US 2016O177276A1 (19) United States (12) Patent Application Publication (10) Pub. No.: US 2016/0177276 A1 L0 et al. (43) Pub. Date: Jun. 23, 2016

(54) SIRP-ALPHA IMMUNOGLOBULINFUSION C07K 6/40 (2006.01) PROTEINS C07K 16/32 (2006.01) C07K 6/28 (2006.01) (71) Applicant: MERCK PATENT GMBH, Darmstadt C07K 6/30 (2006.01) (DE) (52) U.S. Cl. CPC ...... CI2N 9/16 (2013.01); C07K 16/2887 (72) Inventors: Kin-Ming Lo, Lexington, MA (US); (2013.01); C07K 16/30 (2013.01); C07K Nora Zizlsperger, Newton, MA (US); I6/2863 (2013.01); C07K 16/40 (2013.01); Aroop Sircar, Billerica, MA (US) C07K 16/32 (2013.01); C07K 16/18 (2013.01); 21) Appl. No.: 14/827,003 C07K 16/2827 (2013.01);s C07K 16/2803 (21) Appl. No 9 (2013.01); C12Y-301/03048 (2013.01); C07K (22) Filed: Aug. 14, 2015 2317/622 (2013.01); C07K 2319/74 (2013.01); A6 IK 2039/505 (2013.01) Related U.S. Application Data (60) Provisional application No. 62/038,196, filed on Aug. (57) ABSTRACT 15, 2014. Publication Classification The invention discloses immunoglobulin fusion proteins designed to bind both CD47 and a tumor cell antigen. The (51) Int. Cl. immunoglobulin fusion proteins include a SIRPC. moiety that CI2N 9/16 (2006.01) binds CD47 and an antigen binding site for a tumor cell C07K 6/8 (2006.01) antigen. Patent Application Publication Jun. 23, 2016 Sheet 1 of 36 US 2016/0177276 A1

2 EVE CL

Patent Application Publication Jun. 23, 2016 Sheet 2 of 36 US 2016/0177276 A1

Patent Application Publication Jun. 23, 2016 Sheet 3 of 36 US 2016/0177276 A1

FG. K Patent Application Publication Jun. 23, 2016 Sheet 4 of 36 US 2016/0177276 A1

FG.

Patent Application Publication Jun. 23, 2016 Sheet 5 of 36 US 2016/0177276 A1

2 3 4. 5 8 7 8 DAY'S OS.REAVENT

FG. 2A

4 2 7 8 DAYS POS-TREAVEN F.G. 2B Patent Application Publication Jun. 23, 2016 Sheet 6 of 36 US 2016/0177276 A1

KDa N

183 98 62 49 38 FG. 3A 28

4.

8 3 900 80 70) 60 500 40 30 ... 1200 . O 100 90 80 70 60 50 40

30 20 5.465 5.966 O ( ; 2 3 4 5 & 7 8 & 10 1 2 3 4 5 TIME (min) G. 33 Patent Application Publication Jun. 23, 2016 Sheet 7 of 36 US 2016/0177276 A1

20 --Or Anti-CD20-hugG1-huSRPaV2 000 rty's F-SiRPW2 a: sis SRPW2-r f o:mo" : obo Anti-C20 pe 80 -- Anti-CD47 / oak 6 O

8 ...... RED BLOOD CELLS (CD235+) to E. LEUKOCYTES (CD45*)

5

t h Atop top. Antonio SRPCV2 SRPCV2. FC SRPCV2 FG. 4B Patent Application Publication Jun. 23, 2016 Sheet 8 of 36 US 2016/0177276 A1

400 so -o- Anti-CD20-hugG1-huSIRPaV2 : • 'A' FC-huSRPV2 A. g 300 oar- huSIRPav2-Fo A. i 250 -a-. Anti-CD47 f S 200 ti- s a 150 f 100 5 :e : 0. . G. 4C

90 8) --Or Anti-CD20-huigG-huSRPaV2 : sexy. FC-SRPoV2 1.

70 -- Anti-CD20 1 i?S 60a - - Anti-CD47 s 50 is 40 2. is 30 e s' & 88 w st

, 000 Patent Application Publication Jun. 23, 2016 Sheet 9 of 36 US 2016/0177276 A1

EDAs SURVIVA. DAYS wy w CONTRO. At mem. Anti-CD20 3S or Anti-CD47 143 {{ YYYYYYYYYYYYYYYY PPPPPPPPPPPPPPPPAwww.wa. A v. A was w8A wa &w A w w 8A wir aw Ove Anti-C2ianti-CD4F NO

--- e k

?s 75 - 8 5 50 - y r - 25 1 a

v g 20 40 80 8 OO 20 4 6 8 AYS POS iOR NOCAON F.G. 5A

RA 8 mmm's iANSRWVA DAYS a - p SOTYPE CONTRO 3 80 on RXAN 8 w -" cooroo Anti-C47 37 6 (... as Os Anti-CD2O:Anti-C47 55 Patent Application Publication Jun. 23, 2016 Sheet 10 of 36 US 2016/0177276 A1

Ka

88

F.G. 6A Patent Application Publication Jun. 23, 2016 Sheet 11 of 36 US 2016/0177276 A1

anti-EGFR-huigG1-SRPow

5.561 6,025 (. 1 2 3 4 5 & 7 & 8 to 1 2 3 4 5 TIME (min)

380 6.8 i. 81 anti-EGFR-hulgG1-(G4S): SIRPoV2 2400 2200 2}} 1800 160040 20 000 8 6 4- 5,473 5.987 ; ; ; ; ; ; ; ; , , , , , , TIME (min) F.G. 6B Patent Application Publication Jun. 23, 2016 Sheet 12 of 36 US 2016/0177276 A1

(ii) anti-EGFR-hulgG1-(G4S4-SiRPoV2

:0 5.451 5985

2 3 4. " " " " ...... TIME (min)

2800 8.75- anti-EGFR-hulgG1-(G4S)5-SIRPoV2

is TIVE (min) FIG. 6B (CONT) Patent Application Publication Jun. 23, 2016 Sheet 13 of 36 US 2016/0177276 A1

1000 -k- Anti-EGFR-hugg1-h;SRPaw -- Anti-EGFR-hugg1-G4S)3-huSiRPaV2 & Anti-EGFR-hugG-G4S4-huSRPaV2 800 --- Anti-EGFR-hugg1-G4S5-huSiRPaV2 ...y. FC-SRPOW2 Mir SRPCW2-Fi 600 - - Anti-CED47 ... Atti-EGFR

a a.s. 40 es

2 -: w

200 ->k

x

O.01 . 0.1 O OO 300 ni FG, 7A

90 ...... 80 ------RE BOOD CELS (CD235+) 70. E. EUKOCYTES (CD45+) 60 ...... g 50 - is 40-rrier to r as 30 - ...... : ...... :::::::::...... 3...... : ...... 20.... : ...... : ... :...... :. . .

O ...... : ...... ::::::::...... : ...... : ...... O+ hu Atop to of SRPW2 SRPx2-C hSRPOW G. 73 Patent Application Publication Jun. 23, 2016 Sheet 14 of 36 US 2016/0177276 A1

--C - Anti-EGFR-huigG-huSRPaV2 120 vyy FC-SRPCV2 wiro Sirf2-FC

g O m:m Angpt mm. Anti-EGFR i. 3 80

it 60 s

e 40

C O,000 O, OO 0000 Patent Application Publication Jun. 23, 2016 Sheet 15 of 36 US 2016/0177276 A1

2000000 - -o- Anti-EGFR-huigG-(G4S4-huSRPaV2 orro hSir POV2-r 1800000 ...A.. Fc-husiRPv2 ---O- - 800000 - obo Anti-EGFR o r 4000 200000 000000 800000 600000 400000 200000

O R 0.000 0.00 3.0 O. 10 00 Patent Application Publication Jun. 23, 2016 Sheet 16 of 36 US 2016/0177276 A1

00). -- Fe-SiRPo (20 ag) r^ - FC-SRPo (200 g) we anti-EGFR-Fe-SRPo (20 kg) ...O.. anti-EGFR-FC-SRPoi. (200 g)

0.

t 50 100 50 200 250 TIME (h) G. 9A

EDAN SRWWA DAYS O ope - SOTYPE CONTRO 28 oboe C225 3. 80 wim SRPa-FC 3. -- C225+ SRPa-Fo 4. 60 w C a C225iSR-a-F 40

40

20

to 2 3 4 so 70 DAYS OS.REAVEN START G. 9B Patent Application Publication Jun. 23, 2016 Sheet 17 of 36 US 2016/0177276 A1

FG. OA

2 3 4 5 6 7 & 8 to 12 13 14 15 TIME (min) FG. OB Patent Application Publication Jun. 23, 2016 Sheet 18 of 36 US 2016/0177276 A1

2000. --O- Anti-HER2-huigG4-h;SRPav2 180 - sexx as F-SRPo V2 .. othro h;SRPCV2-F2 e:

600 -- Anti-CD47 oboe Anti-ER2 4. O -:

4.

80 - / 2 800 - 30 200 -

O ::::::: 3. . O, . COO f FG. A

80...... RED BLOOD CELLS (CD235t) 7 - ...... s' EUKOCYTES (CD45+)

8 O

F. Anti-CD47 "Antier. Anti-ER SRC.V2 SRPCV2-C SR-V2 G. 13 Patent Application Publication Jun. 23, 2016 Sheet 19 of 36 US 2016/0177276 A1

400 - O - Anti-HER2-hulgG-huSRPaV2 ...yo. FC-hSRPOW2 20 on Anti-CD47 --- obo Anti-ER2 exo-boo"

800

800

400

200

F.G. 11C Patent Application Publication Jun. 23, 2016 Sheet 20 of 36 US 2016/0177276 A1

2200 2 200 30 800 47 8. 500 4. 30 25. 3. 800 700 8. 500 4. 3. 2 552 6,044 84.79,524 00 O 2 3 4 5 & 7 8 to 1 2 3 4 15 TIME (min) FG. 2B Patent Application Publication Jun. 23, 2016 Sheet 21 of 36 US 2016/0177276 A1

" -o- Anti-PDL1-hulgG1-muSIRPa. * -e-A. Anti-PDL4FC-muSIRPe. 10- -- huigG1 control i

0. ------Gioboo oooo- O. 4 100 0000 Patent Application Publication Jun. 23, 2016 Sheet 22 of 36 US 2016/0177276 A1

Ka

88 98 62 49 38 FG. AA 28

280 260 2400 22 200 8 3. S. 4 2 {{0 800 60 400 2 5,486 5.984 ( ------; 2 3 4 5 & 7 & 8 to 12 3 4 5 TIME (min) FG. 4B Patent Application Publication Jun. 23, 2016 Sheet 23 of 36 US 2016/0177276 A1

900 wk- Anti-EGFR-huigG1-huSERPaV1 80 -x- Anti-EGFR-higG1-huSRPawt (N110C) OO c. Anti-EGFR-huggi-(G4S)4-huSRPaV2 A. Fo-SERPOV2 6 orro SRPoV2-Fe: -Do Anti-CD47 500 -- Anti-EGFR 4. 30 200 {{

O 8:: aaaa...: s 8 : 0.00 O.) O. O 00 000 Patent Application Publication Jun. 23, 2016 Sheet 24 of 36 US 2016/0177276 A1

FG. 6A

5,583 5.988 Y " ------" " " " " " " " " " " "," "is TIME (min) F.G. 18B Patent Application Publication Jun. 23, 2016 Sheet 25 of 36 US 2016/0177276 A1

--> huSRPoV2-FChuigG1}-anti-EGFR(Fab) ..o. Anti-EGFR-huggi-G4S4-huSRPaV2 war. SRPCV2-FC 700 A. Fo-huSRPoV2 600 - - - Anti-CR47 8 r"a oo::co. Anti-EGFr 300 g ...A is 400 - A A' g ... is 300

O ------(Bexel : -. 0.00 . O 100C f F.G. 17A

w{v: huSRPoV2-FChuigG1}-anti-EGFR(Fab) 20000 - O - Anti-EGFR-huggi-G4S}4-huSRPaV2 3000 mew SRPCV2-C ---O- 600000 oyye FC-SRFCV2 O K1 wo-born Anti-EGFR 4000 2000 00000 8000 60 -4- 400000 - ? / 2000

geo OOOO . 0. 0. 00 f F.G. 17B Patent Application Publication Jun. 23, 2016 Sheet 26 of 36 US 2016/0177276 A1

EAN SRWWA DAYS wo wo orps SOTYPE CONTRO. 29.5 minor Ani-EGFR 35.5 m-go Ai-EGFr + SRParc 42 x X x Aiii-EGFRSRPa 43. disk ***{ve SR 8Atti-EGFR 40 50

DAYS OS. REAAN SAR

G. 8 Patent Application Publication Jun. 23, 2016 Sheet 27 of 36 US 2016/0177276 A1

Ka 88 98

FG 9A

280 49 3.8 500 4700 60 450 4. 30 c 200 00 90 80 f 6 30 40 30 20 5.576 6.98 6.868 {{ -- (. 1 2 3 4 5 & 7 & 8 to 1 2 3 4 5 TIME (inin) FG. 9B Patent Application Publication Jun. 23, 2016 Sheet 28 of 36 US 2016/0177276 A1

6 09 -o- huSIRPav2-Fcab(HER2) -re orkro SRPW2-F :errars -- Feab{HER2) a/ OO - ... Alti-ER2 - Ati-CD47 80 - 80 -

. yrs O : o s OO

FG. 20A

1200 Kx- huSRPaV2-Fcabtanti-HER2) 1000 -- Anti-ER2-cab w8w Anti-ER2 airo SRPOV2-c worn Anti-CD47 600 400 200

- ? O 00 FG, 2O Patent Application Publication Jun. 23, 2016 Sheet 29 of 36 US 2016/0177276 A1

FIG.21A

(i) 70 80 7.6 A-EGFR.--CCK-SRPaV2 30 40 30 200 430 1000 900 80 700 80 50 i8O 3. 20 5,592 6,164 0. c 1 2 3 4 5 6 7 & 8 to 1 2 3 4 5 TIME (min) FG. 2 B Patent Application Publication Jun. 23, 2016 Sheet 30 of 36 US 2016/0177276 A1

700- ) 600 Anti-EGFR-cis-HCICK-SIRPav2 "6 1500 1400 30 200 40 2000 900 80 70 60 50 AC 30 20 5,592 6.184 40 A (. 1 2 3 4 5 6 i 8 g to 12 3 4 15 4700- ) 80 50 40 1300 20 (0. 24000. 300 8: 70 80 50 40 30: 20- 5.592 6.164 Of

- """""""""""""""------FG. 21B (CONT) it nin Patent Application Publication Jun. 23, 2016 Sheet 31 of 36 US 2016/0177276 A1

ck- Anti-EGFR-hulgG4CK-SERPaV2 80 ... Anti-EGFR-ds-hugGiGK-SiRPo V2 700 -x- Anti-EGFR-ds2-hugGiiCK-SiRPaV2 ----- ... Anti-EGFR-hulgG-huSRPaV2 o 80 -- Anti-EGFR

F.G. 21C Patent Application Publication Jun. 23, 2016 Sheet 32 of 36 US 2016/0177276 A1

F.G. 22A

90 - 6.8 30 17 t 50 40 3

a 120010 1000 9. 80 700 6

i0 2 5506 5.996 ( ; 2 3 4 5 6 & 9 to 1 2 3 4 5 TIME (min) FG. 223 Patent Application Publication Jun. 23, 2016 Sheet 33 of 36 US 2016/0177276 A1

fAN SURVIVA DAYS o pir SOTYPE CONTROL 29.5 -- Anti-EGFR ISIRPa(Q37W) 55 a Anti-EGFRSRPa 43.3 age wo-boo Anti-EGFR 35.5

DAYS OS.REAAN STAR

FG. 23 Patent Application Publication Jun. 23, 2016 Sheet 34 of 36 US 2016/0177276 A1

KDa

88 98 62 43 38 FG. 24A 28

4.

23 -7. 22 2 20 9. 800 700 80 500 se 1300 20

9. 8 60 5 i 30 20 5.45 6.022 7. 0. N V 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 TIME (min) FG. 24B Patent Application Publication Jun. 23, 2016 Sheet 35 of 36 US 2016/0177276 A1

ps SOTYPE CONTROL oxbo Anti-CD2 - O - Anti-CD20 - FC-SRa - - Anti-CD20+ FC-SRPa(Q37W)

50

20 40 50 80 70 80 90 100 DAYS POS-REAMEN START

FG. 25 Patent Application Publication Jun. 23, 2016 Sheet 36 of 36 US 2016/0177276 A1

G. 28

2 3 4. 5,

... k. "

QVC2) SERGAEGREE, NCKEGEFIRV AFRGACPCREE, YNECKECHFPR ,

FR3A 3:23;RE. YNCKEGHFPR,

GE AirC&S RGAGEARE.I.YNCREGIFFRV RECKEG'...'.

rt VS,- - - R&NEESR R&CYCSKSKGS jSES: ENESS; WSS: ENESESS. NE, ESR

aw (WS) &R&N.ESR SS3 KR&N.E.R. KEKG WEEKSEAGES:

NESS W

E: ESSES &NSFSR REAESS US 2016/0177276 A1 Jun. 23, 2016

SRP-ALPHAIMMUNOGLOBULINFUSION 0009. In certain embodiments, the immunoglobulin fusion PROTEINS protein includes a SIRPC. variant with an amino acid sequence at least 85%, at least 90%, or at least 95% identical CROSS-REFERENCE TO RELATED to residues 3-115 of SEQ ID NO:6 or to 3-114 of SEQ ID APPLICATIONS NO:8. 0001. This application claims priority to and the benefit of 0010. In certain embodiments, the immunoglobulin fusion U.S. Provisional Patent Application No. 62/038,196, filed protein includes a SIRPC. variant with an amino acid Aug. 15, 2014, the contents of which are incorporated by sequence at least 80%, at least 85%, at least 90%, or at least reference herein. 95% identical to residues 1-115 of SEQID NO:6 or to 1-114 of SEQID NO:8. FIELD OF THE INVENTION 0011. In some embodiments, the IgV extracellular domain is residues 1-115 of SEQID NO: 6, while in other embodi 0002 This invention relates generally to fusion proteins ments, the IgV extracellular domain is residues 1-114 of SEQ that have the ability to bind CD47 and a surface antigen on a ID NO:8. In some embodiments, the IgV extracellular disease promoting cell. Such as a tumor cell. domain is residues 3-115 of SEQ ID NO: 6 while in other embodiments, the IgV extracellular domain is residues 3-114 BACKGROUND OF THE INVENTION of SEQ ID NO:8. In yet other embodiments, the IgV extra 0003 Macrophages are the principal phagocytes that clear cellular domain is residues 1-114 of SEQ ID NO:193 or diseased cells, such as cancer cells, by phagocytosis. Whether residues 1-115 of SEQID NO:194 or residues 1-115 of SEQ a macrophage phagocytoses a target cellor not depends on the ID NO:195 or residues 1-115 of SEQID NO.196 or residues relative strengths of the pro-phagocytic and anti-phagocytic 1-114 of SEQ ID NO:197 or residues 1-114 of SEQ ID signals. NO:198 or residues 1-115 of SEQ ID NO:199 or residues 0004 Normal, healthy cells are spared from phagocytosis 1-114 of SEQID NO:200 or residues 1-115 of SEQID NO: because the ubiquitously expressed CD47 on normal cells 190. interacts with the signal regulatory protein alpha (SIRPC) on 0012. In other embodiments, the SIRPC. variant has an the macrophage triggering a self"don't eat me' signal. amino acid sequence at least 80%, at least 85%, at least 90%, 0005. However, as cancer cells adapt to enhance their sur or at least 95% identical to residues 1-343 of SEQID NO:6. vival, they subvert normal immune control mechanisms to In other embodiments, the IgV extracellular domain of escape immune surveillance by over-expressing CD47, ren SIRPC. is a wild-type human SIRPC. IgV extracellular dering them resistant to macrophages. For example, CD47 domain. has been shown to be upregulated on human leukemia cells in 0013. In other embodiments, the SIRPC. variant has an order to avoid phagocytosis (Jaiswalet al., Cell, 138:271-285, amino acid sequence at least 80%, at least 85%, at least 90%, 2009). Furthermore, CD47 is highly expressed on human or at least 95% identical to residues 1-114 of SEQID NO:193 acute myeloid leukemia (AML) stem cells and is an adverse or residues 1-115 of SEQ ID NO:194 or residues 1-115 of prognostic factor (Majeti et al., Cell, 138:266-299, 2009). SEQ ID NO:195 or residues 1-115 of SEQ ID NO: 196 or CD47 overexpression as a survival mechanism partly residues 1-114 of SEQID NO:197 or residues 1-114 of SEQ explains why many therapeutic antibodies have limited anti ID NO:198 or residues 1-115 of SEQID NO:199 or residues tumor efficacy despite the fact that antibody-opsonized tumor 1-114 of SEQID NO:200 or residues 1-115 of SEQID NO: cells are expected to engage the activating Fc receptors (FcR) 190. on immune cells to elicitantibody-dependent cellular phago 0014. In certain embodiments, the SIRPC. variant of the cytosis (ADCP) and antibody-dependent cellular cytotoxicity immunoglobulin has a modification to an (ADCC). amino acid at one or more of positions 6, 27, 31, 37, 54, 56. 0006. Accordingly, there is a need in the art for therapies 66, or 72 corresponding to SEQID NO:6 or to SEQID NO:8. that interfere with tumor cells’ ability to avoid phagocytosis The modification may be a substitution, deletion, or insertion through expression of CD47. of an amino acid. In a preferred embodiment, the modifica tion is a Substitution. SUMMARY OF THE INVENTION 0015. In certain embodiments, the SIRPO. variant of the 0007. Described herein are methods and compositions for immunoglobulin fusion protein includes one or more Substi targeting tumor cells with an immunoglobulin fusion protein tutions at positions corresponding to positions 6, 27, 31, 37. specific for both a tumor cellantigen and CD47. Specifically, 54, 56,66, or 72 of SEQID NO:6 or of SEQID NO:8 selected the immunoglobulin fusion protein includes an immunoglo from the group consisting of: V6I, V27I, A27I, 131R, I31T, bulin moiety that is specific for a tumor cell antigen and has a Q37W, Q37H, E54P H56P S66Q, L66A, and M72R. In one second moiety that is specific for CD47. embodiment, the substitution corresponds to V6I. In another 0008. In one aspect, the invention is directed to SIRPO. embodiment, the substitution corresponds to V27I or A27I. In immunoglobulin fusion proteins. The fusion protein includes another embodiment, the substitution corresponds to I31R. In an IgV extracellular domain of SIRPC. or a SIRPC. variant another embodiment, the substitution corresponds to I31T. In having an amino acid sequence at least 80% identical to another embodiment, the substitution corresponds to Q37W. residues 3-115 of SEQID NO:6 or to 3-114 of SEQID NO:8. In another embodiment the substitution corresponds to The fusion protein also includes an immunoglobulin mol Q37H. In another embodiment, the substitution corresponds ecule or portion thereof that binds to a surface antigen on a to E54P. In another embodiment, the substitution corresponds disease promoting cell. In one embodiment, the disease pro to H56P. In another embodiment, the substitution corre moting cell is a tumor cell and the Surface antigen is a tumor sponds to S66Q or L66Q. In another embodiment, the sub antigen. stitution corresponds to M72R. US 2016/0177276 A1 Jun. 23, 2016

