Abstract 2138: Creating next-generation biologics using a novel chemistry platform technology

Bright Peak Therapeutics is developing a portfolio of differentiated biotherapeutics using chemistry for applications in immuno-oncology and autoimmune diseases. Our unique chemical protein synthesis and engineering platform allows us to fine-tune cytokines and other proteins to interrogate and modulate biological functions by incorporating new functional modifications. Standard recombinant bacterial or cellular expression systems used to produce proteins are largely restricted to using canonical amino acids, which limits access to diverse modifications that can bestow additional functional properties. With chemical protein synthesis technology, canonical and non-canonical modifications including conjugation handles can be easily introduced, ultimately enabling a medicinal chemistry approach for engineering cytokine structures. Enhanced cytokines with differentiated biology developed using this approach can be further elaborated by conjugating to a diverse array of molecules.

We first applied our technology platform to identify BPT-143, a rationally designed enhanced IL-2 variant currently in IND-enabling studies. BPT-143 is engineered to have enhanced binding to IL2Rβ and no binding to IL2Rα for improved efficacy and safety independent of the conjugation to a half-life extending 30 kDa PEG. The chemical synthesis technology is robust, reproducible, and scalable. We are applying our platform to enhance a number of other cytokines for use in immuno-oncology and autoimmune diseases.

Additionally, our synthetically engineered cytokines can be easily conjugated to monoclonal antibodies as ‘payloads’ using a distinct chemical conjugation technology. A rapid and simple chemical process allows site-selective conjugation of our engineered cytokines to existing antibodies ‘as-is’ to generate novel immunocytokines (IC). This ‘off-the-shelf’ approach is orthogonal to recombinant fusion methods to create ICs and does not require complex recombinant protein expression optimization and lengthy cell-line development. Moreover, it allows rapid screening of cytokine payloads in a structure-activity relationship (SAR) format to identify dual-targeting ICs with precisely tailored properties to generate the desired biological effect. We have prepared a number of ICs including anti-PD-1/IL-2 ICs with various drug-antibody ratio (DAR) and conjugation sites within the antibody. We will provide an overview of the platform technology and present highlights of its application for discovery and development of designer therapeutic cytokines and ICs.

Citation Format: Vijaya R. Pattabiraman, Matilde Arévalo Ruiz, Régis Boehringer, Benoit Hornsperger, Roy Meoded, Robert C. Tam, Bertolt Kreft. Creating next-generation biologics using a novel chemistry platform technology [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2138.

FLAIR2 improves LesionTOADS automatic segmentation of multiple sclerosis lesions in non-homogenized, multi-center, 2D clinical magnetic resonance images

Background

Accurate segmentation of MS lesions on MRI is difficult and, if performed manually, time consuming. Automatic segmentations rely strongly on the image contrast and signal-to-noise ratio. Literature examining segmentation tool performances in real-world multi-site data acquisition settings is scarce.

Objective

FLAIR², a combination of T₂-weighted and fluid attenuated inversion recovery (FLAIR) images, improves tissue contrast while suppressing CSF. We compared the use of FLAIR and FLAIR² in LesionTOADS, OASIS and the lesion segmentation toolbox (LST) when applied to non-homogenized, multi-center 2D-imaging data.

Methods

Lesions were segmented on 47 MS patient data sets obtained from 34 sites using LesionTOADS, OASIS and LST, and compared to a semi-automatically generated reference. The performance of FLAIR and FLAIR² was assessed using the relative lesion volume difference (LVD), Dice coefficient (DSC), sensitivity (SEN) and symmetric surface distance (SSD). Performance improvements related to lesion volumes (LVs) were evaluated for all tools. For comparison, LesionTOADS was also used to segment lesions from 3 T single-center MR data of 40 clinically isolated syndrome (CIS) patients.

Results

Compared to FLAIR, the use of FLAIR² in LesionTOADS led to improvements of 31.6% (LVD), 14.0% (DSC), 25.1% (SEN), and 47.0% (SSD) in the multi-center study. DSC and SSD significantly improved for larger LVs, while LVD and SEN were enhanced independent of LV. OASIS showed little difference between FLAIR and FLAIR², likely due to its inherent use of T₂w and FLAIR. LST replicated the benefits of FLAIR² only in part, indicating that further optimization, particularly at low LVs is needed. In the CIS study, LesionTOADS did not benefit from the use of FLAIR² as the segmentation performance for both FLAIR and FLAIR² was heterogeneous.

Conclusions

In this real-world, multi-center experiment, FLAIR² outperformed FLAIR in its ability to segment MS lesions with LesionTOADS. The computation of FLAIR² enhanced lesion detection, at minimally increased computational time or cost, even retrospectively. Further work is needed to determine how LesionTOADS and other tools, such as LST, can optimally benefit from the improved FLAIR² contrast.

•

FLAIR² improves automatic MS lesion segmentation with LesionTOADS compared to FLAIR.

•

Segmentation similarity improves for higher lesion volumes, particularly for FLAIR².

•

FLAIR² provides greater sensitivity independent of lesion volume than FLAIR alone.

•

Other segmentation tools need further optimization to fully benefit from FLAIR².

•

FLAIR² provides immediate benefits at 1.5 T and visually improves segmentation at 3 T.

The impact of intensity variations in T1-hypointense lesions on clinical correlations in multiple sclerosis

Background: The correlations between T1-hypointense lesion (‘black hole’) volume and clinical measures have varied widely across previous studies. The degree of hypointensity in black holes is associated with the severity of tissue damage, but the impact on the correlation with disability is unknown.

Objectives: To determine how variations in the intensity level used for lesion classification can impact clinical correlation, specifically with the Expanded Disability Status Scale (EDSS), and whether using a restricted range can improve correlation.

Methods: A highly automated image analysis procedure was applied to the scans of 24 multiple sclerosis (MS) patients with well-distributed EDSS scores to compute their black hole volumes at nine different levels of intensity relative to the reference intensities sampled in normal-appearing white matter (NAWM) and cerebrospinal fluid (CSF). Two methods of volume computation were used.

Results: The black hole volume–EDSS Spearman correlations ranged between 0.49–0.73 (first method) and 0.54–0.74 (second method). The strongest correlations were observed by only including the voxels with maximum intensities at 30–40% of the CSF to NAWM range.

Conclusions: Intensity variations can have a large impact on black hole–EDSS correlation. Restricting the measurement to a subset of the darkest voxels may yield stronger correlations.

