + 22 KN 2 Kn(b) - 2 MNS 19 - 4 RH 10 V + 10 DI 1 Di(a) - 22 KN 3 McC(a) + 2 MNS 31 Or - 4 RH 11 G - 10 DI 2 Di(b) + 22 KN 4 Sl(a) + 2 MNS 37 ERIK - 4 RH 20 VS + 11 YT 1 Yt(a) + 22 KN 5 Yk(a) +/- 2 MNS 38 Os(a) - 4 RH 36 Be(a) - 11 YT 2 Yt(b) - 22 KN 6 McC(b) - 2 MNS 39 ENEP + 4 RH 53 JAHK - 13 SC 1 Sc1 + 22 KN 7 Vil - 2 MNS 40 ENEH + 4 RH 57 JAL-CEST + 13 SC 2 Sc2 - 22 KN 8 Sl3 + 2 MNS 41 HAG - 6 KEL 1 K - 13 SC 4 Rd - 22 KN 9 KCAM + 2 MNS 42 ENAV + 6 KEL 2 k + 13 SC 5 STAR + 23 IN 1 In(a) - 2 MNS 43 MARS - 6 KEL 3 Kp(a) - 13 SC 6 SCER + 23 IN 2 In(b) + 2 MNS 45 ENEV + 6 KEL 4 Kp(b) + 13 SC 7 SCAN + 24 OK 1 Ok(a) + 2 MNS 47 SARA - 6 KEL 6 Js(a) - 14 DO 1 Do(a) + 2 MNS 3 S -/- 6 KEL 7 Js(b) + 14 DO 2 Do(b) + 2 MNS 4 s +/- 6 KEL 14 K14 + 14 DO 4 Hy + 2 MNS 5 U + 6 KEL 21 Kp(c) - 14 DO 5 Jo(a) + U+w Result FYnull Result Weak Partial D Result MIP Typing Accuracy (Pilot Design and Dataset) 11 Blood Groups 95 Samples 36 Antigens Accuracy 3,344 / 3,380 = 98.9% MIP Copy Number Analysis Summary

• Developed an inexpensive MIP blood group genotyping capture assay • Adapted bloodTyper to type from MIP based NGS sequencing results with 99% accuracy • Probes will be optimized and additional probes designed • MIP based copy number analysis to be added Acknowledgments

Department of Pathology

Jeff Golden Ozkan Aydemir Les Silberstein Patrick Marsh Rich Kaufman Nicholas Hathaway Helen Mah Yong Zhao Maria Aguad Jeffrey A. Bailey Jon Michael Uy Robin Smeland-Wagman

Connie Westhoff Sunitha Vege Changxin Shen Questions?

William Lane, Ozkan Aydemir, Changxin Shen, Sunitha Vege, Nicholas Hathaway, Yong Zhao, Connie Westhoff, Jeffrey Bailey

Department of Pathology Prediction of Red Blood Cell Phenotype from Exome Next Generation Sequencing Data

Celina Montemayor, Bhaveshkumar Delvadia, Oscar A. Montemayor, Nasha Elavia, Spencer E. Grissom, Katie L. Lewis, Debrean Loy, Rizaldy Cacanindin, Steven McLaughlin, Marina Bueno, Sharon Adams, John D. Roback, Harold Smith, Leslie Biesecker, and Harvey G. Klein. The views expressed here do not necessarily represent the view of the National Institutes of Health, the Department of Health and Human Services, or the U.S. Federal Government. Stabentheiner et al. Vox Sanguinis 2011; 100: 381-8

Rieneck et al. Transfusion 2013; 53: 2892-8

Lane et al. Transfusion 2016; 56: 743-54

Fichou et al. Vox Sanguinis 2016; 111: 418-24

Möller et al. Blood Advances 2016; 1: 240-9

Schoeman et al. Transfusion 2017; 57: 1078-88

Chou et al. Blood Advances 2017; 1: 1414-22

Jakobsen et al. Transfusion Medicine 2017; epub

Schoeman et al. Transfusion 2018; 58: 284-493

Lane et al. Lancet Hematology 2018; 5: e241–51

Orzinska et al. Blood Transfusion 2018; 16: 285-92

Wu et al. Transfusion 2018; 58:2232-2242

Wheeler et al. Curr Opin Hematol 2018; 25:609-515

Wheeler et al. Genet Med 2018; epub RBC Antigen Genotyping

Patients that have been recently transfused

Patients with antibodies to high or low-frequency antigens for which serology is limited, particularly if historical and not demonstrating

