Spontaneous death of rat chloroleukaemia cells induced by an endogenous growth inhibitor

Abstract

Objectives

When rat chloroleukaemia (CHL) cells are grown undisturbed in a confined space, a genomic long interspersed nuclear element (LINE) is transcriptionally activated at a relatively low population density, followed by the retrotransposition of LINE and population death. This death programme is fundamentally different from conventional cell death pathways.

Materials and methods

This work is essentially based on the re‐analysis of relevant, old experimental data. Elemental analysis of a highly purified, long‐stored inhibitor sample was performed. Genomic sequence searches were performed using the basic local alignment search tool (BLAST).

Results

This death programme is initiated by an endogenous inhibitor secreted by CHL cells. The inhibitor is almost certainly identical to the pentapeptide pyroGlu‐Glu‐Asp‐Cys‐Lys, shown to be a cell line‐specific inhibitor of normal granulocytic cells. The inhibitor is derived from a highly conserved short open reading frame in mammalian genomes.

Conclusions

Although spontaneous population death may be a biological oddity restricted to rat CHL cells, we suggest that this death programme is responsible for the eradication of cancer cells following treatment with an inhibitor administered exogenously.

1. INTRODUCTION

The rate of cell loss in tumours, especially carcinomas, is often only slightly less than that of cell production,1, 2, 3, 4, 5, 6 leading to the hypothesis that increasing apoptosis and cell loss factors to nearly 100% would provide stable tumours that patients could live with. However, cell loss and apoptosis are not synonymous, and although the idea of balanced cancer growth might work if apoptosis is the predominant mode of tumour cell loss, that does not appear to be the case.7 Furthermore, it is essential that the mechanism inducing apoptosis is selective of malignant cells and should not affect normal cells.

We realized that such a mechanism exists in one malignant cell line, rat chloroleukaemia (CHL) cells, in which complete or nearly complete cell loss is induced, and the predominant or only mode of cell loss is caused by a unique death programme. When CHL cells are allowed to grow undisturbed in a confined space, genomic long interspersed nuclear element (LINE) sequences are transcriptionally spontaneously activated at relatively low cellularity, later resulting in the retrotransposition of LINE sequences and population death.8, 9 These findings have been confirmed by Kirilyuk & Tolstonog et al.,10 who also identified a functional LINE retrotransposon from a pool of CHL cell cDNAs. They determined the activity of this retrotransposon in HeLa and CHL cells, demonstrating its mobilization in CHL cells.

From a philosophical viewpoint, evolution of a spontaneously initiated endogenous mortal programme in an immortal cell line is intriguing.8, 9, 10 First, all members of the cell population commit suicide in vitro, with no benefit to the population itself or to the host organism. Second, the mechanism of this programme is fundamentally different from conventional cell death pathways, such as apoptosis, autophagy and programmed necrosis.11, 12 Thus, the programme is initiated in vitro for no apparent reason or need, but the programme can be reverted by a simple exogenous action at any time before onset of the actual cell killing process. It is also relevant to note that reversion of the programme can be repeated an indefinite number of times.

We have re‐examined the spontaneous amplification of genomic LINE sequences and demonstrated that this process is triggered by an endogenous inhibitor of cell growth. The inhibitor is almost certainly chemically identical to the small growth inhibitory peptide pyroGlu‐Glu‐Asp‐Cys‐Lys, which has been isolated and identified from human blood granulocytes.13, 14 We also indicate that the pentapeptide appears to be derived from a short open reading frame (sORF) that is highly conserved in mammalian genomes.

2. MATERIALS AND METHOHDS

This work is essentially based on the re‐analysis of relevant, old experimental data. Elemental analysis (CHNS) of a long‐stored, highly purified sample of a proliferating granulocyte inhibitor,15 isolated by Weddel Pharmaceuticals from bovine blood granulocytes,16 was performed in the Labtium Jyväskylä laboratory, which is associated with the VTT Technical Research Centre of Finland. Genomic derivation of the inhibitor was accomplished by sequence searches using the basic local alignment search tool (BLAST).17

3. RESULTS

3.1. Growth curve analysis

Only one reasonably detailed CHL cell growth curve8 is analysed here, but the general pattern has been confirmed and extended numerous times by ourselves and by other groups.8, 9, 10

Mathematical modelling and growth curve analysis was inferred as described below. In an undisturbed culture in a confined space, CHL cell growth can be accurately simulated (correlation coefficient r = .9957) by the exponential function for approximately 50 hours (Figure 1). The population doubling time is 12.5 hours (in Figure 1, the exponential growth phase covers four population doublings). However, the CHL cell growth rate begins to decrease, and growth follows the Gompertz function18 until a sudden premature stop is displayed (Figure 2). Cell population growth in a confined space is commonly believed to slow when nutrient availability is limited. In the present case, however, the growth rate slows as an endogenous growth inhibitor, secreted by CHL cells, accumulates in the culture medium. As shown in Figure 2, LINE transcription is suddenly activated when a growth rate decrease is detected and follows the Gompertz function, indicating that growth inhibition, and thus accumulation of the inhibitor, precedes the activation of LINE transcription rather than vice versa.