Tolerance to noninherited maternal MHC antigens in mice

The phenomenon of tolerance to noninherited maternal Ags (NIMA) is poorly understood. To analyze the NIMA effect C57BL/6 (H-2(b/b)) males were mated with B6D2F(1) (H-2(b/d)) females, whereby 50% of the offspring are H-2(b/b) mice that have been exposed to maternal H-2(d) alloantigens. Controls were H-2(b/b) offspring of C57BL/6 mothers, either inbred C57BL/6 mice or F(1) backcross mice from breedings with H-2(b/d) fathers. We found that 57% of the H-2(b/b) offspring of semiallogeneic (H-2(b/d)) mothers accepted fully allogeneic DBA/2 (H-2(d/d)) heart grafts for >180 days, while similar transplants were all rejected by day 11 in controls (p < 0.0004). Foster nursing studies showed that both oral and in utero exposure to NIMA are required for this tolerogenic effect. An effect of NIMA was also found to extend the survival of skin grafts from a semiallogeneic donor (p < 0.02). Pretransplant analysis of splenocytes showed a 40-90% reduction of IL-2-, IL-5-, and IFN-gamma-producing T cells responding to H-2(d)-expressing APC in NIMA(d)-exposed vs control mice. Injection of pregnant BALB/c-dm2 (H-2L(d)-negative) female mice i.v. with H-2L(d)(61-80) peptide profoundly suppressed the offspring's indirect pathway alloreactive CD4(+) T cell response to H-2L(d). These results suggest that the natural exposure of the fetus and newborn to maternal cells and/or soluble MHC Ags suppresses NIMA-allospecific T cells of the offspring, predisposing to organ transplant tolerance in adult mice.

Comparative analysis of anti-hepatitis C virus activity and gene expression mediated by alpha, beta, and gamma interferons

A direct comparison of the inhibitory effects of alpha, beta, and gamma interferons (IFNs) on replication of a hepatitis C virus subgenomic replicon in a hepatoma cell line revealed similarities in antiviral potency. However, alternate IFN-induced antiviral mechanisms were suggested following observations of striking differences between IFN-gamma and IFN-alpha/beta with respect to strength and durability of the antiviral response and the magnitude and pattern of IFN-mediated gene expression.

Mechanisms of action of ribavirin in antiviral therapies

Although ribavirin was originally synthesized over 30 years ago and has been used to treat viral infections as monotherapy (respiratory syncytial virus and Lassa fever virus) or with interferon-alpha (IFN-alpha) as combination therapy (hepatitis C virus), the precise mechanism of its therapeutic activities remains controversial. In this review we focus on two main biological properties of ribavirin: its indirect and direct antiviral activities (with particular emphasis on its efficacy against chronic hepatitis C infection). Each property could individually or collectively account for its clinical efficacy against viral infections. First, with emphasis on the evidence for indirect activities of ribavirin, we will review the clinical observations that suggest that the immunomodulatory properties of ribavirin can in part account for its antiviral activities in vivo. We will then describe the mode of ribavirin's direct antiviral activities. These direct activities can be ascribed to several possible mechanisms, including the recently described activity as an RNA mutagen, a property that may be important in driving a rapidly mutating RNA virus over the threshold to 'error catastrophe'.

Transcriptional regulation of CD28 expression by CD28GR, a novel promoter element located in exon 1 of the CD28 gene

CD28 provides an essential costimulatory signal required for Ag-mediated T cell activation via the TCR. Although accumulating evidence exists for the signaling properties of CD28, less is known regarding the regulation of CD28 expression. In this study, we have identified a novel promoter element of CD28, CD28GR (GGGGAGGAGGGG), which is located between +181 and +192 in exon 1 of the CD28 gene. Mutations within the 12-bp CD28GR sequence abolished its transcriptional activity. CD28GR contains a Sp1/EGR-1 binding site, which was found to act as the predominant functional element for regulating CD28 gene expression in Jurkat cells. Exon 1/CD28GR-driven transcription in Jurkat cells was augmented by cotransfection with Sp1 or EGR-1 expression plasmid. Similar augmentation was also shown with pharmacologic activation. This study is the first to identify a regulatory element that is critical for conferring constitutive and activation-induced transcriptional activation of the CD28 gene. Furthermore, our results proposed potential involvement of Sp1 in regulating CD28 expression. The linkage between Sp1 and the expression of CD28 has important implications in how viral infections, such as HIV, can induce immunosuppression.

Induction of T-cell response to cryptic MHC determinants during allograft rejection

The presentation of MHC peptides by recipient and donor antigen presenting cells is an essential element in allorecognition and allograft rejection. MHC proteins contains two sets of determinants: the dominant determinants that are efficiently processed and presented to T cells, and the cryptic determinants that are not presented sufficiently enough to induce T-cell responses in vivo. In transplanted mice, initial T-cell response to MHC peptides is consistently limited to a single or a few immunodominant determinants on donor MHC molecule. However, in this article we show that under appropriate circumstances the hierarchy of determinants on MHC molecules can be disrupted. First, we observed that gamma IFN can trigger de novo presentation of cryptic self-MHC peptides by spleen cells. Moreover, we showed that allotransplantation is associated with induction of T-cell responses to formerly cryptic determinants on both syngeneic and allogeneic MHC molecules. Our results suggest that cross-reactivity and inflammation are responsible for the initiation of these auto- and alloimmune responses after transplantation.

Synthesis and cytokine modulation properties of pyrrolo[2, 3-d]-4-pyrimidone nucleosides

A series of pyrrolo[2,3-d]pyrimidone nucleosides were synthesized and evaluated for their ability to enhance Type 2 cytokines and to suppress Type 1 cytokines in human T cells activated in vitro. Compounds 16b, 16c, 16d, 18c, and 19b induced substantial enhancement of IL-4 (a Type 2 cytokine) levels while three compounds (16b, 16c, and 16f) showed significant suppression of IFNgamma (a Type 1 cytokine) levels. The results revealed a strict structural requirement for the nucleoside-mediated enhancement of IL-4. Modifications of the ribofuranose moiety of the nucleosides either abolished or dramatically reduced the activity. Both the 5'-hydroxy and 5-carboxamidine are crucial for the activity. Of the few nucleoside analogues that demonstrated enhancement on Type 2 cytokine production, 7-(beta-D-ribofuranosyl)pyrrolo[2, 3-d]-4-pyrimidone-5-carboxamidine (16c) showed a dramatic enhancement (>200%) of IL-4 levels and a significant enhancement (36%) of IL-5 levels. Moreover, this compound showed substantial suppression of the Type 1 cytokines, IFNgamma (42%), IL-2 (54%), and TNFalpha (55%). Similarly, compound 16b showed a substantial enhancement of IL-4 (46%) and suppression of IL-2 (35%), IFNgamma (30%), and TNFalpha (26%). To our knowledge, these are the first nucleoside analogues that induce a Type 2 cytokine bias in T cells. The cytokine modulation property of 16c and 16b merits the therapeutic evaluation of these compounds in treating diseases in which immunopathology is associated with polarized Type 1 cytokine responses.

