Post-Chemotherapy and Cytokine Pretreated Marrow Stromal Cell Layers Suppress Hematopoiesis from Normal Donor CD34 Cells

Home , Stromal cell

Bone Marrow Transplantation, (1998) 22, 457–468  1998 Stockton Press All rights reserved 0268–3369/98 $12.00 http://www.stockton-press.co.uk/bmt Post-chemotherapy and cytokine pretreated marrow stromal cell layers suppress hematopoiesis from normal donor CD34+ cells

GN Schwartz1, MK Warren2, SW Rothwell3, J Zujewski1, DC Halverson1, KH Cowan1, A Tolcher1, J O’Shaughnessy1 and RE Gress1

1Department of Experimental Transplantation and Immunology, National Cancer Institute, Bethesda, MD; 2Otsuka Pharmaceutical, Inc., Rockville, MD; and 3Department of Hematology, Walter Reed Army Institute of Research, Washington, DC, USA

Summary: and increased incidence of graft failure after autologous marrow transplantation have also been linked to prior Marrow stromal layers were used to investigate the chemotherapy.6–8 Similarly, low yields of CD34+ cells and potential role of negative regulators produced by the hematopoietic progenitor/stem cells during peripheral blood marrow microenvironment as one potential cause of mobilization regimens correlated with the type and timing hematopoietic suppression after chemotherapy and of prior chemotherapy.9–11 The causes for these deficiencies cytokines. Stromal layers were established from mar- have not been ascertained. row of normal or prechemotherapy donors and breast The continuous replacement of blood cells in normal cancer patients after hematological recovery from one adults is provided by committed and primitive progenitor cycle of 5-fluorouracil, leucovorin, doxorubicin, and cells that are heterogeneous with respect to their self- cyclophosphamide and GM-CSF or PIXY321 (GM- renewal, proliferative and differentiation capacities.12–14 CSF/IL-3 fusion protein). Normal donor CD34+ cells Normally, hematopoietic progenitors reside in the marrow, were placed in contact with stromal layers, and the and blood cell production is regulated by complex interac- number of colony-forming units for granulocytes and tions of progenitor cells with the marrow microenviron- macrophages (CFU-GM) was determined. There were ment.15–17 The stromal cell layer, an adherent layer of cells 25–79% fewer CFU-GM in post-chemotherapy stromal that forms in marrow long-term cultures (LTC) simulates layer cocultures than in no chemotherapy cocultures. the marrow microenvironment and supports the mainte- With neutralizing antibody to TNF-␣ the number of nance and proliferation of hematopoietic progenitor CFU-GM in no chemotherapy and post-chemotherapy cells.15,16 stromal cocultures was, respectively, 96 ؎ 7% (n = 5) Previous results demonstrated that marrow from some -and 142 ؎ 8% (n = 5) of the number with no antibody groups of patients who recovered from high-dose chemo treatment. PIXY321 and GM-CSF pretreated stromal therapy or chemotherapy followed by autologous transplan- layers also suppressed production of CFU-GM. Anti- tation had decreased numbers of hematopoietic progenitors TNF-␣ promoted an increase in CFU-GM numbers such as colony-forming units for granulocytes and macro- from GM-CSF, but not PIXY321, pretreated stromal phages (CFU-GM).18–23. LTC were also shown to have a cocultures. The results demonstrate that post-chemo- reduced capacity to maintain the production of CFU-GM therapy marrow stromal layers were deficient in sup- and burst-forming units for erythroid cells (BFU-E).7,18–22,24 porting in vitro hematopoiesis and suggest that negative Bone marrow injury was evident after recovery from the regulators induced by chemotherapy and cytokines may first cycle of chemotherapy and even when peripheral blood be one cause for this defect. counts and the number of committed progenitors and Keywords: bone marrow; hematopoiesis; stem cells; CD34+ cells in the marrow were within the normal range.22 stromal cells; TNF-␣ Hematopoietic stimulatory cytokines have not been shown to ameliorate suppressive or toxic effects of chemotherapy or radiation on the number of either committed or primitive hematopoietic progenitors in the marrow22,25,26 and in some Dose-intensive chemotherapy regimens improve the cases may have additional suppressive effects on marrow response rates in patients with some refractory malig- recovery.22,26 Damage to either the hematopoietic 1,2 nancies. Myelosuppression, however, is still a major progenitor/stem cell compartment or the marrow microen- dose-limiting effect. Clinically, patients may develop pro- vironment or both has been implicated as a potential cause gressively lower platelet and neutrophil nadirs and have for suppressive effects of bone marrow function observed longer delays in blood cell recovery with each additional after chemotherapy.7,19,21,23,24,26,27 3–5 cycle of chemotherapy. Prolonged times for engraftment A variety of stimulatory and negative factors that regulate the proliferation and differentiation of hematopoietic progenitors are expressed in the marrow microenvironment.6,28–30 Correspondence: Dr GN Schwartz, Building 10, Room 12N226, Depart- ment of Experimental Transplantation and Immunology, Medicine Branch, Two of the possible causes for hematopoietic deficiencies National Cancer Institute, Bethesda, MD 20892, USA observed after chemotherapy are either a decrease in pro- Received 29 December 1997; accepted 6 May 1998 duction of stimulatory factors or an increase in the production Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 458 of inhibitory factors by the marrow microenvironment. For metastatic breast cancer approximately 6 weeks following example, increased production of hematopoietic negative high-dose cyclophosphamide 4 g/m2 i.v. on day 1 and G-CSF regulators such as TGF-␤,IFN-␥,IFN-␣,TNF-␣, and IL-4 10 ␮g/kg daily subcutaneous injection on day 2 and con- was implicated as a cause for the reduced capacity of HIV tinued for 2–4 days after the total WBC was Ͼ2 × 109/l. and CMV infected stromal cell layers, and stromal cell layers of marrow from patients with some malignancies, to support hematopoiesis from normal donor hematopoietic progeni- Preparation of cell suspensions 31–34 tors. The purpose of the present study was to determine Bone marrow aspirates were obtained from the posterior ␣ ␤ whether increases in TNF- ,TGF- or IL-4 from marrow iliac crests. Five to 20 ml of bone marrow were collected stromal layers was a possible cause for the reduced capacity in a 20 ml syringe containing preservative-free heparin of post-chemotherapy LTC to maintain the production of (Sigma, St Louis, MO, USA) and then diluted in Dulbec- committed progenitors. co’s phosphate-buffered saline without Ca+2 or Mg+2 (DPBS; GIBCO BRL, Gaithersburg, MD, USA) containing penicillin–streptomycin (GIBCO). Marrow cells with a Materials and methods density of р1.077 g/ml were collected and washed three times after separation on a Ficoll-sodium diatrizoate Normal donors and patients (Lymphocyte Separation Medium; Organon Teknika, Dur- Normal donors and patients were required to give informed ham, NC, USA) gradient. A portion of the cells was resus- × 6 consent as participants of National Cancer Institute Insti- pended to 2 10 cells/ml in enriched Iscove’s modified 35 tutional Review Board approved protocols. Normal donors Dulbecco’s medium (IMDM) with 10% fetal bovine were not receiving any medication and had normal peripheral serum (Hyclone, Logan, UT, USA) and incubated in tissue ° blood counts at the time of marrow harvest. Patients with culture flasks at 37 C and with 5% CO2. The next day, stage II, III and IV breast cancer were administered FLAC nonadherent cells were collected and washed. Viable cell chemotherapy which consisted of 5-fluorouracil 300 mg/m2 concentrations were determined by hemacytometer counts daily i.v. on days 1–3 (total dose 900 mg/m2)1hafterleuco- and trypan blue dye exclusion. Umbilical cord blood was 36 vorin calcium 500 mg/m2 daily i.v. on days 1–3 (total dose collected and processed as previously described. 1500 mg/m2), doxorubicin 17 mg/m2 daily i.v. on days 1–3 2 2 (total dose 51 mg/m ) and cyclophosphamide 500 mg/m Enrichment of CD34+ bone marrow cells daily i.v. for days 1–3 (total dose 1500 mg/m2), and was followed by either no cytokine, GM-CSF (sargramostim, CD34+ cells were selected by positive immunomagnetic Leukine; Immunex, Seattle, WA, USA) 250 ␮g/m2 daily sub- selection using a high-gradient magnetic separation column cutaneously, or PIXY321 (375 ␮g/m2 twice daily subcutane- and MiniMacs CD34 progenitor cell isolation kit (Miltenyi ously from days 5 to 19 of each 21 day cycle or until hemato- Biotec, Auburn, CA, USA). The purity of the recovered logical recovery.5 After the first cycle of FLAC, marrow was cells was determined by staining with CD34 (anti-HPCA- obtained from patients once their blood counts had recovered 2) phycoerythrin (PE; Becton Dickinson, San Jose, CA, to an ANC Ͻ1.5 × 109/l and platelets Ͻ90 × 109/l, and cyto- USA), and flow cytometry analyses demonstrated that kine administration was discontinued for at least 48 h.22 Mar- purity was у90%. CD34+ cells were suspended in 7.5% row was also obtained from one patient 1 year following five DMSO (Sigma) in 47% heat inactivated fetal bovine serum cycles of FLAC + PIXY321, and from another patient with and IMDM, cryopreserved in a controlled rate freezer, and