Figure 1.

Chart lines (black and red) of CHL cells grown undisturbed in a suspension culture in a confined space. Both chart lines indicate two independent cultures at each time point. Initial growth is accurately simulated by the exponential function N = 2.1 exp(0.05557t) for four cell cycles (grey line; correlation coefficient r = .9957)

Figure 2.

Chart lines (black and red) of CHL cell growth as shown in Figure 1. After cell counts deviate from the exponential function, growth is simulated by the Gompertz function (grey line; N = 125 exp(‐24.5) exp(−0.05557t)) until the premature cessation of growth occurs due to retrotransposition of the LINE (red chart line). Expression of LINE transcripts at different cell count levels is shown (transcript blots are copies from figure 4 in Ref. 8). Note that the specific growth rates are equal in both Gompertz and exponential functions (c = 0.05557)

Medium exchange does not increase resources or stimulate population growth back to the initial exponential function, but rather only dilution of the population density restores the initial exponential growth rate.8, 9, 10 Furthermore, and most importantly, reduced cell counts revert the increased LINE transcription level8, 9, 10 back to baseline. The sudden increase in LINE transcription (approximately 5‐fold) occurs at or close to the Gompertz growth function inflection point (Figure 2), and reversion of the increased LINE transcription by reduced cell counts prevents onset of the actual cell killing process (Figure 2, time limit for population rescue is approximately 70 hours). Note that the medium exchange negates the inhibitor concentration only temporarily because high concentrations are quickly restored by the unaltered population cellularity.

An association between CHL cell population behaviour and growth inhibitor effects was also suggested by time‐lapse photography. Occasionally, a CHL cell was found to be blocked in the cell cycle, apparently in the G1 phase as judged by cell size. The blockage was reversible because the inhibited cell was shown to enter mitosis at a later time, and progenies of this cell proliferated normally at a fast rate. Death of the blocked cell was not observed during the exponential or Gompertz growth phases, but all blocked cells appeared dead approximately 2 days after reaching the LINE retrotransposition phase. These findings were most commonly observed in cultures wherein CHL cells were grown in a chemically defined medium because cells attached loosely to the bottom of the Petri dish in this medium and were thus easy to follow. Based on the cell morphology, both apoptotic and necrotic death mechanisms were present. Note that in vitro CHL cell cultures do not contain other cell lines, such as macrophages, which results in a mixed cytology of dying cells.

Detailed population kinetics of CHL cells are shown in Table 1,19, 20 and they suggest that the dominant growth feature allowing observation of the intriguing CHL cell behaviour is the absence of cell loss during the growth phase. When cell loss is present during the growth phases, long‐term in vitro CHL cell cultures fail.

Table 1.

Detailed kinetic analysis of CHL cell growth. The values in the year 1973 refer to diffusion chamber cultures in vivo,19 and the values in 1981 refer to suspension cultures in vitro.20 Note, in particular, differences in cell loss and growth fraction parameters

Parameter	1973	1981
Population doubling time (h)	24‐58	15.3
Potential doubling time (h)	21	15.3
Cell loss factor (%)	12‐64	0
Duration of cell cycle (h)	13.1 ± 4.2	15.5 ± 3.4
Duration of G1 phase (h)	3.3 ± 3.6	4.3 ± 3.0
Duration of S phase	7.7 ± 2.1	9.2 ± 1.5
Duration of G2 phase	2.1 ± 0.7	2.0 ± 0.5
Growth fraction	0.53	1.0
Pulse labelling index (%)	33	56
Mitotic index (%)	2.0	3.9

3.2. CHL cells also secrete the inhibitor in vivo

Secretion of an endogenous growth inhibitor by CHL cells is also displayed in vivo.21 Schematic representation of the experimental setup is illustrated in Figure 3, and a compilation of the main results is shown in Table 2. CHL tumours grown locally secrete an inhibitor, which inhibits the growth of target cells in diffusion chambers at distant locations, and the effect appears to be cell line‐specific. As expected, high concentrations of the inhibitor are also observable in the sera of CHL rats.22

Figure 3.