Synthesis and cytokine modulation properties of pyrrolo[2, 3-d]-4-pyrimidone nucleosides

A series of pyrrolo[2,3-d]pyrimidone nucleosides were synthesized and evaluated for their ability to enhance Type 2 cytokines and to suppress Type 1 cytokines in human T cells activated in vitro. Compounds 16b, 16c, 16d, 18c, and 19b induced substantial enhancement of IL-4 (a Type 2 cytokine) levels while three compounds (16b, 16c, and 16f) showed significant suppression of IFNgamma (a Type 1 cytokine) levels. The results revealed a strict structural requirement for the nucleoside-mediated enhancement of IL-4. Modifications of the ribofuranose moiety of the nucleosides either abolished or dramatically reduced the activity. Both the 5'-hydroxy and 5-carboxamidine are crucial for the activity. Of the few nucleoside analogues that demonstrated enhancement on Type 2 cytokine production, 7-(beta-D-ribofuranosyl)pyrrolo[2, 3-d]-4-pyrimidone-5-carboxamidine (16c) showed a dramatic enhancement (>200%) of IL-4 levels and a significant enhancement (36%) of IL-5 levels. Moreover, this compound showed substantial suppression of the Type 1 cytokines, IFNgamma (42%), IL-2 (54%), and TNFalpha (55%). Similarly, compound 16b showed a substantial enhancement of IL-4 (46%) and suppression of IL-2 (35%), IFNgamma (30%), and TNFalpha (26%). To our knowledge, these are the first nucleoside analogues that induce a Type 2 cytokine bias in T cells. The cytokine modulation property of 16c and 16b merits the therapeutic evaluation of these compounds in treating diseases in which immunopathology is associated with polarized Type 1 cytokine responses.

The ribavirin analog ICN 17261 demonstrates reduced toxicity and antiviral effects with retention of both immunomodulatory activity and reduction of hepatitis-induced serum alanine aminotransferase levels

The demonstrated utility of the nucleoside analog ribavirin in the treatment of certain viral diseases can be ascribed to its multiple distinct properties. These properties may vary in relative importance in differing viral disease conditions and include the direct inhibition of viral replication, the promotion of T-cell-mediated immune responses via an enhanced type 1 cytokine response, and a reduction of circulating alanine aminotransferase (ALT) levels associated with hepatic injury. Ribavirin also has certain known toxicities, including the induction of anemia upon chronic administration. To determine if all these properties are linked, we compared the D-nucleoside ribavirin to its L-enantiomer (ICN 17261) with regard to these properties. Strong similarities were seen for these two compounds with respect to induction of type 1 cytokine bias in vitro, enhancement of type 1 cytokine responses in vivo, and the reduction of serum ALT levels in a murine hepatitis model. In contrast, ICN 17261 had no in vitro antiviral activity against a panel of RNA and DNA viruses, while ribavirin exhibited its characteristic activity profile. Importantly, the preliminary in vivo toxicology profile of ICN 17261 is significantly more favorable than that of ribavirin. Administration of 180 mg of ICN 17261 per kg of body weight to rats by oral gavage for 4 weeks generated substantial serum levels of drug but no observable clinical pathology, whereas equivalent doses of ribavirin induced a significant anemia and leukopenia. Thus, structural modification of ribavirin can dissociate its immunomodulatory properties from its antiviral and toxicologic properties, resulting in a compound (ICN 17261) with interesting therapeutic potential.

Monocyclic L-nucleosides with type 1 cytokine-inducing activity

A series of 1,2,4-triazole L-nucleosides were synthesized and evaluated for their ability to stimulate type 1 cytokine production by activated human T cells in direct comparison to the known active agent ribavirin. Among the compounds prepared, 1-beta-L-ribofuranosyl-1,2,4-triazole-3-carboxamide (5, ICN 17261) was found to be the most uniformly potent compound. Conversion of the 3-carboxamide group of 5 to a carboxamidine functionality resulted in 1-beta-L-ribofuranosyl-1,2,4-triazole-3-carboxamidine hydrochloride (10), which induced cytokine levels comparable to 5 for two of the three type 1 cytokines examined. Modification of the carbohydrate moiety of 5 provided compounds of reduced activity. Significantly, ICN 17261 offers interesting immunomodulatory potential for the treatment of diseases where type 1 cytokines play an important role.

Inhibition of CD28 expression by oligonucleotide decoys to the regulatory element in exon 1 of the CD28 gene

Ligation of CD28 provides a costimulatory signal essential for Ag-mediated T cell activation via the TCR. Previously we demonstrated that inhibition of human and murine CD28 expression by a guanosine (G)-rich oligonucleotide (ODN), GR1, led to immunosuppression in vitro and in vivo. The bioactivity of GR1 was dependent on a G-rich DNA sequence motif consisting of two G tetrads separated by four nucleotides, (G4N4G4). We have shown recently that a G-rich region, designated CD28GR, in exon 1 of the CD28 gene is such a motif and is a positive regulatory element that binds the transcription factors Sp1 and EGR-1. Here we showed that the bioactivity of GR1 and the related GR2 correlated with the sequence-specific formation of distinct nuclear protein complexes and a high degree of ODN secondary structure. In addition, these ODN blocked transcription factor binding to CD28GR (also in a sequence-specific manner) and prevented CD28GR from driving transcription of a reporter gene. Interestingly, GR1 potently inhibited CD28, but not the expression of other Sp1- and EGR-1-regulated genes, an effect associated with lower Sp1 protein binding affinity of GR1 and GR2 compared with that of canonical Sp1 sites. These data show that DNA sequences that contain the G-rich sequence motif, G4N4G4, such as GR1 and GR2, can functionally mimic the regulatory protein binding ability of CD28GR. Thus, GR1 and GR2 act as molecular decoys to selectively interfere with transcriptional regulation of the CD28 gene.