0016. In certain embodiments, the SIRPC. variant of the 0033 p. a substitution at a position corresponding to immunoglobulin fusion protein has a modification to an position 92 selected from V92A, V92C, V92D, V92E, amino acid at one or more of positions corresponding to V92G, V92I, V92M, V92N, V92Q, V92R, V92S, or positions 4, 6, 27.31,35, 37, 47, 52,53,54, 56,66, 67, 68,72, V92T and/or 92 or 94 of SEQ ID NO:6 or of SEQID NO:8. The modifi 0034 q. a Substitution at a position corresponding to cation may be a Substitution, deletion, or insertion of an position 94 wherein the substitution is F94L. amino acid. In a preferred embodiment, the modification is a 0035. In some embodiments, the SIRPC. variant has a substitution. modification, preferably a substitution, that decreases the 0017. In certain embodiments, the SIRPO. variant of the binding affinity of the SIRPC. variant for CD47 as compared immunoglobulin fusion protein includes one or more of the to wild-type SIRPC. In yet other embodiments, the SIRPC. following substitutions: variant has a modification, preferably a Substitution, that 0018 a. a substitution at a pposition correspondingp 9. to increases the binding affinity of the SIRPC. variant for CD47 position 4 wherein the substitution is L4V: as compared to wild-type SIRPC. 0019 b. a substitution at a position corresponding to 0036. In certain embodiments, the immunoglobulin mol position 6 selected from V6A, V6C, V6D, V6E, V6G, ecule is an intact antibody, while in other embodiments, the V6I, V6L V6M, V6N, V6Q, V6S, or V6T: immunoglobulin molecule is an antigenbinding portion of an 0020 c. a substitution at a position corresponding to antibody. In yet another embodiment, the immunoglobulin position 27 selected from A27C, A27D, A27G, A27H, molecule is a portion of an antibody that is an antibody A27I, A27K, A27L, A27N, A27Q, A27R, A27S, A27T, variable domain. In some embodiments, the antibody variable or A27V; or V27A, V27C, V27D, V27G, V27H, V27I, domain is an antigen-binding fragment such as an Fab., Fab'. V27K, V27L, V27N, V27Q, V27R, V27S, or V27T: F(ab'). Fv, scEv, single chain antibody, minibody, diabody, or 0021 d. a substitution at a position corresponding to single-domain antibody (nanobody). In one embodiment, the position 31 selected from I31A, I31C, I31E, I31 K, I31Q, intact antibody is an anti-EGFR antibody, such as . I31 R, I31T, or I31V: 0037. In other embodiments, immunoglobulin molecule is 0022 e. a Substitution at a position corresponding to an antigenbinding portion of an antibody that is an Fc region, position 35 selected from P35A, P35C, P35E, P35G, wherein the Fc region is engineered to contain an antigen P35N, P35Q, or P35S: binding site, for example, an Fcab moiety. In some embodi 0023 f. a substitution at a position corresponding to ments, when the immunoglobulin molecule is an Fcab, the position 37 selected from Q37A, Q37C, Q37E, Q37G, SIRPC, or SIRPC, variant is connected by its N-terminus to the Q37H, Q37K, Q37L, Q37M, Q37N, Q37R, Q37S, immunoglobulin molecule, whereas in other embodiments Q37T, or Q37W: the SIRPC. or SIRPC. variant is connected to the immunoglo 0024 g. a substitution at a position corresponding to bulin molecule via its C-terminus. 0038. In some embodiments, the SIRPO or SIRPC. variant position 47 selected from E47A, E47C, E47D, E47F, is connected by its N-terminus to the immunoglobulin mol E47G, E47H, E47I, E47K, E47L, E47M, E47N, E47O, ecule, whereas in other embodiments the SIRPO. or SIRPO. E47R, E47S, E47T, E47V, E47W, or E47Y: variant is connected via its C-terminus to the immunoglobulin 0025 h. a substitution at a position corresponding to molecule. In other embodiments, when the immunoglobulin position 52 selected from Q52A, Q52C, Q52E, Q52H or molecule is an intact antibody, the SIRPC. or SIRPC. variant is Q52M; connected to the C-terminus of the heavy chain, or the C-ter 0026 i. a substitution at a position corresponding to minus of the light chain, and optionally, via a linker. In other position 53 wherein the substitution is K53R; embodiments, when the immunoglobulin molecule is an 0027 j. a substitution at a position corresponding to intact antibody, the SIRPC. or SIRPC. variant is connected to position 54 selected from E54D or E54P; the N-terminus of the heavy chain, or the N-terminus of the 0028 k. a substitution at a position corresponding to light chain, and optionally, via a linker. position 56 selected from H56A, H56C, H56D, H56E, 0039. In yet other embodiments, the immunoglobulin H56F, H56G, H56I, H56K, H56L, H56M, H56N, H56P. molecule or portion thereof is connected to the SIRPC. or H56Q, H56R, H56S, H56T, H56V, H56W, or H56Y. SIRPC. variant via a linker moiety. The linker moiety may be 0029. 1. a substitution at a position corresponding to fused to the SIRPOL or SIRPO variant at either the N-terminus position 66 selected from L66A, L66C, L66D, L66E, or C-terminus of the SIRPC. moiety. L66F, L66G, L66H, L66I, L66K, L66M, L66N, L66P. 0040. In yet other embodiments, the immunoglobulin L66Q, L66S, L66T, L66V. L66W, or L66Y; or S66A, molecule orportion thereof is connected via its N-terminus to S66C, S66D, S66E, S66F, S66G, S66H, S66I, S66K, the SIRPC. or SIRPC. variant, optionally via a linker moiety, S66L, S66M, S66N, S66P, S66Q, S66T, S66V, S66W, or while in other embodiments, the immunoglobulin molecule S66Y: or portion thereof is connected via its C-terminus to the 0030 m. a substitution at a position corresponding to SIRPC. or SIRPC. variant, optionally via a linker moiety. In position 67 selected from T67A, T67C, T67D, T67E, other embodiments, the SIRPC. or SIRPC. variant is connected T67F, T67G, T67H, T67I, T67L, T67M, T67N, T67Q, to the N-terminus of an antibody light chain or a portion T67R, T67S, T67V, T67W, or T67Y, thereof, while in another embodiment, the SIRPC. or SIRPO. 0031 in. a substitution at a position corresponding to variant is connected to the C-terminus of an antibody light position 68 wherein the substitution is K68R chain or a portion thereof. In other embodiments, the SIRPC. 0032 o. a substitution at a position corresponding to or SIRPC. variant is connected to the N-terminus of an anti position 72 selected from M72A, M72C, M72D, M72E, body heavy chain or a portion thereof, while in another M72F, M72G, M72H, M72I, M72K, M72L, M72N, embodiment, the SIRPC. or SIRPC. variant is connected to the M72Q, M72R, M72S, M72T, M72V, M72W, or M72Y: C-terminus of an antibody heavy chain or a portion thereof. A US 2016/0177276 A1 Jun. 23, 2016

linker between SIRPC. or a SIRPC. variant and an immuno light chain of the anti-EGFR antibody or antigen binding globulin molecule or portion thereof is contemplated in some portion thereof, optionally via a linker. embodiments. 0050. In a further embodiment, the SIRPO-anti-EGFR 0041. In certain embodiments, the tumor antigen to which immunoglobulin fusion protein includes an IgV extracellular the immunoglobulin molecule or portion thereof binds is domain of SIRPC. or of a SIRPC. variant having a modification selected from HER2, HER3, EGFR, CD20, GD2, PD-L1, and at one or more of positions corresponding to positions 6, 27. CD19. 31,37, 54.56, 66 or 72 of SEQID NO:6 or SEQID NO:8. In 0042. In another embodiment, the invention is directed to one embodiment, the modification is a Substitution corre a SIRPC. immunoglobulin fusion protein that includes an IgV sponding to Q37W. In another embodiment, the modification extracellular domain of SIRPC. or of a SIRPC. variant having is one or more substitutions corresponding to Substitutions an amino acid sequence at least 80% identical to residues selected from V6I, C27I, A27I, I31R, Q37W, Q37H, E54P. 1-115 of SEQID NO: 190; and an immunoglobulin molecule H56P, S66Q, L66Q, and M72R. orportion thereof that binds to a Surface antigen on a disease 0051. In a further embodiment, the invention is directed to promoting cell. The disease promoting cell may be a tumor an immunoglobulin fusion protein having an immunoglobu cell and the Surface antigen may be a tumor antigen. lin molecule orportion thereofthat binds a tumor cell antigen 0043. In one embodiment, the SIRPC. variant has an amino and a CD47 binding moiety comprising an IgV extracellular acid sequence at least 85%, at least 90%, or at least 95% domain of SIRPC. or a SIRPC. variant having an amino acid identical to residues 1-115 of SEQID NO:190. sequence at least 80% identical to residues 3-115 of SEQID 0044. In another embodiment, the SIRPC. variant has a NO:6 or 3-114 of SEQID NO:8. The fusion protein has a % modification to an amino acid at one or more positions cor red blood cell (RBC) binding mean fluorescence intensity responding to positions 6, 27, 31, 37, 54, 56,66, or 72 of SEQ (MFI) of 35% or less when % RBC binding MFI to an anti ID NO:190. In a further embodiment, the modification is CD47 antibody is calibrated at 100%. The anti-CD47 anti selected from the group consisting of V6I;V27I; A27I; I31R: body is B6H12/huIgG1. The fusion protein also binds to I31T: Q37W; Q37H; E54P; H56P; S66Q: L66Q; and M72R. CD47 on a non-red blood cell. The non-red blood cell, in one 0045. In another embodiment, the invention is directed to embodiment, is a tumor cell. a SIRPC. immunoglobulin fusion protein that includes an 0052. In some embodiments, the immunoglobulin fusion anti-EGFR antibody or an antigen binding portion thereof protein has an IgV extracellular domain of SIRPC. or a SIRPC. and an IgV extracellular domain of SIRPC. or of a SIRPC. variant having an amino acid sequence at least 85%, at least Variant having an amino acid sequence at least 80% identical 90%, at least 95% identical to residues 3-115 of SEQID NO:6 to residues 3-115 of SEQ ID NO:6 or to 3-114 of SEQ ID or 3-114 of SEQ ID NO:8. NO:8. 0053. In some embodiments, the immunoglobulin fusion 0046. In a further embodiment, the SIRPO or of a SIRPO. protein has an IgV extracellular domain of SIRPC. or a SIRPC. variant has an amino acid sequence at least 85% identical, at variant having an amino acid sequence at least 80%, at least least 90% identical, or at least 95% identical to residues 3-115 85%, at least 90%, at least 95% identical to residues 1-115 of of SEQID NO:6 or to 3-114 of SEQID NO:8. SEQID NO:6 or 1-114 of SEQID NO:8. 0047. In yet another embodiment, the SIRPC. or of a 0054. In some embodiments, the tumor cell antigen is SIRPC. variant has an amino acid sequence at least 80% EGFR. In some embodiments, the immunoglobulin molecule identical, at least 85% identical, at least 90% identical, or at is an intact antibody. For example, the intact antibody is an least 95% identical to residues 1-115 of SEQID NO:6 or to anti-EGFR antibody in some embodiments, while in certain 1-114 of SEQID NO:8. embodiments the anti-EGFR-antibody is cetuximab. 0048. In a further embodiment, the anti-EGFR antibody or 0055. In one embodiment, the fusion protein has a % RBC antigenbinding portion thereof contains the heavy chain vari binding MFI of less than 30%, less than 25%, less than 20%, able region and light chain variable region from an antibody less than 15%, less than 10%, less than 5%, less than 4%, less selected from cetuximab, , nimotuzumumab, than 3%, less than 2%, or less than 1%. In one embodiment, , , , or . In the % RBC binding MFI is less than 10%. yet another embodiment, the anti-EGFR antibody or antigen 0056. In yet another embodiment, the fusion protein has a binding portion thereof contains the complementarity deter % RBC binding MFI of between 0-1%, 0-2%, 0-3%, 0-4%, mining regions from an antibody selected from cetuximab, 1-2%, 1-3%, 1-4%, 2-3%, 2-4%, 3-4%, 3-7%, 3-10% or panitumumab, nimotuzumumab, matuZumab, futuximab, 5-10%. imgatuZumab, or necitumumab. In yet another embodiment, 0057. In yet another embodiment, the fusion protein has a the anti-EGFR antibody or antigen binding portion thereof % RBC binding MFI of 5% or less, 4% or less, 3% or less, 2% contains the heavy chain variable region and the light chain or less, or 1% or less. variable region from an antibody selected from cetuximab, 0058. In a further embodiment, the antibody moiety is an panitumumab, nimotuzumumab, matuZumab, futuximab, anti-EGFR antibody. For example, the anti-EGFR antibody is imgatuZumab, or necitumumab. In yet a further embodiment, cetuximab, whereas in another embodiment, the anti-EGFR the anti-EGFR antibody is selected from cetuximab, panitu antibody is panitumumab, nimotuzumumab, matuZumab, mumab, nimotuzumumab, matuZumab, futuximab, imgatu futuximab, imgatuZumab, or necitumumab. Zumab, or necitumumab. In another embodiment, the anti 0059. In another aspect, the invention includes nucleic EGFR antibody is cetuximab. acids encoding the SIRPC. immunoglobulin fusion proteins 0049. In another embodiment, the SIRPO. or SIRPO. vari described herein. Because immunoglobulin fusion proteins ant is connected to the N-terminus of the heavy or light chain described herein may require assembly of two or more pep of the anti-EGFR antibody or antigenbinding portion thereof, tide chains, the invention contemplates the nucleic acids optionally via a linker. In another embodiment, the SIRPC. or required to encode the individual peptide chains that upon SIRPC. variant is connected to the C-terminus of the heavy or expression assemble to form the fusion protein. In another US 2016/0177276 A1 Jun. 23, 2016 aspect, the invention includes a cell comprising a nucleic acid (top) encodes the heavy chain variable domain of antibody or the nucleic acids encoding an immunoglobulin fusion pro (VH) followed by heavy chain constant domains (CH1, hinge tein as described herein. In yet another aspect, the invention (H)-CH2-CH3). DNA construct 2 (bottom) encodes the light includes a method of producing an immunoglobulin fusion chain variable domain of antibody (VL) followed by light protein by maintaining such a cell under conditions that per chain constant domain (CL) genetically fused via an optional mit expression of the nucleic acid or nucleic acids encoding linker (L) to SIRPC. an immunoglobulin fusion protein of the invention. 0069 FIG. 1G is a schematic drawing of an antibody 0060. In a further aspect, the invention is directed to phar SIRPC. showing the tetrameric structure comprising the two maceutical compositions that include pharmaceutically polypeptide components encoded by the DNA construct effective amounts of an immunoglobulin fusion protein shown in FIG.1F. described herein including a pharmaceutically acceptable 0070 FIG. 1H is a schematic drawing of DNA constructs carrier. for the expression of a SIRPO-antibody. DNA construct 1 0061. In yet a further aspect, the invention is directed to (top) encodes SIRPC. genetically fused via an optional linker methods of treating cancer by administering an effective (L) to heavy chain constant domains (hinge (H)-CH2-CH3) amount of an immunoglobulin fusion protein described genetically fused via an optional linker (L) to the heavy chain herein. The cancers that can be treated include breast, col orectal, lung, pancreatic, endometrial, ovarian, gastric, pros variable domain of antibody (VH) followed by heavy chain tate, renal, cervical, myeloma, lymphoma, leukemia, thyroid, constant domain 1 (CH1), and an upper hinge region (H). uterine, bladder, neuroendocrine, head and neck, liver, DNA construct 2 (bottom) encodes the light chain variable nasopharyngeal, testicular, Small cell lung cancer, non-small domain of antibody (VL) followed by light chain constant cell lung cancer, , basal cell skin cancer, Squamous domain (CL). cell skin cancer, dermatofibrosarcoma protuberans, Merkel 0071 FIG. 1I is a schematic drawing of a SIRPC.-antibody cell carcinoma, glioblastoma, glioma, sarcoma, mesothe showing the tetrameric structure comprising the two polypep lioma, or myelodisplastic syndromes. tide components encoded by the DNA construct shown in FIG. 1H. BRIEF DESCRIPTION OF THE DRAWINGS 0072 FIG. 1J is a schematic drawing of a DNA construct 0062 FIGS. 1A-O schematically illustrate the different for the expression of a SIRPC-Fc-schv are shown. DNA DNA and protein constructs of the fusion proteins of the construct encodes SIRPC. genetically fused via an optional invention. linker (L) to heavy chain constant domains (hinge (H)-CH2 0063 FIG. 1A shows two schematic diagrams of DNA CH3) genetically fused via an optional linker (L) to the heavy constructs for the expression of an antibody-SIRPC. fusion chain variable domain of antibody (VH) genetically fused via protein. DNA construct 1 (top) encodes the heavy chain vari an optional linker (L) to the light chain variable domain of able domain of the antibody (VH) followed by heavy chain antibody (VL). constant domains (CH1, hinge (H)-CH2-CH3) genetically (0073 FIG.1K is a schematic drawing of a SIRPO-Fc-scFv. fused via an optional linker (L) to SIRPC. DNA construct 2 showing the dimeric structure comprising the polypeptide (bottom) encodes the light chain variable domain of the anti component encoded by the DNA construct shown in FIG. 1.J. body (VL) followed by light chain constant domain (CL). 0074 FIG.1L is a schematic drawing of a DNA construct 0064 FIG. 1B is a schematic drawing of an antibody for the expression of a schv-Fc-SIRPC. are shown. The DNA SIRPC. fusion protein having a tetrameric structure where the construct encodes the heavy chain variable domain of anti two polypeptide components (i.e., light chain and heavy body (VH) genetically fused via an optional linker (L) to the chain) are each encoded by a DNA construct shown in FIG. light chain variable domain of antibody (VL) genetically 1A. fused via an optional linker (L) to heavy chain constant 0065 FIG.1C shows two schematic drawings of Fc fusion domains (hinge (H)-CH2-CH3) genetically fused via an proteins. They are from left to right (1) SIRPO-Fc and (2) optional linker (L) to SIRPC. Fc-SIRPO. (0075 FIG. 1M is a schematic drawing of a scv-Fc 0066 FIG.1D shows three DNA constructs for the expres SIRPC. showing the dimeric structure comprising the sion of tetravalent bispecific antibodies. DNA construct 1 polypeptide component encoded by the DNA construct (top) encodes the heavy chain variable domain of a first anti shown in FIG. 1L. body (VH(1)) followed by the heavy chain constant domains 0076 FIG. 1N is a schematic drawing of a DNA construct (CH1, hinge (H)-CH2-CH3) genetically fused via an optional for the expression of a SIRPO-Fcab are shown. DNA con linker (L) to the light chain variable domain of second anti struct encodes SIRPC. genetically fused via an optional linker body (VL(2)) followed by the light chain constant domain (L) to heavy chain constant domains (hinge (H)-CH2-CH3) (CL). DNA construct 2 (middle) encodes the light chain vari with constant domain 3 modified to bind antigen. able domain of the first antibody (VL(1)) followed by light chain constant domain (CL). DNA construct 3 (bottom) (0077 FIG. 10 is a schematic drawing of a SIRPO-Fcab encodes the heavy chain variable domain of the second anti showing the dimeric structure comprising the polypeptide body (VH(2)) followed by heavy chain constant domain 1 component encoded by the DNA construct shown in FIG.1N. (CH1), and an upper hinge region (H). Interchain disulfide bonds are depicted as short bars between 0067 FIG. 1E is a schematic drawing of a tetravalent two polypeptide chains. The linker is optional. bispecific antibody (TetBiAb) having a hexameric structure (0078 FIGS. 2A-B are line graphs showing the effect of where the three polypeptide components are encoded by the anti-CD47 B6H12 on red blood cell counts (FIG. 2A) and DNA constructs shown in FIG. 1D. hematocrit levels (FIG. 2B) in cynomolgus monkeys. 0068 FIG.1F is a schematic drawing of DNA constructs (0079 FIGS. 3A-B show the analysis of the expression of for the expression of an antibody-SIRPC. DNA construct 1 the two polypeptides of anti-CD20-hulgG1-SIRPCV2 by US 2016/0177276 A1 Jun. 23, 2016

SDS-PAGE (FIG. 3A) and assembly of the full tetrameric SIRPCV2-Fc(huIgG1)-anti-EGFR(Fab) in an ADCC assay molecule by size exclusion chromatography (SEC) (FIG.3B) using A549 target cells and engineered Jurkat effector cells as described in Example 2. (FIG. 17B), as described in Example 10. 0080 FIGS. 4A-D show binding of anti-CD20-hulgG1 0094 FIG. 18 shows the survival of mice after treatment SIRPCV2 to cells expressing CD47 (CD47-transfected CHO with SIRPOV2-Fc(hulgG1)-anti-EGFR(Fab) in an orthoto cells, FIG. 4A; leukocyte-enriched whole blood, FIG. 4B) or pic A549 lung tumor model (inverted filled triangle: isotype expressing both CD20 and CD47 (Raji cells, FIG. 4C: Nama control; filled circle: anti-EGFR; cross: anti-EGFR and lwa cells, FIG. 4D). SIRPO-Fc; open circle: anti-EGFR-hulgG1-SIRPCV2; filled I0081 FIGS. 5A-B show the survival of mice after injec diamond: SIRPCV2-Fc(hulgG1)-anti-EGFR(Fab)). tion with Daudi cells (FIG. 5A) or with Raji cells (FIG. 5B) (0095 FIGS. 19A-B show the analysis of the expression of and treatment with the tetravalent bispecific anti-CD20/anti the polypeptide of SIRPOV2-Fcab(HER2) by SDS-PAGE CD47. (FIG. 19A) and assembly of the full dimeric molecule by size 0082 FIGS. 6A-B show the analysis of the expression of exclusion chromatography (SEC) (FIG. 19B) as described in the two polypeptides of anti-EGFR-hulgG1-SIRPC. by SDS Example 15. PAGE (FIG. 6A) and assembly of the full tetrameric molecule (0096 FIGS. 20A-B show binding of SIRPCV2-Fcab by size exclusion chromatography (SEC) (FIG. 6B) as (HER2) to cells expressing CD47 (CD47-transfected CHO described in Example 4. cells, FIG. 20A) or expressing both HER2 and CD47 (BT474 I0083 FIGS. 7A-C show binding of anti-EGFR-hulgG1 cells, FIG.20B). SIRPO. proteins to cells expressing CD47 (CD47-transfected (0097 FIGS. 21A-C show the analysis of the expression of CHO cells, FIG. 7A: leukocyte-enriched whole blood, FIG. the two polypeptides of anti-EGFR-hulgG1/anti-EGFR-LC 7B) or expressing both EGFR and CD47 (A549 cells, FIG. SIRPCV2 by SDS-PAGE (FIG.21A) and assembly of the full 7C). tetrameric molecule by size exclusion chromatography I0084 FIG. 8 shows the in vitro activity of anti-EGFR (SEC) (FIG.21B) as described in Example 9, and the binding hulgG1-SIRPCV2 in an ADCC assay using A549 target cells of anti-EGFR-hulgG1/anti-EGFR-LC-SIRPOV2 to cells and engineered Jurkat effector cells. expressing CD47 (CD47-transfected CHO cells, FIG. 21C). I0085 FIGS. 9A-B show the pharmacokinetic analysis of (0098 FIGS. 22A-B shows the analysis of the expression anti-EGFR-hulgG1-SIRPOV2 in mice (FIG.9A) and the sur of the two polypeptides of anti-EGFR-hulgG1-SIRPCV2 vival of mice after treatment with anti-EGFR-hulgG1-SIR (Q37W) by SDS-PAGE (FIG.22A) and assembly of the full PO.V2 in an orthotopic A549 lung tumor model (FIG. 9B; tetrameric molecule by size exclusion chromatography inverted filled triangle: isotype control; filled circle: anti (SEC) (FIG.22B) as described in Example 18. EGFR: filled triangle: SIRPC-Fc; filled diamond: anti-EGFR 0099 FIGS. 23 shows the survival of mice after treatment and SIRPO-Fc; open circle: anti-EGFR-hulgG1-SIRPCV2). with anti-EGFR-hulgG1-SIRPCV2 or anti-EGFR-hulgG1 I0086 FIGS. 10A-B show the analysis of the expression of SIRPCV2(Q37W) in an orthotopic A549 lung tumor model. the two polypeptides of anti-HER2-hulgG1-SIRPOV2 by 0100 FIGS. 24A-B show the analysis of the expression of SDS-PAGE (FIG. 10A) and assembly of the full tetrameric the two polypeptides of anti-CD20-hulgG1-SIRPCV2 molecule by size exclusion chromatography (SEC) (FIG. (Q37W) by SDS-PAGE (FIG. 24A) and assembly of the full 10B) as described in Example 5. tetrameric molecule by size exclusion chromatography I0087 FIGS. 11A-C show binding of anti-HER2-hulgG1 (SEC) (FIG. 24B) as described in Example 19. SIRPCV2 to cells expressing CD47 (CD47-transfected CHO 0101 FIG. 25 shows the survival of mice after treatment cells, FIG. 11A; leukocyte-enriched whole blood, FIG. 11B) with the combination of anti-CD20 and Fc-SIRPOV2 or orexpressing both HER2 and CD47(BT474 cells, FIG.11C). Fc-SIRPCV2(Q37W) in a Daudi disseminated lymphoma 0088 FIGS. 12A-B show the analysis of the expression of model. the two polypeptides of anti-PD-L1-hulgG1-muSIRPC. by 0102 FIG. 26 shows an alignment of the IgV domain of SDS-PAGE (FIG. 12A) and assembly of the full tetrameric known human SIRPC. alleles: IgV (V1) (residues 1-115 of molecule by size exclusion chromatography (SEC) (FIG. SEQID NO:6), IgV (V2) (SEQID NO:8), IgV (V3) (SEQID 12B) as described in Example 7. NO.193), IgV (V4) (SEQ ID NO: 194), IgV (V5) (SEQ ID I0089 FIG. 13 shows binding of anti-PD-L1-hulgG1-mu NO:195), IgV (V6) (SEQ ID NO:196), IgV (V7) (SEQ ID SIRPC. to A20 cells expressing PD-L1 and CD47. NO:197), IgV (V8) (SEQ ID NO:198), IgV (V9) (SEQ ID 0090 FIGS. 14A-B show the analysis of the expression of NO:199), and IgV (V10) (SEQ ID NO:200). the two polypeptides of anti-EGFR-hulgG1-SIRPO.(N110O) by SDS-PAGE (FIG. 14A) and assembly of the full tetrameric DETAILED DESCRIPTION OF THE INVENTION molecule by size exclusion chromatography (SEC) (FIG. 0103) The present invention is directed to immunoglobu 14B) as described in Example 8. linfusion proteins with enhanced tumor targeting and effector 0091 FIG. 15 shows binding of anti-EGFR-hulgG1 functions. Generally, immunoglobulin fusion proteins of the SIRPO.(N110O) to cells expressing CD47 (CD47-transfected invention include a CD47 binding agent moiety and an immu CHO cells). noglobulin moiety. The immunoglobulin moiety binds to a 0092 FIGS. 16A-B show the analysis of the expression of Surface antigen on a disease promoting cell, while the CD47 the two polypeptides of SIRPCV2-Fc(hulgG1)-anti-EGFR binding agent moiety binds to CD47 on the same cell. In one (Fab) by SDS-PAGE (FIG. 16A) and assembly of the full embodiment, the invention involves genetically joining a tetrameric molecule by size exclusion chromatography tumor-specific immunoglobulin moiety to a moiety that binds (SEC) (FIG. 16B) as described in Example 10. CD47. The preferred CD47 binding agent is SIRPC. or a 0093 FIGS. 17A-B show binding of SIRPoV2-Fc variant of SIRPC. Accordingly, certain embodiments of the (hulgG1)-anti-EGFR(Fab) to cells expressing CD47 (CD47 invention are directed to immunoglobulin fusion proteins that transfected CHO cells, FIG. 17A) and the in vitro activity of bind to tumor cells expressing CD47. US 2016/0177276 A1 Jun. 23, 2016