Table 1 Recovery of committed and primitive progenitors in post-chemotherapy marrow CD34+ cells

Source of CD34+ cells No. per 104 CD34+ cellsa Calculated No. per ml of marrow aspirateb

CFU-GM (n) LTC-IC (n) CD34+ cells CFU-GM (×10−3) LTC-IC (×10−3) (×10−5)

Normal donors 292 ± 25 (7) 102 ± 21 (6) NDc ND ND Before FLAC 238 ± 39 (9) 75 ± 15 (3) 3.5 ± 0.9 6.9 ± 1.3 1.9 ± 0.8 After FLACd +No cytokine 184 ± 26 (5) 47 ± 10 (2) 4.2 ± 1.2 7.1 ± 1.6 3.4 ± 1.6 +GM-CSF 265 ± 24 (5) 57 ± 16 (2) 3.7 ± 0.7 9.9 ± 1.8 2.7 ± 0.1 +PIXY321 150 ± 29 (6) 27 ± 14 (2) 1.3 ± 0.4e 1.8 ± 0.4e 0.2 ± 0.01

aMarrows were enriched for CD34+ cells (purity у91%), cultured in methylcellulose with PHA-LCM for detection of CFU-GM, and in limiting dilution over the M2-B104 murine stromal cell line for detection of LTC-IC. Values are the mean Ϯ s.e.m. of mean values from n different normal donor and patient samples. bThe number of nucleated cells and the percentage of CD34+ cells in unseparated marrow cells were used to calculate the number of CD34+ cells, CFU- GM and LTC-IC per ml marrow aspirate. cNot determined. dMarrows were obtained after hematological recovery from the first or second cycle of FLAC chemotherapy. eBy Wilcoxon–Mann–Whitney U statistic for unequal sample sizes, values are significantly different (P Ͻ 0.05) from values from before FLAC chemotherapy and after FLAC + No cytokine marrows. Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 459 stored in the vapor phase of liquid nitrogen. Cells were Coculture of normal donor CD34+ cells with stromal cell quickly thawed in a 37°C water bath, washed twice with layers excess IMDM, incubated for 1.5–2 h at 37°C, 5% CO , 2 For stromal contact cocultures, 1 or 2 × 104 CD34+ cells washed again, and counted. were placed directly on the irradiated stromal cell layers. From 1 to 4 weeks later, nonadherent and adherent cells Establishment of bone marrow stromal cell layers were harvested from two LTC as previously described.22,34,37 Briefly, the adherent cells were detached Stromal cell layers were established basically as previously from the culture wells by incubating the cell layers at 37°C 34,37 × 6 described by culturing 2 10 cells with a density of with 0.25% trypsin and 1 mm EDTA (GIBCO) or a 0.1% р 2 1.077 g/ml in 2.0 cm wells of Nunclone 4-Well Multid- solution of type I collagenase (Sigma) prepared in DPBS ishes (Nunc, Naperville, IL, USA). The cells were diluted without Ca+2 and Mg+2 or serum. The cells were washed, in 1 ml of Myeloid long term culture medium (Stem Cell Technologies, Vancouver, Canada) containing freshly 10000 prepared 1 × 10–6 m hydrocortisone-21-hemisucccinate Nonadherent cells ° (Sigma) and incubated at 37 C with 5% CO2 and room air. Weekly, 500 ␮l of the medium and nonadherent cells were Adherent cells removed and replaced with 500 ␮l of fresh medium. After 4 weeks, the stromal layers were exposed to 15 Gy of ␥- 1000 irradiation (1 Gy/min) from a 137Cs Gammacell 40 irradi- * ator. This dose of radiation eliminated progenitor cells sur- viving in the stromal cell layers (data not shown). For cytokine pretreatment studies the media were changed the next day, and DPBS or 20 ng/ml PIXY321 (Immunex) or GM- 100 * CSF (R&D Systems, Minneapolis, MN, USA) were added to the cultures. Two days later the layers were washed at least three times to remove cytokine; after an additional 3 Number of CFU-GM/Culture 4 + days 2 × 10 CD34 normal donor marrow cells were cocul- 10 tured with the stromal layers. Pre-P27 Post-P25 Patient marrow stromal layer

10000 150 Nonadherent cells Total CFU-GM Adherent cells Total BFU-E

1000 100 *

100 50 * * * * Number of BFU-E/Culture

% of number in pretherapy LTC * * * 10 0 * * 0 1 2 3 4 5 Pre-P27 Post-P25 Weeks in long term cultures (LTC) Patient marrow stromal layer

Figure 1 Effect of FLAC chemotherapy on production of CFU-GM and Figure 2 Effect of post-chemotherapy stromal layers on production of BFU-E in bone marrow long-term cultures (LTC). LTC were established committed progenitors from normal donor CD34+ cells. Stromal layers from marrow obtained from patients prior to start of therapy and after were established from marrow obtained from one patient prior to FLAC hematological recovery from the first cycle of FLAC chemotherapy. chemotherapy (ie P25) and one patient after hematological recovery from Weekly, nonadherent and adherent cells were harvested and subcultured the first cycle of FLAC + PIXY321 (ie P25). Umbilical cord blood CD34+ in methylcellulose supplemented with PHA-LCM and EPO to stimulate cells were cocultured in direct contact with the irradiated stromal layers. colony formation from CFU-GM and BFU-E. The number of CFU-GM One week later, the nonadherent and adherent cells from two stromal layer and BFU-E in post-chemotherapy marrow used to establish LTC was cocultures were collected, washed, counted and subcultured in methylcel- either similar to or greater than the number in prechemotherapy marrows. lulose supplemented with PHA-LCM and EPO to stimulate colony forma- The values are the median calculated from the percentage of CFU-GM tion from CFU-GM and BFU-E. The values are the mean Ϯ s.d. of CFU- and BFU-E from post-chemotherapy LTC relative to prechemotherapy GM and BFU-E numbers per coculture calculated from the number of LTC from the same patient (n = 13 for weeks 1 and 4; n = 11 for weeks colonies detected in three methylcellulose cultures. Day 0 CD34+, CFU- 2 and 3; and n = 8 for week 5). *By Wilcoxon matched-pairs signed rank GM and BFU-E numbers in cord blood added to stromal layers were, test of significance, the number of CFU-GM and BFU-E in post-chemo- respectively, 1 × 104, 947 Ϯ 210 and 720 Ϯ 72. *By Student’s two tailed therapy LTC was significantly different (P Ͻ 0.05) from numbers in pre- t-test, significantly different (P р 0.01) from the number of CFU-GM chemotherapy LTC. detected in prechemotherapy stromal layer cocultures. Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 460 Table 2 CFU-GM from different CD34+ targets cocultured with mar- Limiting dilution assay for detection of long-term culture- row stromal layers initiating cells (LTC-IC)