Inhibition of CD28 Expression by Oligonucleotide Decoys to the Regulatory Element in Exon 1 of the CD28 Gene

Ligation of CD28 provides a costimulatory signal essential for Ag-mediated T cell activation via the TCR. Previously we demonstrated that inhibition of human and murine CD28 expression by a guanosine (G)-rich oligonucleotide (ODN), GR1, led to immunosuppression in vitro and in vivo. The bioactivity of GR1 was dependent on a G-rich DNA sequence motif consisting of two G tetrads separated by four nucleotides, (G4N4G4). We have shown recently that a G-rich region, designated CD28GR, in exon 1 of the CD28 gene is such a motif and is a positive regulatory element that binds the transcription factors Sp1 and EGR-1. Here we showed that the bioactivity of GR1 and the related GR2 correlated with the sequence-specific formation of distinct nuclear protein complexes and a high degree of ODN secondary structure. In addition, these ODN blocked transcription factor binding to CD28GR (also in a sequence-specific manner) and prevented CD28GR from driving transcription of a reporter gene. Interestingly, GR1 potently inhibited CD28, but not the expression of other Sp1- and EGR-1-regulated genes, an effect associated with lower Sp1 protein binding affinity of GR1 and GR2 compared with that of canonical Sp1 sites. These data show that DNA sequences that contain the G-rich sequence motif, G4N4G4, such as GR1 and GR2, can functionally mimic the regulatory protein binding ability of CD28GR. Thus, GR1 and GR2 act as molecular decoys to selectively interfere with transcriptional regulation of the CD28 gene.

Contact hypersensitivity responses following ribavirin treatment in vivo are influenced by type 1 cytokine polarization, regulation of IL-10 expression, and costimulatory signaling

We previously described the promotion of type 1 cytokine responses by the nucleoside analogue, ribavirin, in human T cells in vitro. In this study, we examined whether type 1 cytokine polarization by ribavirin in vivo could promote contact hypersensitivity (CHS) responses to dinitrofluorobenzene, a type 1 cytokine-mediated immune response. Unexpectedly, although type 1 cytokine responses were enhanced following ribavirin treatment in vitro and in vivo, the magnitude of CHS responses in BALB/c and C57BL/6 mice was influenced more by a second ribavirin-regulated pathway. The key regulatory molecule in this pathway was IL-10. Ribavirin-mediated suppression of IL-10 in BALB/c mice was associated with increased B7-2 expression and enhanced CHS responses, whereas enhanced IL-10 levels, following ribavirin administration, led to increased B7-1 expression and impaired CHS responses in C57BL/6 mice. The effect of ribavirin on the expression of B7 molecules and on CHS responses was neutralized by IL-10 administration in BALB/c and by anti-IL-10 Ab in C57BL/6. Thus, ribavirin controlled CHS responses directly through the modulation of IL-10 expression, and in vivo outcome was dictated by the preferential expression of either B7-1, an inappropriate costimulatory molecule in CHS, or B7-2, the predominant costimulatory molecule in CHS. Replacing dinitrofluorobenzene priming with IFN-alpha stimulation, we showed that the ribavirin-regulated pathway could function independent of Ag priming. Altogether, these data showed that, although ribavirin treatment induced a type 1 cytokine bias in contact allergen-primed BALB/c and C57BL/6 mice, in vivo CHS responses were dependent on ribavirin-mediated regulation of both IL-10 and preferential costimulatory signaling.

Contact Hypersensitivity Responses Following Ribavirin Treatment In Vivo Are Influenced by Type 1 Cytokine Polarization, Regulation of IL-10 Expression, and Costimulatory Signaling

We previously described the promotion of type 1 cytokine responses by the nucleoside analogue, ribavirin, in human T cells in vitro. In this study, we examined whether type 1 cytokine polarization by ribavirin in vivo could promote contact hypersensitivity (CHS) responses to dinitrofluorobenzene, a type 1 cytokine-mediated immune response. Unexpectedly, although type 1 cytokine responses were enhanced following ribavirin treatment in vitro and in vivo, the magnitude of CHS responses in BALB/c and C57BL/6 mice was influenced more by a second ribavirin-regulated pathway. The key regulatory molecule in this pathway was IL-10. Ribavirin-mediated suppression of IL-10 in BALB/c mice was associated with increased B7-2 expression and enhanced CHS responses, whereas enhanced IL-10 levels, following ribavirin administration, led to increased B7-1 expression and impaired CHS responses in C57BL/6 mice. The effect of ribavirin on the expression of B7 molecules and on CHS responses was neutralized by IL-10 administration in BALB/c and by anti-IL-10 Ab in C57BL/6. Thus, ribavirin controlled CHS responses directly through the modulation of IL-10 expression, and in vivo outcome was dictated by the preferential expression of either B7-1, an inappropriate costimulatory molecule in CHS, or B7-2, the predominant costimulatory molecule in CHS. Replacing dinitrofluorobenzene priming with IFN-α stimulation, we showed that the ribavirin-regulated pathway could function independent of Ag priming. Altogether, these data showed that, although ribavirin treatment induced a type 1 cytokine bias in contact allergen-primed BALB/c and C57BL/6 mice, in vivo CHS responses were dependent on ribavirin-mediated regulation of both IL-10 and preferential costimulatory signaling.

Increased potency of an aptameric G-rich oligonucleotide is associated with novel functional properties of phosphorothioate linkages

We previously showed that inhibition of the expression of CD28 (an essential immune receptor on T cells) mediated by a phosphorothioate (PS)-modified aptameric oligodeoxynucleotide (ODN) sequence, GR1, resulted in reduced T cell responses in vitro and in vivo. Using GR1 sequences differing only in the amount of terminal PS linkages (chimeric SO-ODN), the present study demonstrated that even after a substantial reduction in PS linkages, this 18-mer ODN sequence could still confer functionality in the ODN-mediated inhibition of CD28 expression. We showed that secondary structure and full retention of the ability to form a specific protein-ODN complex and to increase cellular uptake in activated Jurkat T cells were critical parameters in the determination of the magnitude of bioactivity of chimeric SO-ODN. We report that a chimeric SO-ODN with terminal PS linkages that total 9 (ICN 17221) or 12 (ICN 17263) was sufficient to inhibit CD28 expression and suppress in vivo inflammatory ear responses to contact allergen in mice with similar potency to the 17-thioate S-ODN (ICN 16064). Interestingly, all chimeric SO-ODN showed similar in vitro nuclease resistance. These data suggest alternate functional properties for PS linkages, unrelated to nuclease resistance, in enhancing the bioactivity of a G-rich aptamer.