Schematic representation of the assay setup. Target cells are grown in diffusion chambers implanted in the peritoneal cavity of host rats

Table 2.

Effect of an inhibitor, produced by a CHL tumour growing locally in a rat, on different target cells grown in diffusion chambers.21 Growth of the target cells was estimated from cell counts in tumour‐bearing rats compared to those in controls and from pulse labelling by³H‐thymidine (M ± SE). Control animals were normal rats or rats in which the tumour regressed spontaneously or was surgically removed (Figure 3). N represents the number of animal pairs studied. All effects shown, except those of non‐granulocytic cells, are statistically significant

Target cells	Cell count		3H‐TdR uptake
	N	CHL rat, % of control	N	CHL rat, % of control
Chloroleukaemia cells	30	72.1 ± 8.6	10	60.5 ± 11.2
Proliferating normal granulocytic cells (rat and mouse bone marrow)	38	62.5 ± 13.3	35	50.4 ± 6.3
Non‐granulocytic cells (mastocytoma and HeLa)	21	111.2 ± 13.7	21	98.3 ± 12.9

3.3. Chemical nature of the inhibitor

The chemical nature of the inhibitor remains a crucial concern. Repeated short‐term extractions of normal granulocytic cell populations (production of conditioned medium by incubating intact cells in physiological saline or Hanks’ balanced salt solution) have served as the main mechanism for generating starting material for isolating and studying growth inhibitors of normal granulocytic cells,15 and this procedure has also been used in studies involving CHL cells.23, 24

Conditioned media obtained from CHL cell suspensions have been tested in different assay systems together with similar extracts obtained from normal granulocytes.21, 22, 23, 24 The biological effects of an inhibitor obtained from both cell types were qualitatively identical and cell line‐specific, but direct comparisons of the chemical properties of the two inhibitors were never made beyond observations of the inhibitor elution volumes during gel filtration. Inhibitors from both sources had a relative elution volume of V_e/V₀ = 2.73 on Sephadex G‐7524 and an identical relative elution volume on Sephadex G‐25 (V_e/V₀ = 2.05; unpublished result). Notably, both the inhibitor isolated by Paukovits25 and our inhibitor behave identically during gel filtration in our assay system.26

Thus, the inhibitor is almost certainly identical to the small peptide pyroGlu‐Glu‐Asp‐Cys‐Lys, which had been isolated and identified from human blood granulocytes.13, 14 CHNS analysis of our highly purified inhibitor isolated from bovine blood granulocytes is shown in Table 3.16 This preparation was stored in a plastic vial at room temperature for 37 years prior to analysis. The elemental composition calculated for the pure pentapeptide is also shown in Table 3. The preparation made by Weddel Pharmaceuticals was not completely pure, and most of the cysteine moieties presumably decomposed during the 37 years of storage. Furthermore, the fact that the Weddel preparation was eluted on a Sephadex G‐10 column with a relative elution volume of V_e/V₀ = 1.1716 indicates that the inhibitor weighs less than 700 daltons (the molecular weight of the identified pentapeptide is 604 daltons). Thus, the likelihood of the two inhibitors being identical is high.

Table 3.

Elemental composition of a preparation containing a semi‐purified granulocytic inhibitor. The composition was measured from a semi‐purified sample (Weddel preparation) obtained by repeated 1‐hour incubations of intact bovine blood granulocytes in Hanks’ balanced salt solution.16 The elemental composition of the pentapeptide (pyroGlu‐Glu‐Asp‐Cys‐Lys; C₂₃H₃₆N₆O₁₁S), isolated and identified from human blood granulocytes,13, 14 is shown for comparison. The compositions of both preparations are scaled to the same level with respect to nitrogen. The low sulphur content of the Weddel preparation is believed to be artificial, caused by the decomposition of cysteine during 37 years of storage

Element	Weddel preparation weight %	Pentapeptide weight %
Carbon	26.1	23.0
Hydrogen	3.9	3.0
Nitrogen	7.0	7.0
Sulphur	0.3	2.5