Patients receiving certain monoclonal therapies (daratumumab, anti-CD47) RBC Antigen Genotyping

Patients that have been recently transfused

Patients with antibodies to high or low-frequency antigens for which serology is limited, particularly if historical and not demonstrating

Patients receiving certain monoclonal therapies (daratumumab, anti-CD47)

NGS High-throughput RBC Antigen Genotyping

Patients that have been recently transfused

Patients with antibodies to high or low-frequency antigens for which serology is limited, particularly if historical and not demonstrating

Patients receiving certain monoclonal therapies (daratumumab, anti-CD47)

NGS High-throughput Detection of more/all known variants

Detection of novel variants NIH Clinical Center NGS

RBC phenotype NGS

RyLAN RBC phenotype

Red cell and Leukocyte Antigen prediction from NGS NGS

RyLAN RBC phenotype

Exome Whole genome Targeted NGS NGS NGS RyLAN architecture

RyLAN *RyLAN_Q variant Quality database filters (MongoDB) (MongoDB)

Genotype & NGS predicted .bam file RyLAN RBC (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis RyLAN architecture Open Source Version control

RyLAN *RyLAN_Q variant Quality database filters (MongoDB) (MongoDB)

Genotype & NGS predicted .bam file RyLAN RBC (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis RyLAN architecture Open Source Scalability Flexibility RyLAN *RyLAN_Q variant Quality database filters (MongoDB) (MongoDB)

Genotype & NGS predicted .bam file RyLAN RBC (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis

JSON format

Montemayor et al, in press

RyLAN architecture QUAL DP AO/DP Mapping quality RyLAN *RyLAN_Q QR variant Quality QA database filters (MongoDB) (MongoDB)

Genotype & NGS predicted .bam file RyLAN RBC (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis RyLAN architecture Open Source Scalability Flexibility RyLAN *RyLAN_Q variant Quality database filters (MongoDB) (MongoDB)

Genotype & NGS predicted .bam file RyLAN RBC (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis

RyLAN architecture Open Source Scalability Flexibility RyLAN *RyLAN_Q Security variant Quality database filters (MongoDB) (MongoDB)

Genotype & NGS predicted .bam file RyLAN RBC (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis RyLAN output

RyLAN *RyLAN_Q allele Quality database filters (MongoDB) (MongoDB)

Genotype & NGS Custom RyLAN predicted .bam file .vcf converter RBC (Python) phenotype RyLAN output

RyLAN *RyLAN_Q allele Quality database filters Support Complex Queries(MongoDB) - Individual(MongoDB) and Cohort Level

Genotype & NGS Custom RyLAN predicted .bam file .vcf converter RBC (Python) phenotype RyLAN output

RyLAN *RyLAN_Q allele Quality database filters (MongoDB) (MongoDB)

Genotype & NGS Custom RyLAN predicted .bam file .vcf converter RBC (Python) phenotype RyLAN output

> Give me a quick > Give me a quick view view of your of your predictions and RyLAN *RyLAN_Q predictions and filtered calls for allele Quality filtered calls for participant X database filters participant X (MongoDB) (MongoDB)

Genotype & NGS Custom RyLAN predicted .bam file .vcf converter RBC (Python) phenotype RyLAN output

> Give me a quick > How many weak Kidd alleles are found view of your RyLAN *RyLAN_Q in Cohort X, in how predictions and allele Quality many participants, and filtered calls for database filters how many are in participant X (MongoDB) (MongoDB) homozygous state?