Ribavirin polarizes human T cell responses towards a Type 1 cytokine profile

BACKGROUND/AIMS

The therapeutic benefit of ribavirin, a nucleoside analog, in the treatment of chronic HCV infection is seen even in the absence of any apparent direct antiviral effect. We surmised that ribavirin may act by eliciting altered virus-specific immune responses. Because antiviral immunity is predominantly mediated by cytotoxic T cells and antiviral cytokines, we sought to determine whether ribavirin could promote antiviral (Type 1) cytokine expression in human T cells.

METHODS

Isolated human T cells were activated in vitro with enterotoxin B or with phorbol ester plus ionomycin. Cytokine ELISAs were performed on culture supernatants, cytokine mRNA was detected following RT-polymerase chain reaction of T cell RNA, and T cell proliferation measured using MTT assay.

RESULTS

Ribavirin enhanced a Type 1 (IL-2, IFNgamma, TNFalpha) while suppressing a Type 2 cytokine response (IL-4, IL-5 and IL-10), at the level of both protein and mRNA expression. Ribavirin mediated comparable effects on cytokine expression both following activation of specific T cell subpopulations with superantigen and following activation of a larger percentage of T cells via pharmacologic means. The in vitro effect on cytokine expression following ribavirin treatment was comparable in both CD4+ or CD8+ T cell subsets and was observed in a dose range that promoted T cell proliferation.

CONCLUSIONS

These data support the view that ribavirin promotes a Type 1 cytokine-mediated immune response, a property which may account in part for its ability to enhance the antiviral activity of interferon-alpha in the treatment of chronic HCV infection.

The presentation of self and allogeneic MHC peptides to T lymphocytes

The presentation of donor-derived MHC peptides by recipient APCs to T cells is an essential component of the rejection of allografts (indirect allorecognition). Initial alloreactive T cell response is confined to a few well processed and presented dominant determinants on donor MHC. However, during long-term graft rejection, T cell response spreads to formerly poorly presented cryptic allogeneic MHC peptides. This phenomenon is likely to play an important role in the amplification and the perpetuation of the rejection process. Additionally, we present evidence that T cell repertoire selection to allogeneic MHC peptides is acquired via recognition of self-MHC peptides presented in the thymus during ontogeny. Supporting this view, we have shown that indirect alloresponses can lead to self-T cell tolerance breakdown to cross-reactive determinants on self-MHC molecules or alternatively that sensitization of recipients to self-MHC peptides can lead to accelerated graft rejection. It is therefore essential to determine the factors which govern the processing and presentation of self and allogeneic MHC molecules and to elucidate the mechanisms regulating subsequent T cell responses in order to design antigen-specific based immune therapies in transplantation.

Indirect T-cell allorecognition: perspectives for peptide-based therapy in transplantation

Indirect allorecognition is an important component of allotransplant rejection. Although the initial indirect alloresponse is limited to a few dominant determinants on donor major histocompatibility complex (MHC) molecules, subsequent spreading to additional determinants on recipient and donor antigens is common. Gilles Benichou and colleagues discuss the mechanisms by which immunodominance is acquired or disrupted in indirect alloresponses, and examine the implications for the design of peptide-based selective immunotherapy in transplantation.

Oligonucleotide-mediated inhibition of CD28 expression induces human T cell hyporesponsiveness and manifests impaired contact hypersensitivity in mice

Ligation of CD28 provides a costimulatory signal essential for Ag-mediated T cell activation via the TCR. Blocking CD28 ligation can inhibit cytokine expression and elicits a state of T cell hyporesponsiveness. In this study, we examined the effect of inhibiting CD28 expression on in vitro and in vivo T cell responses. To address this, we have synthesized a series of G-rich phosphorothioate oligonucleotides that inhibited activation-induced transcription and cell surface expression of CD28 on human T cells. CD28 blockade was selective, as expression of other activation-induced receptors was unaffected by oligonucleotide treatment. Using strategic changes to base composition, we identified a minimal 12-mer sequence, containing two sets of four contiguous guanosines separated by 3 to 5 bases, which conferred activity in vitro. Furthermore, inhibition of CD28 expression mediated by one representative active oligonucleotide, GR1, resulted in a concomitant dose-dependent diminution of anti-CD3/PMA-induced cytokine (IL-2, IFN-gamma, IL-8) production. Inhibition of IL-2 synthesis was dependent on CD28 expression, as GR1 failed to abrogate activated IL-2 production in a CD28-deficient T cell line, HUT 78. The inhibitory activity of GR1 reduced T cell proliferative responses in MLR and induced Ag-specific T cell hyporesponsiveness to alloantigens. Finally, s.c. administration of GR1 impaired in vivo contact hypersensitivity responses in mice and was associated with substantially decreased CD28 and IFN-gamma mRNA expression in lymph node cells. Collectively, our studies show the tolerogenic potential of oligonucleotide-mediated CD28 inhibition on T cell activation, in vitro and in vivo.

Oligonucleotide-mediated inhibition of CD28 expression induces human T cell hyporesponsiveness and manifests impaired contact hypersensitivity in mice

Ligation of CD28 provides a costimulatory signal essential for Ag-mediated T cell activation via the TCR. Blocking CD28 ligation can inhibit cytokine expression and elicits a state of T cell hyporesponsiveness. In this study, we examined the effect of inhibiting CD28 expression on in vitro and in vivo T cell responses. To address this, we have synthesized a series of G-rich phosphorothioate oligonucleotides that inhibited activation-induced transcription and cell surface expression of CD28 on human T cells. CD28 blockade was selective, as expression of other activation-induced receptors was unaffected by oligonucleotide treatment. Using strategic changes to base composition, we identified a minimal 12-mer sequence, containing two sets of four contiguous guanosines separated by 3 to 5 bases, which conferred activity in vitro. Furthermore, inhibition of CD28 expression mediated by one representative active oligonucleotide, GR1, resulted in a concomitant dose-dependent diminution of anti-CD3/PMA-induced cytokine (IL-2, IFN-gamma, IL-8) production. Inhibition of IL-2 synthesis was dependent on CD28 expression, as GR1 failed to abrogate activated IL-2 production in a CD28-deficient T cell line, HUT 78. The inhibitory activity of GR1 reduced T cell proliferative responses in MLR and induced Ag-specific T cell hyporesponsiveness to alloantigens. Finally, s.c. administration of GR1 impaired in vivo contact hypersensitivity responses in mice and was associated with substantially decreased CD28 and IFN-gamma mRNA expression in lymph node cells. Collectively, our studies show the tolerogenic potential of oligonucleotide-mediated CD28 inhibition on T cell activation, in vitro and in vivo.