0104 CD47 is ubiquitously expressed on all human cells. preferred embodiment, the CD47 binding agent is SIRPC. Although tumor cells, especially cancer stem cells, express SIRPC. variant, or an affinity optimized variant of SIRPC. higher levels of CD47, hematopoietic stem cells also overex 0.108 “SIRPC.” refers to wild-type signal-regulatory pro press CD47. Therefore, SIRPO, a SIRPC. variant, or another tein alpha or an amino acid sequence of a recombinant or CD47 binding agent with sufficient binding affinity for CD47 non-recombinant polypeptide having the amino acid is unlikely to discriminate between cancer cells and normal sequence of wild-type signal-regulatory protein alpha or a cells, including red blood cells, which are present in circula native or naturally occurring allelic variant of signal-regula tion at 5 billion cells/mL and occupy about half of the blood tory protein alpha. In one embodiment, SIRPC. is a wild-type Volume. Accordingly, the immunoglobulin fusion proteins of mammalian SIRPC, whereas in a preferred embodiment, the present invention are designed to have high affinity bind SIRPC. is a wild-type human SIRPC. The amino acid ing to a tumor antigen on tumor cells and low binding affinity sequence for the mature form of the predominant wild type for CD47. This results in weak binding of the immunoglobu human SIRPC. (SIRPOV1) is provided in Table 4 as SEQ ID lin fusion protein to CD47 on normal cells. This way, the NO:6. In one embodiment, SIRPC. includes a signal immunoglobulin fusion proteins are designed to not target sequence, whereas in another embodiment, SIRPC. refers to normal cells, thereby circumventing the toxicity to normal the mature form of the protein. cells resulting from ubiquitous expression of CD47 and also 0109. According to one embodiment, a SIRPC. is a SIRPC. preventing the ubiquitously expressed CD47 on normal cells extracellular domain, i.e., a SIRPC. protein engineered to from becoming a drug sink for anti-CD47 therapy which exclude the transmembrane and cellular domain. In another would otherwise result in unfavorable pharmacokinetics that embodiment, a SIRPC. includes at least the extracellular would be incompatible with the once weekly or once domain. In one embodiment, the SIRPC. protein is a human biweekly regiment typical for antibody therapy. For this rea SIRPC. extracellular domain. The sequence of the wild-type son, it is preferred that the fusion moiety bind CD47 with low SIRPOV1's extracellular domain is residues 1-343 of SEQID affinity while still blocking CD47's ability to interact with NO:6. SIRPO 0110. In yet anotherembodiment, a SIRPC. is a SIRPC. IgV 0105. The immunoglobulin fusion proteins of the inven domain of the extracellular domain. In one embodiment, a tion (1) have high binding affinity for a tumor-specificantigen SIRPC. IgV domain is a human SIRPC. IgV domain. For on a tumor cell; (2) achieve enhanced tumor targeting through example, in one embodiment, a SIRPC. IgV domain is resi additional avidity provided by low affinity binding of the dues 1-115 of SEQID NO:6, while in another embodiment, a fusion moiety to CD47 on the same tumor cell; (3) elicit SIRPO, IgV is residues 1-114 of SEQID NO:8, while in yet potent ADCP and ADCC through a combination of blockade another embodiment, a SIRPC. IgV is residues 3-115 or SEQ of the CD47 “don’t eat me' signal and Fc-dependent activa ID NO:6, while in yet another embodiment a SIRPC. IgV is tion of immune effector cells; and (4) avoid toxicity by low residues 3-114 of SEQID NO:8. In another embodiment, a affinity binding to CD47 on normal cells, including red blood SIRPC. IgV domain is residues 1-114 of SEQID NO.193. In cells. Importantly, the ADCC/ADCP induced toxicity to nor another embodiment, a SIRPC. IgV domain is residues 1-115 mal cells is further lowered when the CD47 binder, e.g., of SEQ ID NO:194. In another embodiment, a SIRPC. IgV SIRPC. or a SIRPC. variant, is joined to the Fc region of the domain is residues 1-115 of SEQ ID NO:195. In another antibody moiety in a configuration that is not optimal for FcR embodiment, a SIRPC. IgV domain is residues 1-115 of SEQ engagement, e.g., the X-Fc configuration, which has an ID NO:196. In another embodiment, a SIRPC. IgV domain is amino-to-carboxyl orientation similar to that of a native anti residues 1-114 of SEQID NO:197. In another embodiment, a body, elicits higher ADCC activity than the Fc-X configura SIRPC. IgV domain is residues 1-114 of SEQID NO.198. In tion. another embodiment, a SIRPC. IgV domain is residues 1-115 0106. According to the present invention, the targeting of SEQ ID NO:199. In another embodiment, a SIRPC. IgV specificity of the immunoglobulin fusion protein is driven domain is residues 1-114 of SEQID NO:200. In yet another primarily by the binding of an antibody moiety to its cognate embodiment, a SIRPC. includes at least the SIRPC. IgV tumor-specific antigen rather than CD47, a ubiquitously domain of the extracellular domain. expressed antigen. Moreover, the targeting specificity is fur 0111. The invention also includes “variants' of SIRPC. A ther enhanced by an avidity effect provided by binding in cis “variant of SIRPO is defined as a SIRPO. amino acid of the fusion partner to CD47 overexpressed on the tumor sequence that is altered by one or more amino acids as com cell. Successful bispecific targeting in cis depends critically pared to wild-type SIRPC. The variant may have “conserva on the relative receptor density and the physical location of tive' changes, wherein a Substituted amino acid has similar the two targets on the cell Surface. Such enhanced tumor structural or chemical properties, e.g., replacement of leucine targeting should offer a better therapeutic index in terms of with isoleucine. More rarely, a variant can have “nonconser both Superior efficacy and safety, when compared to anti Vative' changes, e.g., replacement of a glycine with a tryp CD47 antibodies and other CD47 blockade agents. tophan. Similar minor variations can also include amino acid deletions or insertions, or both. CD47 Binding Agents 0.112. In one embodiment, SIRPC. variants include 0107 Immunoglobulin fusion proteins of the invention polypeptides that have at least about 70%, 75%, 80%, 81%, include a moiety that is capable of binding CD47. In one 82%, 83%, 84%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, embodiment, CD47 binding agents include antibodies to 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or more CD47. In other embodiments, CD47 binding agents are non sequence identity with wild-type SIRPC. antibody proteins or molecules that have binding affinity for 0113. In another embodiment, SIRPC. variants include CD47, for example, ligands that bind the CD47 receptor. For polypeptides that have at least about 70%, 75%, 80%, 81%, example, in one embodiment, the CD47 binding agent por 82%, 83%, 84%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, tion of the fusion protein is an anti-CD47 antibody. In a 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or more US 2016/0177276 A1 Jun. 23, 2016

sequence identity with a wild-type SIRPC. extracellular 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, domain. In one embodiment, SIRPC. variants include 98%, 99%, or more sequence identity with the IgV domain of polypeptides that have at least about 70%, 75%, 80%, 81%, a residues 1-115 of SEQID NO:199. In another embodiment, 82%, 83%, 84%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, SIRPC. variants include polypeptides that have at least about 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or more 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, sequence identity with residues 1-343 of SEQID NO:6. 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or more sequence identity with the IgV domain of 0114. In yet another embodiment, SIRPC. variants include a residues 1-114 of SEQID NO:200. In another embodiment, polypeptides that have at least about 70%, 75%, 80%, 81%, SIRPC. variants include polypeptides that have at least about 82%, 83%, 84%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or more 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, sequence identity with the IgV domain of a wild-type SIRPC. 98%, 99%, or more sequence identity with the IgV domain of extracellular domain, for example, residues 1-115 of SEQID a residues 1-114 of SEQID NO:190. SEQ ID NO:190 is a NO:6 in one embodiment, residues 1-114 of SEQID NO:8 in consensus sequence of ten known human SIRPC. IgV another embodiment, residues 1-114 of SEQID NO: 193 in domains. yet another embodiment, 1-115 of SEQ ID NO: 194 in yet another embodiment, 1-115 of SEQ ID NO: 195 in yet 0115 To determine the percent identity of two amino acid another embodiment, 1-115 of SEQ ID NO: 196 in yet sequences or of two nucleic acids, the sequences are aligned another embodiment, 1-114 of SEQ ID NO: 197 in yet for optimal comparison purposes (e.g., gaps can be intro another embodiment, 1-114 of SEQ ID NO: 198 in yet duced in the sequence of a first amino acid or nucleic acid another embodiment, 1-115 of SEQ ID NO: 199 in yet sequence for optimal alignment with a second amino acid or another embodiment, or 1-114 of SEQ ID NO:200 in yet nucleic acid sequence). The percent identity between the two another embodiment. In one particular embodiment, SIRPC. sequences is a function of the number of identical positions variants include polypeptides that have at least about 70%, shared by the sequences (i.e., '% homology (H of identical 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, 88%, positions/total it of positions)times 100). The determination 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, of percent homology between two sequences can be accom 99%, or more sequence identity with the IgV domain of a plished using a mathematical algorithm. A non-limiting residues 1-115 of SEQ ID NO:6. In another embodiment, example of a mathematical algorithm utilized for the com SIRPC. variants include polypeptides that have at least about parison of two sequences is the algorithm of Karlin and Alts 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, chul, (1990) Proc. Natl. Acad. Sci. USA, 87:2264-68, modi 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, fied as in Karlin and Altschul, (1993) Proc. Natl. Acad. Sci. 98%, 99%, or more sequence identity with the IgV domain of USA, 90:5873-77. Such an algorithm is incorporated into the a residues 1-114 of SEQ ID NO:8. In another embodiment, NBLAST and XBLAST programs of Altschulet al., (1990).J. SIRPC. variants include polypeptides that have at least about Mol. Biol., 215:403-10. BLAST nucleotide searches can be 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, performed with the NBLAST program, score=100, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, wordlength=12. BLAST protein searches can be performed 98%, 99%, or more sequence identity with the IgV domain of with the XBLAST program, score=50, wordlength=3. To a residues 1-114 of SEQID NO:193. In another embodiment, obtain gapped alignments for comparison purposes, Gapped SIRPC. variants include polypeptides that have at least about BLAST can be utilized as described in Altschul et al., (1997) 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, Nucleic Acids Research, 25(17):3389-3402. When utilizing 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, BLAST and Gapped BLAST programs, the default param 98%, 99%, or more sequence identity with the IgV domain of eters of the respective programs (e.g., XBLAST and a residues 1-115 of SEQID NO:194. In another embodiment, NBLAST) can be used. SIRPC. variants include polypeptides that have at least about 0116. In some embodiments, a SIRPC. variant includes 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, one or more mutations in the variable domain of the extracel 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, lular domain as compared to wild-type SIRPC. Mutations 98%, 99%, or more sequence identity with the IgV domain of contemplated by the invention are described below and are a residues 1-115 of SEQID NO:195. In another embodiment, also provided in Tables 1 and 2 found in Example 16 below, as SIRPC. variants include polypeptides that have at least about well as in Table 3 found in Example 17 below. 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, 0117. In one embodiment, a SIRPC. variant of the inven 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, tion has a modification to an amino acid a position corre 98%, 99%, or more sequence identity with the IgV domain of sponding to one or more of positions 6, 27, 31, 37, 54, 56,66, a residues 1-115 of SEQID NO:196. In another embodiment, or 72 corresponding to SEQID NO:6 or to SEQID NO:8 or SIRPC. variants include polypeptides that have at least about to SEQ ID NO:190. The modification may be a deletion, 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, substitution, or insertion. In a preferred embodiment, the 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, modification is a substitution. In other embodiments, the 98%, 99%, or more sequence identity with the IgV domain of SIRPC. variant has a modification to an amino acid at two or a residues 1-114 of SEQID NO:197. In another embodiment, more, at three or more, at four or more, at five or more, at six SIRPC. variants include polypeptides that have at least about or more, at seven or more, or at eight positions corresponding 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, to positions 6, 27, 31, 37, 54, 56,66, or 72 corresponding to 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, SEQID NO:6 or to SEQID NO:8 or to SEQID NO:190. 98%, 99%, or more sequence identity with the IgV domain of 0118. In a further embodiment, a SIRPC. variant of the a residues 1-114 of SEQID NO.198. In another embodiment, invention has one or more substitutions as follows: V6I,V27I, SIRPC. variants include polypeptides that have at least about A27I, 131R, I31T, Q37W, Q37H, E54P H56P S66Q, L66A, 70%, 75%, 80%, 81%, 82%, 83%, 84%, 85%, 86%, 87%, and M72R. In one embodiment, the substitution corresponds US 2016/0177276 A1 Jun. 23, 2016

to V6I. In another embodiment, the substitution corresponds L66F, L66G, L66H, L66I, L66K, L66M, L66N, L66P. to V27I or A27I. In another embodiment, the substitution L66Q, L66S, L66T, L66V. L66W, or L66Y; or S66A, corresponds to I31R. In another embodiment, the substitution S66C, S66D, S66E, S66F, S66G, S66H, S66I, S66K, corresponds to I31T. In another embodiment, the substitution S66L, S66M, S66N, S66PS66Q, S66T, S66V, S66W, or corresponds to Q37W. In another embodiment the substitu S66Y: tion corresponds to Q37H. In another embodiment, the sub 0.133 m. a substitution at a position corresponding to stitution corresponds to E54P. In another embodiment, the position 67 selected from T67A, T67C, T67D, T67E, substitution corresponds to H56P. In another embodiment, T67F, T67G, T67H, T67I, T67L, T67M, T67N, T67Q, the substitution corresponds to S66Q or L66Q. In another T67R, T67S, T67V, T67W, or T67Y, embodiment, the substitution corresponds to M72R. 0.134 in a Substitution at a position corresponding to 0119. In one embodiment, a SIRPC. variant of the inven position 68 wherein the substitution is K68R tion has a modification to an amino acid at one or more 0.135 o. a substitution at a position corresponding to positions corresponding to positions 4, 6, 27, 31, 35, 37, 47. position 72 selected from M72A, M72C, 52, 53, 54, 56,66, 67, 68,72, 92 or 94 of SEQID NO:6 or of 0.136 M72D, M72E, M72F, M72G, M72H, M72I, M72K, SEQ ID NO:8 or of SEQID NO:190. The modification may M72L, M72N, M72Q, M72R, M72S, M72T, M72V, M72W, be a deletion, substitution, or insertion. In a preferred embodi or M72Y: ment, the modification is a substitution. In other embodi 0.137 p. a substitution at a position corresponding to ments, the SIRPC. variant has a modification to an amino acid position 92 selected from V92A, V92C, V92D, V92E, at two or more, at three or more, at four or more, at five or V92G, V92I, V92M, V92N, V92Q, V92R, V92S, or more, at six or more, at seven or more, at eight or more, at nine V92T and/or or more, atten or more, at eleven or more, at twelve or more, O138 C ... a Substitution at a pposition correspondingp 9. to at thirteen or more, at fourteen or more, at fifteen or more, at position 94 wherein the substitution is F94L. sixteen or more, or at seventeen positions corresponding to (0.139. To determine the affinity of a SIRPo variant for positions 4, 6, 27.31,35, 37, 47, 52,53,54, 56,66, 67, 68,72, binding CD47, surface plasmon resonance (SPR) may be 92 or 94 of SEQID NO:6 or of SEQID NO:8 or of SEQID used. In an exemplary protocol, purified goat anti-human IgG NO:190. Fc (Jackson Immuno Research Laboratories) is immobilized 0120. In a further embodiment, a SIRPC. variant of the onto the CM5 chip using amine coupling chemistry using a invention has one or more Substitutions as follows: Biacore 4000 instrument (GE Healthcare). Biacore CM-5 0121 a. a substitution at a position corresponding to chips, ethanolamine, NHS/EDC coupling reagents and buff position 4 wherein the substitution is L4V: ers are obtained from Biacore (GE Healthcare). The immo 0.122 b. a substitution at a position corresponding to bilization steps are carried out at a flow rate of 30 ul/min in position 6 selected from V6A, V6C, V6D, V6E, V6G, HEPES buffer (20 mM HEPES, 150 mM NaCl, 3.4 mM V6I, V6L V6M, V6N, V6Q, V6S, or V6T: EDTA and 0.005% P20 surfactant). The sensor surfaces are 0123 c. a Substitution at a position corresponding to activated for 7 min with a mixture of NHS (0.05 M) and EDC position 27 selected from A27C, A27D, A27G, A27H, (0.2M). The goat anti-human IgGFc is injected at a concen A27I, A27K, A27L, A27N, A27Q, A27R, A27S, A27T, tration of ~30 g/ml in 10 mM sodium acetate, pH 5.0, for 7 or A27V; or V27A, V27C, V27D, V27G, V27H, V27I, min. Ethanolamine (1 M. pH 8.5) is injected for 7 minto block V27K, V27L, V27N, V27Q, V27R, V27S, or V27T: any remaining activated groups. An average of 12,000 0.124 d. a substitution at a position corresponding to response units (RU) of capture antibody is immobilized on position 31 selected from I31A, I31C, I31E, I31 K, I31Q, each flow cell. Kinetic binding experiments are performed I31 R, I31T, or I31V: using the same HEPES buffer (20 mM HEPES, 150 mM 0.125 e. a substitution at a position corresponding to NaCl, 3.4 mM EDTA and 0.005% P20 surfactant) and are position 35 selected from P35A, P35C, P35E, P35G, equilibrated at 25°C. Kinetic data are collected by injecting P35N, P35Q, or P35S: SIRPC. variants at 0.5 and 1 g/ml for two minutes at a flow 0126 f. a substitution at a position corresponding to posi rate of 30 ul/min, followed by a buffer wash for 30s at the tion 37 selected from Q37A, Q37C, Q37E, Q37G, Q37H, same flow rate. Human CD47-His is bound at different con Q37K, Q37L, Q37M, Q37N, Q37R, Q37S, Q37T, or Q37W: centrations for 3 min followed by a dissociation step for 10 0127 g. a substitution at a position corresponding to min at the 30 ul/min flow rate. The data are fit using a 1:1 position 47 selected from E47A, E47C, E47D, E47F, Langmuir binding model with the BIA evaluation software. E47G, E47H, E47I, E47K, E47L, E47M, E47N, E47O, Kinetic rate constants are determined from the fits of the E47R, E47S, E47T, E47V, E47W, or E47Y: association and dissociation phases, and the K is derived I0128 h. a substitution at a position corresponding to from the ratio of these constants. position 52 selected from Q52A, Q52C, Q52E, Q52H or 0140. In an alternative method to determine the avidity of Q52M; a SIRPC. variant for binding CD47, a cell binding assay may I0129 i. a substitution at a position corresponding to be used. In an exemplary protocol, 2x10 Chinese hamster position 53 wherein the substitution is K53R; ovary (CHO) cells transfected with CD47 per well are incu 0.130 j. a substitution at a position corresponding to bated with varying concentrations of antibodies diluted in position 54 selected from E54D or E54P; PBS+1% FBS in a 96 well plate for 60 min on ice. After 0131 k. a substitution at a position corresponding to washing with PBS+1% FBS, cells are incubated with FITC position 56 selected from H56A, H56C, H56D, H56E, F(ab')2 goat Anti-Human IgG, Fcy (Jackson ImmunoRe H56F, H56G, H56I, H56K, H56L, H56M, H56N, H56P. search, West Grove, Pa.) diluted 1:200 in PBS+1% FBS for 60 H56Q, H56R, H56S, H56T, H56V, H56W, or H56Y. min on ice. After washing again, cells are fixed with 1% I0132 1. a substitution at a position corresponding to formaldehyde in PBS. Cells are analyzed by flow cytometry position 66 selected from L66A, L66C, L66D, L66E, (MACSQuant, Miltenyi Biotec, Cologne, Germany). An US 2016/0177276 A1 Jun. 23, 2016

EC50 is calculated by fitting data to a sigmoidal curve (log bulin portion is an Fc region fused to the heavy chain variable (agonist) vs. response Variable slope (four parameters)) domain of antibody (VH) followed by heavy chain constant with Graph Pad Prism. domain 1 (CH1), and an upper hinge region (H). 0141 When these mutations are introduced into SIRPC. or 0144. According to the present invention, the fusion moi a fusion protein comprising SIRPC, the resulting variant gen ety, for example, SIRPC, can be genetically joined to the erally has enough SIRPC. biological activity to be useful as a immunoglobulin in a way that does not adversely affect the therapeutic protein. In some embodiments, the biological binding of the antibody to the tumor antigen. Preferably, the activity of the SIRPo variant is at least 0.01 fold, 0.03 fold, fusion moiety is joined to the Fc in a configuration that is not 0.06 fold, 0.1 fold, 0.3 fold, 0.6 fold, 1 fold, 3 fold, 5, fold, 6 optimal for FcR engagement, resulting in diminished ADCC/ fold, 10 fold, 20 fold, 30 fold, 40 fold, 50 fold, 60 fold or 100 ADCP activities on normal cells. In some embodiments, the fold of the biological activity of wild type SIRPC. or a fusion Fc region retains or has modified FcR binding capability. protein containing wild-type SIRPC. Biological activity of Human antibody fragments (e.g., parental and optimized the SIRPC. variants of the invention can be tested in an in-vitro variants) can be engineered to contain certain constant (i.e., or in-vivo assay. In-vitro assays to determine the biological Fc) regions with a specified effector function (e.g., antibody activity of SIRPC. on a cell expressing CD47 are well estab dependent cellular cytotoxicity (ADCC)). Human constant lished in the art. For example, the biological activity may be regions are known in the art. determined in a leukocyte transmigration assay, as described 0145 Fragments of antibodies that have the same or com by Liu et al. (J. Mol. Bio., 365:680, 2007). parable binding characteristics to those of the whole antibody may also be present. Such fragments may contain one or both Immunoglobulin Moieties Fab fragments or the F(ab')2 fragment. The antibody frag 0142. As used herein, the term “antibody’ means an intact ments may contain all six CDRs of the whole antibody, antibody (e.g., an intact ). In some although fragments containing fewer than all of Such regions, embodiments, an 'antibody includes an antigen-binding such as three, four or five CDRs, are also functional. fragment of an antibody. Antigen-binding fragments include 0146 According to the present invention, the immunoglo Fab, Fab., F(ab')2, Fv, single chain antibodies (e.g., sclv), bulin moiety of the immunoglobulin fusion protein in one minibodies, diabodies, and single-domain antibodies embodiment is an intact antibody. In yet another embodi (“sdAb’ or “nanobodies' or “camelids'). In yet other ment, the immunoglobulin moiety is a tetravalent bispecific embodiments, an antibody includes an intact antibody or antibody. This embodiment is exemplified, for example, in antigen-binding fragment of an antibody (e.g., a phage dis FIG. 1E. In certain embodiments, the SIRPC. moiety is fused play antibody including a fully human antibody, a semisyn to the C-terminus of the heavy chain, for example, in FIG. 1B thetic antibody or a fully synthetic antibody) that has been or FIG. 1M, while in other embodiments, the SIRPC. moiety optimized, engineered or chemically conjugated. Examples is fused to the C-terminus of the light chain, for example, in of antibodies that have been optimized are affinity-matured FIG. 1G. The SIRPC. moiety may be fused to the immuno antibodies. Examples of antibodies that have been engineered globulin moiety by an optional linker. are Fc optimized antibodies, and multispecific antibodies 0.147. In still a further embodiment, the immunoglobulin (e.g., bispecific antibodies). An antibody conjugated to a moiety is an Fc engineered to include at its C-terminus VH toxin moiety is an example of a chemically conjugated anti and CH1 domains, with the light chain binding to the heavy body. In some embodiments, antibodies may be IgG1, IgG2. chain variable region, for example, in FIG. 1I. In some IgG3, IgG4, IgM, IgE. Ig), or IgA. embodiments, the SIRPC. moiety is fused to the N-terminus of 0143. As used herein, the term “immunoglobulin' or the Fc portion, optionally via a linker. In other embodiments, “immunoglobulin molecule' means an antibody or antigen the SIRPC. moiety is fused to the C-terminus of the light chain binding fragment of an antibody as defined herein, but also or the C-terminus of heavy chain, optionally via a linker. includes portions of an antibody, such as a heavy chain or 0.148. In yet another embodiment, the immunoglobulin light chain variable or constant region. Examples of portions moiety is an Fc joined at its C-terminus to an schv. This of an immunoglobulin include CH1, CH2, CH3, hinge, VH1, embodiment is exemplified, for example, in FIG. 1K. In such CL and VL domains as well as an Fc region. Further, “immu an embodiment, the SIRPC. moiety may be fused to the N-ter noglobulin' includes an Fc fragment or region that has been minus of the Fc region, optionally via a linker. In yet another engineered to include an antigen binding site (“Fcab’’). For embodiment, the immunoglobulin moiety is an Schv joined at example, in one embodiment, the immunoglobulin fusion its C-terminus to an Fc region. This embodiment is exempli protein of the invention includes an “Feab' moiety with bind fied, for example, in FIG. 1M. In such an embodiment, the ing specificity for a tumor antigen. “Feabs' are discussed in SIRPC. moiety may be fused the C-terminus of the Fc region, the art, for example, in WO2008/003103 and WO2012/ optionally via a linker. 007167. Further, Example 15 provides an example of an 0149. In a further embodiment, the immunoglobulin moi immunoglobulin fusion protein of the invention where the ety is an Fc region engineered to include an antigen binding immunoglobulin moiety is an Fcab to HER2. In some domain. For example, in Some embodiments, the Fc region is embodiments, “immunoglobulin' includes engineered anti an Fcab region which is an Fc region engineered, for example, bodies where further variable or constant heavy or light chain to include an antigen binding domain in the CH3 domain. regions are added to an otherwise intact antibody, or where This embodiment, is exemplified, for example in FIG. 10. In variable or constant regions are relocated or rearranged from some embodiments, the SIRPC. moiety may be fused to the an original position to a new position within an antibody. For N-terminus of the Fc or Fcab region, optionally via a linker, example, FIG. 1E shows an immunoglobulin that is a tetrava while in other embodiments, the SIRPC. moiety may be fused lent bispecific antibody, i.e., an antibody engineered to to the C-terminus, optionally via a linker. include a second variable region. One example of relocation 0150. In yet another embodiment, the immunoglobulin or rearrangement is shown in FIG. 1 I where the immunoglo moiety is a Fab. In yet another embodiment, the immunoglo US 2016/0177276 A1 Jun. 23, 2016