Genotype & NGS Custom RyLAN predicted .bam file .vcf converter RBC (Python) phenotype RyLAN output

Genotype & NGS Custom RyLAN predicted .bam file .vcf converter RBC (Python) phenotype RyLAN architecture Open Source Scalability Flexibility Security Reproducibility RyLAN architecture Open Source Scalability Flexibility Security Reproducibility RyLAN architecture Open Source Scalability Flexibility Security Reproducibility RyLAN architecture

Docker / Singularity Container RyLAN architecture

RyLAN coder

Docker / Singularity Container

RyLAN architecture Open Source Scalability Flexibility Security Reproducibility Parallelizable RyLAN architecture Open Source Scalability Flexibility Security Reproducibility Parallelizable RyLAN architecture Open Source Scalability Flexibility Security Reproducibility Parallelizable NHGRI 07-HG002

1018 participants, exome sequencing performed 2008-2014 NHGRI 07-HG002 RyLAN architecture

RyLAN *RyLAN_Q variant Quality database filters (MongoDB) (MongoDB)

Genotype & NGS predicted .bam file RyLAN RBC (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis NHGRI 07-HG002 RyLAN architecture

RyLAN *RyLAN_Q variant Quality database filters (MongoDB) (MongoDB)

Genotype & Serologic NGS predicted and genetic .bam file RyLAN RBC validation (Python) phenotype

Freebayes Copy gvcf number (basepair resolution) analysis Queried genes RBCs

A4GALT ACHE GYPC BCAM ERMAP RHAG CD55 KEL ART4 GBGT1 CR1 FUT3 AQP1 ABCG2 CD44 FUT7 ICAM4 ABCB6 CD151 ACKR1 FUT1 SLC29A1 GCNT2 SLC14A1 FUT2 B3GALNT1 KLF1 SLC4A1 XK

325 SNVs and 43 indels interpreted by RyLAN

Cohort Metrics

Average depth at the genomic positions of interest: 86.3 (0-747)

Average QUAL value for a variant call: 1,164 (0.02-6808)

Scale chr18: Your Sequence from Blat Search UCSC Genes (RefSeq, GenBank, CCDS, Rfam, tRNAs & Comparative Genomics) SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 1000 Genomes Phase 3 Integrated Variant Calls: SNVs, Indels, SVs

JK*01W.01 c.130G>A p.Glu44Lys

Image from UCSC Browser

1000 Genomes Project Phase 3 Paired-end Accessible Regions - Pilot Criteria 1000 Genomes Project Phase 3 Paired-end Accessible Regions - Strict Criteria

RefSeq Genes Non-Human RefSeq Genes

Publications: Sequences in Scientific Articles

Gene Expression in 53 tissues from GTEx RNA-seq of 8555 samples (570 donors)

H3K27Ac Mark (Often Found Near Active Regulatory Elements) on 7 cell lines from ENCODE

DNaseI Hypersensitivity Clusters in 125 cell types from ENCODE (V3) Transcription Factor ChIP-seq (161 factors) from ENCODE with Factorbook Motifs 100 vertebrates Basewise Conservation by PhyloP

Multiz Alignments of 100 Vertebrates

Simple Nucleotide Polymorphisms (dbSNP 150) Found in >= 1% of Samples

Repeating Elements by RepeatMasker Simple Nucleotide Polymorphisms (dbSNP 147)

SNPedia all SNPs (including empty pages)

Simple Nucleotide Polymorphisms (dbSNP 150)

Human mRNAs from GenBank

Basic Gene Annotation Set from GENCODE Version 19 Scale chr18: Your Sequence from Blat Search UCSC Genes (RefSeq, GenBank, CCDS, Rfam, tRNAs & Comparative Genomics) SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 SLC14A1 1000 Genomes Phase 3 Integrated Variant Calls: SNVs, Indels, SVs

JK*01W.01 Jka/Jkb c.130G>A p.Glu44Lys

Image from UCSC Browser

1000 Genomes Project Phase 3 Paired-end Accessible Regions - Pilot Criteria 1000 Genomes Project Phase 3 Paired-end Accessible Regions - Strict Criteria

RefSeq Genes Non-Human RefSeq Genes

Publications: Sequences in Scientific Articles

Gene Expression in 53 tissues from GTEx RNA-seq of 8555 samples (570 donors)

H3K27Ac Mark (Often Found Near Active Regulatory Elements) on 7 cell lines from ENCODE

DNaseI Hypersensitivity Clusters in 125 cell types from ENCODE (V3) Transcription Factor ChIP-seq (161 factors) from ENCODE with Factorbook Motifs 100 vertebrates Basewise Conservation by PhyloP

Multiz Alignments of 100 Vertebrates

Simple Nucleotide Polymorphisms (dbSNP 150) Found in >= 1% of Samples

Repeating Elements by RepeatMasker Simple Nucleotide Polymorphisms (dbSNP 147)