Source of stromal layer No. of CFU-GM in 1 week stromal Long-term culture-initiating cells (LTC-IC) determinations coculturesa were made by culturing marrow CD34+ cells in limiting dilution over the M2-B104 murine stromal cell line (ATCC, ND-1 CD34+ P29 CD34+ Rockville, MD, USA) in wells of 96-well culture plates targets targets (Nunclone) as previously described.38 Cells from the liquid cultures were washed and then diluted in Myelocult long ND-1 (normal donor) 266 ± 81 469 ± 184 P29 (prechemotherapy) 238 ± 42 410 ± 31 term culture medium (Stem Cell Technologies) with × 3 + P31 (prechemotherapy) 195 ± 15 391 ± 0 4 10 CD34 cells/ml as the highest concentration and then diluted 1:1.6 for three to five more dilutions. Twenty a2 × 104 CD34+ cells were cocultured in contact with a normal donor and wells with an established stromal cell layer were cocultured two prechemotherapy stromal cell layers. One week later, cells from two with 120 ␮l of each cell dilution. Cultures were maintained stromal layer cocultures were harvested, pooled and subcultured in methyl- ° in a humidified 37 C incubator with 5% CO2 and 95% cellulose supplemented with PHA-LCM to stimulate colony formation ␮ from CFU-GM. room air. Weekly, 60 l of medium and nonadherent cells Values are the mean Ϯ s.d. total number of CFU-GM calculated from were replaced with an equal volume of medium. After 5 CFU-GM colony numbers from three methylcellulose cultures. weeks, the number of wells containing cobblestone areas of у20 translucent cells and the number of negative wells were determined and used to calculate the number of LTC- IC per liquid culture. counted, and then subcultured in methylcellulose with 10% PHA-LCM or with 10 ng/ml IL-3, 10 ng/ml GM-CSF and 50 ng/ml SCF (Stem Cell Technologies) to stimulate colony Antibody treatment of bone marrow cocultures growth from CFU-GM. For BFU-E determinations, 3 U/ml ␣ ␤ of EPO were added to the methylcellulose. One ml of Neutralizing antibodies to TNF- (goat IgG), TNF- (goat ␤ methylcellulose medium containing cells was placed in a IgG), IL-4 (goat IgG) and TGF- (chicken IgG) (all from 35-mm2 gridded tissue culture dish (Nunc) and then main- R&D Systems) were resuspended in Dulbecco’s phosphate tained in a humidified 37°C incubator with 5% CO and buffered saline (DPBS; GIBCO) with 0.1% deionized bov- 2 ␮ 95% room air. After 14 days, colonies containing у50 gra- ine serum albumin. Antibodies (20 g/ml) were added to nulocytes or macrophages or both and colonies containing stromal layers on day 0, 2 and 4 after coculture of + 34 у50 erythroid cells were counted as derived from CFU- CD34 cells. GM and BFU-E, respectively. In stromal noncontact cocul- × 4 + tures 1–2 10 CD34 cells in 6.5-mm nontissue culture- Detection of TNF-␣ activity treated 0.1 mm pore Transwells (Costar, Cambridge, MA, USA) were placed over the irradiated stromal layers. After Enzyme-linked immunoassays were used to detect TNF-␣ ° 1 week of incubation at 37 C and 5% CO2 and 95% room (Immunotech, Westbrook, ME, USA and R&D Systems), air, cells in the Transwells were removed, washed, counted, MIP-1␣ (R&D Systems), TGF-␤ (R&D Systems) and IFN- and subcultured in methylcellulose for detection of CFU- ␥ (R&D Systems) activity in media from stromal layers. GM. Some layers were incubated for 18 h with 10 ␮g/ml of

Table 3 Effect of post-chemotherapy stromal layers on progenitors from normal donor CD34+ cellsa

Marrow source No. of CFU-GM and BFU-E in 1 week stromal layer coculturesb

Prechemotherapyc Post-chemotherapyc Post-chemotherapyc

CFU-GM P20 P18 P14 Nonadherent cells 472 ± 31 574 ± 26d 374 ± 105 Adherent cells 577 ± 76 304 ± 48d 535 ± 9 % Total in prechemotherapy 100 84 77 BFU-E Nonadherent cells 322 ± 50 273 ± 28 283 ± 58 Adherent cells 500 ± 71 350 ± 49d 819 ± 28d % Total in prechemotherapy 100 76 134

aNormal donor umbilical cord blood CD34+ cells were cocultured in contact with pre- and post-chemotherapy stromal cell layers. One week later, cells from two cocultures each were harvested, pooled, and subcultured in methylcellulose supplemented with PHA-LCM and EPO to stimulate colony formation from CFU-GM and BFU-E. Day 0 CD34+, CFU-GM and BFU-E numbers in cord blood added to stromal layers were 2 × 104, 440 Ϯ 72 and 457 Ϯ 10, respectively. bValues are mean Ϯ s.d. calculated from the number of BFU-E colonies from three methylcellulose cultures. cPatient marrow sample P20 was obtained prior to chemotherapy and P18 and P14 were obtained after hematological recovery from FLAC + PIXY321. dBy Student’s two-tailed t-test signiﬁcantly different (P Ͻ 0.05) from number in prechemotherapy stromal layer cocultures. Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 461 150 Table 4 ELISA determinations of negative regulators in supernatants from long-term cultures (LTC)

Source of marrow for LTC Concentration of factors in LTC supernatants (pg/ml)a

100 TNF-␣ MIP-1␣ TGF-␤ INF-␥

Medium Alone 18 ± 12 7 ± 2 2046 ± 156 ND No chemotherapy Normal donor – A 23 ± 8NTNTNT * * ± ± 50 Normal donor – B ND 20 6 3242 21 ND Normal donor – C ND 23 ± 2 2558 ± 174 ND P02 (Prechemotherapy) 4 ± 013± 1 3904 ± 99 ND Post-chemotherapy