bulin moiety is a Fab'. In yet another embodiment, the immu 0.155. In another embodiment, the CD47 binding agent, noglobulin moiety is a F(ab')2. In yet another embodiment, which in one embodiment may be a SIRPC. or a SIRPC. the immunoglobulin moiety is an Fv. In yet another embodi variant, is joined to the antibody or immunoglobulin portion ment, the immunoglobulin moiety is an ScFv. In yet another via a polypeptide linker sequence that connects the C-termi embodiment, the immunoglobulin moiety is a minibody. In nus of the CD47 binding agent with the N-terminus of the yet another embodiment, the immunoglobulin moiety is a antibody or immunoglobulin portion of the fusion protein. diabody. In yet a further embodiment, the immunoglobulin 0156 The invention also contemplates that the CD47 moiety is a single-domain antibody (nanobody). binding agent portion of the fusion protein and the antibody or 0151. Any of the immunoglobulin moieties disclosed immunoglobulin portion of the fusion protein is a chemical herein may be linked to the CD47 binding agent at the N-ter linker. minus of the immunoglobulin moiety or at the C-terminus of the immunoglobulin moiety. In embodiments where the Tumor Cell Antigens immunoglobulin moiety is an intact antibody or includes both 0157. The immunoglobulin fusion proteins of the inven a heavy chain (or a portion of a heavy chain) and a light chain tion are designed to be specific for a tumor cell antigen in (or a portion of a light chain) the CD47 binding agent is addition to being specific for CD47. As described above, the preferably attached to the C-terminus of the heavy chain; immunoglobulin fusion proteins of the invention may achieve however, it is also contemplated that the CD47 binding agent the desired outcome of avoiding binding CD47 on healthy may be attached to the C-terminus of the light chain, or to the cells by being bispecific for both CD47 and a tumor cell N-terminus of the heavy chain, or to the N-terminus of the antigen. Accordingly, the immunoglobulin fusion proteins of light chain. In embodiments where the immunoglobulin moi the invention can be specific for any tumor antigen. Exem ety is an Fc oran Fcab, the CD47 binding agent is preferably plary tumor cell antigens include but are not limited to: attached at the N-terminus of the Fc portion, although in some 4-1 BB, 4F2, a-LEWISy, A2aR, AATK, ACKR, ACVR, embodiments, the CD47 binding agent is attached at the ADCYAP1R1, ADIPOR1, ADIPOR2, ADORA1, ADORA2, C-terminus of the Fc portion. In one embodiment, the CD47 ADORA3, ADR, AGTR, AHR, ALK, AMHR2, ANGPT1, binding agent is fused to the N terminus of the immunoglo ANGPT2, ANGPT4, APLNR, APRILR, AR, AVPR1A, bulin moiety. In another embodiment, the CD47 binding AVPR1B, AVPR2, AXL, B7.1, B7.2, B7-DC, B7-H1, B7-H2, agent is fused to the C terminus of the immunoglobulin moi B7-H3, B7-H4, B7RP1, BAFF, BAFFR, BAI1, BAI2, ety. BDKRB1, BDKRB2, BMPR1A, BMPR1B, BMPR2, BRD8, 0152 Antibodies known in the art that could be useful in BRS3, BTLA, C3AR1, CSAR1, CSAR2, CALCR, CASR, creating the immunoglobulin fusion proteins of the invention CCKAR, CCKBR, CCR1, CCR10, CCR2, CCR3, CCR4, include anti-EGFR antibodies such as Cetuximab, Panitu CCR5, CCR6, CCR7, CCR8, CCR9, CCRL1, CCRL2, CD2, mumab, , Matuzumab, Futuximab, Modotux CD3, CD4, CD5, CD6, CD7, CD11, CD15, CD18, CD19, imab, Imgatuzumab, Necitumumab; anti-CD20 antibodies CD20, CD22, CD24, CD25, CD27, CD28, CD30, CD33, such as , , , Ibritumo CD37, CD38, CD3E, CD40, CD40L, CD43, CD44, CD47, mab tiuxetan, , Ocrelizumab, CD52, CD54, CD55, CD56, CD66, CD70, CD73, CD74, I-131, , ; and anti-HER2 antibodies CD80, CD86, CD97, CD112, CD123, CD133, CD137, Such as , , ; and anti CD137L, CD152, CD154, CD155, CD161, CD163, CD166, PD-L1 antibodies such as , Avelumab, and Dur CD172, CD200, CD200R, CD206, CD244, CD300, CEA, valumab. In another embodiment, the invention contemplates CEACAM3, CELSR, CHRM1, CHRM2, CHRM3, CHRM4, modifications to the known heavy and/or light chain CHRM5, CIITA, CMKLR1, CNR1, CNR2, CNTFR, sequences of the aforementioned antibodies so long as those CRHR1, CRHR2, CRIM1, CRLF1, CRLF2, CRLF3, modified antibodies retain the unique heavy and light chain CSPG4, CSF1R, CSF2R, CSF3R, CTLA4, CX3CR1, complementarity determining regions or enough of the CXCR1, CXCR2, CXCR3, CXCR4, CXCR5, CXCR6, complementarity determining regions to retain binding speci CXCR7, CYSLTR1, CYSLTR2, Cripto, DARC, DDR1, ficity for the antigen of the non-modified antibody. DDR2, Digoxin, DII4, DRD1, DRD2, DRD3, DRD4, DRD5, DTR4, EDA2R, EDAR, ED-B, EDNRA, EDNRB, EGFR, Linker Sequences EGFRVIII, ELTD1, EMR1, EMR2, EMR3, EMR4P ENG, EPCAM, EPHR, Episialin, EPOR, ERBB2, ERBB3, 0153. The immunoglobulin fusion proteins of the inven ERBB4, ESR1, ESR2, ESRR, F2R, F4/80, FAS, FCER2, tion may include a linker sequence that joins the CD47 bind FCGR1, FDF, FFAR, FGFR1, FGFR2, FGFR3, FGFR4, ing agent portion of the fusion protein with the antibody or FGFRL1, Fibrin, FKBP, FLT1, FLT3, FLT4, FN14, FOLR1, immunoglobulin portion of the fusion protein. A preferred FPR1, FSHR, FZD1, FZD10, FZD2, FZD3, FZD4, FZD5, linker is a Gly-Ser flexible linker of variable length. For FZD6, FZD7, FZD8, FZD9, G-28, GABBR, GAL9, GALR. example, the linker sequence is (Gly Ser), where according GCGR, GD2, GD3, GDNFR, GHR, GHRHR, GHSR, GIPR, to various embodiments, n=1,2,3,4,5,6,7,8,9, or 10. In one GITR, GLP1R, GLP2R, GM3, GM-CSFR, GNRHR, embodiment, n=4. A Gly-Ser flexible linker of variable length GPBAR1, GPER, Gr-1, Hapten, HCAR, HCRTR, HER1, may be introduced to optimize the targeting and effector HER2, HER3, HER4, HLA-DR10, HLA-DRB, HLA-G, functions. HPV 16, HPVE6, HPVE7, HMGB1, HMW-MAA, HRH, 0154) In one embodiment, the CD47 binding agent, which HTR1, HTR2, HTR3, HTR4, HTR5, HTR6, HTR7, HIVE, in one embodiment may be a SIRPC. or a SIRPC. variant, is ICOS, IDO, IFNAR, IFNGR, IFNLR, IGF1R, IGF2R, joined to the antibody or immunoglobulin portion via a IL1OR, IL11R, IL12R, IL13R, IL15R, IL17R, IL18R1, polypeptide linker sequence that connects the N-terminus of IL18RAP, IL1R, IL1RL, IL2OR, IL21R, IL22R, IL23R, the CD47 binding agent with the C-terminus of the antibody IL27RA, IL28RA, IL2R, IL31RA, IL35R, IL3RA, IL4R, of immunoglobulin portion of the fusion protein. IL5RA, IL6R, IL7R, IL9R, ILT2, ILT3, ILT4, ILT5, INSR, US 2016/0177276 A1 Jun. 23, 2016

INSRR, IRAK, IRP-2, ITGA1, ITGA10, ITGA11, ITGA2, 0162 For example, using a technique Such as flow cytom ITGA2B, ITGA3, ITGA4, ITGA5, ITGA6, ITGA7, ITGA8, etry to identify the mean fluorescence intensity (MFI) of% ITGA9, ITGAD, ITGAE, ITGAL, ITGAM, ITGAV, ITGAX, red blood cell binding to a CD47 antibody as a control and ITGB1, ITGB2, ITGB3, ITGB4, ITGB5, ITGB6, ITGB7, setting that value at 100%, the relative % red blood cell ITGB8, KAR, KIR, KISS1R, KIT, L6-antigen, LAG3, binding MFI of the fusion protein can be measured. In one LAIR1, LEPR, Lewis Y. LGR4, LGR5, LGR6, LHCGR, embodiment, the immunoglobulin fusion protein of the LIFR, LIR, LMTK2, LMTK3, LPAR, LPHN, LTE4R, LTBR, invention has less than 35%, less than 30%, less than 25% less LTK, Lysozyme, MAGE-1, MAGE-3, MAS1, MAS1L, than 20%, less than 15%, less than 10%, less than 9%, less MC1R, MC2R, MC3R, MC4R, MC5R, MCHR, MERTK, than 8%, less 7%, less than 6%, less than 5%, less than 4%, mesothelin, MET, MFG-E8, MIR, MIG, MLNR, MPL, less than 3%, less than 2%, or less than 1% red blood cell MRGPRD, MRGPRE, MRGPRF, MRGPRG, MRGPRX1, binding MFI compared to the control antibody which has its MRGPRX2, MRGPRX3, MRGPRX4, MST1R, MTNR1, RBC MFI calibrated at 100%. In other embodiments, the 96 MUC1, MUC16, MUSK, NAIP, NCAM1, NGFR, NIP-cap, RBC MFI of the immunoglobulin fusion protein of the inven NKG2A, NKp46, NLRC, NLRP, NLRX1, NMBR, NMUR1, tion is between 0-1%, 0-2%, 0-3%, 0-4%, 0-5%, 0-6%, 0-7%, NMUR2, NOD1, NOD2, NPBWR1, NPBWR2, NPFFR, 0-8%, 0-9%, 0-10%, 0-15%, 0-20%, 0-25%, 0-30%, 0-35%, NPR1, NPR2, NPR3, NPSR1, NPY1R, NPY2R, NPY4R, 1-2%, 1-3%. 1-4%, 1-5%, 1-6%, 1-7%, 1-8%, 1-9%, 1-10%, NPY5R, NPY6R, NROB1, NROB2, NR1D1, NR1D2, 1-15%, 1-20%, 1-25%, 1-30%, 1-35%, 2-3%, 2-4%, 2-5%, NR1H2, NR1H3, NR1H4, NR1H5P, NR1I2, NR1 I3, 2-6%. 2-7%, 2-8%, 2-9%, 2-10%, 3-4%, 3-5%, 3-6%, 3-7%, NR2C1, NR2C2, NR2E1, NR2E3, NR2F1, NR2F2, NR2F6, 3-8%, 3-9%, 3-10%, 4-5%, 4-6%, 4-7%, 4-8%, 4-9%, 4-10%, NR3C1, NR3C2, NR4A1, NR4A2, NR4A3, NR5A1, 5-6%, 5-7%, 5-8%, 5-9%, 5-10%, 5-15%, 5-20%, 5-25%, NR5A2, NR6A1, NRP1, NRP2, NTRK, NTSR2, OGFR, 5-30%, 5-35%, 6-7%. 6-8%, 6-9%, 6-10%, 7-8%, 7-9%, OPR, OSMR, OX40, OX40L, OXER1, OXGR1, OXTR, 7-10%, 8-9%, 8-10%, 9-10%, 10-15%, 10-20%, 10-25%, P2RY, PD-1, PDGFR, PD-L1, PGR, PGRMC2, Phosphati 10-30%, or 10-35%, when the 96 RBC MFI of the anti-CD47 dylserine, PLAP, PLAUR, PLXN, PPAR, PRLHR, PRLR, antibody is calibrated at 100%. In yet another embodiment, PODXL, PROKR, PSCA, PSMA, PTAFR, PTGDR, PTGDS, the '% RBC MFI of the immunoglobulin fusion protein of the PTGER, PTGFR, PTGIR, PTH1R, PTH2R, PTPR, QRFPR, invention is 35% or less, 30% or less, 25% or less, 20% or less, RANK, RAR, RELT, RET, ROBO, ROR, ROS1, RXFP, 15% or less, 10% or less, 9% or less, 8% or less, 7% or less, RXR, RYK, S100A8, S100A9, S1PR, SCTR, SERPINB1, 6% or less, 5% or less, 4% or less, 3% or less, 2% or less, or Siglec-F, SDC, SLAM7, SMO, SORT1, SPOCK2, SSEA-1, 1% or less, when the % RBC MFI of the anti-CD47 antibody SSTR, ST2, STYK1, SUCNR1, TAAR, TACR, TAG72, is calibrated at 100%. In one embodiment, the anti-CD47 TBXA2R, TEK, TGFBR, THR, TIE1, TIGIT, TIM1, TIM2, antibody used as the control is B6H12/hulgG1 whose light TIM3, TIM4, TLR1, TLR10, TLR2, TLR3, TLR4, TLR5, and heavy chain amino acid sequences are found in Table 4 as TLR6, TLR7, TLR8, TLR9, TNC, TNFRSF11A, SEQ ID NO: 146 (light chain) and SEQ ID NO:148 (heavy TNFRSF11B, TNFRSF 13B, TNFRSF14, TNFRSF17, chain). The fusion protein also binds to CD47 on a non-red TNFRSF19, TNFRSF1A, TNFRSF1B, TNFRSF21, blood cell. The non-red blood cell, in one embodiment, is a TNFRSF25, TNFRSF6B, TPRA1, TRAIL-R1, TRAIL-R2, tumor cell. TRAIL-R3, TRAIL-R4, TRHR, TSHR, TUBB3, TWEAKR, 0163 The invention also contemplates methods of identi TYRO3, UTS2R, VCAM1, VDR, VEGFR, VEGFR2, fying fusion proteins of the invention based on their hemag VIPR1, VIPR2, VISTA, and XCR1. glutination profiles. An example of testing for hemagglutina 0158 Methods of preparing antibodies to a tumor cell tion of immunoglobulin fusion proteins of the invention to antigen are known in the art and include hybridoma based erythrocytes is provided in Example 18 below. methods, phage display based methods, and yeast display. 0159 Further, it is possible to create new antibodies to any Use of Fusion Proteins tumor cell antigen according to known methods and use the antibodies generated thereby to create immunoglobulin 0164. The fusion proteins disclosed herein can be used to fusion proteins of the invention. treat various forms of cancer. A non-limiting of list of cancers for which the immunoglobulin fusion proteins of the inven Properties of Immunoglobulin Fusion Proteins tion may be used to treat include Adrenal Cancer, Anal Can cer, Bile Duct Cancer, Bladder Cancer, Bone Cancer, Brain/ 0160. As described herein, it is desirable that the immu CNS, Basal Cell Skin Cancer, Breast Cancer, Cancer of noglobulin fusion proteins of the invention, while having the Unknown Primary, Castleman Disease, Cervical Cancer, ability to bind to CD47 on disease producing cells, such as Colorectal Cancer, Endometrial Cancer, Esophagus Cancer, tumor cells, do not bind, or at least bind at acceptably low Dermatofibrosarcoma Protuberans, Ewing Family Of levels to CD47 on normal cells and, in particular, red blood Tumors, Eye Cancer, Gallbladder Cancer, Gastrointestinal cells (erythrocytes), such that any levels of CD47 binding by Carcinoid Tumors, Gastrointestinal Stromal Tumor (GIST), the immunoglobulin fusion protein remain at acceptable lev Gastric Cancer, Gestational Trophoblastic Disease, Glioma, els for therapeutic use of the fusion protein. Glioblastoma, Head and Neck Cancer, Hodgkin Disease, 0161 Accordingly, the invention contemplates that immu Kaposi Sarcoma, Kidney Cancer, Laryngeal and Hypopha noglobulin fusion proteins of the invention may be character ryngeal Cancer, Leukemia, Acute Lymphocytic Leukemia ized by their relative binding to red blood cells (erythrocytes) (ALL), Acute Myeloid Leukemia (AML), Chronic Lympho as compared to binding of a control to red blood cells. An cytic Leukemia (CLL), Chronic Myeloid Leukemia, Chronic example of testing the relative binding of immunoglobulin Myelomonocytic Leukemia (CMML), Liver Cancer, Lung fusion proteins of the invention to erythrocytes is provided in Cancer. Non-Small Cell Lung Cancer, Small Cell Lung Can Example 17 below. cer, Lung US 2016/0177276 A1 Jun. 23, 2016

0.165 Carcinoid Tumor, Liver Cancer, Lymphoma, Lym sion media, coatings, isotonic and absorption delaying phoma of the Skin, Malignant Mesothelioma, Merkel Cell agents, and the like, that are compatible with pharmaceutical Carcinoma, Melanoma, Multiple Myeloma, Myeloma, administration. The use of Such media and agents for phar Myelodysplastic Syndrome, Nasal Cavity and Paranasal maceutically active Substances is known in the art. Sinus Cancer, Nasopharyngeal Cancer, Neuroendocrine Can 0171 Pharmaceutical compositions containing fusion cer, Neuroblastoma, Non-Hodgkin Lymphoma, Oral Cavity proteins, such as those disclosed herein, can be presented in a and Oropharyngeal Cancer, Osteosarcoma, Ovarian Cancer, dosage unit form and can be prepared by any Suitable method. Pancreatic Cancer, Penile Cancer, Pituitary Tumors, Prostate A pharmaceutical composition should be formulated to be Cancer, Renal Cancer, Retinoblastoma, Rhabdomyosar compatible with its intended route of administration. coma, Salivary Gland Cancer, Sarcoma, Sarcoma—Adult Examples of routes of administration are intravenous (IV), Soft Tissue Cancer, Squamous Cell Skin Cancer, Small Intes intradermal, inhalation, transdermal, topical, transmucosal, tine Cancer, Stomach Cancer, Testicular Cancer, Thymus and rectal administration. The pharmaceutical compositions Cancer, Thyroid Cancer, Uterine Cancer, Uterine Sarcoma, are intended for parenteral, intranasal, topical, oral, or local Vaginal Cancer, Vulvar Cancer, Waldenstrom Macroglobu administration, such as by a transdermal means, for therapeu linemia, and Wilms Tumor. tic treatment. The pharmaceutical compositions can be 0166 The cancer cells are exposed to a therapeutically administered parenterally (e.g., by intravenous, intramuscu effective amount of the fusion protein so as to inhibit prolif lar, or Subcutaneous injection), or by oral ingestion, or by eration of the cancer cell. In some embodiments, the fusion topical application or intraarticular injection at areas affected proteins inhibit cancer cell proliferation by at least 40%, 50%, by the vascular or cancer condition. Additional routes of 60%, 70%, 80%, 90%, 95%, 98%, 99%, or 100%. administration include intravascular, intra-arterial, intratu 0167. In some embodiments, the fusion protein is used in mor, intraperitoneal, intraventricular, intraepidural, as well as therapy. For example, the fusion protein can be used to inhibit nasal, ophthalmic, intrascleral, intraorbital, rectal, topical, or tumor growth in a mammal (e.g., a human patient). In some aerosol inhalation administration. embodiments, use of the fusion protein to inhibit tumor 0172. The invention provides compositions for parenteral growth in a mammal includes administering to the mammala administration that comprise the above mention agents dis therapeutically effective amount of the fusion protein. In Solved or Suspended in an acceptable carrier, preferably an other embodiments, the fusion protein can be used for inhib aqueous carrier, e.g., water, buffered water, saline, PBS, and iting proliferation of a tumor cell. the like. The compositions may contain pharmaceutically (0168 As used herein, “treat,” “treating,” and “treatment” acceptable auxiliary substances as required to approximate mean the treatment of a disease in a mammal, e.g., in a human. physiological conditions, such as pH adjusting and buffering This includes: (a) inhibiting the disease, i.e., arresting its agents, tonicity adjusting agents, wetting agents, detergents development; and (b) relieving the disease, i.e., causing and the like. The invention also provides compositions for regression of the disease state. oral delivery, which may contain inert ingredients such as 0169 Generally, a therapeutically effective amount of binders or fillers for the formulation of a tablet, a capsule, and active component is in the range of 0.1 mg/kg to 100 mg/kg, the like. Furthermore, this invention provides compositions e.g., 1 mg/kg to 100 mg/kg, e.g., 1 mg/kg to 10 mg/kg. The for local administration, which may contain inert ingredients amount administered will depend on variables such as the Such as solvents or emulsifiers for the formulation of a cream, type and extent of disease or indication to be treated, the an ointment, and the like. overall health of the patient, the in vivo potency of the anti 0173 A preferred route of administration for fusion pro body, the pharmaceutical formulation, and the route of teins is IV infusion. Useful formulations can be prepared by administration. The initial dosage can be increased beyond methods well known in the pharmaceutical art. For example, the upper level in order to rapidly achieve the desired blood see Remington's Pharmaceutical Sciences, 18th ed. (Mack level or tissue level. Alternatively, the initial dosage can be Publishing Company, 1990). Formulation components suit Smaller than the optimum, and the dosage may be progres able for parenteral administration include a sterile diluent sively increased during the course of treatment. The optimal Such as water for injection, Saline Solution, fixed oils, poly dose can be determined by routine experimentation. For ethylene glycols, glycerine, propylene glycol or other syn parenteral administration a dose between 0.1 mg/kg and 100 thetic Solvents; antibacterial agents such as benzyl alcohol or mg/kg, alternatively between 0.5 mg/kg and 50 mg/kg, alter methyl paraben; antioxidants such as ascorbic acid or sodium natively, between 1 mg/kg and 25 mg/kg, alternatively bisulfite; chelating agents such as EDTA; buffers such as between 2 mg/kg and 10 mg/kg, alternatively between 5 acetates, citrates orphosphates; and agents for the adjustment mg/kg and 10 mg/kg is administered and may be given, for oftonicity Such as Sodium chloride or dextrose. example, once weekly, once every other week, once every 0.174 For intravenous administration, suitable carriers third week, or once monthly per treatment cycle. include physiological saline, bacteriostatic water, Cremophor 0170 For therapeutic use, a fusion protein of the invention ELTM (BASF, Parsippany, N.J.) orphosphate buffered saline is preferably combined with a pharmaceutically acceptable (PBS). The carrier should be stable under the conditions of carrier. As used herein, “pharmaceutically acceptable carrier' manufacture and storage, and should be preserved against means buffers, carriers, and excipients suitable for use in microorganisms. The carrier can be a solvent or dispersion contact with the tissues of human beings and animals without medium containing, for example, water, ethanol, polyol (for excessive toxicity, irritation, allergic response, or other prob example, glycerol, propylene glycol, and liquid polyethylene lem or complication, commensurate with a reasonable ben glycol), and Suitable mixtures thereof. efit/risk ratio. The carrier(s) should be “acceptable' in the 0.175 Pharmaceutical formulations preferably are sterile. sense of being compatible with the other ingredients of the Sterilization can be accomplished, for example, by filtration formulations and not deleterious to the recipient. Pharmaceu through sterile filtration membranes. Where the composition tically acceptable carriers include buffers, solvents, disper is lyophilized, filter sterilization can be conducted prior to or US 2016/0177276 A1 Jun. 23, 2016

following lyophilization and reconstitution. Aqueous solu mL, i.e., a 40% decrease, by Day 5 (FIG. 2A), together with tions may be packaged for use as-is, or lyophilized, the lyo a corresponding reduction in the hematocrit level (FIG. 2B). philized preparation being combined with a sterile aqueous Therefore, treatment with anti-CD47 antibodies can lead to carrier prior to administration. The pH of the preparations severe anemia. typically will be between 3 and 11, more preferably between 5 and 9 or between 6 and 8, and most preferably between 7 Example 2 and 8, such as 7 to 7.5. The resulting compositions in solid form may be packaged in multiple single dose units, each Anti-CD20-hulgG1-SIRPC. Immunoglobulin Fusion containing a fixed amount of the above-mentioned agent or Protein agents, such as in a sealed package of tablets or capsules. 0179 20A) Construction and Expression of anti-CD20 huIgG1-SIRPo. Fusion Protein Production 0180. The generation of an exemplary anti-CD20 0176 The immunoglobulin fusion proteins of the inven hulgG1-SIRPC. is based on the anti-CD20 2B8 (rituximab) tion are generally produced recombinantly, using mammalian monoclonal antibody (Reffetal, Blood83:435, 1994) and the cells containing a nucleic acid or nucleic acids engineered to SIRPO. protein (Jiang et al. JBC 274: 559, 1999). The DNA express the fusion protein. If an immunoglobulin fusion pro and protein sequence of the Fab light chain for 2B8 are tein of the invention requires two or more polypeptide chains provided in SEQID NO: 1 and SEQID NO:2, respectively. to be expressed in order for assembly of the fusion protein, The DNA and protein sequence of the Fab heavy chain for then the invention contemplates that nucleic acids encoding 2B8 are provided in SEQID NO:3 and SEQID NO:4, respec each of the polypeptide chains be contained within a cell or tively. The DNA and protein sequence of SIRPC. allele V1 are cells in order to facilitate recombinant production of the provided in SEQID NO: 5 and SEQID NO:6, respectively. immunoglobulin fusion protein. For example, in one embodi The DNA and protein sequence of the IgV domain of SIRPC. ment, a nucleic acid encodes a heavy chain orportion thereof allele V2 are provided in SEQID NO: 7 and SEQID NO:8, of an immunoglobulin fusion protein of the invention while respectively. Anti-CD20-hulgG1-SIRPCV2 was generated another nucleic acid encodes a light chain or portion thereof by linking the C-terminus of the anti-CD20 heavy chain of an immunoglobulin fusion protein of the invention. Either polypeptide to the IgV domain of SIRPCV2 via a (G4S). the nucleic acid encoding the heavy chain or portion thereof linker. or the nucleic acid encoding the light chain orportion thereof 0181 For expression of anti-CD20-huIgG1-SIRPCV2, also includes a nucleic acid sequence encoding the CD47 the following two gene constructs were assembled by stan binding moiety, for example, a SIRPC. moiety, as described dard recombinant DNA techniques and cloned into the mam herein. In a further embodiment, a cell contains a nucleic acid malian expression vector pTT5 (containing the mouse light encoding a heavy chain or portion thereof of an immunoglo chain signal peptide sequence for secretion) as in FIG. 1A: (1) bulin fusion protein of the invention and contains a nucleic Construct VH(anti-CD20)-CH1-H-CH2-CH3-(G4S)-SIR acid encoding a light chain or portion thereof of an immuno PCV2 (SEQ ID NO:11) encoding the following elements: globulin fusion protein of the invention. Either the nucleic anti-CD20 heavy chain variable domain followed by human acid encoding the heavy chain or portion thereof or the heavy chain constant domains 1-3 isotype IgG1 followed by nucleic acid encoding the light chain or portion thereof of an a (G4S) linker and SIRPOV2 and (2) Construct VL(anti immunoglobulin fusion protein of the invention also contains CD20)-CL (SEQ ID NO:1.) encoding the anti-CD20 light a nucleic acid encoding a CD47 binding moiety, Such as a chain variable domain followed by human kappa light chain SIRPC. moiety as described herein. constant domain. The corresponding amino acid sequences 0177 Although exemplary methods of fusion protein for these two exemplary constructs are shown in SEQ ID expression and production are described in, for example, NO:12 and SEQID NO:2, respectively. Examples 2 and 4 below, a wide variety of suitable vectors, 0182. The set of two vectors for anti-CD20-hulgG1-SIR cell lines and protein production methods have been used to PCV2 expression was co-transfected transiently into Expi293 produce biopharmaceuticals and could be used in the synthe cells using Expi293fectin (Life Technologies, Grand Island, sis of the fusion proteins of the invention. Such methods are N.Y.). The protein was purified in a single step by protein A within the knowledge of the skilled artisan. affinity chromatography. Expression of the two polypeptides and assembly of the full tetrameric molecule were confirmed EXAMPLES on SDS-PAGE and SEC. For SDS-PAGE, the purified protein samples were reduced with DTT and run on NuPAGE MES Example 1 4-12% Gel, 200V for 35 min, followed by Coomassie stain Effect of anti-CD47 B6H12 on Red Blood Cells in ing. The two major bands on the gel had the expected MW and Cynomolgus Monkeys the correct stoichiometirc ratio with >95% purity (FIG. 3A). In FIG.3A, lane 1 shows the molecular weight (MW) marker 0.178 The in vivo effect of anti-CD47 B6H12 monoclonal and lane 2 shows the expected MW (63. 23 kDa) and the antibody (chimeric B6H12-human IgG4 (Lindberg et al. JBC correct stoichiometric ratio (1:1) of the two polypeptides of 269: 1567, 1994)) on red blood cells (RBC) was evaluated in anti-CD20-hulgG1-SIRPOV2. For SEC, the purified protein cynomolgus monkeys. A group of 3 monkeys received a samples were analyzed on a TSK-GEL Super SW3000 SEC single intravenous dose of B6H12 each at 12 mg/kg on Day 0. column 4.6x300mm (Tosoh Biosciences, Tokyo, Japan) that Blood samples were withdrawn on-10 (ten days before injec was equilibrated with 50 mM sodium phosphate, 400 mM tion to obtain baseline level), 0, 1, 3, 5 and 7 days after the sodium perchlorate, pH 6.3+0.1 and 38+2.0 mS/cm. Size single intravenous dose for RBC count and hematocrit (HCT) exclusion chromatography showed a peak at the expected determination. FIGS. 2A-B shows that there was a strong MW of about 172 kDa for the monomeric anti-CD20 reduction of red blood cells from 5.8x10 to 3.7x10 RBC/ huIgG1-SIRPoV2 (FIG. 3B). US 2016/0177276 A1 Jun. 23, 2016