Self determinant selection and acquisition of the autoimmune T cell repertoire

Autologous proteins are continuously processed and presented in the form of peptides associated with self major histocompatibility (MHC) molecules at the surface of antigen-presenting cells for interaction with autoreactive T cells. During thymic selection, the presentation of self peptides is an essential element in the establishment of the T cell repertoire. Developing T cells which recognize self peptide/self MHC complexes with sufficient affinity are clonally deleted. However, we and others have recently demonstrated that a variety of self peptides, despite their high binding affinity to MHC molecules, never reach the threshold of presentation to ensure negative selection (cryptic self peptides). This mechanism may have been selected to avoid excessive purging of T cell repertoire during ontogeny. However, T cells directed to cryptic self determinants represent a continuous threat for the initiation of autoimmunity in adults. Supporting this view, recent studies have documented the involvement of cryptic self peptide presentation in different autoimmune diseases. In this article, we examine the factors that govern the selection of self peptides for presentation to autoreactive T cells in vivo and discuss their contribution to both the induction and the maintenance of self tolerance. In addition, we analyze the mechanisms by which the hierarchy of determinants on a self protein can be disrupted, thereby leading to the presentation of previously cryptic self peptides and the induction of an autoimmune T-cell-mediated process.

T cell tolerance is influenced by concomitant T cell recognition of cross-reactive self-peptides

Although the current dogma of T cell recognition stresses its exquisite specificity, T cell clones selected for a given peptide can recognize other sequentially or structurally related peptides. Here, we have examined the immunogenicity and tolerogenicity of various self-peptides derived from region 61-80 of different MHC class I proteins co-expressed in the same mouse. Following immunization of B10.A mice (K(k), A(k), E(k), L(d), D(d)) with self-L(d) 61-80 peptide, vigorous MHC class II-restricted T cell proliferation was elicited after restimulation with either the immunogen or with self-K(k) 61-80 but not with self-D(d) 61-80. Furthermore, adult B10.A mice, tolerized with L(d) 61-80 prior to immunization with L(d) 61-80 did not respond to challenge with L(d) 61-80 and the cross-reactive K(k) 61-80. However, following K(k) 61-80 immunization, L(d) 61-80-tolerized mice responded to K(k) 61-80 but not to L(d) 61-80. Thus, tolerance induction to L(d) 61-80 resulted in the elimination/inactivation of L(d) 61-80-reactive T cells including the subpopulation that cross-reacted with K(k) 61-80. However, T cells that recognized K(k) 61-80 exclusively were preserved. Moreover, we showed that immunization with K(k) 61-80 resulted in tolerance breakdown to the cross-reactive, dominant self-peptide D(b) 61-80 in B10.A(4R) mice (K(k), A(k), L(d),D(b)). Together, these results show that the autoimmune T cell repertoire is influenced by the concomitant recognition of different cross-reactive self-peptides within the same individual.

T cell tolerance is influenced by concomitant T cell recognition of cross-reactive self-peptides

Although the current dogma of T cell recognition stresses its exquisite specificity, T cell clones selected for a given peptide can recognize other sequentially or structurally related peptides. Here, we have examined the immunogenicity and tolerogenicity of various self-peptides derived from region 61-80 of different MHC class I proteins co-expressed in the same mouse. Following immunization of B10.A mice (K(k), A(k), E(k), L(d), D(d)) with self-L(d) 61-80 peptide, vigorous MHC class II-restricted T cell proliferation was elicited after restimulation with either the immunogen or with self-K(k) 61-80 but not with self-D(d) 61-80. Furthermore, adult B10.A mice, tolerized with L(d) 61-80 prior to immunization with L(d) 61-80 did not respond to challenge with L(d) 61-80 and the cross-reactive K(k) 61-80. However, following K(k) 61-80 immunization, L(d) 61-80-tolerized mice responded to K(k) 61-80 but not to L(d) 61-80. Thus, tolerance induction to L(d) 61-80 resulted in the elimination/inactivation of L(d) 61-80-reactive T cells including the subpopulation that cross-reacted with K(k) 61-80. However, T cells that recognized K(k) 61-80 exclusively were preserved. Moreover, we showed that immunization with K(k) 61-80 resulted in tolerance breakdown to the cross-reactive, dominant self-peptide D(b) 61-80 in B10.A(4R) mice (K(k), A(k), L(d),D(b)). Together, these results show that the autoimmune T cell repertoire is influenced by the concomitant recognition of different cross-reactive self-peptides within the same individual.

Induction of T cell responses to a self-antigen following allotransplantation

T cell tolerance to self-antigens is established through the recognition by immature T cells of dominant self-peptides presented in association with self-MHC molecules in the developing thymus (negative selection). The self-peptide Dd 61-80 is dominant in syngeneic BALB/c mice (H2d). T cell tolerance to Dd 61-80 in this mouse strain resulted in the absence of T cell proliferation following in vivo priming with Dd 61-80 peptide. Here, we show that transplantation of BALB/c mice with allogeneic B10.A (H2a) splenocytes led to an autoimmune T cell response toward the dominant self-peptide Dd 61-80. No T cell responses to Dd 61-80 peptide were observed after transplantation of C57BL/6 (H2b) splenocytes into BALB/c recipients. In addition, we provide evidence indicating that the breakdown of tolerance to Dd 61-80 self-peptide resulted from the presentation of the donor crossreactive peptide Kk 61-80 at the surface of recipient antigen-presenting cells. Taken together, our results suggest that following allotransplantation, T cell responses to donor antigens could spread to crossreactive determinants on self-proteins, thus perpetuating and amplifying the rejection process and presumably initiating tissue-specific autoimmune disorders.