Number of CFU-GM/Culture * P01 (FLAC + PIXY321) 4 ± 2 20 2779 ± 147 ND P02 (FLAC + GM-CSF) 116 ± 7b 14 NT ND P04 (FLAC + PIXY321) 28 ± 2 14 2965 ± 180 ND 0 P06 (FLAC + GM-CSF) 12 ± 4 14 3421 ± 85 ND ND54 P20 P25 P31 P34 (FLAC + No cytokine)c 45 ± 0b 14 1920 ± 89 ND Source of marrow stromal layers a Ϯ Figure 3 Effect of post-chemotherapy stromal layers on production of Values are the mean s.d. of concentrations from two aliquots of culture CFU-GM from normal donor CD34+ cells. Stromal layers were established supernatant collected at time of weekly media change. b Ͻ from marrow obtained from one normal donor (ie ND54) and three Significantly greater (P 0.05) than concentration in medium or super- patients after hematological recovery from the first cycle of FLAC + GM- natants from no chemotherapy stromal layers. c ␣ CSF (ie P20) or FLAC + PIXY321 (ie P25 and P31). 2 × 104 marrow Neutralizing antibody to TNF- promoted an increase in CFU-GM num- + bers in contact cocultures of post-chemotherapy P34 stromal layers and CD34 cells were cocultured in direct contact with the irradiated stromal + layers. Four weeks later, the nonadherent and adherent cells were col- normal donor CD34 cells (Table 6), but not in prechemotherapy P34 lected, pooled, washed, counted and subcultured in methylcellulose sup- stromal layer cocultures (Table 5). = = plemented with IL-3, GM-CSF and SCF to stimulate colony formation NT not tested; ND not detectable. from CFU-GM. The values are the mean Ϯ s.d. of CFU-GM numbers per coculture calculated from the number of colonies detected in three methylcellulose cultures. *By Student’s two tailed t-test, significantly different determined as described by St. Groth.40 P values and a pro- (P р 0.01) from the number of CFU-GM detected in normal donor stromal gram which performs the calculations for LTC-IC were layer cocultures (ie ND54). established using Microsoft Excel 5.0. Lipopolysaccharide, E. coli (DIFCO, Detroit, MI, USA). ␣ Immunofluorescent microscopy was used to detect TNF- Results on the surface of cells in the stromal layers. Cells from × 4 irradiated stromal layers were harvested, and 4 10 cells Reduced progenitor cell numbers in marrow from were subcultured in eight-chambered glass slides (Nunc). patients treated with FLAC + PIXY321 Three to 4 days later the cells were fixed in 2% formal- dehyde and 0.1% gluteraldehyde for 10 min, washed three Marrows obtained before chemotherapy and after hematol- times in PBS, and then blocked with a 1:10 dilution of goat ogical recovery from the first or second cycles of FLAC serum for 30 min at 37°C. Cells in each chamber were chemotherapy were enriched for CD34+ cells (у91% incubated for 1 h with 50 ␮l of 500 ␮g/ml neutralizing purity) and then cultured for detection of the committed antibody to TNF-␣ (R&D Systems), washed three times for progenitors, CFU-GM, and the primitive progenitors, LTC- 5 min each in PBS, and then incubated for another hour at IC. The frequency or number of CFU-GM from 1 × 104 37°C with 50 ␮lof5␮g/ml anti-mouse IgG fluorescein CD34+ cells was similar for cells from all three treatment antibody (Pierce, Rockford, IL, USA). After washing three arms and was not significantly different from the frequency times in PBS, the chambers and gaskets were removed from in prechemotherapy or normal donor CD34+ cells (Table 1). the slides, 15 ␮l of Vectashield (Vector Laboratories, Bur- Due to the reduced number of CD34+ cells in lington, CA, USA) were deposited on each stromal area. FLAC + PIXY321 marrows, there was, however, a reduced The slides were coverslipped, sealed, and stored at 4°C number of CFU-GM per ml of marrow aspirate that was until ready to be viewed in a Leitz Orthoplan microscope not observed in FLAC + GM-CSF or FLAC + No cytokine (Leica, Deerfield, IL, USA). marrows. The results also suggest that there was also a reduced number of LTC-IC in FLAC + PIXY321 marrows. These results indicate that PIXY321 administered after Statistics FLAC chemotherapy had suppressive effects on the recov- The Student’s two-tailed t-test was used to test for signifi- ery of both committed and primitive progenitors. cant differences between treated and untreated groups. Values between different groups were considered to be sig- Decreased production of CFU-GM and BFU-E in post- nificantly different for P Ͻ 0.05. Nonparametric Wilcoxon chemotherapy marrow long-term cultures (LTC) matched-pairs rank test of significance was used to compare progenitor cell numbers in marrow prechemotherapy and LTC were used to assess the effect of chemotherapy on post-chemotherapy LTC.39 The multiplicity of LTC-IC was the production of the committed progenitors, CFU-GM and Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 462 Table 5 Effect of neutralizing antibodies on CFU-GM from normal Decreased CFU-GM production from normal donor donor CD34+ cells cocultured in contact with normal donor or prechemo- CD34+ cells cocultured with post-chemotherapy stromal therapy stromal layers layers Source of stromal layers No. of CFU-GM in post-chemotherapy Marrow stromal layers were used to determine whether stromal layer cocultures (% of no. with no a changes in the marrow microenvironment were a potential antibody treatment) cause for the reduced hematopoietic capacity of post- chemotherapy LTC. Normal donor and prechemotherapy No antibody anti-TNF-␣ anti-TNF-␤ CD34+ cells were cocultured in contact with one normal P27 (Prechemotherapy) donor and two prechemotherapy stromal layers (Table 2). Nonadherent cells 329 ± 52 471 ± 15 276 ± 55 The number of CFU-GM detectable after 1 week of cocul- Adherent cells (100) (143)b (84) ture was similar for all three stromal layers and varied 907 ± 271 996 ± 55 889 ± 227 depending on the source of CD34+ cells. The effect of post- (100) (110) (98) chemotherapy stromal layers on production of CFU-GM P34 (Prechemotherapy) Nonadherent cells 51 ± 18 41 ± 27 NDc was therefore compared to normal donor or prechemo- Adherent cells (100) (80) ND therapy stromal layers cocultured on the same day with the 376 ± 31 362 ± 83 same source of CD34+ cells. (100) (96) The number of CFU-GM and BFU-E produced from nor- ND54B (normal donor) 700 ± 33 823 ± 4ND + ND57C (normal donor) (100) (117)b ND mal donor CD34 cells cocultured with pre- and post- ND58A (normal donor) 1083 ± 88 952 ± 131 ND chemotherapy stromal layers was determined. After 1 week, (100) (88) there were significantly fewer CFU-GM in both the adher- 2647 ± 189 2059 ± 130 ent and nonadherent fraction of post-chemotherapy P25 (100) (78) stromal layer cocultures than in prechemotherapy P27 stromal layer cocultures (Figure 2). There was a 1.3-fold aIrradiated stromal layers were cocultured with normal donor CD34+ cells (different source of target cells for each stromal layer coculture), and neu- increase in total CFU-GM numbers in prechemotherapy tralizing antibodies were added on 0, 2, 4 and 7 days after addition of P27 stromal cocultures, however, the number from P25 CD34+ cells. On day 7 or 8 the cells were harvested, washed and subcul- post-chemotherapy stromal cocultures had decreased by tured in methylcellulose for the detection of CFU-GM. Nonadherent and 45%. Similarly, there was a decreased production of BFU- adherent cells from normal donor cocultures were pooled before assay for E. Fewer CFU-GM and BFU-E were also observed in post- CFU-GM. Values are the mean Ϯ s.d. calculated from colony numbers from three methylcellulose cultures each. chemotherapy P18 stromal layer cocultures than in pre- bBy Student’s two tailed t-test, the number of CFU-GM in antibody treat- chemotherapy P20 stromal layer cocultures (Table 3). In ment group was significantly different (P Ͻ 0.05) from the number in the another experiment, after 4 weeks of coculture the total no antibody treated control cocultures. c number of CFU-GM in P20, P25 and P31 post-chemo- Not determined. therapy stromal layer cocultures was only 21% (P = 0.004), 57% (P = 0.01) and 59% (P = 0.01), respectively, of the number in the normal donor stromal coculture (ie ND54) (Figure 3). A reduced number of CFU-GM was observed in both 1 (Figure 2) and 4 week P25 post-chemotherapy BFU-E, from more primitive hematopoietic progenitors. + LTC were established from marrow obtained from 19 stromal cocultures cocultured with different CD34 target patients prior to chemotherapy and after their hematological cells and even when multiple recombinant factors were recovery from the first or second cycle of FLAC chemo- used to stimulate colony formation. These results demon- therapy. For 13 of those patients, the number of CFU-GM strate that some post-chemotherapy stromal layers had a and BFU-E in post-chemotherapy marrow used to establish reduced capacity to support the maintenance or production LTC was either similar to or greater than the number in of committed progenitors. prechemotherapy marrows (two out of two FLAC + No cytokine, seven out of eight FLAC + GM-CSF and four out ␣ + Detection of TNF- in post-chemotherapy LTC and of nine FLAC PIXY321). The number of progenitors pro- stromal layer cocultures duced in LTC of pre- and post-chemotherapy marrows from those 13 patients was compared (Figure 1). Within the first In preliminary studies, culture supernatants from several 2 weeks, post-chemotherapy LTC had approximately 70% LTC were collected 1 week after medium change and fewer progenitors than prechemotherapy LTC. During the assayed for TNF-␣, MIP-1␣, TGF-␤, and IFN-␥; four fac- 5 weeks of culture there were progressively fewer CFU- tors shown to have some suppressive effects on in vitro GM and BFU-E in post-chemotherapy LTC than in pre- hematopoiesis (Table 4). Increased concentrations of TNF- chemotherapy LTC of marrow from the same patients. ␣ were detected in culture supernatants from two (ie P02 Results in a previous report,22 demonstrated that post- and P34) out of five post-chemotherapy stromal layers. For chemotherapy LTC had a reduced capacity for production example, for P02, TNF-␣ increased from 4 Ϯ 0 pg/ml in of committed progenitors. Results in the present report prechemotherapy cultures to 116 Ϯ 7 pg/ml in culture further demonstrate a similar reduced capacity for pro- supernatants from post-chemotherapy stromal layers. No duction of committed progenitors even when the number increased concentrations of MIP-1␣, TGF-␤, or IFN-␥ were of CFU-GM and BFU-E in marrow used to initiate LTC detected in post-chemotherapy LTC supernatants collected was similar in pre- and post-chemotherapy marrows. 1 week after media replacement. Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 463 Table 6 Effect of neutralizing antibodies on CFU-GM from normal donor CD34+ cells cocultured in contact with post-chemotherapy stromal layers

Source of stromal layers No. of CFU-GM in post-chemotherapy stromal layer cocultures (% of no. with no antibody treatment)a

No antibody anti-TNF-␣ anti-TNF-␤ anti-IL-4

P20 (FLAC + GM-CSF) Nonadherent cells 1042 ± 143 (100) 650 ± 186 (62) NDc ND Adherent cells 1367 ± 224 (100) 2544 ± 94 (186)b ND ND P25 (FLAC + PIXY321) 462 ± 126 (100) 507 ± 61 (117) 587 ± 107 (127) 213 ± 61 (46)b P34 (FLAC + No cytokine) 408 ± 84 (100) 603 ± 18 (148)b ND 389 ± 88 (95) P38 (1 year FLAC + PIXY321) 1142 ± 94 (100) 1707 ± 212 (149)b 1147 ± 329 (100) 2094 ± 163 (183)b P60 (CTX + G-CSF) 1148 ± 124 (100) 1889 ± 269 (164)b 1426 ± 213 (124) 1569 ± 246 (137) aIrradiated stromal layers were cocultured with normal donor CD34+ cells (different source of target cells for each experiment), and neutralizing antibodies were added on 0, 2, 4 and 7 days after addition of CD34+ cells. On day 7 or 8 the cells were harvested, washed, and subcultured in methylcellulose for the detection of CFU-GM. Values are the mean Ϯ s.d. calculated from colony numbers from three methylcellulose cultures each. bBy Student’s two-tailed t-test, the number of CFU-GM in antibody treatment group was signiﬁcantly different (P Ͻ 0.05) from the number in the no antibody treated control cocultures. cNot determined.