0183. In addition, anti-CD20 and anti-CD47 in a standard (0189 The results show that anti-CD20-hulgG1-SIRPOV2 monoclonal antibody format (anti-CD20 hulgG1 and anti binding to both Raji and Namalwa cells was somewhat CD47 hulgG1) and SIRPC. in a Fc-fusion protein format enhanced compared to control molecules (FIGS. 4C-D), Sug (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) were generated gesting an avidity effect. Because fluorophore labeling abol as controls to compare with the anti-CD20-hulgG1-SIRPC. ished binding of SIRPC. to CD47, binding was assayed by format. anti-Fc detection. However, it has previously been found that 2(B)(i) Binding of anti-CD20-hulgG1-SIRPC. to CD47 anti-Fc binding to an Fc region that has an additional protein Expressed on Cells moiety attached to its C-terminus is diminished compared to 0184 The ability of anti-CD20-hulgG1 SIRPCV2 to bind the same Fc region without an additional moiety in a cell to CD47 overexpressed on the cell surface was measured, and binding assay (data not shown), Suggesting, without wishing compared to the control molecules. 2x10 Chinese hamster to be bound by theory, that anti-Fc binding to the Fc fusion ovary (CHO) cells transfected with CD47 per well were incu protein may be sterically hindered in that context. Thus, the bated with varying concentrations of proteins diluted in PBS+ observation of similar binding of anti-CD20-hulgG1-SIR 1% FBS in a 96 well plate for 60 min on ice. After washing PCV2 and anti-CD20 by anti-Fc likely underestimates the with PBS+1% FBS, cells were incubated with FITC F(ab')2 amount of anti-CD20-hulgG1-SIRPCV2 cell binding and goat Anti-Human IgG, Fcy (Jackson ImmunoResearch, West indeed suggests an avidity effect for anti-CD20-hulgG1-SIR Grove, Pa.) diluted 1:200 in PBS+1% FBS for 60 minonice. After washing again, cells were fixed with 1% formaldehyde PCV2 binding to the cells. The ability of anti-CD20-hulgG1 in PBS. Cells were analyzed by flow cytometry (MAC SIRPCV2 to harness the avidity of binding to the tumor cells SQuant, Miltenyi Biotec, Cologne, Germany). by binding to two tumor targets on the same cell may result in 0185. The results show that anti-CD20-hulgG1-SIR more specific targeting and less side effects in vivo. PCV2, anti-CD47, SIRPOV2-Fc and Fc-SIRPOV2 bound to CD47 expressed on CD47-tranfected CHO cells, but anti Example 3 CD20 did not bind because CD20 is not expressed (FIG. 4A). Fc-SIRPCV2 and anti-CD47 bound with similar EC50's (7-8 Anti-CD20/anti-CD47 Bispecific Antibody nM), but SIRPOV2-Fcbound better (EC50=3 nM). However, (0190 3(A) Description of anti-CD20/anti-CD47 the median fluorescence was highest for anti-CD47, followed 0191 The generation of an exemplary tetravalent bispe by SIRPOV2-Fc, and lowest for Fc-SIRPCV2 and anti cific antibody (TetBiAb) against CD20 and CD47 is based on CD20-hulgG1-SIRPoV2. the anti-CD202B8 (rituximab) monoclonal antibody (Reffet 0186. The ability of anti-CD20-huIgG1-SIRPCV2 to bind al, Blood 83:435, 1994) and the anti-CD47 B6H12 mono to CD47 expressed on the cell surface of blood cells was clonal antibody (Lindberget al. JBC 269: 1567, 1994). In the measured, and compared to the control molecules. Fresh anti-CD20/anti-CD47 TetBiAb against CD20 and CD47, the whole blood from healthy human donors was enriched for C-terminus of the anti-CD20 heavy chain polypeptide is leukocytes with dextran precipitation and was washed with PBS+1% FBS. 2x10 leukocyte-enriched human whole linked to the N-terminus of the anti-CD47 Fab light chain via blood cells per well were incubated with 50 lug/ml proteins a G4S linker (FIG. 1D+E). The construction, expression, and diluted in PBS+1% FBS in a 96 well plate for 60 min on ice. binding properties of anti-CD20/anti-CD47 are described in After washing with PBS+1% FBS, cells were incubated with International Patent Application Publication No. WO2014/ a 1:200 dilution of FITC F(ab')2 goat Anti-Human IgG, Fcy 144357. (Jackson ImmunoResearch, West Grove, Pa.), a 1:100 dilu 3(B) In Vivo Biological Activities of anti-CD20/anti-CD47 tion of PE mouse anti-human CD235a (BD Biosciences, San (0192. The utility of anti-CD20/anti-CD47, as a proxy for Jose, Calif.), and a 1:100 dilution of eFluor 450 mouse anti anti-CD20-hulgG1-SIRPO, was shown by the following in human CD45 (eBioscience, San Diego, Calif.) in PBS+1% vivo experiments. In disseminated lymphoma models, SCID FBS for 60 minonice for protein detection and cell sorting by mice were injected i.v. either with 5x10° CD20+human flow cytometry. After washing again, cells were fixed with 1% Daudi lymphoma cells or with 1x10°CD20+human Rajilym formaldehyde in PBS. Cells were analyzed by flow cytometry phoma cells, followed by i.p. injection of 25 ug/mouse of an (MACSQuant, Miltenyi Biotec, Cologne, Germany). antibody isotype control, 25 ug/mouse of anti-CD20, 25 0187. The results show that anti-CD47 bound to CD47 ug/mouse of anti-CD47, or 42 g/mouse of anti-CD20/anti expressed on erythrocytes and leukocytes, but anti-CD20 CD47, which is the equimolar amount of tetravalent bispe hulgG1-SIRPCV2-Fc and Fc-SIRPCV2 only bound to CD47 cific antibody. All the groups (n=10-11) received treatment expressed on leukocytes and not to CD47 expressed on eryth twice per week for 3 weeks, and results were reported as rocytes (FIG. 4B). general health, e.g., paralysis, which precedes death by 10-14 2(B)(ii) Demonstration of Avidity of anti-CD20-hulgG1 days, and survival of mice. Treatment with anti-CD20/anti SIRPC. by Binding Both Antigens Expressed on Cells CD47 tetravalent bispecificantibody was found to be superior 0188 Binding of anti-CD20-huIgG1-SIRPCV2 to CD20 to the two monotherapies in both Daudi (FIG. 5A) and Raji and CD47 on the cell surface was measured on human Ramos (FIG. 5B) tumor models. Similar results are expected for an B cell lymphoma cells that overexpress CD20 and express anti-CD20-hulgG1-SIRPO. CD47, and human Namalwa B cell lymphoma cells that over express CD47 and express CD20, 2x10 Raji or Namalwa Example 4 cells per well were incubated with varying concentrations of proteins diluted in PBS+1% FBS in a 96 well plate for 60 min Anti-EGFR-hulgG1-SIRPC. Immunoglobulin Fusion on ice. After washing with PBS+1% FBS, cells were incu Protein bated with FITC F(ab')2 goat Anti-Human IgG, Fcy (Jackson ImmunoResearch, West Grove, Pa.) diluted 1:200 in PBS+ 0193 4(A) Construction and Expression of anti-EGFR 1% FBS for 60 min on ice. After washing again, cells were huIgG1-SIRPo. analyzed by flow cytometry (MACSQuant, Miltenyi Biotec, 0194 The generation of an exemplary anti-EGFR Cologne, Germany). hulgG1-SIRPC. is based on the anti-EGFRC225 (cetuximab) US 2016/0177276 A1 Jun. 23, 2016

monoclonal antibody (Kawamoto, PNAS80:1337, 1983) and were confirmed on Sodium dodecyl Sulfate-polyacrylamide the SIRPC. protein (Jiang et al. JBC 274: 559, 1999). The gel electrophoresis (SDS-PAGE) and size exclusion chroma DNA and protein sequence of the Fab light chain for C225 are tography (SEC). For SDS-PAGE, the purified protein provided in SEQID NO:13 and SEQID NO:14, respectively. samples were reduced with DTT and run on NuPAGE MES The DNA and protein sequence of the Fab heavy chain for 4-12% Gel, 200V for 35 min, followed by Coomassie stain C225 are provided in SEQ ID NO:15 and SEQ ID NO:16, ing. The two major bands on the gel had the expected molecu respectively. The DNA and protein sequence of SIRPC. allele lar weights (MW) and the correct stoichiometric ratio with V1 are provided in SEQID NO:5 and SEQID NO:6, respec >95% purity (FIG. 6A). In FIG. 6A, lane 1 shows the molecu tively. The DNA and protein sequence of the IgV domain of lar weight (MW) marker, lane 2 shows the expected MW (64. SIRPC. allele V2 are provided in SEQID NO: 7 and SEQID 23 kDa) and the correct stoichiometric ratio (1:1) of the two NO:8, respectively. An exemplary anti-EGFR-hulgG1-SIR polypeptides of anti-EGFR-hulgG1-SIRPC.V1, lane 3 shows PC.V1 was generated by linking the C-terminus of the anti the expected MW (64. 23 kDa) and the correct stoichiometric EGFR heavy chain polypeptide to the IgV domain of SIR ratio (1:1) of the two polypeptides of anti-EGFR-hulgG1 PC.V1 via a (G4S) linker An exemplary anti-EGFR-hulgG1 (G4S)-SIRPOV2, lane 4 shows the expected MW (64, 23 SIRPCV2 was generated by linking the C-terminus of the kDa) and the correct stoichiometric ratio (1:1) of the two anti-EGFR heavy chain polypeptide to the IgV domain of polypeptides of anti-EGFR-hulgG1-(G4S)-SIRPCV2, and SIRPCV2 via a (G4S), (G4S), or (G4S) linker. lane 5 shows the expected MW (64. 23 kDa) and the correct (0195 For expression of the anti-EGFR-hulgG1-SIR stoichiometric ratio (1:1) of the two polypeptides of anti PC.V1, the following two gene constructs were assembled by EGFR-huIgG1-(G4S)s-SIRPCV2. For SEC, the purified pro standard recombinant DNA techniques and are cloned into tein samples were analyzed on a TSK-GEL Super SW3000 the mammalian expression vector pTT5 (containing the SEC column 4.6 300 mm (Tosoh Biosciences, Tokyo, Japan) mouse light chain signal peptide sequence for secretion) as in that was equilibrated with 50 mM sodium phosphate, 400 FIG. 1A: (1) Construct VH(anti-EGFR)-CH1-H-CH2-CH3 mM sodium perchlorate, pH 6.3+0.1 and 38+2.0 mS/cm2. (G4S)-SIRPOV1 (SEQ ID NO:17) encoding the following Size exclusion chromatography showed a peak at the elements: anti-EGFR heavy chain variable domain followed expected MW of about 173 kDa for each of the monomeric by human heavy chain constant domains 1-3 isotype IgG1 anti-EGFR-hulgG1-SIRPO. proteins (FIG. 6B(i)-(iv)). followed by a (G4S) linker and the IgV domain of SIRPOV1 0.198. In addition, anti-EGFR and anti-CD47 in a standard and (2) Construct VL(anti-EGFR)-CL (SEQ ID NO:13) monoclonal antibody format (anti-EGFR huIgG1 and anti encoding the anti-EGFR light chain variable domain fol CD47 hulgG1) and SIRPC. in an Fc-fusion protein format lowed by human kappa light chain constant domain. The (SIRPOV2-Fc, Fc-SIRPCV2, and Fc-SIRPoV2CC) (FIG. corresponding amino acid sequences for these two constructs 1C) were generated as controls to compare with the anti are shown in SEQID NO:18 and SEQID NO:14 respectively. EGFR-huIgG1-SIRPC. format. Fc-SIRPCV2CC is an Fc-fu (0196. For expression of the anti-EGFR-hulgG1-SIR sion protein in which the SIRPC. ECD moiety has the IgV PCV2 with varying linker lengths, the following four gene domain of allele V2 connected to the IgC domains of allele VI constructs were assembled by standard recombinant DNA (SEQ ID NO:192). techniques and cloned into the mammalian expression vector 4.(B)(i) Binding of anti-EGFR-huIgG1-SIRPC. to CD47 pTT5 (containing the mouse light chain signal peptide Expressed on Cells sequence for secretion) as in FIG. 1A: (1) ConstructVH(anti (0199 The ability of the exemplary anti-EGFR-hulgG1 EGFR)-CH1-H-CH2-CH3-(G4S)3-SIRPCV2 (SEQ ID SIRPC. to bind to CD47 overexpressed on the cell surface was NO:69), encoding the following elements: anti-EGFR heavy measured, and compared to the control molecules. 2x10 chain variable domain followed by human heavy chain con CHO cells transfected with CD47 per well were incubated stant domains 1-3 isotype IgG1 followed by a (G4S)3 linker with varying concentrations of antibodies diluted in PBS+1% and the IgV domain of SIRPCV2, (2) Construct VH(anti FBS in a 96 well plate for 60 min on ice. After washing with EGFR)-CH1-H-CH2-CH3-(G4S)4-SIRPCV2 (SEQ ID PBS+1% FBS, cells were incubated with FITC F(ab')2 goat NO:19), encoding the following elements: anti-EGFR heavy Anti-Human IgG, Fcy (Jackson ImmunoResearch, West chain variable domain followed by human heavy chain con Grove, Pa.) diluted 1:200 in PBS+1% FBS for 60 min on ice. stant domains 1-3 isotype IgG1 followed by a (G4S)4 linker After washing again, cells were fixed with 1% formaldehyde and the IgV domain of SIRPCV2, (3) Construct VH(anti in PBS. Cells were analyzed by flow cytometry (MAC EGFR)-CH1-H-CH2-CH3-(G4S)5-SIRPCV2 (SEQ ID SQuant, Miltenyi Biotec, Cologne, Germany). NO:71), encoding the following elements: anti-EGFR heavy (0200. The results show that the anti-EGFR-hulgG1 chain variable domain followed by human heavy chain con SIRPO. proteins, anti-CD47, SIRPO-FcV2 and Fc-SIRPCV2 stant domains 1-3 isotype IgG1 followed by a (G4S)5 linker bound to CD47 expressed on CD47-transfected CHO cells, and the IgV domain of SIRPCV2, and (4) ConstructVL(anti but anti-EGFR did not bind because EGFR is not expressed EGFR)-CL (SEQ ID NO:13), encoding the following ele (FIG. 7A). Anti-EGFR-huIgG1-hulgG1-huSIRPOV2 bound ments: anti-EGFR light chain variable domain followed by to CD47 expressed on CHO cells better than anti-EGFR human kappa light chain constant domain. The corresponding huIgG1-huSIRPOV1. The length of the linker in between Fc amino acid sequences for these four constructs are shown in and SIRPC. did not change the affinity of SIRPC. for CD47 SEQ ID NO:70, 20, 72 and SEQID NO:14 respectively. expressed on cells. It was also found that SIRPO-FcV2 bound (0197) Each set of two vectors for anti-EGFR-hulgG1 to CD47 expressed on CD47-transfected CHO cells better SIRPCV2 expression was co-transfected transiently into than Fc-SIRPOV2, which bound better than Fc-SIRPOV2CC Expi293 cells using Expi293fectin (Life Technologies, Grand (data not shown). Island, N.Y.). The proteins were purified in a single step by 0201 The ability of the exemplary anti-EGFR-hulgG1 protein A affinity chromatography. Expression of the two SIRPC. to bind to CD47 expressed on the cell surface of blood polypeptides and assembly of the full tetrameric molecules cells was measured, and compared to the control molecules. US 2016/0177276 A1 Jun. 23, 2016

Fresh whole blood from healthy human donors was enriched mouse, or approximately 1 or 10 mg/kg). On the dosing day, for leukocytes with dextran precipitation and was washed 36 mice were randomly assigned to 6 groups (N=6/group), in with PBS+1% FBS. 2x10 leukocyte-enriched human whole which each group received one dose? one molecule, intrave blood cells per well were incubated with 50 lug/ml proteins nously via mouse tail vein, respectively. Mouse body weight diluted in PBS+1% FBS in a 96 well plate for 60 min on ice. was recorded. Mice received the same dose/article (N=6) After washing with PBS+1% FBS, cells were incubated with were further divided into two subgroup (n=3). Four time a 1:200 dilution of FITC F(ab')2 goat Anti-Human IgG, Fcy blood withdrawals were taken from each Subgroup, i.e., one (Jackson ImmunoResearch, West Grove, Pa), a 1:100 dilution subgroup at 1 h, 24h, 72 hand 168 h, whereas another at 7 hr, of PE mouse anti-human CD235a (BDBiosciences, San Jose, 48 hr, 120 hr and 240 hr. At the indicated time points, small Calif.), and a 1:100 dilution of eFluor 450 mouse anti-human blood samples were taken using a heparinized micro glass CD45 (eBioscience, San Diego, Calif.) in PBS+1% FBS for capillary and collected in tubes coated with heparinto prevent 60 min on ice for protein detection and cell sorting by flow clotting. After centrifugation to remove the cells, the concen cytometry. After washing again, cells were fixed with 1% tration of the proteins in plasma was determined by ELISA, formaldehyde in PBS. Cells were analyzed by flow cytometry assayed by capture with anti-human IgG H+L (Jackson (MACSQuant, Miltenyi Biotec, Cologne, Germany). Immunoresearch, West Grove, Pa.), followed by detection 0202 The results show that anti-CD47 bound to CD47 with peroxidase-conjugated anti-human IgG Fc (Jackson expressed on erythrocytes and leukocytes, but anti-EGFR Immunoresearch, West Grove, Pa.) to detect anti-EGFR hulgG1-SIRPCV2, SIRPO-Fc and Fc-SIRPC. only bound to huIgG1-SIRPo. CD47 expressed on leukocytes and not to CD47 expressed on (0206. The results show that anti-EGFR-hulgG1-SIR erythrocytes (FIG. 79). PCV2 was cleared similar to an antibody (FIG.9A). After an 4.(B) (ii) Demonstration of Avidity of anti-EGFR-hulgG1 initial 2-fold drop in the distribution phase at the first time SIRPC. by Binding Both Antigens Expressed on Cells point, the plasma concentrations declined linearly according 0203 The ability of the exemplary anti-EGFR-hulgG1 to a circulating half-life of about 8 days. Exposure was dose SIRPC. to bind with avidity to EGFR and CD47 on the cell dependent with an AUC after 10 days of 2059 hug/ml for the Surface was measured on human A549 epidermoid carcinoma 20 ug dose and 13164 hug/ml for the 200 ug dose. Clearance cells that overexpress EGFR and express CD47. 2x10 A549 was dose dependent with 0.49 mL/h.kg for the 20 ug dose and cells per well were incubated with varying concentrations of 0.76 mL/h.kg for the 200 ug dose. antibodies diluted in PBS+1% FBS in a 96 well plate for 60 4(D)(i) In Vitro Biological Activities of anti-EGFR-hulgG1 min on ice. After washing with PBS+1% FBS, cells were SIRPO incubated with FITC F(ab')2 goat Anti-Human IgG, Fcy 0207. The in vitro biological activity of anti-EGFR (Jackson ImmunoResearch, West Grove, Pa.) diluted 1:200 in hulgG1-SIRPC. is shown in an antibody-dependent cell-me PBS+1% FBS for 60 minonice. Cells were analyzed by flow diated cytotoxicity (ADCC) assay, 6x10' human A549 epi cytometry (MACSQuant, Miltenyi Biotec, Cologne, Ger dermoid carcinoma cells were transferred to each well of a many). 96-well plate and incubated overnight at 37° C. The media 0204 The results show that the binding of anti-EGFR from the cells was removed and replaced with serial dilutions hulgG1-SIRPCV2 to A549 cells was similar to that of anti of the recombinant antibodies for concentrations between EGFR (FIG. 7C). Because anti-Fc detection likely under 0.02-1600 ng/ml. After a 15-30 min incubation at 37° C., estimates the binding to cells of Fc fusion proteins having an 1.5x10 effector cells (engineered Jurkat cells stably express additional C-terminal moiety (as described for Example ing the FcyRIIIa receptor, V158 (high affinity) variant, and an 2B(ii)), the observation of similar binding of anti-EGFR NFAT response element driving expression of firefly hulgG1-SIRPC. and anti-EGFR suggests that avidity effects luciferase (Promega Madison, Wis.)) were added to each well for anti-EGFR-hulgG1-SIRPC. binding to the cells. It is note of plates containing antibodies and A549 cells (effector-to worthy that in contrast to the binding to CD47 overexpressed target cells ratio 2.5:1). After a 24-hour incubation, ADCC on CHO cells, the binding of anti-CD47 to the endogenous activity was measured via luciferase activity by adding Bio CD47 on the A549 tumor cells was much better than that of Glo reagent (Promega Madison, Wis.) and measuring lumi SIRPO-Fc, which was in turn much better than that of Fc nescence after a 15 min incubation. SIRPC. Moreover, without wishing to be bound by theory, (0208 Anti-EGFR-hulgG1-SIRPOV2 was found to have avidity likely contributes to the increased biological activities significantly higher ADCC activity than anti-EGFR, SIR of anti-EGFR-huIgG1-SIRPCV2 fusion protein compared to PoV2-Fc and Fc-SIRPCV2 (FIG. 8). Without wishing to be anti-EGFR observed with respect to both the enhanced bound by theory, the much-enhanced in vitro activity of anti ADCC in vitro (FIG. 8) and enhanced anti-tumor activity in EGFR-huIgG1-SIRPCV2 compared to anti-EGFR is most vivo (FIG.9B). The ability of anti-EGFR-hulgG1-SIRPC. to likely caused by the avidity-driven binding to CD47 once the harness the avidity of binding to the tumor cells by binding to higher-affinity interaction between EGFR and the fusion pro two tumor targets on the same cell may result in more specific tein on the same cell has occurred, resulting in simultaneous targeting and less side effects in vivo. binding of two tumor targets on the same cell. 4(C) Pharmacokinetic Analysis of anti-EGFR-hulgG1 4(D)(ii) In Vivo Biological Activities of anti-EGFR-hulgG1 SIRPO SIRPO 0205 Pharmacokinetic analysis of anti-EGFR-hulgG1 (0209. The utility of anti-EGFR-hulgG1-SIRPC. is shown SIRPC. was measured after injection into mice, and compared by an in Vivo experiment. In an orthotopic lung tumor model, to the control molecules. 36 healthy female C57BL/6 mice (8 NOD-SCID mice were injected i.v. with 2.5x10 human weeks of age) from Charles River Laboratories were allowed A549-luc epidermoid carcinoma cells, followed by i.p. injec for acclimation for at least 3 days. Two dosing levels for each tion of 250 ug/mouse of an antibody isotype control, 250 of molecule (single IV bolus) were given to the mice in a ug/mouse of anti-EGFR. 250 g/mouse of anti-CD47, 132 volume of 100 uL/mouse (equivalent to 20 or 200 ug per ug/mouse of SIRPO-Fc, combination of 250 ug/mouse of US 2016/0177276 A1 Jun. 23, 2016 anti-EGFR and 132 ug/mouse of SIRPO-Fc, or 292 g/mouse The protein was purified in a single step by protein A affinity of anti-EGFR-hulgG1-SIRPCV2, which is the equimolar chromatography. Expression of the two polypeptides and amount of fusion protein. All the groups (n=7) received treat assembly of the full tetrameric molecule were confirmed on ment twice a week for 3 weeks, and results were reported as Sodium dodecyl Sulfate-polyacrylamide gel electrophoresis bioluminescent signals from lungs, general health, e.g. (SDS-PAGE) and size exclusion chromatography (SEC). For paralysis, which preceded death by 10-14 days, and survival SDS-PAGE, the purified protein samples were reduced with of mice. DTT and run on NuPAGEMES 4-12% Gel, 200V for 35min, 0210 Treatment with anti-EGFR-hulgG1-SIRPCV2 was followed by Coomassie staining. The two major bands on the found to be superior to the two monotherapies (FIG. 9B). gel had the expected molecular weights (MW) and the correct Specifically, based on median survival days, anti-EGFR stoichiometric ratio with >95% purity (FIG. 10A). In FIG. hulgG1-SIRPCV2 was significantly more efficacious than 10A, lane 1 shows the molecular weight (MW) marker and anti-EGFR (40 days vs 31 days, p=0.0175). Moreover, based lane 2 shows the expected MW (64. 23 kDa) and the correct on overall Survival, the fusion protein was somewhat more stoichiometric ratio (1:1) of the two polypeptides of anti efficacious than the combination of the two monotherapies, HER2-hulgG1-SIRPCV2. For SEC, the purified protein despite the fact that SIRPC.-Fc binds the target A549 cell samples were analyzed on a TSK-GEL Super SW3000 SEC better than the Fc-SIRPC. Without wishing to be bound by column 4.6x300mm (Tosoh Biosciences, Tokyo, Japan) that theory, the enhanced anti-tumor activity of anti-EGFR was equilibrated with 50 mM sodium phosphate, 400 mM hulgG1-SIRPCV2 is most likely caused by the avidity-driven sodium perchlorate, pH 6.3+0.1 and 38+2.0 mS/cm2. Size binding to CD47 once the higher-affinity interaction between exclusion chromatography showed a peak at the expected EGFR and the fusion protein on the same cell has occurred, MW of about 173 kDa for the monomeric anti-HER2 resulting in enhanced ADCP and ADCC in vivo. Simulta huIgG1-SIRPoV2 (FIG. 10B). neously blocking both the EGF/EGFR and the SIRPC/CD47 0215. In addition, anti-HER2 and anti-CD47 in a standard interactions, thus preventing signaling of these two pathways, monoclonal antibody format (anti-HER2 hulgG1 and anti may also contribute to the enhanced anti-tumor activities in CD47 huIgG1) and SIRPC. in an Fc-fusion protein format V1VO. (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) were generated as controls to compare with the anti-HER2-hulgG1-SIRPC. Example 5 format. 5(B)(i) Binding of anti-HER2-hulgG1-SIRPC. to CD47 Anti-HER2-hulgG1-SIRPC. Immunoglobulin Fusion Expressed on Cells Proteins 0216) The ability of anti-HER2-hulgG1-SIRPC. to bind to 0211 5(A) Construction and Expression of anti-HER2 CD47 expressed on the cell Surface was measured, and com hulgG1-SIRPo. pared to the control molecules. 2x10 CHO cells transfected 0212. The generation of an exemplary anti-HER2 with CD47 per well were incubated with varying concentra hulgG1-SIRPC. is based on the anti-HER24D5 (trastuzumab) tions of antibodies diluted in PBS+1% FBS in a 96 well plate monoclonal antibody (Carteretal, PNAS 89: 4285, 1992)and for 60 min on ice. After washing with PBS+1% FBS, cells the SIRPC. protein (Jiang et al. JBC 274: 559, 1999). The were incubated with FITCF(ab')2 goat Anti-Human IgG, Fcy DNA and protein sequence of the Fab light chain for 4D5 are (Jackson ImmunoResearch, West Grove, Pa.) diluted 1:200 in provided in SEQID NO:21 and SEQID NO:22, respectively. PBS+1% FBS for 60 minonice. Cells were analyzed by flow The DNA and protein sequence of the Fab heavy chain for cytometry (MACSQuant, Miltenyi Biotec, Cologne, Ger 4D5 are provided in SEQ ID NO:23 and SEQ ID NO:24, many). respectively. The DNA and protein sequence of the IgV 0217. The results show that anti-HER2-hulgG1-SIR domain of SIRPC. allele V2 are provided in SEQID NO: 7 and PCV2, anti-CD47, SIRPOV2-Fc and Fc-SIRPOV2 bound to SEQ ID NO:8, respectively. Anti-HER2-hulgG1-SIRPCV2 CD47 overexpressed on transfected CHO cells, but anti was generated by linking the C-terminus of the anti-HER2 HER2 did not bind because HER2 is not expressed (FIG. heavy chain polypeptide to the IgV domain of SIRPOV2 via 11A). Again, SIRPO-Fc showed a higher median fluores a (G4S) linker. cence than Fc-SIRPO. 0213 For expression of anti-HER2-huIgG1-SIRPCV2, 0218. The ability of anti-HER2-hulgG1-SIRPC. to bind to the following two gene constructs were assembled by stan CD47 expressed on the cell surface of blood cells was mea dard recombinant DNA techniques and cloned into the mam sured, and compared to the control molecules. Fresh whole malian expression vector pTT5 (containing the mouse light blood from healthy human donors was enriched for leuko chain signal peptide sequence for secretion) as in FIG. 1A: (1) cytes with dextran precipitation and was washed with PBS+ Construct VH(anti-HER2)-CH1-H-CH2-CH3-(G4S)-SIR 1% FBS. 2x10 leukocyte-enriched human whole blood cells PoV2 (SEQ ID NO:25), encoding the following elements: per well were incubated with 50 lug/ml proteins diluted in anti-HER2 heavy chain variable domain followed by human PBS+1% FBS in a 96 well plate for 60 min on ice. After heavy chain constant domains 1-3 followed by a (G4S). washing with PBS+1% FBS, cells were incubated with a linker and IgV domain of SIRPCV2 and (2) Construct 1:200 dilution of FITC F(ab')2 goat Anti-Human IgG, Fcy VL(anti-HER2)-CL (SEQID NO:21), encoding the follow (Jackson ImmunoResearch, West Grove, Pa.), a 1:100 dilu ing elements: anti-HER2 light chain variable domain fol tion of PE mouse anti-human CD235a (BD Biosciences, San lowed by human kappa light chain constant domain. The Jose, Calif.), and a 1:100 dilution of eFluor 450 mouse anti corresponding amino acid for these two constructs are shown human CD45 (eBioscience, San Diego, Calif.) in PBS+1% in SEQID NO:26 and SEQID NO:22 respectively. FBS for 60 minonice for protein detection and cell sorting by 0214. The two vectors were co-transfected transiently into flow cytometry. After washing again, cells were fixed with 1% Expi293 cells using Expi293fectin (Life Technologies, Grand formaldehyde in PBS. Cells were analyzed by flow cytometry Island, N.Y.) for expression of anti-HER2-hulgG1-SIRPC. (MACSQuant, Miltenyi Biotec, Cologne, Germany). US 2016/0177276 A1 Jun. 23, 2016