SNPedia all SNPs (including empty pages)

Simple Nucleotide Polymorphisms (dbSNP 150)

Human mRNAs from GenBank

Basic Gene Annotation Set from GENCODE Version 19

Sample individual output LU*B/LU*B 97% DO*B/DO*B 40% KEL*02/KEL*02 (k/k) 91% DO*A/DO*B KEL*02/KEL*01(K/k) 48% 8% DO*A/DO*A LU*B/LU*A 12% KEL*01/KEL*01(K/K) 3% 1% LU*A/LU*A 0.1%

SC*01/SC*01 CO*A/CO*A YT*A/ YT*A 99.6% 94% 94%

CO*A/CO*B YT*A/ YT*B 6% 6% SC*01/SC*02 0.4% YT*B/ YT*B 0.4% Serologic Validation Serologic Validation

106 samples collected Serologic Validation

106 samples collected anti-K1 anti-Fya CLIA-certified testing anti-Fyb anti-JKa anti-JKb Serologic Validation

106 samples collected anti-K1 anti-Fya CLIA-certified testing anti-Fyb anti-JKa anti-JKb

100% concordance in 103 samples (515 reactions) 3 discrepancies Discrepancies

Predicted by RyLAN: Serology results: 1 K1-; Fy(a-b+); Jk(a-b+) K1-; Fy(a+b-); Jk(a-b+) Discrepancies

Predicted by RyLAN: Serology results: 1 K1-; Fy(a-b+); Jk(a-b+) K1-; Fy(a+b-); Jk(a-b+)

confirmed serologically as Fy(a-b+) Discrepancies

Predicted by RyLAN: Serology results: 1 K1-; Fy(a-b+); Jk(a-b+) K1-; Fy(a+b-); Jk(a-b+)

confirmed serologically as Fy(a-b+)

Predicted by RyLAN: Serology results: 2 K1-; Fy(a+b+); Jk(a+wb+) K1-; Fy(a+b+); Jk(a-b+) Discrepancies

Predicted by RyLAN: Serology results: 1 K1-; Fy(a-b+); Jk(a-b+) K1-; Fy(a+b-); Jk(a-b+)

confirmed serologically as Fy(a-b+)

Predicted by RyLAN: Serology results: 2 K1-; Fy(a+b+); Jk(a+wb+) K1-; Fy(a+b+); Jk(a-b+)

Heterozygous SLC14A1 c.130G>A Discrepancies

Predicted by RyLAN: Serology results: 1 K1-; Fy(a-b+); Jk(a-b+) K1-; Fy(a+b-); Jk(a-b+)

confirmed serologically as Fy(a-b+)

Predicted by RyLAN: Serology results: 2 K1-; Fy(a+b+); Jk(a+wb+) K1-; Fy(a+b+); Jk(a-b+)

Heterozygous SLC14A1 c.130G>A

anti-Jka

anti-Jkb Discrepancies

Predicted by RyLAN: Serology results: 1 K1-; Fy(a-b+); Jk(a-b+) K1-; Fy(a+b-); Jk(a-b+)

confirmed serologically as Fy(a-b+)

Predicted by RyLAN: Serology results: 2 K1-; Fy(a+b+); Jk(a+wb+) K1-; Fy(a+b+); Jk(a-b+)

Heterozygous SLC14A1 c.130G>A

Predicted by RyLAN: Serology results: 3 K1+; Fy(a-b+); Jk(a+wb+) K1-; Fy(a-b+); Jk(a-b+)

Heterozygous SLC14A1 c.130G>A Serologic Validation Additional Variants Additional Variants