Neutralizing antibodies were added to stromal layer Table 7 Stimulation of TNF-␣ release from cytokine pretreated stro- cocultures to determine whether TNF-␣ might be one cause mal layers for the suppressive effects of post-chemotherapy stromal layers. Treatment with anti-TNF-␣ had little effect on the Cytokine pretreatmenta Concentration of TNF-␣ (pg/ml of media)b number of CFU-GM from normal donor CD34+ cells in noncontact cocultures with normal donor, pre-, or post- ND57C ND73 stromal ND241 stromal layer layer stromal layer chemotherapy stromal layers (Figure 4). In contrast, in ␣ post-chemotherapy stromal contact cocultures, anti-TNF- DPBS (no factor control) 676 ± 43 343 ± 36 384 ± 13 treatment promoted a 1.4-fold increase in the total number GM-CSF 2765 ± 56c 550 ± 32c 1896 ± 33c + of CFU-GM from normal donor CD34 cells. A significant PIXY321 1243 ± 27c 767 ± 18c 2192 ± 116c increase was observed for four out of five post-chemo- Medium alone 19 ± 8NDND therapy stromal layer cocultures. Neutralizing antibody to TNF-␤ (another goat IgG antibody) did not promote an aStromal layers were established from marrow of normal donors, irradiated to eliminated residual hematopoietic progenitors, and then incubated with increase in CFU-GM numbers in either pre- (Table 5) or DPBS, GM-CSF, or PIXY321 for 2 days, washed and 3 days later medium post-chemotherapy (Table 6) stromal layer cocultures. was changed and LPS added to the cultures. Neither anti-TGF-␤ nor anti-LIF promoted an increase in bSupernatants from stromal layer cultures were collected 18 h after media CFU-GM numbers from P20 and P34 post-chemotherapy change or LPS stimulation. Values are the mean ± s.d. of two or four stromal layer cocultures (data not shown). Anti-IL-4 pro- measurements. cBy Student’s two-tailed t-test significantly different (P Ͻ 0.01) from the moted an increase in CFU-GM numbers in one (ie, P38) concentration detected in DPBS treated stromal layers. out of three post-chemotherapy stromal layer cocultures ND = not detected. (Table 6). These results suggest that increased levels of TNF-␣ and IL-4 might be one cause for the reduced capacity of post-chemotherapy stromal layers to support were approximately 1.6- to 5.7-fold greater than concen- hematopoiesis. trations detected in noncytokine pretreated stromal layers. These results demonstrate that cells in GM-CSF and Decreased numbers of CFU-GM from normal donor PIXY321 pretreated stromal layers had an increased + capacity to release TNF-␣. CD34 cells cocultured with cytokine pretreated stromal layers The effect of cytokine pretreated stromal layers on the number of CFU-GM from normal donor CD34+ cells was determ- Immunofluorescence studies were done to determine ined. In stromal noncontact cocultures, stromal layers preincu- whether GM-CSF and PIXY321 had an effect on the pro- bated with 2, 20 or 200 ng/ml of either GM-CSF or PIXY321 duction or release of TNF-␣ from normal donor stromal did not have suppressive effects on cell or CFU-GM numbers layers (Figure 5). Only low levels of TNF-␣ were detect- from CD34+ target cells (n = 10, data not shown). In contrast, able on the surfaces of cells in unstimulated cytokine pre- contact cocultures of normal donor CD34+ cells with stromal treated and LPS-stimulated noncytokine pretreated stromal layers pretreated with 20 ng/ml PIXY321 was reduced to layers (Figure 5a and b). Increased expression of TNF-␣ 73 Ϯ 5% (mean Ϯ s.e.m. for n = 3 different donor stromal was observed on most cells after LPS stimulation of GM- layers) of the number of CFU-GM in cocultures with noncyto- CSF (Figure 5c and d), PIXY321 (Figure 5e and f) and IL- kine pretreated stromal layers (Table 8). A reduced number 3 pretreated stromal layers (data not shown). TNF-␣ was of CFU-GM was also observed in cocultures with GM-CSF also detected in media from LPS-stimulated stromal layers pretreated stromal layers of marrow from two out of four (Table 7). The concentrations of TNF-␣ in media from donors. Anti-TNF-␣ did not promote an increase in CFU-GM GM-CSF and PIXY321 pretreated stromal layers, however, numbers from PIXY321 pretreated stromal layers. When there Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 464 200 No chemotherapy

Post-chemotherapy * 150

100

50 % Number in untreated

0 Noncontact Contact Type of stromal cocultures

Figure 4 Effect of anti-TNF-␣ on CFU-GM numbers from normal donor CD34+ cells cocultured with marrow stromal cell layers. Stromal layers were established from marrow obtained from normal donors or patients prior to chemotherapy and from patients after hematological recovery from FLAC or CTX + G-CSF treatment. Normal donor CD34+ cells were cocultured either in Transwells (ie stromal noncontact) or directly on (ie stromal contact) irradiated stromal layers. Neutralizing antibody to TNF-␣ was added to the cultures on days 0, 2 and 4 after coculture with CD34+ cells. Cells were harvested on day 7 or 8 and subcultured in methylcellulose supplemented with PHA-LCM to stimulate colony formation from CFU- Figure 5 Upregulation of TNF-␣ release in cytokine pretreated normal Ϯ GM. The mean s.e.m. percentage of CFU-GM numbers in cultures after donor stromal layers. Established stromal layers were irradiated to elimin- ␣ anti-TNF- treatment compared to numbers in untreated cocultures was ate residual hematopoietic progenitors and 1 week later the layers were calculated. Values are from three stromal noncontact and five stromal con- incubated with DPBS, GM-CSF or PIXY321. Two days later the layers tact prechemotherapy cocultures and from eight stromal noncontact (three were washed three times with fresh media to remove remaining cytokine + + + FLAC No cytokine, three FLAC PIXY321 and two FLAC GM-CSF) and the following week the cells were detached and subcultured in glass + and five stromal contact post-chemotherapy (one FLAC No cytokine, chambered slides. Once the cells had spread (approximately 3 days) the + + + two FLAC PIXY321, one FLAC GM-CSF, one CTX G-CSF) post- cells were stimulated with LPS and then 18 h later the layers were labeled chemotherapy cocultures. *By Student’s two tailed t-test, significantly dif- with anti-TNF-␣ and secondary fluorescent antibody. (a) and (b) show = ferent (P 0.005) from the number of CFU-GM detected in untreated stro- the immunofluorescent and corresponding differential interference contrast mal layer cocultures. image, respectively, of cells incubated with DPBS + LPS, (c) and (d) show cells incubated with GM-CSF + LPS and (e) and (f) show cells incubated with PIXY321 + LPS. White diffuse areas are positive for TNF-␣ and the bright spots are autofluorescent areas within the cells. Magnification × 117. were fewer CFU-GM in GM-CSF pretreated stromal layer cocultures than in DPBS treated layers then there was an increase in CFU-GM numbers with anti-TNF-␣ treatment that macrophage and erythroid progenitors. Reduced numbers was not observed in PIXY321 pretreated cultures (Figure 6a). of CFU-GM were also detected in GM-CSF and PIXY321 For the two other donor marrows, there were fewer and pretreated stromal layers. The results indicate that TNF-␣ larger CFU-GM-derived colonies after anti-TNF-␣ treat- was one possible cause for some of the suppressive effects ment (Figure 6b). Deterioration in integrity of GM-CSF of post-chemotherapy and GM-CSF pretreated stromal lay- pretreated stromal layers was partially prevented by treat- ers, but not for the suppressive effects of PIXY321 pre- ment with anti-TNF-␣, but had no apparent effect on the treated stromal layers. deterioration of PIXY321 pretreated stromal layers. These A previous study demonstrated that long-term cultures results suggest that TNF-␣ mediated some of the hematopo- (LTC) of marrow obtained after FLAC chemotherapy pro- ietic suppressive effects of GM-CSF pretreated stromal lay- duced 50–90% fewer erythroid and granulocyte–macro- ers, but was not the primary cause for the hematopoietic phage progenitors than prechemotherapy LTC from the suppressive effects of PIXY321 pretreated stromal layers. same patients.22 This decrease was observed even when the number of committed progenitors had recovered to prechemotherapy numbers and whether or not the patients had Discussion received either GM-CSF or PIXY321 to enhance their blood cell recovery. Damage to the progenitor/stem cell In the present work, the effect of normal donor, GM-CSF compartment or the marrow microenvironment or both has and PIXY321 pretreated normal donor and patient post- been implicated as potential causes for the reduced capacity chemotherapy marrow stromal layers on the proliferation of post-chemotherapy LTC to support hematopo- of normal donor CD34+ cells was investigated. The results iesis.7,19,21,23,24,26,27 demonstrate that post-chemotherapy marrow stromal layers In the present study, stromal layers established from were deficient in supporting the production of granulocyte– post-chemotherapy marrow were cocultured with normal Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 465 200 a Table 8 Effect of cytokine pretreated stromal layers on CFU-GM No antibody numbers from normal donor CD34+ cells Anti-TNF-α Normal donor No. of CFU-GM after 1 week (% of number in 150 marrow source DPBS pretreated)a