0219. The results show that anti-CD47 bound to CD47 human kappa light chain constant domain. The corresponding expressed on erythrocytes and leukocytes, but anti-HER2 amino acid for these two constructs are shown in SEQ ID huIgG1-SIRPoV2, SIRPoV2-Fc and Fc-SIRPo V2 only NO:32 and SEQID NO:28 respectively. bound to CD47 expressed on leukocytes and not to CD47 0224. In addition, anti-GD2 and anti-CD47 in a standard expressed on erythrocytes (FIG. 11B). monoclonal antibody format (anti-GD2 hulgG1 and anti 5(B)(ii) Binding Avidity of anti-HER2-hulgG1-SIRPC. on CD47 huIgG1) and SIRPC. in a Fc-fusion protein format Cells Expressing Both Antigens (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) are generated as 0220. The ability of anti-HER2-huIgG1-SIRPC. to bind controls to compare with the anti-GD2-hulgG1-SIRPC. for with avidity to HER2 and CD47 on the cell surface was mat. measured on human BT474 mammary gland/breast adeno carcinoma cells that overexpress HER2 and express CD47. Example 7 2x10 BT474 cells per well were incubated with varying concentrations of anti-HER2-hulgG1-SIRPC. Fc-SIRPC. Anti-PD-L1-hulgG1-SIRPC. Immunoglobulin Fusion anti-HER2, and anti-CD47 diluted in PBS+1% FBS in a 96 Proteins well plate for 60 min on ice. After washing with PBS+1% 0225 7(A) Construction and Expression of anti-PD-L1 FBS, cells were incubated with FITC F(ab')2 goat Anti-Hu huIgG1-SIRPo. man IgG, Fcy (Jackson ImmunoResearch, West Grove, Pa.), 0226. The generation of an exemplary anti-PD-L1 diluted 1:200 in PBS+1% FBS for 60 min on ice. After wash hulgG1-SIRPC. is based on the anti-PD-L1 monoclonal anti ing again, cells were fixed with 1% formaldehyde in PBS. body avelumab (International Patent Application Publication Cells were analyzed by flow cytometry (MACSQuant, Milte No. WO2013/079174) and the SIRPC. protein (Jiang et al. nyi Biotec, Cologne, Germany). JBC 274: 559, 1999). The DNA and protein sequence of the 0221) The results show that anti-HER2-hulgG1-SIR Fablight chain for anti-PD-L1 are provided in SEQID NO:33 PCV2 binding to BT474 cells is enhanced compared to con and SEQ ID NO:34, respectively. The DNA and protein trol molecules either individually or in combination (FIG. sequence of the Fab heavy chain for anti-PD-L1 are provided 11C), providing evidence for avidity. As explained in previ in SEQID NO:35 and SEQIDNO:36, respectively. The DNA ous examples, anti-Fc based detection likely under-estimates and protein sequence of human IgV domain of SIRPC. allele the binding of the antibody-SIRPC. fusion protein to cells. V2 are provided in SEQID NO:7 and SEQID NO:8, respec Thus, the observation of similar binding of anti-HER2 tively. The DNA and protein sequence of murine SIRPC. are hulgG1-SIRPCV2 and anti-HER2 suggests avidity effects for provided in SEQID NO:37 and SEQID NO:38, respectively. anti-HER2-hulgG1-SIRPC. binding to cells. The ability of Anti-PD-L1-hulgG1-muSIRPC. was generated by linking the anti-HER2-hulgG1-SIRPC. to harness the avidity of binding C-terminus of the anti-PD-L1 heavy chain polypeptide to the to the tumor cells by binding to two tumor targets on the same IgV domain of muSIRPC. via a (G4S) linker. cell may result in more specific targeting and less side effects 0227. For expression of anti-PD-L1-hulgG1-muSIRPC. in vivo. the following two gene constructs were assembled by stan dard recombinant DNA techniques and cloned into the mam Example 6 malian expression vector pTT5 (containing the mouse light Anti-GD2-hulgG1-SIRPC. Immunoglobulin Fusion chain signal peptide sequence for secretion) as in FIG. 1A: (1) Construct VH(anti-PD-L1)-CH1-H-CH2-CH3-(G4S)-mu Proteins SIRPC. (SEQ ID NO:41), encoding the following elements: 0222. The generation of an exemplary anti-GD2-hulgG1 anti-PD-L1 heavy chain variable domain followed by human SIRPC. is based on the anti-GD2 14.18 monoclonal antibody heavy chain constant domains 1-3 followed by a (G4S). (Hank etal, Clin. Cancer Re. 15:5923, 2009) and the SIRPO. linker and IgV domain of muSIRPC. and (2) Construct protein (Jiang et al. JBC 274: 559, 1999). The DNA and VL(anti-PD-L1)-CL (SEQID NO:33), encoding the follow protein sequence of the Fab light chain for 14.18 are provided ing elements: anti-PD-L 1 light chain variable domain fol in SEQID NO:27 and SEQIDNO:28, respectively. The DNA lowed by human kappa light chain constant domain. The and protein sequence of the Fab heavy chain for 14.18 are corresponding amino acid for these two constructs are shown provided in SEQID NO:29 and SEQID NO:30, respectively. in SEQID NO:42 and SEQID NO:34 respectively. The DNA and protein sequence of the IgV domain of SIRPC. 0228. The two vectors were co-transfected transiently into allele V2 are provided in SEQID NO: 7 and SEQID NO:8, Expi293 cells using Expi293fectin (Life Technologies, Grand respectively. Anti-GD2-hulgG1-SIRPCV2 is generated by Island, N.Y.) for expression of anti-PD-L1-hulgG1-mu linking the C-terminus of the anti-GD2 heavy chain polypep SIRPC. The protein was purified in a single step by protein A tide to SIRPCV2 via a (G4S) linker. affinity chromatography. Expression of the two polypeptides 0223 For expression of anti-GD2-hulgG1-SIRPCV2, the and assembly of the full tetrameric molecule were confirmed following two gene constructs are assembled by standard on Sodium dodecyl sulfate-polyacrylamide gel electrophore recombinant DNA techniques and cloned into the mamma sis (SDS-PAGE) and size exclusion chromatography (SEC). lian expression vector pTT5 (containing the mouse light For SDS-PAGE, the purified protein samples were reduced chain signal peptide sequence for secretion) as in FIG. 1A: (1) with DTT and run on NuPAGEMES 4-12% Gel, 200V for 35 Construct VH(anti-GD2)-CH1-H-CH2-CH3-(G4S)-SIR min, followed by Coomassie staining. The two major bands PoV2 (SEQ ID NO:31) encoding the following elements: on the gel had the expected molecular weights (MW) and the anti-GD2 heavy chain variable domain followed by human correct stoichiometric ratio with >95% purity (FIG. 12A). In heavy chain constant domains 1-3 followed by a (G4S). FIG. 12A, lane 1 shows the molecular weight (MW) marker linker and IgV domain of SIRPCV2 and (2) Construct and lane 2 shows the expected MW (64. 23 kDa) and the VL(anti-GD2)-CL (SEQ ID NO:27) encoding the following correct stoichiometric ratio (1:1) of the two polypeptides of elements: anti-GD2 light chain variable domain followed by anti-PD-L1-hulgG1-muSIRPC. For SEC, the purified protein US 2016/0177276 A1 Jun. 23, 2016 samples were analyzed on a TSK-GEL Super SW3000 SEC 0234 For expression of the anti-EGFR-huIgG1-SIR column 4.6x300mm (Tosoh Biosciences, Tokyo, Japan) that PCV1 (N110O), the following two gene constructs were was equilibrated with 50 mM sodium phosphate, 400 mM assembled by standard recombinant DNA techniques and are sodium perchlorate, pH 6.3+0.1 and 38+2.0 mS/cm. Size cloned into the mammalian expression vectorpTT5 (contain exclusion chromatography showed a peak at the expected ing the mouse light chain signal peptide sequence for secre MW of about 173 kDa for the monomeric anti-PD-L1 tion) as in FIG. 1A: (1) Construct VH(anti-EGFR)-CH1-H- hulgG1-muSIRPC. (FIG. 12B). CH2-CH3-(G4S)-SIRPoV1(N110O) (SEQ ID NO:43), 0229. In addition, anti-PD-L1 and anti-CD47 in a standard encoding the following elements: anti-EGFR heavy chain monoclonal antibody format (anti-PD-L1 IgG1 and anti variable domain followed by human heavy chain constant CD47 hulgG1) and SIRPC. in a Fc-fusion protein format domains 1-3 isotype IgG1 followed by a (G4S) linker and (SIRPCV2-Fc, Fc-huSIRPoV2, and Fc-musIRPC) (FIG.1C) IgV domain of SIRPOV1(N110O). 2) Construct VL(anti are generated as controls to compare with the anti-PD-L1 EGFR)-CL (SEQ ID NO:13), encoding the following ele hulgG1-SIRPC. format. ments: anti-EGFR light chain variable domain followed by 7(B) Binding Avidity of anti-PD-L1-hulgG-SIRPC. on Cells human kappa light chain constant domain. The corresponding Expressing Both Antigens amino acid sequences for these two constructs are shown in 0230. The ability of anti-PD-L1-hulgG1-muSIRPC. to SEQ ID NO:44 and SEQID NO:14 respectively. bind with avidity to PD-L1 and CD47 on the cell surface was 0235. The two vectors were co-transfected transiently into measured on mouse A20 B cell lymphoma cells that overex Expi293 cells using Expi293fectin (Life Technologies, Grand press PD-L1 and express CD47. 2x10 A20 cells per well Island, N.Y.) for expression of anti-EGFR-huIgG1-SIRPCV1 were incubated with varying concentrations of anti-PD-L1 (N110O). The protein was purified in a single step by protein hulgG1-muSIRPC, Fc-muSIRPO, and anti-PD-L1 diluted in A affinity chromatography. Expression of the two polypep PBS+1% FBS in a 96 well plate for 60 min on ice. After tides and assembly of the full tetrameric molecule were con washing with PBS+1% FBS, cells were incubated with FITC firmed on Sodium dodecyl sulfate-polyacrylamide gel elec F(ab')2 goat Anti-Human IgG, Fcy (Jackson ImmunoRe trophoresis (SDS-PAGE) and size exclusion chromatography search, West Grove, Pa.), diluted 1:200 in PBS+1% FBS for (SEC). For SDS-PAGE, the purified protein samples were 60 min on ice. After washing again, cells were fixed with 1% reduced with DTT and run on NuPAGE MES 4-12% Gel, formaldehyde in PBS. Cells were analyzed by flow cytometry 200V for 35 min, followed by Coomassie staining The two (MACSQuant, Miltenyi Biotec, Cologne, Germany). major bands on the gel had the expected molecular weights 0231. The results show that anti-PD-L1-hulgG1-mu (MW) and the correct stoichiometric ratio with >95% purity SIRPC. binding to A20 cells was generally enhanced com (FIG. 14A). In FIG. 14A, lane 1 shows the molecular weight pared to the binding of the control molecules (FIG. 13), (MW) marker and lane 2 shows the expected MW (64. 23 providing evidence for avidity. As explained in previous kDa) and the correct stoichiometric ratio (1:1) of the two examples, anti-Fc based detection likely under-estimates the polypeptides of anti-EGFR-huIgG1-SIRPCV1(N110O). For binding of the antibody-SIRPC. fusion protein to cells. Thus, SEC, the purified protein samples were analyzed on a TSK the observation of similar binding of anti-PD-L1-hulgG1 GEL Super SW3000 SEC column 4.6x300 mm (Tosoh Bio muSIRPC. and anti-PD-L1 suggests avidity effects for anti sciences, Tokyo, Japan) that was equilibrated with 50 mM PD-L1-hulgG1-muSIRPC. binding to cells. The ability of sodium phosphate, 400 mM sodium perchlorate, pH 6.3+0.1 anti-PD-L1-hulgG1-muSIRPC. to harness the avidity of bind and 38+2.0 mS/cm. Size exclusion chromatography showed ing to the tumor cells by binding to two tumor targets on the a peak at the expected MW of about 173 kDa for the mono same cell may result in more specific targeting and less side mericanti-EGFR-hulgG1-SIRPOV1(N110O) (FIG. 14B). effects in vivo. 0236. In addition, anti-EGFR and anti-CD47 in a standard Example 8 monoclonal antibody format (anti-EGFR huIgG1 and anti CD47 huIgG1) and SIRPC. in a Fc-fusion protein format Anti-EGFR-hulgG1-SIRPO.(aglycosylated) (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) are generated as Immunoglobulin Fusion Proteins controls to compare with the anti-EGFR-hulgG1-SIRPC. for mat. 0232 8(A) Construction and Expression of anti-EGFR 8(B) Binding of anti-EGFR-hulgG1-SIRPO.(aglycosylated) hulgG1-SIRPC (aglycosylated) to CD47 Expressed on Cells 0233. The generation of an exemplary anti-EGFR hulgG1-SIRPC (aglycosylated) is based on the anti-EGFR 0237. The ability of anti-EGFR-hulgG1-SIRPOV1 C225 (cetuximab) monoclonal antibody (Kawamoto, PNAS (N110O) to bind to CD47 overexpressed on the cell surface 80:1337, 1983) and the SIRPC. protein with N110O mutation was measured, and compared to the control molecules. 2x10 (Jiang et al. JBC 274: 559, 1999). Aglycosylated SIRPC, via CHO cells transfected with CD47 per well were incubated N110O. was shown to bind worse to CD47 than glycosylated with varying concentrations of antibodies diluted in PBS+1% SIRPC. (Ogura et al. JBC 279: 13711, 2004). The DNA and FBS in a 96 well plate for 60 min on ice. After washing with protein sequence of the Fab light chain for C225 are provided PBS+1% FBS, cells were incubated with FITC F(ab')2 goat in SEQID NO:13 and SEQIDNO:14, respectively. The DNA Anti-Human IgG, Fcy (Jackson ImmunoResearch, West and protein sequence of the Fab heavy chain for C225 are Grove, Pa.) diluted 1:200 in PBS+1% FBS for 60 min on ice. provided in SEQID NO:15 and SEQID NO:16, respectively. After washing again, cells were fixed with 1% formaldehyde The DNA and protein sequence of SIRPC. allele V1 are pro in PBS. Cells were analyzed by flow cytometry (MAC vided in SEQ ID NO: 5 and SEQ ID NO:6, respectively. SQuant, Miltenyi Biotec, Cologne, Germany). Anti-EGFR-hulgG1-SIRPOV1(N110O) is generated by link 0238. The results show that anti-EGFR-hulgG1-SIR ing the C-terminus of the anti-EGFR heavy chain polypeptide PoV1(N110O) bound to CD47 expressed on transfected to the IgV domain of SIRPOV1(N110O) via a (G4S) linker. CHO cells (FIG. 15). Unlike prior reports (Ogura et al., JBC US 2016/0177276 A1 Jun. 23, 2016 20

279: 13711, 2004), anti-EGFR-hulgG1-huSIRPo. 0243 For expression of the anti-EGFR-ds2-hulgG1/anti V1(N110O) bound as well to CD47 as anti-EGFR-hulgG1 EGFR-LC-SIRPC, the following two gene constructs were huSIRPOV1. assembled by standard recombinant DNA techniques and cloned into the mammalian expression vectorpTT5 (contain Example 9 ing the mouse light chain signal peptide sequence for secre tion) as in FIG. 1F: (1) Construct VH(anti-EGFR)(ds2 Anti-EGFR-hulgG1/anti-EGFR-LC-SIRPo. Q105C)-CH1-H-CH2-CH3 (SEQ ID NO:75), encoding the Immunoglobulin Fusion Proteins following elements: anti-EGFR heavy chain variable domain with a stabilizing mutation (ds2: Q105C to compensate for 0239 9(A) Construction and Expression of anti-EGFR destabilization of the light chain fusion (Orcutt et al., PEDS hulgG1/anti-EGFR-LC-SIRPO. 23:221, 2010), followed by human heavy chain constant 0240. The generation of an exemplary anti-EGFR domains 1-3 isotype IgG and (2) Construct VL(anti-EGFR) hulgG1/anti-EGFR-LC-SIRPC. is based on the anti-EGFR (ds2-543C)-CL-(G4S)-SIRPCV2 (SEQID NO:79), encod C225 (cetuximab) monoclonal antibody (Kawamoto, PNAS ing the following elements: anti-EGFR light chain variable 80:1337, 1983) and the SIRPO. protein (Jiang et al. JBC 274: domain with a stabilizing mutation (ds2: S43C), followed by 559, 1999). The DNA and protein sequence of the Fab light human kappa light chain constant domain, followed by a chain for C225 are provided in SEQID NO:13 and SEQID (G4S) linker and the IgV domain of SIRPCV2. The corre NO:14, respectively. The DNA and protein sequence of the sponding amino acid sequences for these two constructs are Fab heavy chain for C225 are provided in SEQID NO:15 and shown in SEQID NO:76, and SEQID NO:80 respectively. SEQID NO:16, respectively. The DNA and protein sequence 0244. In addition, anti-EGFR and anti-CD47 in a standard of SIRPC. allele V1 are provided in SEQID NO:5 and SEQID monoclonal antibody format (anti-EGFR huIgG1 and anti NO:6, respectively, and the DNA and protein sequence of the CD47 huIgG1) and SIRPC. in an Fc-fusion protein format IgV domain of SIRPC. allele V2 are provided in SEQID NO:7 (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) are generated as and SEQID NO:8, respectively. In a particular embodiment, controls to compare with the anti-EGFR-hulgG1/anti-EGFR anti-EGFR-huIgG1/anti-EGFR-LC-SIRPC. is generated by LC-SIRPC. format. linking the C-terminus of the anti-EGFR light chain polypep 0245 An alternate method of stabilization was examined tide to the IgV domain of SIRPCV2 via a (G4S) linker, and by introducing the heavy and light chain disulfide pairing also can be fused directly without any linker or with (GXS), format of IgG4, by introducing two mutations (S13 1C and where X, Y=0, 1, 2, 3, 4, 5, 6, 7, 8 or more. C222S in SEQID NO: 46) in the heavy chain CH1 domain. 0241 For expression of anti-EGFR-hulgG1/anti-EGFR The light chain C-terminal cysteine which would have nor LC-SIRPCV2, the following two gene constructs were mally paired with the heavy chain C222 will now form a assembled by standard recombinant DNA techniques and disulfide bond with C131. Additionally this mutated heavy cloned into the mammalian expression vectorpTT5 (contain chain can have either of the ds 1 ords2 mutation for even more ing the mouse light chain signal peptide sequence for secre stability enhancement and paired with the respective (ds 1 or tion) as in FIG. 1F: (1) Construct VH(anti-EGFR)-CH1-H- ds2) light chain. CH2-CH3 (SEQ ID NO:45), encoding the following 0246. Each of the two vectors were co-transfected tran elements: anti-EGFR heavy chain variable domain, followed siently into Expi293 cells using Expi293fectin (Life Tech by human heavy chain constant domains 1-3 isotype IgG and nologies, Grand Island, N.Y.) for expression of anti-EGFR (2) Construct VL(anti-EGFR)-CL-(G4S)-SIRPCV2 (SEQ hulgG1/anti-EGFR-LC-SIRPC. The proteins were purified in ID NO:47), encoding the following elements: anti-EGFR a single step by protein A affinity chromatography. Expres light chain variable domain, followed by human kappa light sion of the two polypeptides and assembly of the full tet chain constant domain, followed by a (G4S) linker and the rameric molecule were confirmed on Sodium dodecyl sulfate IgV domain of SIRPO.V2. The corresponding amino acid polyacrylamide gel electrophoresis (SDS-PAGE) and size sequences for these two constructs are shown in SEQ ID exclusion chromatography (SEC). For SDS-PAGE, the puri NO:46, and SEQID NO:48 respectively. fied protein samples were reduced with DTT and run on 0242 For expression of anti-EGFR-ds 1-hulgG1/anti NuPAGE MES 4-12% Gel, 200V for 35 min, followed by EGFR-LC-SIRPCV2, the following two gene constructs Coomassie staining. The two major bands on the gel had the were assembled by standard recombinant DNA techniques expected molecular weights (MW) and the correct stoichio and cloned into the mammalian expression vectorpTT5 (con metric ratio with >95% purity (FIG. 21A). In FIG. 21A, lane taining the mouse light chain signal peptide sequence for 1 shows the molecular weight (MW) marker, lane 2 shows the secretion) as in FIG.1F: (1) Construct VH(anti-EGFR)(ds 1 expected MW (49, 36 kDa) and the correct stoichiometric G44C)-CH1-H-CH2-CH3 (SEQ ID NO:73), encoding the ratio (1:1) of the two polypeptides of anti-EGFR-hulgG1/ following elements: anti-EGFR heavy chain variable domain anti-EGFR-LC-SIRPo, lane 3 shows the expected MW (49, with a stabilizing mutation (ds 1: G44C) to compensate for 36 kDa) and the correct stoichiometric ratio (1:1) of the two destabilization of the light chain fusion (Orcutt et al., PEDS polypeptides of anti-EGFR-ds 1-hulgG1/anti-EGFR-LC 23:221, 2010), followed by human heavy chain constant SIRPC, and lane 4 shows the expected MW (49, 36 kDa) and domains 1-3 isotype IgG and (2) Construct VL(anti-EGFR) the correct stoichiometric ratio (1:1) of the two polypeptides (ds1-A100C)-CL-(G4S)-SIRPoV2 (SEQ ID NO:77), of anti-EGFR-ds2-huIgG1/anti-EGFR-LC-SIRPC. The wild encoding the following elements: anti-EGFR light chain vari type sequences of the VH and VL domains had the least able domain with a stabilizing mutation (ds 1: A100C), fol number of extra bands on the SDS PAGE, contrary to the lowed by human kappa light chain constant domain, followed expectation that the ds 1 and ds2 mutations would stabilize the by a (G4S) linker and the IgV domain of SIRPOV2. The light chainfusion. For SEC, the purified protein samples were corresponding amino acid sequences for these two constructs analyzed on a TSK-GEL Super SW3000 SEC column 4.6x are shown in SEQID NO:74 and SEQID NO:78 respectively. 300 mm (Tosoh Biosciences, Tokyo, Japan) that was equili US 2016/0177276 A1 Jun. 23, 2016