G. Daniels. Human Blood Groups

rs199665533 Arg77His

delArg35, Leu35 Arg75Cys

G. Daniels. Human Blood Groups

delArg35, Leu35 Arg75Cys

G. Daniels. Human Blood Groups

delArg35, Leu35 Arg75Cys

Serology: Lua = 0 Lub = w+ G. Daniels. Human Blood Groups

Lu12+, Lu24+, Lu22+, Lu(a-b+),Lu21+ Lu5+, Lu17+,Lu4+,Lu8+, Lu14-,Lu25+,Lu16+,Lu6+, Lu9-,Lu20+,Lu7+,Lu26+, Au(a+b-) Lu12+, Lu24+, Lu22+, Lu(a-b+),Lu21+ Lu5+, Lu17+,Lu4+,Lu8+, Lu14-,Lu25+,Lu16+,Lu6+, Lu9-,Lu20+,Lu7+,Lu26+, Au(a-b+) Acknowledgements NIH CC DTM Emory University NISC John Roback, MD Gerry Bouffard Sharon Adams MT, CHS Bhavesh Delvadia, SBB Meghana Vemulapalli Marina Bueno, SBB Rizaldi Cacanindin, SBB Nasha Elavia, MD NIDDK Fleming Institute Harvey Klein, MD Harold Smith, PhD Panagiota Karagianni, PhD Kimberly Levy, SBB Debrean Loy, SBB Steven McLaughlin, SBB NHGRI Silicon Valley Jeffery Miller, MD Leslie Biesecker, MD PhD Oscar Montemayor, MSEE Magdalene Nwokocha, MD Katie Lewis, RN Rick Lima, MSEE Erika Reese, SBB Ilana Miller Danielle Smellie, MD Maxim Tynuv, SBB NIH HPC Staff Kamille West, MD aaBB Boston 13 Oct 2018

Genetic elucidation of XG, the last unresolved blood group system

Presenter: Jill Storry, Ph.D.

Technical Director, Blood Group Immunology Associate Professor of Experimental Transfusion Medicine

Abstract co-authors: Mattias Möller, Yan Quan Lee, Karina Vidovic, Marion Darlison, Linda Björkman, Sven Kjellström, Jill R. Storry, Martin L. Olsson

Division of Hematology Department of Clinical Immunology and Transfusion Medicine, and Transfusion Medicine, Department of Laboratory Medicine, Office for Medical Services, Lund University, Sweden Region Skåne, Sweden Already 10 years ago, only two blood group systems remained to define genetically

Veldhuisen B et al. Blood group genotyping. Vox Sang. 2009; 97:198-206. Molecular mechanisms underlying

the P1/P2 blood group phenotypes

Westman et al. Blood, April, 2018 (RUNX1) Yeh et al. Transfusion, April, 2018 (EGR1)

 Genetic basis of Xga blood group expression

Anti-Xga is not commonly encountered, so why is this important: Last system not resolved → scientific challenge Anti-Xga reagents are scarce/bad → phenotyping problematic Genotyping not possible → time has come

Also, the role of Xg protein on RBCs still unknown It all started with a lecture* at the aaBB…

XK The GATA1 XG X XK

GATA1 factor! XG

*Storry JR. Invited lecture at the Annual Meeting of the aaBB, Boston, 2012 Xga was the first blood group assigned to a chromosome

Discovered by Mann et al. (Lancet, 1962)

Skewed frequencies between genders ~30% of men are Xg(a‒) ~10% of women are Xg(a‒)

Xg protein is lacking on RBCs of those who are Xg(a‒) Phenotypic relationship between Xga and CD99

CD99 is the 2nd antigen in the XG system ~100% of all people are CD99 positive

Reid ME, Lomas-Francis C, Olsson ML. The Blood Group Antigen FactsBook, 3rd ed. 2012 Genetic findings

The PBDX gene was identified to encode Xg glycoprotein (Ellis et al. Nat Genet 1994)

However, no explanation for presence/absence of Xga

CD99 is encoded by the MIC2/CD99 gene

Rare CD99-negative individuals have different deletions in the coding regions of MIC2 (Thornton et al. Vox Sang. 2015)

A hypothetical regulatory site, XGR, was proposed already in 1987 (Goodfellow et al. Ann Hum Genet. 1987) Our hypothesis:

Xga expression is transcriptionally regulated by a single SNP within the XG region, potentially disrupting an erythroid transcription factor binding site The XG gene and its product

SP Extracellular TM Cyto 1 22 143 164 180

16 possible O-glycosylation sites What is the function of these proteins?