* * + * * Stromal CD34 DPBS PIXY321 GM-CSF layer cells pretreated pretreated pretreated 100 * ND54B ND42E 700 ± 33 565 ± 19 397 ± 25 ND57C ND36C (100) (81)b (57)b * ND58A ND57C 1083 ± 88 698 ± 44 1231 ± 43 ND38H ND54A (100) (64)b (114) 50 2647 ± 49 1945 ± 257 3391 ± 383 (100) (73)b (128)b % of number in untreated cultures 2254 ± 133 2229 ± 243 1799 ± 132 (100) (99) (80)b 0 aIrradiated stromal layers were cocultured with 2 × 104 normal donor None GM-CSF PIXY321 + Pretreatment of stromal layer CD34 cells. One week later, nonadherent and adherent cells were harvested, pooled and subcultured in methylcellulose with IL-3, GM-CSF and SCF to stimulate the proliferation of CFU-GM. Values are the mean Ϯ s.d. of CFU-GM per stromal coculture calculated from the number of colonies 200 b from three methylcellulose cultures. No antibody bBy Student’s two-tailed t-test signiﬁcantly different (P Ͻ 0.05) than the number of CFU-GM in DPBS pretreated stromal layer cocultures. Anti-TNF-α

150 donor CD34+ cells. For four out of five different patient marrows, the number of CFU-GM detected in contact post- chemotherapy stromal layer cocultures was only 21–60% 100 of the number in normal donor or prechemotherapy stromal layer cocultures. This decrease was observed after 1 and 4 * * * weeks of culture. Results in the present study demonstrate that production of CFC was suppressed by post-chemo- 50 therapy stromal layers. For example, there was a 1.3-fold increase in the number of CFU-GM from umbilical cord % of number in untreated cultures blood CD34+ cells cocultured for 1 week with prechemotherapy stromal layers. In contrast, the number of CFU- 0 GM from the same target CD34+ cells cocultured with post- None GM-CSF PIXY321 chemotherapy stromal layers was only 55% of the number Pretreatment of stromal layer at day 0. Although only a few patient stromal layers were Figure 6 Anti-TNF-␣ effects on CFU-GM numbers in cocultures of nor- evaluated, results in the present study corroborate obser- mal donor CD34+ cells with GM-CSF and PIXY321 pretreated stromal vations reported by others showing that stromal layers of layers. Established stromal layers were irradiated to eliminate residual marrow exposed in vitro or in vivo to chemotherapeutic hematopoietic progenitors, and the following day the media were changed drugs were deficient in supporting the proliferation of and DPBS or 20 ng/ml PIXY321 or GM-CSF was added to the cultures. 7,19,21,41 Two days later the layers were washed at least three times to remove myeloid as well as lymphoid precursors. cytokine; after an additional 3 days, 2 × 104 CD34+ normal donor marrow Fewer CFU-GM were detected when normal donor cells were cocultured in direct contact with the stromal layers. On days CD34+ cells were cocultured in contact with PIXY321 pre- 0, 2 and 4, DPBS or neutralizing antibody to TNF-␣ was added to the treated stromal layers and 50% of GM-CSF pretreated stro- cocultures. On day 7, the nonadherent and adherent cells from stromal mal layers than were detected in untreated stromal layers. cocultures were collected, washed, counted and subcultured in methylcellulose supplemented with IL-3, GM-CSF and SCF to stimulate colony In addition to the effects of chemotherapy alone, some hem- formation from CFU-GM. The values are the mean Ϯ s.d. number of atopoietic stimulatory factors have been shown to have CFU-GM calculated from three methylcellulose cultures each. (a) Experi- further adverse effects on marrow content and func- ment with reduced numbers of CFU-GM from GM-CSF pretreated stromal tion.22,25,42 For example, cumulative production of CFU- layers (ND54B stromal layer in Table 8). (b) Experiment with similar numbers of CFU-GM in untreated and GM-CSF pretreated stromal layers GM in LTC of marrow obtained from patients still adminis- (ND58A stromal layer in Table 8). *By Student’s two tailed t-test, signifi- tered either GM-CSF or IL-3 after autologous marrow cantly different (P Ͻ 0.05) from the number of CFU-GM detected in stro- transplantation was only approximately one-third the num- mal layer cocultures with no cytokine or antibody treatment; **signifi- ber of CFU-GM produced in LTC of marrow from patients cantly different from number in no antibody treatment of similar not administered cytokine after marrow transplantation.42 pretreated groups. This difference was no longer evident in LTC of marrow obtained from the same patients 4 weeks after cytokine treatment was discontinued. Whether the decreased production of CFC in the LTC was due to fewer hematopoietic progenitors in marrow from cytokine treated patients or the Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 466 result of effects of the cytokines on function of the marrow TNF-␣ concentration in the media was 2- to 6-fold times microenvironment was not reported. Results in the present higher in media from PIXY321 and GM-CSF pretreated study demonstrate that both GM-CSF and PIXY321 had stromal layers than in DPBS pretreated stromal layers. suppressive effects on in vitro hematopoiesis by their inter- Others observed that in vivo or in vitro GM-CSF treatment action with cells of the marrow microenvironment. was associated with an increase in TNF-␣.45–48 For When normal donor CD34+ cells were cocultured in con- example, Perkins et al46 demonstrated that LPS-stimulated tact with PIXY321 pretreated stromal layers the number of peripheral blood monocytes from patients administered CFU-GM was reduced to approximately 73% of the num- GM-CSF released 8-fold more TNF-␣ than monocytes ber in cocultures with untreated stromal layers. This obtained from the same patients prior to GM-CSF. In the decrease was not observed when CD34+ target cells were present study, some of the effects of in vivo administered placed in Transwells over the stromal layers (ie stromal GM-CSF could be reproduced by treating stromal cell lay- noncontact cocultures). Using stromal noncontact cocul- ers with the cytokine in vitro. Macrophages and monocytes tures, previous studies demonstrated that normal donor and contain pools of mRNA for TNF-␣ and the increase in HIV-infected marrow stromal layers produced factors that TNF-␣ after stimulation with LPS is the result of acceler- stimulated the proliferation of progenitors found in normal ated gene transcription and translation of stored, as well donor marrow, peripheral blood and umbilical cord blood as new mRNA.49 One consequence of chemotherapy and cells.35,37,43 Like the results in the present study, direct con- cytokine treatment may be an increase in levels of TNF-␣ tact of the hematopoietic progenitors with the stromal lay- mRNA that are rapidly expressed in response to LPS or ers was necessary to detect suppressive effects of the stro- other factors produced by stromal layers (for example, mal layers on the maintenance and proliferation of IL-1) or CD34+ target cells. hematopoietic progenitors.37,43 Fibroblasts, macrophages, When the number of CFU-GM from normal donor CD34+ endothelial cells, and adipocytes are the primary cell types was reduced in GM-CSF pretreated stromal layer cocultures, found in stromal cell layers that develop in marrow long TNF-␣ antibody treatment promoted an increase to the num- term cultures; a cellular composition that is similar to that ber observed in untreated stromal cocultures. Anti-TNF-␣ observed for the in vivo marrow microenvironment.15,16 however, promoted a reduction in total number, but an Marrow stromal layers express a multitude of factors that increase in the number of large colonies (many were may be important in the positive and negative regulation macroscopic) when the number of CFU-GM was not reduced of hematopoiesis.16,28,29 Downregulation or upregulation of in untreated GM-CSF pretreated stromal layers. In addition the expression or production of some of those factors has to its inhibitory effects, TNF-␣ hasbeenshownalsotohave been implicated as a possible cause for the reduced capacity stimulatory effects on in vitro hematopoiesis, and