Presentation of a self-peptide for in vivo tolerance induction of CD4+ T cells is governed by a processing factor that maps to the class II region of the major histocompatibility complex locus

Self-proteins are regularly processed for presentation to autoreactive T cells in association with both class I and class II major histocompatibility complex (MHC) molecules. The presentation of self-peptides plays a crucial role in the acquisition of T cell repertoire during thymic selection. We previously reported that the self-MHC class I peptide Ld 61-80 was immunogenic in syngeneic B10.A mice (H-2a). We showed that despite its high affinity for self-MHC class II molecules, Ld 61-80 peptide failed to induce elimination of autoreactive CD4+ T cells, presumably due to incomplete processing and presentation in the B10.A's developing thymus (cryptic-self peptide). In this report, we showed that the cryptic phenotype was not an intrinsic property of the self-peptide Ld 61-80 since it was found to be naturally presented and subsequently tolerogenic in BALB/c mice (H-2d) (dominant self-peptide). In addition, the self-peptide Ld 61-80 was found to be immunogenic in different H-2a mice while it was invariably tolerogenic in H-2d mice regardless of their background genes. We observed that Ld 61-80 bound equally well to H-2d and H-2k MHC class II molecules. Also, no correlation was found between the quantity of self-Ld protein and the tolerogenicity of Ld 61-80. Surprisingly, Ld 61-80 was not naturally presented in (H-2d x H-2a) F1 mice, indicating that the H-2a MHC locus contained a gene that impaired the presentation of the self-peptide. Analyses of T cell responses to the self-peptide in several H-2 recombinant mice revealed that the presentation of Ld 61-80 was controlled by genes that mapped to a 170-kb portion of the MHC class II region. This study shows that (a) endogenously processed self-peptides presented by MHC class II molecules are involved in shaping the CD4+ T cell repertoire in the thymus; (b) The selection of self-peptides for presentation by MHC class II molecules to nascent autoreactive T cells is influenced by nonstructural MHC genes that map to a 170-kb portion of the MHC class II region; and (c) the MHC locus of H-2a mice encodes factors that prevent or abrogate the presentation by MHC class II molecules of the self-peptide Ld 61-80. These findings may have important implications for understanding the molecular mechanisms involved in T cell repertoire acquisition and self-tolerance induction.

Biological availability and nuclease resistance extend the in vitro activity of a phosphorothioate-3'hydroxypropylamine oligonucleotide

Augmented biological activity in vitro has been demonstrated in oligonucleotides (oligos) modified to provide nuclease resistance, to enhance cellular uptake or to increase target affinity. How chemical modification affects the duration of effect of an oligo with potent activity has not been investigated directly. We postulated that modification with internucleotide phosphorothioates and 3' alkylamine provided additional nuclease protection which could significantly extend the biological activity of a 26 mer, (T2). We showed this analog, sT2a, could maximally inhibit interferon gamma-induced HLA-DR mRNA synthesis and surface expression in both HeLa and retinal pigmented epithelial cells and could continue to be effective, in the absence of oligo, 15 days following initial oligo treatment; an effect not observed with its 3'amine counterpart, T2a. In vitro stability studies confirmed that sT2a conferred the greatest stability to nucleases and that cellular accumulation of 32P-sT2a in both cell types was also greater than other T2 oligos. Using confocal microscopy, we revealed that the intracellular distribution of sT2a favored greater nuclear accumulation and release of oligo from cytoplasmic vesicles; a pattern not observed with T2a. These results suggest that phosphorothioate-3'amine modification could increase the duration of effect of T2 oligo by altering nuclease resistance as well as intracellular accumulation and distribution; factors known to affect biological availability.

Nocturnal desaturation and serum erythropoietin: a study in patients with chronic obstructive pulmonary disease and in normal subjects

1. To clarify the relationship between nocturnal oxygen desaturation and erythropoietin production in patients with chronic obstructive pulmonary disease, we determined arterial oxygen saturation and serum immunoreactive erythropoietin levels over 24 h in eight patients with chronic obstructive pulmonary disease and in nine normal subjects. 2. In the normal subjects, there was a significant circadian variation in serum erythropoietin levels with the highest mean deviation from the geometric mean at 22.00 hours and the nadir at 05.00 hours. 3. The three patients with chronic obstructive pulmonary disease with the most marked nocturnal desaturation (lowest arterial oxygen saturation < 57%) and most marked daytime hypoxaemia (daytime arterial partial pressure of oxygen < 6 kPa) had raised nocturnal serum erythropoietin levels. In two of these patients, the serum erythropoietin level was raised throughout the 24 h and erythrocyte mass was also raised. In the other patient, the serum erythropoietin level was not raised in five daytime samples and erythrocyte mass was normal. 4. The other five patients with chronic obstructive pulmonary disease with less severe nocturnal hypoxaemia (lowest arterial oxygen saturation range 78-86%) had serum erythropoietin levels (range 14-36 m-i.u./ml) which were indistinguishable from normal (range 12-44 m-i.u./ml) and showed circadian changes which were not significantly different (P = 0.35) from those in the normal subjects. 5. Thus, mild nocturnal oxygen desaturation is not associated with elevation of serum erythropoietin levels, whereas daytime hypoxaemia with associated severe nocturnal desaturation is associated with increased serum erythropoietin levels both by day and by night.

Comparisons of human, rat and mouse erythropoietins by isoelectric focusing: differences between serum and urinary erythropoietins

Isoelectric focusing (IEF) in the pH range 2.5-5.0 has been used to compare the immunoreactive (ir) erythropoietin (Epo) in paired samples of serum and urine from three patients, two with idiopathic aplastic anaemia and one with paroxysmal nocturnal haemoglobinuria and also from three anaemic rats. Serum samples only were also examined from two further patients with aplastic anaemia and from three mice, made anaemic (like the rats) by irradiation and phenylhydrazine treatment. Most of the ir-Epo recovered after IEF was found in the pH range 2.5-3.9. For the sera, the proportion of more acidic ir-Epo with pI less than 3.0 recovered after IEF increased from human to rat to mouse. Human sera contained a greater proportion of ir-Epo with pI greater than 3.4 than rat or mouse sera. For the urines, the distribution of ir-Epo by IEF was similar between human and rat. For both species, the proportion of ir-Epo with pI less than 3.0 recovered after IEF was greater in urine than in the paired serum samples. The Second International Reference Preparation of Human Urinary Epo differed from the Epo in unextracted human urine in that there was a lower proportion of ir-Epo with pI less than 3.0. The differences observed between serum and urinary Epo are of particular interest because only the urinary form of native human Epo has ever been purified, and because this was used to compare native with rDNA-derived Epo.