brated with 50 mM sodium phosphate, 400 mM sodium per followed by the hinge region (EPKSC, SEQ ID NO:51, to chlorate, pH 6.3+0.1 and 38+2.0 mS/cm. Size exclusion allow for a disulfide bridge with the anti-EGFR light chain) chromatography showed a peak at the expected MW of about and (2) Construct VL(anti-EGFR)-CL (SEQ ID NO:13), 172 kDa for the monomeric anti-EGFR-hulgG1/anti-EGFR encoding the following elements: anti-EGFR light chain vari LC-SIRPC. (FIG. 21B). able domain followed by human kappa light chain constant 0247. In addition, anti-EGFR and anti-CD47 in a standard domain. The corresponding amino acid sequences for these monoclonal antibody format (anti-EGFR huIgG1 and anti two constructs are shown in SEQ ID NO:52 and SEQ ID CD47 huIgG1) and SIRPC. in an Fc-fusion protein format NO:14 respectively. (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) are generated as 0252. The two vectors were co-transfected transiently into controls to compare with the anti-EGFR-hulgG1-SIRPC. for Expi293 cells using Expi293fectin (Life Technologies, Grand mat. Island, N.Y.) for expression of SIRPCV2-Fc(hulgG1)-anti 9(B) Binding of anti-EGFR-hulgG1/anti-EGFR-LC-SIRPO. EGFR(Fab). The protein was purified in a single step by to CD47 Expressed on Cells protein A affinity chromatography. Expression of the two 0248. The ability of anti-EGFR-huIgG1/anti-EGFR-LC polypeptides and assembly of the full tetrameric molecule SIRPCV2 to bind to CD47 overexpressed on the cell surface were confirmed on Sodium dodecyl Sulfate-polyacrylamide was measured, and compared to the control molecules. 2x10 gel electrophoresis (SDS-PAGE) and size exclusion chroma CHO cells transfected with CD47 per well were incubated tography (SEC). For SDS-PAGE, the purified protein with varying concentrations of antibodies diluted in PBS+1% samples were reduced with DTT and run on NuPAGE MES FBS in a 96 well plate for 60 min on ice. After washing with 4-12% Gel, 200V for 35 min, followed by Coomassie stain PBS+1% FBS, cells were incubated with FITC F(ab')2 goat ing. The two major bands on the gel had the expected molecu Anti-Human IgG, Fcy (Jackson ImmunoResearch, West lar weights (MW) and the correct stoichiometric ratio with Grove, Pa.) diluted 1:200 in PBS+1% FBS for 60 minonice. >95% purity (FIG. 16A). In FIG. 16A, lane 1 shows the After washing again, cells were fixed with 1% formaldehyde molecular weight (MW) marker and lane 2 shows the in PBS. Cells were analyzed by flow cytometry (MAC expected MW (64. 23 kDa) and the correct stoichiometric SQuant, Miltenyi Biotec, Cologne, Germany). The results ratio (1:1) of the two polypeptides of SIRPO-Fc(hulgG1)- show that anti-EGFR-hulgG1/anti-EGFR-LC-SIRPCV2 anti-EGFR(Fab). For SEC, the purified protein samples were bound to CD47 expressed on transfected CHO cells, but not analyzed on a TSK-GEL Super SW3000 SEC column 4.6x as well as anti-EGFR-hulgG1-SIRPCV2 (FIG. 21C). 300 mm (Tosoh Biosciences, Tokyo, Japan) that was equili brated with 50 mM sodium phosphate, 400 mM sodium per Example 10 chlorate, pH 6.3+0.1 and 38+2.0 mS/cm. Size exclusion chromatography showed a peak at the expected MW of about SIRPO-Fc(hulgG1)-anti-EGFR(Fab) 173 kDa for the monomeric SIRPC-Fc(hulgG1)-anti-EGFR Immunoglobulin Fusion Protein (Fab) (FIG. 16B). 0249 10(A) Construction and Expression of SIRPO-Fc 0253. In addition, anti-EGFR and anti-CD47 in a standard (hulgG1)-anti-EGFR(Fab) monoclonal antibody format (anti-EGFR huIgG1 and anti 0250. The generation of an exemplary SIRPO-Fc CD47 huIgG1) and SIRPC. in a Fc-fusion protein format (hulgG1)-anti-EGFR(Fab) is based on the anti-EGFR C225 (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) are generated as (cetuximab) monoclonal antibody (Kawamoto, PNAS controls to compare with the SIRPC-Fc(hulgG1)-anti-EGFR 80:1337, 1983) and the SIRPO. protein (Jiang et al. JBC 274: (Fab) format. 559, 1999). The DNA and protein sequence of the Fab light 10(B) Binding of SIRPO-Fc(huIgG1)-anti-EGFR(Fab) to chain for C225 are provided in SEQID NO:13 and SEQID CD47 Expressed on Cells NO:14, respectively. The DNA and protein sequence of the (0254 The ability of SIRPO-Fc(hulgG1)-anti-EGFR(Fab) Fab heavy chain for C225 are provided in SEQID NO:15 and to bind to CD47 overexpressed on the cell surface was mea SEQID NO:16, respectively. The DNA and protein sequence sured, and compared to the control molecules. 2x10 CHO of the IgV domain of SIRPC. allele V2 are provided in SEQID cells transfected with CD47 per well were incubated with NO: 7 and SEQID NO:8, respectively. SIRPC-Fc(hulgG1)- varying concentrations of antibodies diluted in PBS+1% FBS anti-EGFR(Fab) was generated by linking SIRPC. to the in a 96 well plate for 60 min on ice. After washing with N-terminus of the Fc heavy chain via a (G4S), linker followed PBS+1% FBS, cells were incubated with FITC F(ab')2 goat by linking anti-EGFR. Fab heavy chain to the C-terminus of Anti-Human IgG, Fcy (Jackson ImmunoResearch, West the Fc heavy chain via a (G4S) linker. Grove, Pa.) diluted 1:200 in PBS+1% FBS for 60 min on ice. 0251 For expression of the SIRPOV2-Fc(hulgG1)-anti After washing again, cells were fixed with 1% formaldehyde EGFR(Fab), the following two gene constructs were in PBS. Cells were analyzed by flow cytometry (MAC assembled by standard recombinant DNA techniques and SQuant, Miltenyi Biotec, Cologne, Germany). cloned into the mammalian expression vectorpTT5 (contain (0255. The results show that SIRPC-Fc(hulgG1)-anti ing the mouse light chain signal peptide sequence for secre EGFR(Fab) bound to CD47 expressed on CD47-transfected tion), as in FIG. 1H: (1) Construct SIRPCV2-(G4S)-H-CH2 CHO cells (FIG. 17A). The binding of SIRPC-Fc(hulgG1)- CH3-(G4S)-VH(anti-EGFR)-CH1 (SEQ ID NO:49), anti-EGFR(Fab) or the control SIRPCV2-Fc was similar to encoding the following elements: the IgV domain of SIR that of anti-CD47, and better than that of anti-EGFR-hulgG1 PCV2 followed by a (G4S) linker and human heavy chain huSIRPOV2 or Fc-SIRPOV2. hinge region with cysteine (which natively forms a disulfide 10(C)(i) In Vitro Biological Activities of SIRPO-Fc(huIgG1)- bond with the light chain) mutated to a serine, (EPKSS, SEQ anti-EGFR(Fab) ID NO:50), followed by constant domains 2 and 3, followed (0256 The in vitro biological activity of SIRPO-Fc by a (G4S) linker, and anti-EGFR heavy chain variable (hulgG1)-anti-EGFR(Fab) was shown in an antibody-depen domain followed by human heavy chain constant domain 1 dent cell-mediated cytotoxicity (ADCC) assay, 6x10' human US 2016/0177276 A1 Jun. 23, 2016 22

A549 epidermoid carcinoma cells were transferred to each SIRPV2C-Fc(huIgG1)-anti-EGFR(scFv) is generated by well of a 96-well plate and incubated overnight at 37°C. The linking the IgV domain of SIRPC. to the N-terminus of the Fc media from the cells was removed and replaced with serial heavy chain via a (G4S) linker followed by linking anti dilutions of the recombinant antibodies for concentrations EGFR(scFv) to the C-terminus of the Fc heavy chain via a between 0.02-1600 ng/ml. After a 15-30 min incubation at (G4S) linker. 37°C., 1.5x10 effector cells (engineered Jurkat cells stably 0261) For expression of the SIRPOV2-Fc(hulgG1)-anti expressing the FcyRIIIa receptor, V158 (high affinity) vari EGFR(scFv) C225, the following gene construct was ant, and an NFAT response element driving expression of assembled by standard recombinant DNA techniques and firefly luciferase (Promega Madison, Wis.)) were added to cloned into the mammalian expression vectorpTT5 (contain each well of plates containing antibodies and A549 cells ing the mouse light chain signal peptide sequence for secre (effector-to-target cells ratio 2.5:1). After a 24-hour incuba tion), as in FIG. 1J: Construct SIRPCV2-(G4S)-H-CH2 tion, ADCC activity was measured via luciferase activity by CH3-(G4S)-C225(VH)-(G4S)-C225(VL) (SEQ ID adding Bio-Glo reagent (Promega Madison, Wis.) and mea NO:55), encoding the following elements: IgV domain of Suring luminescence after 15 min incubation. SIRPCV2 followed by a (G4S) linker and human heavy 0257 SIRPCV2-Fc(hulgG1)-anti-EGFR(Fab) was found chain hinge region with cysteine (which natively forms a to have lower ADCC activity than anti-EGFR, but higher disulfide bond with the light chain) mutated to a serine, ADCC activity than Fc-SIRPCV2 (FIG. 17B). The orienta (EPKSS, SEQ ID NO:50), followed by constant domains 2 tion of the binding domain relative to the Fc domain dictates and 3, followed by a (G4S) linker and anti-EGFR heavy the ADCC activity. Thus, SIRPO-FcV2 has higher activity chain variable domain, followed by a (G4S) linker and anti than Fc-SIRPCV2 (FIG. 17B) and anti-EGFR has higher EGFR light chain variable domain. The corresponding amino activity than Fc-anti-EGFR(Fab) (data not shown, but no acid sequence for this construct is shown in SEQID NO:56. activity observed). Without wishing to be bound by theory, 0262. In addition, anti-EGFR and anti-CD47 in a standard since the anti-EGFR(Fab) moiety of SIRPCV2-Fc(hulgG1)- monoclonal antibody format (anti-EGFR huIgG1 and anti anti-EGFR(Fab) bound cells with higher affinity than the CD47 hulgG1), anti-EGFR in a schv format (anti-EGFR SIRPC. moiety, it may position the Fc in an orientation for (scFv)), and SIRPC. in a Fc-fusion protein format (SIRPOV2 optimal ADCC activity. Fc and Fc-SIRPCV2) (FIG.1C) were generated as controls to 10(C)(ii) In Vivo Biological Activities of SIRPO-Fc compare with the SIRPCV2-Fc(hulgG1)-anti-EGFR(scEv) (hulgG1)-anti-EGFR(Fab) format. 0258. The utility of SIRPCV2-Fc(hulgG1)-anti-EGFR (Fab) is shown by an in vivo experiment. In an orthotopic lung Example 12 tumor model, NOD-SCID mice were injected i.v. with 2.5.x 10' human A549-luc epidermoid carcinoma cells, followed Anti-EGFR(scFv)-Fc(hulgG1)-SIRPC. by i.p. injection of 400 ug/mouse of an antibody isotype Immunoglobulin Fusion Proteins control, 250 g/mouse of anti-EGFR, combination of 250 0263. The generation of an exemplary anti-EGFR(scFv)- ug/mouse of anti-EGFR and 136 ug/mouse of SIRPCV2-Fc. Fc(hulgG1)-SIRPC. is based on the anti-EGFR(scFv) C225 298 g/mouse of anti-EGFR-huIgG1-SIRPCV2, or 298 (US Pat. No. 7,820,165) and the SIRPC. protein (Jiang et al., ug/mouse of SIRPCV2-Fc(hulgG1)-anti-EGFR(Fab), which JBC 274: 559, 1999). The DNA and protein sequence of is the equimolar amount of fusion protein. All the groups anti-EGFR(scFv) C225 are provided in SEQ ID NO:53 and (n=7) received treatment twice a week for 3 weeks, and SEQID NO:54, respectively. The DNA and protein sequence results were reported as bioluminescent signals from lungs, of the IgV domain of SIRPC. allele V2 are provided in SEQID general health, e.g. paralysis, which preceded death by 10-14 NO: 7 and SEQID NO:8, respectively. Anti-EGFR(scEv)-Fc days, and Survival of mice. (hulgG1)-SIRPCV2 is generated by linking anti-EGFR 0259 Treatment with SIRPOV2-Fc(hulgG1)-anti-EGFR (scFv) to the N-terminus of the Fc heavy chain via a (G4S). (Fab) fusion protein was found to be slightly inferior to the linker followed by linking the IgV domain of SIRPCV2 to the combination and anti-EGFR-hulgG1-SIRPOV2 (median sur C-terminus of the Fc heavy chain via a (G4S) linker. vival 40 days, 42 days and 43.5 days respectively, FIG. 18), 0264. For expression of the anti-EGFR(scFv) C225-Fc despite having similar binding to EGFR and superior binding (hulgG1)-SIRPCV2, the following gene construct was to CD47. The reduction in ADCC shown in FIG. 17B may assembled by standard recombinant DNA techniques and account for the decreased anti-tumor efficacy of SIRPCV2 cloned into the mammalian expression vectorpTT5 (contain Fc(hulgG1)-anti-EGFR(Fab) compared to anti-EGFR ing the mouse light chain signal peptide sequence for secre hulgG1-SIRPCV2. tion), as in FIG. 1L: Construct C225(VH)-(G4S)-C225 (VL)-(G4S)-H-CH2-CH3-(G4S)- SIRPCV2 (SEQ ID Example 11 NO:57), encoding the following elements: anti-EGFR heavy SIRPO-Fc(hulgG1)-anti-EGFR(scFv) chain variable domain, followed by a (G4S) linker and anti EGFR light chain variable domain followed by a (G4S). Immunoglobulin Fusion Proteins linker and human heavy chain hinge region with cysteine 0260 The generation of an exemplary SIRPO-Fc (which natively forms a disulfide bond with the light chain) (hulgG1)-anti-EGFR(scFv) is based on the SIRPO. protein mutated to a serine, (EPKSS, SEQID NO:50), followed by (Jiang et al. JBC 274: 559, 1999) and the anti-EGFR(scEv) constant domains 2 and 3, followed by a (G4S) linker and the C225 (U.S. Pat. No. 7,820,165). The DNA and protein IgV domain of SIRPCV2. The corresponding amino acid sequence of the IgV domain of SIRPC. allele V2 are provided sequence for this construct is shown in SEQID NO:58. in SEQID NO: 7 and SEQID NO:8, respectively. The DNA 0265. In addition, anti-EGFR and anti-CD47 in a standard and protein sequence of anti-EGFR(scFv) C225 are provided monoclonal antibody format (anti-EGFR huIgG1 and anti in SEQ ID NO:53 and SEQ ID NO:54, respectively. CD47 hulgG1), anti-EGFR in a schv format (anti-EGFR US 2016/0177276 A1 Jun. 23, 2016

(scFv)), and SIRPC. in a Fc-fusion proteinformat (SIRPCV2 assembled by standard recombinant DNA techniques and Fc and Fc-SIRPCV2) (FIG. 1C) were generated as controls to cloned into the mammalian expression vectorpTT5 (contain compare with the anti-EGFR(scEv)-Fc(huIgG1)-SIRPCV2 ing the mouse light chain signal peptide sequence for secre format. tion), as in FIG. 1L: Construct CHRI-19Fv1(VH)-linker CHRI-19Fv1(VL)-(G4S)-H-CH2-CH3-(G4S)-SIRPOV2 Example 13 (SEQ ID NO:63), encoding the following elements: anti CD19 heavy chain variable domain, followed by a (G4S). SIRPO-Fc(hulgG1)-anti-CD19(scFv) linker and anti-CD19 light chain variable domain, followed Immunoglobulin Fusion Proteins by a (G4S) linker and human heavy chain hinge region with 0266 The generation of SIRPC-Fc(hulgG1)-anti-CD19 cysteine (which natively forms a disulfide bond with the light (scFV) is based on the SIRPC. protein (Jiang et al. JBC 274: chain) mutated to a serine, (EPKSS, SEQ ID NO:50), fol 559, 1999) and the anti-CD19(scEv) CHRI-19Fv1 (Nichol lowed by constant domains 2 and 3, followed by a (G4S). son et al, Molecular Immunology 34:1157, 1997). The DNA linker and the IgV domain of SIRPCV2. The corresponding and protein sequence of the IgV domain of SIRPC. allele V2 amino acid sequences for this construct is shown in SEQID are provided in SEQ ID NO: 7 and SEQ ID NO:8, respec NO:64. tively. The DNA and protein sequence of CHRI-19Fv1 are 0271. In addition, anti-CD47 in a standard monoclonal provided in SEQID NO:59 and SEQID NO:60, respectively. antibody format (anti-CD47 huIgG1), anti-CD19 in a schv SIRPCV2-Fc(hulgG1)-anti-CD19(scFv) is generated by format (anti-CD19(sclv)), and SIRPC. in a Fc-fusion protein linking the IgV domain of SIRPCV2 to the N-terminus of the format (SIRPOV2-Fc and Fc-SIRPCV2) (FIG. 1C) are gen Fc heavy chain via a (G4S) linker followed by linking anti erated as controls to compare with the anti-CD19(sclv)-Fc CD19(scFv) to the C-terminus of the Fc heavy chain via a (hulgG1)-SIRPCV2 format. (G4S) linker. 0267 For expression of the SIRPOV2-Fc(hulgG1)-anti Example 15 CD19(scFv), the following gene construct is assembled by standard recombinant DNA techniques and cloned into the SIRPC.-Fcab(HER2) Immunoglobulin Fusion mammalian expression vector pTT5 (containing the mouse Proteins light chain signal peptide sequence for secretion), as in FIG. 1J: Construct SIRPCV2-(G4S)-H-CH2-CH3-(G4S)- 15(A) Construction and Expression of SIRPO-Fcab(HER2) CHRI-19Fv1 (VH)-linker-CHRI-19Fv1 (VL) (SEQ ID (0272. The generation of an exemplary SIRPO-Fcab NO:61), encoding the following elements: SIRPCV2 fol (HER2) is based on the SIRPO. protein (Jiang et al. JBC 274: lowed by a (G4S) linker and human heavy chain hinge region 559, 1999) and the anti-HER2 Fcab H10-03-6 (Wozniak with cysteine (which natively forms a disulfide bond with the Knopp et al., PEDS 23:289, 2010). The DNA and protein light chain) mutated to a serine, (EPKSS, SEQ ID NO:50), sequence of the IgV domain of SIRPC. allele V2 are provided followed by constant domains 2 and 3, followed by a (G4S). in SEQID NO: 7 and SEQID NO:8, respectively. The DNA linker and anti-CD19 heavy chain variable domain, followed and protein sequence of Fcab(HER2) are provided in SEQID by a (G4S) linker and anti-CD19 light chain variable NO:65 and SEQ ID NO:66, respectively. SIRPCV2-Fcab domain. The corresponding amino acid sequences for this (HER2) is generated by linking the IgV domain of SIRPCV2 construct is shown in SEQID NO:62. to the N-terminus of Fcab(HER2) via a (G4S) linker. 0268. In addition, anti-CD47 in a standard monoclonal (0273 For expression of the SIRPCV2-Fcab(HER2), the antibody format (anti-CD47 huIgG1), anti-CD19 in a schv following gene construct was assembled by standard recom format (anti-CD19(sclv)), and SIRPC. in a Fc-fusion protein binant DNA techniques and cloned into the mammalian format (SIRPOV2-Fc and Fc-SIRPCV2) (FIG. 1C) are gen expression vector pTT5 (containing the mouse light chain erated as controls to compare with the SIRPCV2-Fc signal peptide sequence for secretion), as in FIG. 1F: Con (hulgG1)-anti-CD19(scFv) format. struct SIRPOV2-(G4S)-H-CH2-CH3(anti-HER2) (SEQ ID NO:67), encoding the following elements: SIRPCV2 fol Example 14 lowed by a (G4S) linker and human heavy chain hinge region with cysteine (which natively forms a disulfide bond with the Anti-CD19(scFv)-Fc(hulgG1)-SIRPC. light chain) mutated to a serine, (EPKSS, SEQ ID NO:50), Immunoglobulin Fusion Proteins followed by constant domain2 and constant domain3 that has 0269. The generation of an exemplary anti-CD19(scFv)- been modified to bind HER2 via the AB, CD, and EF loops. Fc(hulgG1)-SIRPC. is based on the anti-CD19(scFv) CHRI The corresponding amino acid sequences for this construct is 19Fv1 (Nicholson et al. Molecular Immunology 34:1157, shown in SEQID NO:68. 1997) and the SIRPO. protein (Jiang et al. JBC 274: 559, 0274 The vector was transfected transiently into Expi293 1999). The DNA and protein sequence of CHRI-19Fv1 are cells using Expi293fectin (Life Technologies, Grand Island, provided in SEQID NO:59 and SEQID NO:60, respectively. N.Y.) for expression of SIRPCV2-Fcab(HER2). The protein The DNA and protein sequence of the IgV domain of SIRPC. was purified in a single step by protein A affinity chromatog allele V2 are provided in SEQID NO: 7 and SEQID NO:8, raphy. Expression of the molecule was confirmed on Sodium respectively. Anti-CD19(scFv)-Fc(hulgG1)-SIRPOV2 is dodecyl sulfate-polyacrylamide gel electrophoresis (SDS generated by linking anti-CD19(scFv) to the N-terminus of PAGE) and size exclusion chromatography (SEC). For SDS the Fc heavy chain via a (G4S), linker followed by linking the PAGE, the purified protein sample was reduced with DTT and IgV domain of SIRPOV2 to the C-terminus of the Fc heavy run on NuPAGEMES 4-12% Gel, 200V for 35 min, followed chain via a (G4S) linker. by Coomassie staining. The major band on the gel had the (0270. For expression of the anti-CD19(scFv)-Fc expected molecular weights (MW) with >95% purity (FIG. (hulgG1)-SIRPCV2, the following gene construct is 19A). In FIG. 19A, lane 1 shows the molecular weight (MW) US 2016/0177276 A1 Jun. 23, 2016 24 marker and lane 2 shows the expected MW (40 kDa) of the binding of SIRPC. to CD47, and to predict mutations with SIRPCV2-Fcab(HER2). For SEC, the purified protein the potential to decrease or increase binding affinity of SIRPC. sample was analyzed on a TSK-GEL Super SW3000 SEC to CD47 and identify candidates worth pursuing experimen column 4.6x300mm (Tosoh Biosciences, Tokyo, Japan) that tally. Briefly, the crystal structure of the CD47/SIRPC. com was equilibrated with 50 mM sodium phosphate, 400 mM plex was analyzed to identify SIRPC. residue positions pre sodium perchlorate, pH 6.3+0.1 and 38+2.0 mS/cm. Size dicted to affect CD47 binding. exclusion chromatography showed a peak at the expected 0281 Computational mutagenesis was performed on the MW of about 80 kDa for the monomeric SIRPOV2-Fcab selected set of SIRPC. positions to arrive at a value for the difference in binding energy of various putative mutations as (HER2) (FIG. 19B). compared to wild-type SIRPO, and a threshold value was set 0275. In addition, anti-HER2 and anti-CD47 in a standard to categorize mutations predicted to have either reduced affin monoclonal antibody format (anti-HER2 hulgG1 and anti ity or increased affinity for CD47 relative to wild-type SIRPC. CD47 hulgG1), Fcab(HER2), and SIRPC. in a Fc-fusion pro Because the threshold value setting for the designation of tein format (SIRPCV2-Fc and Fc-SIRPCV2) (FIG. 1C) were reduced or increased affinity SIRPC. variants overlapped, generated as controls to compare with the SIRPCV2-Fcab there was also significant overlap in the computationally pre (HER2) format. dicted mutations listed in Table 1 and Table 2 (see below). 15(B)(i) Binding of SIRPC-Fcab(HER2) to CD47 Expressed on Cells TABLE 1 0276. The ability of SIRPoV2-Fcab(HER2) to bind to CD47 expressed on the cell Surface was measured, and com SIRPC mutations predicted to reduce CD47 binding pared to the control molecules. 2x10 CHO cells transfected SIRPoV1 SEQ ID NO: 6 with CD47 per well were incubated with varying concentra (SIRPoV2 O tions of antibodies diluted in PBS+1% FBS in a 96 well plate if different) SEQID NO: 8 Computationally Designed Mutations for 60 min on ice. After washing with PBS+1% FBS, cells Residue Residue # (Total) in Either SIRPCV1 or SIRPCV2 were incubated with FITCF(ab')2 goat Anti-Human IgG, Fcy V 6 A, C, D, E, G, I, L, M, N, Q, S, T (12) (Jackson ImmunoResearch, West Grove, Pa.) diluted 1:200 in A (V) 27 C, D, G, H, L, N, S, T, V (9) I 31 A, C, E, K, Q, R., T V (8) PBS--1% FBS for 60 min on ice. After washing again, cells P 35 A, C, E, G, Q, S (6) were fixed with 1% formaldehyde in PBS. Cells were ana Q 37 A, C, E, G, H, K, L, M, N, R, S, T (12) lyzed by flow cytometry (MACSQuant, Miltenyi Biotec, E 47 A, C, D, F, G, H, I, K, L, M, N, Q, R, S, Cologne, Germany). T, V, W, Y (18) Q 52 A, C, E, M (4) (0277. The results show that SIRPCV2-Fcab(HER2), anti E S4 D (1) CD47, and SIRPO-Fc bound to CD47 expressed on trans H 56 A, C, D, E, F, G, I, K, L, M, N, P, Q, R, fected CHO cells, but anti-HER2 and Fcab(HER2) did not S, T, V, W, Y (19) bind because HER2 is not expressed (FIG. 20A). L (S) 66 A, C, D, E, F, G, H, I, M, N, P, Q, S, T, V, W, Y (17) 15(B) (ii) Binding Avidity of SIRPC-Fcab(HER2) on Cells T 67 A, C, D, E, F, G, H, I, L., M, N, Q, R, S, Expressing Both Antigens V, W, Y (17) M 72 A, C, D, E, F, G, H, I, K, L, N, Q, R, S, 0278. The ability of SIRPoV2-Fcab(HER2) to bind with T, V, W, Y (18) avidity to HER2 and CD47 on the cell surface was measured V 92 A, C, D, E, G, I, M, N, Q, R, S, T (12) on human BT474 mammary gland/breast adenocarcinoma cells that overexpress HER2 and express CD47. 2x10 BT474 cells per well were incubated with varying concentra TABLE 2 tions of SIRPCV2-Fcab(HER2), SIRPoV2-Fc, Fcab(HER2), anti-HER2, and anti-CD47 diluted in PBS+1% FBS in a 96 SIRPC. mutations that are predicted to have tighter CD47 binding well plate for 60 min on ice. After washing with PBS+1% compared to wild-type. FBS, cells were incubated with FITC F(ab')2 goat Anti-Hu SIRPOV1 SEQ ID man IgG, Fcy (Jackson ImmunoResearch, West Grove, Pa.), (SIRPCV NO: 6 or diluted 1:200 in PBS+1% FBS for 60 min on ice. After wash 2 if SEQID Computationally Designed ing again, cells were fixed with 1% formaldehyde in PBS. different) NO: 8 Mutations in Either SIRPCV1 or Cells were analyzed by flow cytometry (MACSQuant, Milte Residue Residue # SIRPCV2 nyi Biotec, Cologne, Germany). V 6 A, D, I (0279. The results show that SIRPCV2-Fcab(HER2) bind A (V) 27 A, G, I, K, Q, R, S, T I 31 C, K, RT ing to BT474 cells was enhanced compared to the binding of P 35 G., N, Q, S Fcab(HER2), particularly at lower concentrations (FIG. Q 37 A, G, H, W 20B), providing a strong evidence for avidity. The ability of E 47 G, S, W.Y SIRPCV2-Fcab(HER2) to harness the avidity of binding to Q 52 E, H the tumor cells by binding to two tumor targets on the same E S4 P H 56 C, I, PY cell may result in more specific targeting and less side effects L (S) 66 A, C, D, E, F, H, K, L, M, N, P, Q, V, W in vivo. T 67 D, E, F, N, Q, W, Y M 72 A, C, D, E, F, G, H, I, K, L, N, Q, R, S, W.Y Example 16 V 92 N Computational Methods to Identify SIRPC. Residues Affecting CD47 Binding 0282. 33 SIRPC. variants containing single point muta 0280 Computational methods familiar to those skilled in tions, mainly from Table 2, were selected for further experi the art were used to identify SIRPC. residues that may affect mental characterization. (See Table 3 in Example 17). US 2016/0177276 A1 Jun. 23, 2016