the results of stromal layers to support in vitro hematopoiesis that is are dependent upon the cell lineage and maturation stage of observed with some malignancies, viral infections and poor the hematopoietic progenitors.50,51 In addition, TNF-␣ stimu- marrow recovery after marrow transplantation.19,30–34,44 Pre- lates the release of other factors.52–54 Thevariableeffectsof vious studies implicated TNF-␣, IL-4 and IFN-␣ as factors GM-CSF pretreated stromal layers with and without treat- partially responsible for the reduced capacity of HIV- ment with anti-TNF-␣ on the number and proliferative infected stromal layers to support the proliferation of nor- capacity of CFU-GM may reflect some of the different effects mal donor CD34+ cells.33,34 of TNF-␣ on populations of mature and primitive progenitors Like the results in a previous study,34 there was no sig- and stromal cells. nificant effect of TNF-␣ neutralizing antibody treatment on In contrast to the effects observed in GM-CSF pretreated the total number of CFU-GM produced in cocultures of stromal layer cocultures, TNF-␣ antibody treatment had no normal donor CD34+ cells with normal donor or prechemo- effect on the reduced number of CFU-GM observed in therapy stromal layers. In contrast, anti-TNF-␣ promoted a PIXY321 pretreated stromal layer cocultures. Similarly, 1.3-fold to 1.6-fold significant increase in CFU-GM num- deterioration in integrity of GM-CSF pretreated stromal bers in stromal layer cocultures for four out of five post- layers was partially prevented by treatment with anti-TNF- chemotherapy marrows. This increase was observed for ␣, but neither anti-TNF-␣ nor anti-IL-4 treatment delayed stromal layers of marrow from patients whether or not they the deterioration of PIXY321 pretreated stromal layers. had been administered cytokines after chemotherapy. An These results indicate that other factors or conditions were increased number of CFU-GM was observed after anti-IL- responsible for the inhibitory effects of PIXY321 pretreated 4 in only one out of three patient post-chemotherapy stro- stromal layers. In addition to possible upregulation of other mal layer cocultures. Interestingly, that patient developed known or unknown negative regulators of hematopoiesis, prolonged thrombocytopenia and neutropenia during FLAC other causes for the negative effects of PIXY321 pretreat- chemotherapy, and her blood counts were still below nor- ment include insufficient levels of membrane bound stimu- mal a year after treatment27 (unpublished observations). latory or survival factors with alterations in extracellular These results suggest that one consequence of chemo- matrix and hematopoietic progenitor cell interactions.16,17,31 therapy was an increase in production or release of TNF- Results in the present report demonstrate that post- ␣ from cells of the marrow microenvironment. chemotherapy as well as GM-CSF and PIXY321 pretreated Only low levels of TNF-␣ were detectable on the cell marrow stromal layers are deficient in their ability to sup- surface or in culture media from normal donor untreated or port hematopoiesis from normal donor CD34+ cells. The cytokine pretreated stromal layers. After LPS stimulation results also indicate that an increase in release or production there was an increase in cell surface expression of TNF-␣ of TNF-␣ is one cause for the early hematopoietic suppress- in GM-CSF and PIXY321 pretreated stromal layers. The ive effects of post-chemotherapy and GM-CSF pretreated Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 467 stromal layers. TNF-␣ is a multifunctional factor that has blood stem cell and bone marrow recipients. Bone Marrow direct effects on hematopoietic progenitors.51 However, Transplant 1997; 19: 557–563. either alone or in combination with other factors, TNF-␣ 9 Tarella C, Caracciolo D, Gavarotti P et al. Circulating pro- also induces the release or production of IP-10, Mig, IL-8 genitors following high-dose sequential (HDS) chemotherapy and MIP-1␣ from monocytes/macrophages, fibroblasts, or with G-CSF: short intervals between drug courses severely 52,53 impair progenitor mobilization. Bone Marrow Transplant endothelial cells, three of the cell types that comprise 1995; 16: 223–228. the marrow microenvironment.15,16 Mig (monokine ␥ ␣ 10 Akard LP, Wiemann M, Thompson JM et al. Impaired stem inducible by interferon- ), MIP-1 , (monokine inducible cell collection by consecutive courses of high dose-mobilizing by interfer-␥) on IP-10 and IL-8 are several of the chemo- chemotherapy using cyclophosphamide, etoposide, and G- kines shown to suppress the proliferation or production of CSF. J Hematother 1996; 5: 271–277. hematopoietic progenitors.38,55–56 What role these chemo- 11 Drake M, Ranaghan L, Morris TCM et al. Analysis of effect kines may have in producing the suppressive effects of of prior therapy on progenitor cell yield: use of a chemo- post-chemotherapy and cytokine pretreated stromal layers therapy scoring system. Br J Haematol 1997; 98: 745–749. is not known. Further defining the mechanisms responsible 12 Sutherland HJ, Eaves AC, Eaves CJ. Quantitative assays for for the deficiencies in the functional integrity of the marrow human hemopoietic progenitor cells. In: Gee AP (ed) Bone microenvironment observed after chemotherapy and cyto- Marrow Processing and Purging. CRC Press: Boston, 1991, pp 155–171. kine treatment may be important in understanding the 13 Civin CI, Gore SD. Antigenic analysis of hematopoiesis: a causes underlying the atypical hematopoiesis observed after review. J Hematother 1993; 2: 137–144. dose-intensive chemotherapy and autologous transplan- 14 Verfaille CM, Miller JS. CD34+/CD33− cells reselected from tation. macrophage inflammatory-1␣ + interleukin-3-supplemented ‘stromal-noncontact’ cultures are highly enriched for long- term bone marrow culture initiating cells. Blood 1994; 84: Acknowledgments 1442–1449. 15 Allen TD, Dexter TM. The essential cells of the hemopoietic We thank Ms Janet Szabo and Ms Xiao-Yi Yan for their skillful microenvironment. Exp Hematol 1984; 12: 517–521. technical assistance. 16 Mayani H, Guilbert LJ, Janowska-Wieczorek A. Biology of the hemopoietic microenvironment. Eur J Haematol 1992; 49: 225–233. 17 Yoder MC, Williams DA. Minireview. Matrix molecule inter- References actions with hematopoietic stem cells. Exp Hematol 1995; 23: 961–967. 1 Hyrniuk WM, Levine MN, Levine L. The implications of dose 18 Bhavnani M, Morris-Jones PH, Testa NG. Children in long- intensity in chemotherapy of metastic breast cancer. J Clin term remission after treatment for acute lymphoblastic leu- Oncol 1984; 2: 1281–1288. kaemia show persisting haemopoietic injury in clonal and 2 Fisher DC, Peters WP. Advances in the clinical use of gra- long-term cultures. Br J Haematol 1989; 71: 37–41. nulocyte colony-stimulating factor and granulocyte–macro- 19 Greenberger JS. Toxic effects on the hematopoietic microen- phage stimulating factor to intensify cancer chemotherapy. vironment. Exp Hematol 1991; 19: 1101–1109. Curr Opin Hematol 1996; 1: 221–227. 20 Betticher DC, Huxol H, Muller R et al. Colony growth in 3 Bishop JF. Platelet support and the use of cytokines. Stem cultures from bone marrow and peripheral blood after curative Cells 1994; 12: 370–377. treatment for leukemia and severe aplastic anemia. Exp Hema- 4 O’Shaughnessy JA, Venzon DJ, Gossard M et al. A phase I tol 1993; 21: 1517–1521. study of sequential versus concurrent interleukin-3 and granulocyte–macrophage colony stimulating factor in advanced 21 Novitzky N, Mohamed R. Alterations in both the hematopo- breast cancer patients treated with FLAC (5-fluorouracil, leu- ietic microenvironment and the progenitor cell population fol- covorin, doxorubicin, cyclophosphamide) chemotherapy. low the recovery from myeloablative therapy and bone mar- Blood 1995; 86: 2913–2921. row transplantation. Exp Hematol 1995; 23: 1661–1666. 