Xg and CD99 proteins are 48% homologous Similar to glycophorins Large N-terminal portions heavily O-glycosylated A single transmembrane domain

Xg is relatively RBC-specific whilst CD99 is expressed in many different cell types

CD99 shown to be an adhesion molecule Roles in immunology, cancer etc

The function of the 149-aa Xg glycoprotein is unknown Based on homology – may have similar role as CD99 but on RBCs Three-pronged bioinformatics strategy

Compare historical frequencies with SNP frequencies in XG region from 1000G (Nature, 2015) and Erythrogene (Möller et al. Blood Adv. 2016)

SNP

Expression quantitative trait Transcription factor binding loci (eQTL) from GTEx portal prediction with JASPAR (Nature, 2017) (Nucl Acids Res. 2018)

Blood samples from 158 blood donors anonymized other than for gender: Xga phenotyping, FACS, qPCR, EMSA, luciferase etc A SNP upstream of XG correlates with the expected phenotype distribution

Among 2,612 investigated genetic variants in the XG region, one specific SNP (rs311103), ~4 kb upstream of the transcription start site, was identified to have the strongest correlation to the expected distribution. This SNP had the best statistical fit to data in three superpopulations

• Lowest χ2: rs311103 Independently, the same SNP was found to influence XG mRNA expression the most in blood The implicated SNP abolishes a potential GATA motif XG ACKR1

GATA1 binding (murine) Causes Fy(a‒b‒) phenotype in Africans SNP genotyping by allelic discrimination

G/G G/C

C/C ?!

H2O

C/G?

G/C? This GATA disruption abolishes XG transcripts Wildtype-homozygous women are Xga strong

Low/No transcripts Dosage effect?

= Xg(a+) = Xg(a‒) EMSA shifts and supershifts indicate that GATA1 binds to wild-type but not mutant motif GATA1 peptides identified by MS/MS (24% coverage) A luciferase reporter assay shows this GATA binding site to exert clear enhancer effects

Luc+ Control Rluc

HEL rs311103 genotype also correlated well with CD99 expression level by FACS

Marion Darlison. M.Sc. Thesis, Lund University 2018 Conclusions

 We have explained why a third of all men and 10% of all women lack the Xg protein on their RBCs.

 Genotyping for rs311103 predicts Xga status and correlates with CD99 expression levels.

 Challenges in “G/C” males (X/Y) need to be addressed.

 The unknown function of Xg protein and its effects in health and disease remains to be clarified. Blood – 19 July 2018 Celebrations in the lab! The same Taiwanese group also came to similar conclusions on the Xg system

Blood Advances – 14 Aug 2018 Thanks to the Blood Group @ LU Our research aims to uncover new roles of the red blood cell surface in health and disease, with a special focus on the polymorphic Dept. of Hematology & Transfusion Mmedicine molecules known as blood groups. Clinical Immunology & Transfusion Medicine Department of Laboratory Medicine LabMedicine, Office for Medical Services

Linda Björkman Annika Hult Karina Vidovic Martin L Olsson Magnus Jöud Anja Nylander Philaiphon (Pat) M.D. Ph.D., staff scientist Ph.D.,staff scientist, Professor, M.D., Ph.D. M.D., Ph.D. student M.D., Ph.D. student Jongruamklang Lab manager Principal investigator M.Sc., Ph.D. student

Yan Quan Lee Julia Westman Åsa Hellberg Abdul Ghani Alattar Marion Darlison Jennifer Ricci Hagman Linn Stenfelt Mattias Möller Ph.D., Post-doc Ph.D., Post-doc Ph.D., Reference M.D., Ph.D. student M.Sc. student M,Sc., Ph.D. student M.Sc., Ph.D. student M.D., Ph.D. student (currently in Santa laboratory coordinator Bioinformatician Barbara, CA, USA)

Our work is funded by:

ALF

GATA1 identified by MS/MS (as no. 3 on top 10 list binding to wt, not mutant)

Möller M, Lee YQ, Vidovic K, Björkman L, Kjellström S, Storry JR, Olsson ML. Disruption of a GATA-1 binding motif upstream of the XG/PBDX gene abolishes erythroid Xga expression and elucidates the last unresolved blood group system (Blood, 2018). GATA1 forms a complex with LDB1 when binding to the Xga-defining SNP

Biotinylated probe Wild-type GATA Nuclear extract - + + + +

Antibody - - GATA1 LDB1 TRPS1 LDB1 complexes with GATA1

Love PE, Warzecha C, Li L. Ldb1 complexes: the new master regulators of erythroid gene transcription. Trends in Genetics. 2014;30(1):1–9.