An immunological cross-reactant of erythropoietin in serum which may invalidate EPO radioimmunoassay

Atypically high radioimmunoassay estimates of erythropoietin in serum samples from two out of a series of 130 normal Norwegian children have been found to be invalid. Invalidity was due to non-parallelism of radioimmunoassay (RIA) response dilution lines given by these two sera when compared with the second International Reference Preparation (IRP). In contrast, immunoreactive erythropoietin in other sera usually shows parallelism with the IRP. Fractionation by gel filtration of the sera showing non-parallelism in the RIA showed them to contain an erythropoietin cross-reactant which was eluted before both an 125I recombinant human erythropoietin marker and normal erythropoietin. In family studies, non-parallelism of RIA response dilution lines and the large molecular size erythropoietin cross-reactant was found in serum from one sibling but not in five other family members of one of the two index cases. Radioimmunoassays based on two different antisera differed in recognition of the abnormal cross-reactant of erythropoietin.

Sexual dimorphism of erythropoietin-degrading activity in mouse submaxillary gland extracts

In the course of investigation of submaxillary gland (SG) extracts from mice as a possible source of extra-renal erythropoietin (EPO) we have extended our previous studies of the degradation of EPO added to SG and kidney extracts. The discrepancy between estimates of EPO obtained with two radioimmunoassays (RIAs) differing only in time of incubation with 125I-labeled recombinant human EPO (r-HuEPO) (20 h and 72 h) has been used as an indicator of tracer degradation occurring during the RIA incubation. Degradation of 125I-labeled r-HuEPO by male mouse SG extracts was not prevented by addition of inhibitors of monodeiodinases or proteolytic enzymes. Degradation of added 125I-labeled r-HuEPO was monitored using gel filtration fast protein liquid chromatography. SG extracts from male and androgen-treated female mice both degraded tracer r-HuEPO to a greater extent than extracts from female mice. Tracer degradation increased with time and tissue concentration and could give rise to invalid estimates of EPO in SG extracts by RIA. In contrast, none of the kidney extracts degraded r-HuEPO. Recovery of mouse serum EPO added to and incubated with male mouse SG or kidney extracts was 13% and 93%, respectively, estimated by RIA under conditions that excluded degradation of the RIA tracer antigen.

Development of radioimmunoassays for human erythropoietin using recombinant erythropoietin as tracer and immunogen

Since the estimation of erythropoietin (EPO) by radioimmunoassay (RIA) has been limited by the availability of highly purified urinary (U) human (Hu) EPO, we investigated the use of recombinant (R)-HuEPO as a replacement for U-HuEPO in the preparation of 125I-tracer and high affinity antisera. In each of two validated RIAs developed using U-HuEPO-derived reagents, dose-response lines and potency estimates for samples were compared when 125I-U-HuEPO and anti-U-HuEPO antisera were sequentially replaced to give assay variants using R-HuEPO-derived reagents. Two U-HuEPO preparations, the 2nd International Reference Preparation and CAT-1, and R-HuEPO were variously used as standards. The samples tested included clinically relevant human sera and partially purified preparations of U-HuEPO and R-HuEPO. In each RIA and for each assay variant tested, samples gave dose-response lines whose slopes did not differ significantly from that of the standard. For each of the two variant RIAs, potency estimates for any sample were consistent and, where examined, RIA potency estimates agreed with in vivo bioassay determinations. These results, obtained independently in two laboratories, indicate that RIAs having appropriate specificity and sensitivity for the estimation of EPO in clinical samples can be developed using reagents derived from R-HuEPO.

Serum immunoreactive erythropoietin in hypoxic lung disease with and without polycythaemia

We studied 20 patients with chronic airflow obstruction, 10 patients without polycythaemia and 10 patients with compensatory polycythaemia having respectively mean red cell mass 24.7 (SD 4.2) and 47.8 (SD 7.5) ml/kg, mean daytime PaO2 7.6 and 6.9 kPa, mean FEV1 0.85 and 0.821. Groups were matched for severity of daytime arterial hypoxaemia but nocturnal arterial oxygen desaturation was more severe in the patients with polycythaemia than in those without. We also studied six additional patients with chronic airflow obstruction and polycythaemia and 19 normal controls. Estimates of serum immunoreactive erythropoietin (siEp) in those without polycythaemia were 19 m-i.u./ml (geometric mean) with 95% confidence range 11-35 m-i.u./ml and stable during 3 months. In those with polycythaemia they were similar and consistent in five and, in the other five, higher on at least one occasion. There was no significant difference between siEp in daytime (12.00 hours to 16.00 hours) and morning (07.00 hours) samples but geometric mean estimates of erythropoietin in paired daytime and morning samples were higher and more variable in patients with polycythaemia than in those without. The geometric mean estimate of siEp in all patients with chronic airflow obstruction and polycythaemia was greater than in normal subjects but, despite secondary polycythaemia, siEp could be in the range for normal subjects. In the patients with polycythaemia we were unable to predict the finding of normal or elevated siEp. Changes in siEp after erythrapheresis (10-26% reduction in packed cell volume) were observed in the 10 patients with polycythaemia and in one without. One month after erythrapheresis, packed cell volume remained below and siEp was above initial pretreatment levels, implying an erythropoietin secretory response and that the development of secondary polycythaemia had induced a fall in siEp.

Serum immunoreactive erythropoietin and erythropoiesis in protein-energy malnutrition

Immunoreactive erythropoietin was estimated in the sera of 23 Nigerian children with protein-energy malnutrition (PEM) and 14 healthy Nigerian children of similar age attending a well baby clinic. The geometric mean estimate for this parameter was 262 mIU/ml (observed range 39-1340 mIU/ml; 95% confidence range 25-1738 mIU/ml) in the children with PEM and 80 mIU/ml (observed range 43-257 mIU/ml; 95% confidence range 27-241 mIU/ml) in the health children. Erythropoietin levels were above the 95% confidence range for the healthy children in 14 of the cases of PEM. There was a statistically significant inverse correlation between the haemoglobin levels of the children with PEM and the logarithm of immunoreactive serum erythropoietin estimates (r = -0.73; P less than 0.001). By contrast, statistically significant correlations were not found between the logarithm of erythropoietin estimates and either the percentage of erythroblasts in the marrow, the M/E ratio or the logarithm of the absolute blood reticulocyte count. These data suggest that there is no abnormality of erythropoietin production in PEM and that the anaemia seen in this condition results from an impairment of erythropoiesis. A stepwise multiple regression analysis revealed a positive correlation between the logarithm of the erythropoietin level and the logarithm of the concentration of circulating neutrophil metamyelocytes plus myelocytes and we speculate on the aetiology of this finding.