5 O’Shaughnessy JA, Tolcher A, Riseberg D et al. Prospective, 22 Schwartz GN, Hakim F, Zujewski J et al. Early suppressive randomized trial of 5-fluorouracil, leucovorin, doxorubicin, effects of chemotherapy and cytokine treatment on committed and cyclophosphamide chemotherapy in combination with versus primitive haemopoietic progenitors in patient bone interleukin-3/granulocyte–macrophage colony-stimulating fac- marrow. Br J Haematol 1996; 92: 537–547. tor (GM-CSF) fusion protein (PIXY321) versus GM-CSF in 23 Novitzky N, Mohammed R. Alterations in the progenitor cell patients with advanced breast cancer. Blood 1996; 87: population follow recovery from myeloablative therapy and 2205–2211. bone marrow transplantation. Exp Hematol 1997; 25: 471– 6 Brandwein JM, Callum J, Sutcliffe SB et al. Analysis of fac- 477. tors affecting hematopoietic recovery after autologous bone 24 Domenech J, Roingeard F, Binet C. The mechanisms involved marrow transplantation for lymphoma. Bone Marrow Trans- in the impairment of hematopoiesis after autologous bone mar- plant 1990; 6: 291–294. row transplantation. Leuk Lymphoma 1197; 24: 239–256. 7 Straetmans, N, Ma DD, Herman P et al. Long-term culture of 25 Schwartz GN, Neta R, Vigneulle RM et al. Recovery of hema- autologous transplanted bone marrow for acute myeloid leu- topoietic colony-forming cells in irradiated mice pretreated kaemia: evidence for an in vitro haemopoietic defect and lack with interleukin-1 (IL-1). Exp Hematol 1988; 16: 752–757. of correlation with the speed of engraftment. Bone Marrow 26 Horung RL, Lango DL. Hematopoietic stem cell depletion by Transplant 1995; 15: 421–428. restorative growth factor regimens during repeated high dose 8 Bentley SA, Brecher ME, Powell E et al. Long-term cyclophosphamide therapy. Blood 1992; 88: 77–83. engraftment failure after marrow ablation and autologous 27 Warren MK, Zujewski J, Rose WL et al. Early suppressive hematopoietic reconstitution: differences between peripheral effects of chemotherapy on recovery of bone marrow megaka- Post-chemotherapy stromal layers suppress hematopoiesis GN Schwartz et al 468 ryocyte precursors: possible relationship to platelet recovery. and blood after autologous transplantation with mafosfamide- Stem Cells 1996; 12: 31–37. purged marrow in lymphoid malignancies. Bone Marrow 28 Eaves CJ, Cashman JD, Kay RJ et al. Mechanisms that regu- Transplant 1994; 14: 253–259. late the cell cycle status of very primitive hematopoietic cells 43 Verfaillie CM. Direct contact between human primitive hema- in long-term human marrow cultures. II. Analysis of positive topoietic progenitors and bone marrow stroma is not required and negative regulators produced by stromal cells with the for long-term in vitro hematopoiesis. Blood 1992; 79: 2821– adherent layer. Blood 1991; 78: 110–117. 2826. 29 Sensebe L, Mortensen BT, Fixe P et al. Cyokines active on 44 Migliaccio AR, Migliaccio G, Johnson G et al. Comparative granulomonopoiesis: release and consumption by human mar- analysis of hematopoietic growth factors released by stromal row myeloid stromal cells. Br J Haematol 1997; 98: 274–282. cells from normal donors or transplanted patients. Blood 1990; 30 Roingeard F, Domenech J, Dayan A et al. Cytokine pro- 75: 305–312. duction in long term cultures after autologous bone marrow 45 Rabinowitz J, Petros WP, Stuart AR, Peters WP. Characteriz- transplantation. Bone Marrow Transplant 1995; 15: 741–747. ation of endogenous cytokine concentrations after high-dose 31 Duhrsen U, Hossfeld DK. Review article: stromal abnormali- chemotherapy with autologous bone marrow support. Blood ties in neoplastic bone marrow diseases. Ann Hematol 1996; 1993; 81: 2452–2459. 73: 53–70. 46 Perkins RC, Vadhan-Raj S, Scheule RK et al. Effects of con- 32 Taichman RS, Nassiri MR, Reilly MJ et al. Infection and rep- tinuous high dose rhGM-CSF infusion on human monocyte lication of human cytomegalovirus in bone marrow stromal activity. Am J Hematol 1993; 43: 279–285. cells: effects on the production of IL-6, MIP-1alpha,and TGF- 47 Williams MA, Kouroumoussis I, Syndercombe-Court D et al. beta1. Bone Marrow Transplant 1997; 19: 471–480. Administration of recombinant human granulocyte–macro- 33 Schwartz GN, Kessler S, Francis ML et al. Negative regu- phage colony-stimulating factor after chemotherapy regulates lators may mediate hematopoietic suppression in HIV-infected the expression and secretion of monocyte tumor necrosis fac- bone marrow long term cultures. In: Guigon M (ed). The tor (TNF) and TNF receptors p55 and p75. Blood 1995; 86: Negative Regulation of Hematopoiesis. John Libbey Eurotext: 4234–4242. London: 1993, pp 403–405. 48 Sisson SD, Dinarello CA. Production of interleukin-1␣, 34 Schwartz GN, Kessler SW, Szabo JM et al. Negative regu- interleukin-1␤ and tumor necrosis factor by human mono- lators may mediate some of the inhibitory effects of HIV-1 nuclear cells stimulated with granulocyte–macrophage colony- infected stromal cell layers on erythropoiesis and myelopoiesis stimulating factor. Blood 1988; 72: 1368–1374. in human bone marrow long term cultures. J Leuk Biol 1995; 49 Kovacs EJ. Control of IL-1 and TNF-␣ production at the level 57: 948–955. of second messenger pathways. In: Kimball ES (ed). Cytokines 35 Schwartz GN. Use of Gel-Well culture chambers as a liquid and Inflammation. CRC Press: Boca Raton ,1991, pp 91–104. culture system to measure responses of hematopoietic colony- 50 Khoury E, Lemoine FM, Baillou C et al. Tumor necrosis fac- forming cells to growth factors. Int J Cell Cloning 1989; 7: tor alpha in human long-term bone marrow cultures: distinct 360–372. effects on nonadherent and adherent progenitors. Exp Hematol 36 Warren MK, Rose WL, Beall LD, Cone J. CD34+ cell expan- 1992; 20: 991–997. sion and expression of lineage markers during liquid culture 51 Caux C, Saeland S, Favre C etal. Tumor necrosis factor-␣ of human progenitor cells. Stem Cells 1995; 13: 167–174. strongly potentiates interleukin-3 and granulocyte–macro- 37 Schwartz GN, Kessler SW, Rothwell SW et al. Inhibitory phage colony-stimulating factor-induced proliferation of effects of HIV-1-infected stromal cell layers on the production human CD34+ hematopoietic progenitor cells. Blood 1990; 75: of myeloid progenitor cells in human long-term bone marrow 2292–2298. cultures. Exp Hematol 1994; 22: 1288–1296. 52 Baggiolini B, Dewald B, Moser B. Human chemokines: an 38 Schwartz GN, Liao F, Gress RE, Farber JM. Suppressive update. Adv Rev Immunol 1997; 15: 675–705. effects of recombinant human monokine induced by IFN- 53 Ebnet K, Simon MM, Shaw S. Regulation of chemokine gene gamma (rHuMig) chemokine on the number of committed and expression in human endothelial cells by proinflammatory primitive hemopoietic progenitors in liquid cultures of CD34+ cytokines and Borrelia burgdorfei. Ann NY Acad Sci 1996; human bone marrow cells. J Immunol 1997; 159: 895–904. 797: 107–117. 39 Hamburg M. Comparison of two populations using paired 54 Parker AN, Pragnell IB. Inhibitors of haemopoiesis and their observations (the Wilcoxon Matched- Pairs Signed Rank potential clinical relevance. Blood Rev 1995; 9: 226–233. Test). Statistical Analysis for Decision Making, 2nd edn. Har- 55 Bonnet D, Lemoine FM, Najman A, Guigon M. Comparison court Brace Jovanovich: New York, 1977, pp 523–532. of the inhibitory effects of AcSDKP, TNF-␣, TGF-␤ and MIP- 40 St. Groth SF. Review article. The evaluation of limiting 1␣ on marrow-purified CD34+ progenitors. Exp Hematol dilution assays. J Immunol Meth 1982; 49: R11–R23. 1995; 23: 551–556. 41 Gibson LF, Fortney J, Landreth KS et al. Disruption of bone 56 Lu L, Xiao M, Grigsby S et al. Comparative effects of sup- marrow stromal cell function by etoposide. Blood and Bone pressive cytokines on isolated single CD34+++ stem/progenitor Marrow Transplant 1997; 3: 122–132. cells from human bone marrow and umbilical cord blood 42 Albin N, Douay L, Fouillard L et al. In vivo effects of GM- plated with and without serum. Exp Hematol 1993; 21: CSF and IL-3 on hematopoietic cell recovery in bone marrow 1442–1446.