Example 17 diluted in PBS+1% FBS in a 96 well plate for 60 min on ice. After washing with PBS+1% FBS, cells were incubated with Anti-EGFR-hulgG1-SIRPC. Variants FITC F(ab')2 goat Anti-Human IgG, Fcy (Jackson Immu 0283) 17(A) Construction and Expression of anti-EGFR noResearch, West Grove, Pa.) diluted 1:200 in PBS+1% FBS hulgG1-SIRPC. Variants for 60 min on ice. After washing again, cells were fixed with 0284 Antibody-SIRPC. variants were generated in the 1% formaldehyde in PBS. Cells were analyzed by flow context of anti-EGFR-hulgG1-SIRPCV2 described in cytometry (MACSQuant, Miltenyi Biotec, Cologne, Ger Example 4. The mutations in the IgV domain of SIRPC. allele many). An EC50 was calculated for each variant for binding V2 are listed in Table 3 (with reference to SEQ ID NO: 8). to CD47 expressed on CHO cells by fitting data to a sigmoidal Anti-EGFR-hulgG1-SIRPCV2 Variants were generated by curve (log(agonist) vs. response Variable slope (four linking the C-terminus of the anti-EGFR heavy chain parameters)) with Graph Pad Prism and reported in Table 3. polypeptide to the IgV domain of variant SIRPCV2 via a 0289. The results show that many anti-EGFR-hulgG1 (G4S) linker. SIRPCV2 variants, including V6I, V27I, I31R, I31T, Q37H, 0285 For expression of each of the anti-EGFR-hulgG1 Q37W, H56P, and S66Q, bound to CD47 expressed on trans SIRPCV2 Variants, the following two gene constructs were fected CHO cells with greater affinity than wild-type anti assembled by standard recombinant DNA techniques and EGFR-huIgG1-SIRPCV2 (Table 3), whereas variants E54P cloned into the mammalian expression vectorpTT5 (contain and M72R bound with similar affinity. As expected, the posi ing the mouse light chain signal peptide sequence for secre tive controls anti-CD47, 1D4, AS2, and AS1 also bound to tion) as in FIG. 1A: (1) Construct VH(anti-EGFR)-CH1-H- CD47 with greater affinity and the negative control anti CH2-CH3-(G4S)-SIRPCV2 (SEQ ID NO:19) with the EGFR did not bind because EGFR is not expressed. The sequence being modified to encode the particular mutation(s) results show a single point mutation in SIRPC. is sufficient to listed in Table 3 for each variant; the construct encoded the increase the affinity of SIRPC. for CD47. following elements: anti-EGFR heavy chain variable domain 0290. In order to compare the intrinsic binding affinity, i.e. followed by human heavy chain constant domains 1-3 isotype minimize the avidity effect due to bivalent engagement that IgG1 followed by a (G4S) linker and IgV domain of variant occurs at high receptor density, the binding of anti-EGFR SIRPCV2 and (2) Construct VL(anti-EGFR)-CL (SEQ ID hulgG1-SIRPCV2 Variants and control molecules to cells NO:13), encoding the following elements: anti-EGFR light expressing low levels of CD47 was determined. 2x10 chain variable domain followed by human kappa light chain CD47' human Ramos lymphoma cells per well were incu constant domain. The corresponding amino acid sequences bated with varying concentrations of antibodies diluted in for these two constructs are shown in SEQID NO:20 (which PBS+1% FBS in a 96 well plate for 60 min on ice. After sequence must be modified to include the particular muta washing with PBS+1% FBS, cells were incubated with FITC tions listed in Table 3) and SEQID NO:14 respectively. F(ab')2 goat Anti-Human IgG, Fcy (Jackson ImmunoRe 0286. The set of two vectors for each of the anti-EGFR search, West Grove, Pa.) diluted 1:200 in PBS+1% FBS for 60 hulgG1-SIRPCV2 Variants were co-transfected transiently min on ice. After washing again, cells were fixed with 1% into Expi293 cells using Expi293fectin (Life Technologies, formaldehyde in PBS. Cells were analyzed by flow cytometry Grand Island, N.Y.) for expression of each of the anti-EGFR (MACSQuant, Miltenyi Biotec, Cologne, Germany). An hulgG1-SIRPCV2 Variants. The proteins were purified in a EC50 was calculated for each variant for binding to CD47 single step by protein A affinity chromatography. Expression expressed on Ramos cells by fitting data to a sigmoidal curve of the two polypeptides and assembly of the full tetrameric (log(agonist) vs. response Variable slope (four param molecule was confirmed on size exclusion chromatography eters)) with Graph Pad Prism and reported in Table 3. (SEC). For SEC, the purified protein samples were analyzed 0291. The results show that many anti-EGFR-hulgG1 on a TSK-GEL Super SW3000 SEC column 4.6 300 mm SIRPCV2 variants, including V6I, V27I, I31R, I31T, Q37H, (Tosoh Biosciences, Tokyo, Japan) that was equilibrated with Q37W, E54P H56P S66Q, and M72R, bound to CD47 50 mM sodium phosphate, 400 mM sodium perchlorate, pH expressed on Ramos cells with greater affinity than wild-type 6.3+0.1 and 38+2.0 mS/cm. Size exclusion chromatography anti-EGFR-hulgG1-SIRPOV2 (Table 3), but the differences showed a peak at the expected MW of about 173 kDa for the in EC50 values between the variants were greater compared monomeric anti-EGFR-hulgG1-SIRPCV2. The percentage to the differences seen with CD47' cells. As expected, the of the monomeric peak relative to all SEC peaks was reported positive controls anti-CD47, 1D4, AS2, and AS1 also bound for each variant in Table 3. to CD47 with greater affinity and the negative control anti 0287. In addition, wild-type anti-EGFR-hulgG1-SIR EGFR did not bind to Ramos cells because EGFR is not PoV2 (“WT), chimeric antibody B6H12/hulgG1 (“anti expressed. CD47), and anti-EGFR-hulgG1-SIRPCV2 with multiple 0292. The ability of the higher-affinity anti-EGFR mutations (“1D4” (V27I/K53R/S66T/K68R/F103V), hulgG1-SIRPCV2 Variants to bind to CD47 expressed on the (Weiskopf, Science 341:88, 2013): “AS2’ (K53R/S66T/ cell Surface of blood cells was measured, and compared to the K68R); and “AS1 (L4V/V27I/I31T/K53R/S66T/K68R/ control molecules. 2x10 fresh whole blood cells from F94L)) were generated as positive controls and anti-EGFR healthy human donors per well were incubated with 50 g/ml was generated as a negative control. of proteins diluted in PBS+1% FBS in a 96 well plate for 60 17(B) (i) Binding of anti-EGFR-hulgG1-SIRPC. Variants to min on ice. After washing with PBS+1% FBS, cells were CD47 Expressed on Cells incubated with a 1:200 dilution of FITC F(ab')2 goat Anti 0288. The ability of anti-EGFR-hulgG1-SIRPOV2 vari Human IgG, Fcy (Jackson ImmunoResearch, West Grove, ants to bind to CD47 overexpressed on the cell surface was Pa.) to detect binding of anti-EGFR-hulgG1-SIRPCV2 Vari measured, and compared to the control molecules. 2x10 ants and a 1:100 dilution of PE mouse anti-human CD235a CHO cells transfected to express high levels of CD47 per well (BD Biosciences, San Jose, Calif.) to select for erythrocytes were incubated with varying concentrations of antibodies in PBS+1% FBS for 60 minonice. After washing again, cells US 2016/0177276 A1 Jun. 23, 2016 26 were fixed with 1% formaldehyde in PBS. Cells were ana -continued lyzed by flow cytometry (MACSQuant, Miltenyi Biotec, Cologne, Germany). The median fluorescence intensity CD47LO (MFI) at 50 g/ml of each anti-EGFR-hulgG1-SIRPCV2 CD47#7 cells % RBC variant was determined and reported in Table 3. In addition, % cells ECSO ECSO RBC MFI of the degree of binding to erythrocytes was expressed as a % of Protein Monomer (nM) (nM) MFI anti-CD47 the anti-CD47 MFI (100x(MFI of protein)/(MFI of anti K68I 97 >100 K68T 97 >100 CD47)). M72I 96 >100 0293. The results confirmed that anti-CD47 bound to M72N 90 40 CD47 expressed on erythrocytes, but anti-EGFR-hulgG1 M72R 87 8 84 O.3 196 SIRPCV2 did not (Table 3), as shown before (FIG. 7B). M72W 96 >100 V92N 8O 72 Several of the anti-EGFR-hulgG1-SIRPCV2 variants, KS3-K68I 97 NB including V6I,V27I, I31T, Q37H, E54P and M72R, retained K53N - K68E 98 NB lack of binding to erythrocytes, similar to wild-type anti K53O+ K68T 97 NB EGFR-hulgG1-SIRPCV2 (3% or less of anti-CD47 binding). K53T -- K68A 95 NB However, other variants, including I31 Rand S66Q, had some KS3V - K68H 97 NB level of binding to erythrocytes (12% and 21% of anti-CD47 binding, respectively), as did the positive controls 1D4 0295) To potentially improve the therapeutic index of an (53%), AS2 (12%), and AS1 (37%). Q37W and H56P only anti-EGFR-huIgG1-SIRPC. fusion protein in the treatment of bound weakly to erythrocytes (4% of anti-CD47 binding). cancer, it may be desirable to select an anti-EGFR-hulgG1 0294 Table 3: List of anti-EGFR-hulgG1-SIRPOV2 vari SIRPC. variant with an optimal increase in binding to CD47 ants, showing percent monomer by SEC, the EC50 (nM) of on CD47' and CD47 cells compared to anti-EGFR binding to CD47 expressed on CD47' cells (CD47-trans hulgG1-SIRPC, and a relative lack of binding to erythrocytes fected CHO cells) and CD47 cells (Ramos cells), the MFI (particularly compared to anti-CD47). For example, a variant (mean fluorescence intensity) of the proteins at 50 lug/ml may be chosen with an about 5-fold to an about 30-fold bound to human erythrocytes (RBC), and the % of RBC increase in binding to CD47' cells compared to wild-type binding relative to anti-CD47 MFI (calculated as (100x(MFI anti-EGFR-huIgG1-SIRPC, and an about 30% or less, an about 10% or less, an about 5% or less, oran about 3% or less of protein)/(MFI of anti-CD47)). Wild type anti-EGFR binding to erythrocytes as compared to anti-CD47. Fulfilling hulgG1-SIRPCV2 (“WT), positive control chimeric anti such criteria, the biological activity of exemplary variant body B6H12/huIgG1 (“anti-CD47), and negative control anti-EGFR-hulgG1-SIRPOV2(Q37W) was further charac anti-EGFR are in bold and the higher affinity SIRPO. positive terized. It is contemplated that to improve the therapeutic controls (“1D4”, “AS2’ and “AS1) are in italics. index of an antibody-SIRPC. fusion protein targeting a differ ent tumor antigen, such as CD20 or HER2, analogous criteria CD47LO may be used to choose the optimal SIRPC. variant. CD47#7 cells % RBC % cells ECSO ECSO RBC MFI of Example 18 Protein Monomer (nM) (nM) MFI anti-CD47 Anti-EGFR-hulgG1-SIRPO.(Q37W) WT 97 8 85 0.4 2% Anti-CD47 96 7 6 23.3 100% Anti-EGFR 99 NB NB 0.3 1% 0296 18(A) Construction and Expression of anti-EGFR iD4 97 3 2 12. 52% huIgG1-SIRPo.(Q37W) AS2 97 3 2 2.9 12% 0297 Anti-EGFR-huIgG1-SIRPCV2(Q37W) vectors ASI 94 7 3 8.7 37% from example 17 were co-transfected transiently into V6I 96 3 9 0.4 196 Expi293 cells using Expi293fectin (Life Technologies, Grand V27I 97 2 2 0.7 3% Island, N.Y.) for expression of anti-EGFR-huIgG1-SIRPCV2 V27Q 93 >100 I31R 91 2 14 2.9 12% (Q37W). The protein was purified in a single step by protein I31T 97 4 13 O.S 296 A affinity chromatography. Expression of the two polypep P35G 97 18 tides and assembly of the full tetrameric molecule were con P35N 97 >100 firmed on Sodium dodecyl sulfate-polyacrylamide gel elec Q37A 96 >100 trophoresis (SDS-PAGE) and size exclusion chromatography Q37H 96 6 71 O.3 296 Q37V 97 13 (SEC). For SDS-PAGE, the purified protein samples were Q37W 95 2 3 1 4% reduced with DTT and run on NuPAGE MES 4-12% Gel, E47Y 97 10 200V for 35 min, followed by Coomassie staining. The two Q52E 88 >100 major bands on the gel had the expected molecular weights Q52H 90 >100 (MW) and the correct stoichiometric ratio with >95% purity E54P 91 8 81 O.3 196 H56P 97 2 13 O.9 4% (FIG.22A). In FIG.22A, lane 1 shows the molecular weight H56Y 90 12 (MW) marker and lane 2 shows the expected MW (64. 23 S66E 90 33 kDa) and the correct stoichiometric ratio (1:1) of the two S66H 86 25 polypeptides of anti-EGFR-huIgG1-SIRPCV2(Q37W). For S66Q 90 5 2 4.9 21% SEC, the purified protein samples were analyzed on a TSK S66W 97 >100 T67E 91 >100 GEL Super SW3000 SEC column 4.6 300 mm (Tosoh Bio T67W 91 >100 sciences, Tokyo, Japan) that was equilibrated with 50 mM K68A 97 >100 sodium phosphate, 400 mM sodium perchlorate, pH 6.3+0.1 K68E 97 >100 and 38+2.0 mS/cm. Size exclusion chromatography showed K68H 99 56 a peak at the expected MW of about 173 kDa for the mono mericanti-EGFR-hulgG1-SIRPOV2(Q37W) (FIG.22B). US 2016/0177276 A1 Jun. 23, 2016 27

0298. In addition, anti-EGFR and anti-CD47 in a standard Example 19 monoclonal antibody format (anti-EGFR huIgG1 and anti CD47 hulgG1) and SIRPC. in a Fc-fusion protein format Anti-CD20-hulgG1-SIRPO.(Q37W) (SIRPCV2-Fc, Fc-SIRPoV2, and Fc-SIRPoV2(Q37W)) (FIG. 1C) and wild-type anti-EGFR-hulgG1-SIRPOV2 were 0303) 19(A) Construction and Expression of anti-CD20 generated as controls to compare with anti-EGFR-hulgG1 huIgG1-SIRPo.(Q37W) SIRPoV2(Q37W). (0304 Exemplary anti-CD20-hulgG1-SIRPOV2(Q37W) 18(B) In Vitro Biological Activities of anti-EGFR-hulgG1 is based on the anti-CD20-hulgG1-SIRPCV2 described in SIRPo(Q37W) Example 2. Anti-CD20-hulgG1-SIRPOV2(Q37W) was gen 0299. The ability for anti-EGFR-hulgG1-SIRPOV2 erated by linking the C-terminus of the anti-CD20 heavy (Q37W) to cause erythrocytes to hemagglutinate was deter chain polypeptide to the IgV domain of variant SIRPCV2 mined, and compared to the control molecules. 30-50 ul of containing the Q37W mutation via a (G4S) linker. fresh human whole blood cells per well were incubated with (0305 For expression of anti-CD20-huIgG1-SIRPCV2, 1, 3, 10 and 30 g/ml of test proteins in a total volume of 100 the following two gene constructs were assembled by stan ul of HBSS+0.5% BSA in a 96 well plate at 37° C. for 2-4 dard recombinant DNA techniques and cloned into the mam hours. The plates were centrifuged and the Supernatant was malian expression vector pTT5 (containing the mouse light removed. The cell pellets were resuspended with 100 ul of PBS. Wells were ranked between full solubilization of RBCs chain signal peptide sequence for secretion) as in FIG. 1A: (1) (no hemagglutination), partial pellet and Solubilization (+he Construct VH(anti-CD20)-CH1-H-CH2-CH3-(G4S)-SIR magglutination) and dense pellet of cells with no solubiliza PoV2(Q37W) (SEQ ID NO:11 altered by the SIRPo allele tion (++hemagglutination). V2 encoding mutation Q37W) encoding the following ele 0300. The results confirmed that anti-CD47 causes eryth ments: anti-CD20 heavy chain variable domain followed by rocytes to hemagglutinate (+hemagglutination at 3 g/mland human heavy chain constant domains 1-3 isotype IgG1 fol ++hemagglutination at 10 and 30 ug/ml), but anti-EGFR lowed by a (G4S) linker and the IgV domain of variant hulgG1-SIRPCV2 at all the concentrations tested did not SIRPCV2 with mutation at Q37W and (2) ConstructVL(anti hemagglutinate, correlating with the lack of erythrocyte bind CD20)-CL (SEQ ID NO: 1) encoding the anti-CD20 light ing shown in FIG. 7B (data not shown). Anti-EGFR-hulgG1 chain variable domain followed by human kappa light chain SIRPCV2(Q37W) at all the concentrations tested also did not constant domain. The corresponding amino acid sequences cause erythrocytes to hemagglutinate, despite the increased for these two constructs are shown in SEQ ID NO:12, addi binding of anti-EGFR-hulgG1-SIRPCV2(Q37W) to erythro tionally containing the SIRPC. allele V2 mutation Q37W, and cytes. This data provides further Supporting evidence that SEQ ID NO:2, respectively. anti-EGFR-hulgG1-SIRPOV2(Q37W) may achieve a better (0306 The set of two vectors for anti-CD20-hulgG1-SIR therapeutic index by increasing binding without increasing PCV2(Q37W) expression was co-transfected transiently into erythrocyte-related toxicity. Expi293 cells using Expi293fectin (Life Technologies, Grand 18(C) In Vivo Biological Activities of anti-EGFR-hulgG1 Island, N.Y.). The protein was purified in a single step by SIRPo(Q37W) protein A affinity chromatography. Expression of the two 0301 The utility of anti-EGFR-hulgG1-SIRPOV2 polypeptides and assembly of the full tetrameric molecule (Q37W) is shown by an in vivo experiment. In an orthotopic were confirmed on SDS-PAGE and SEC. For SDS-PAGE, the lung tumor model, NOD-SCID mice were injected i.v. with purified protein samples were reduced with DTT and run on 2.5x10' human A549-luc epidermoid carcinoma cells, fol NuPAGE MES 4-12% Gel, 200V for 35 min, followed by lowed by i.p. injection of 400 ug/mouse of an antibody iso Coomassie staining. The two major bands on the gel had the type control, 250 g/mouse of anti-EGFR, 298 g/mouse of expected MW and the correct stoichiometric ratio with >95% anti-EGFR-huIgG1-SIRPCV2, or 298 ug/mouse of anti purity (FIG. 24A). In FIG. 24A, lane 1 shows the molecular EGFR-hulgG1-SIRPCV2(Q37W), which is the equimolar weight (MW) marker and lane 2 shows the expected MW (63. amount of fusion protein. All the groups (n=8) received treat 23 kDa) and the correct stoichiometric ratio (1:1) of the two ment twice a week for 3 weeks, and results were reported as polypeptides of anti-CD20-hulgG1-SIRPCV2. For SEC, the bioluminescent signals from lungs, general health, e.g. purified protein samples were analyzed on a TSK-GEL Super paralysis, which preceded death by 10-14 days, and survival SW3000 SEC column 4.6x300 mm (Tosoh Biosciences, of mice. Tokyo, Japan) that was equilibrated with 50 mM sodium 0302 Treatment with anti-EGFR-hulgG1-SIRPOV2 phosphate, 400 mM sodium perchlorate, pH 6.3+0.1 and (Q37W) fusion protein was found to be superior to the two 38+2.0 mS/cm2. Size exclusion chromatography showed a monotherapies and anti-EGFR-hulgG1-SIRPCV2 (FIG. 23). peak at the expected MW of about 172 kDa for the mono Introduction of a single Q37W mutation in the SIRPCV2 mericanti-CD20-hulgG1-SIRPOV2(Q37W) (FIG. 24B). moiety of the fusion protein improved the median survival 0307. In addition, anti-CD20 and anti-CD47 in a standard days from 43.5 days for the wild-type anti-EGFR-hulgG1 monoclonal antibody format (anti-CD20 hulgG1 and anti SIRPCV2 to 55 days, and the difference is highly significant CD47 huIgG1) and SIRPC. in a Fc-fusion protein format (p=0.0019). The results clearly show that increasing the affin (SIRPOV2-Fc, Fc-SIRPoV2, and Fc-SIRPoV2(Q37W)) ity of SIRPC. for CD47 resulted in enhanced anti-tumor effi (FIG. 1C) and wild-type anti-CD20-hulgG1-SIRPCV2 from cacy, which, without wishing to be bound by theory, can most Example 2 were generated as controls to compare with anti easily be explained by the enhanced avidity-driven CD47 CD20-hulgG1-SIRPoV2(Q37W). binding, and targeting of the A549 cells for elimination by immune cells. The wild-type anti-EGFR-huIgG1-SIRPCV2 19(B) In Vivo Biological Activity of the Combination of in turn was more efficacious than the anti-EGFR antibody, anti-CD20 with Fc-SIRPCV2(Q37W) prolonging the median survival day from 35.5 to 43.5 days 0308 As an indication for the improved biological activity (p=0.0187), confirming what was observed in a previous of anti-CD20-hulgG1-SIRPOV2(Q37W) as compared to experiment (FIG. 9B). Thus, this data provides further Sup anti-CD20-hulgG1-SIRPOV2, a disseminated lymphoma porting evidence that anti-EGFR-hulgG1-SIRPCV2(Q37W) model in mouse was used to test the combination of anti may achieve a better therapeutic index by improving efficacy CD20 with either Fc-SIRPCV2 or Fc-SIRPoV2(Q37W). without increasing erythrocyte-related toxicity. SCID mice were injected i.v. with 5x10° CD20+ human