3.4. Generation of the inhibitor

BLAST searches for non‐redundant human proteins revealed that only one canonical human protein, phospholipase C, contains the peptide segment QEDCK. However, a granulocyte‐specific growth inhibitor being a phospholipase C cleavage product is highly unlikely. Recent innovations in molecular biology have revealed a novel class of small bioactive regulatory peptides that derive from sORFs,27, 28, 29, 30, 31 and the present inhibitor could presumably be a member of this class. BLAST searches were successful, and a highly conserved 60‐nucleotide sORF was identified (Table 4). Conservation of a sequence across species indicates that the sequence has functional value and has been maintained through evolution despite speciation. The probability of finding a sequence by chance in the current database of mammalian genomes that is identical to the consensus sORF (Table 4) is practically zero (e value is 1e‐20), and the probability of finding a single sequence with 85% identity to the consensus sequence by chance is approximately 5e‐10 (ie, 0.0000000005). Because sequences at the 85% identity level were found in more than 50 species in the current mammalian database, the identified sORF is not random but rather a highly conserved sequence.

Table 4.

Highly conserved short open reading frames (sORFs) in mammalian genomes presumably encoding the precursor peptide. The segment encoding the mature inhibitor is shown in bold. Sequences are given for ten selected species that span the mammalian evolutionary tree (the divergence time for elephantulus and humans is 105 million years44). The presumed consensus sequence and putative peptide are also indicated. The probable start signal of the precursor sequence is shown in bold

Species	Genomic sequence of precursor peptide (conserved sORF)
Elephantulus	ATACTCTTGCATGTAAAGATGAAAATGTATCCAAAGGAAATACAGGAAGATTGCAAGTAA
Armadillo	ATCCTCAACTATGTAAGGATGGAAATTAATCCAAAGGAAAAGCAGGAAGATTGCAAGTAA
Cachelot	ATGCCCACTGTTTGTAAGATTAAAAATGCATCCAAGGAAAAACAGGAAGATTGCAAGTAA
Horse	GTGCCCAATGTATGCAAGAATGAAAATGTCTCCAAGGAGAAACAGGAAGATTGCAAGTAA
Cat	CCGATGTATGTATGTAAGAATGAAAATGTATCCAGGGAAAAACAGGAAGATTGCAAGTAA
Polar bear	ATGCCCAATGCATGTAAGAATGAAAATGTATCCAAGGAAAAACAGGAAGATTGCAAGTAA
Mouse	ATGCCCAATGCATGTAAGAATGAAAATGTATCCAAGGAAAAACAGGAAGATTGCAAGTAA
Hamster	ATTGCTAATGTATGTAAAGATGAAAATGTATACAAGGAAAAACAGGAAGATTGCAAGTAA
Tupaia	ATCCCCAATGTATCTGAGGATGAAAATGTATCCAAGAAAAAACAGGAAGATTGCAAGTAA
Humana	ATCATCAATGTATGTAAGGATGAAAATGTATCCAAGGAAAAACAGGAAGATTGTcAGTAA
CONSENSUS	ATGCCCAATGTATGTAAGAATGAAAATGTATCCAAGGAAAAACAGGAAGATTGCAAGTAA
PUTATIVE PEPTIDE	M P N V C K N E N V S K E K Q E D C K *

^aBecause of a possibly critical base change in the human DNA sequence (small C) in which K is mutated to Q in the inhibitor sequence, great apes may have a conserved substitutive sORF that encodes an intact inhibitor. This sORF was previously identified.45

Besides being derived from a conserved sORF, characteristic features of the putative precursor peptide include the location of the QEDCK segment at the C‐terminus of the precursor and the common presence of lysine as the P1 amino acid preceding glutamine (Table 4). Note that after apparent cleavage of the precursor peptide at this site (between K and Q), glutamine becomes the N‐terminal amino acid and can then spontaneously convert to pyroglutamate. Table 4 shows ten examples of genomic sequences putatively encoding the precursor peptide; animal species were selected to span the mammalian evolutionary tree. As shown in Table 4, non‐ATG sequences serve as initiation codons of the precursor sequence in many species, which appears to be a common property in many established sORFs.31

It was later discovered that also a 284 nucleotides long ultraconserved mammalian element32 contains another probable sORF (plus strand on human chromosome 8, complement strand to exostosin‐1 gene), which has the QEDCK segment at the C‐terminus of the putative precursor peptide. Needleman‐Wunsch alignment (BLAST) of the two putative peptides (MPNVCKNENVSKEKQEDCK* and IAFFGLHVSCLGQEDCK*) gives the NW score 10 with identities 40% (positives 50%).

4. DISCUSSION

The findings described herein demonstrate that rat chloroleukaemia cells spontaneously secrete a granulocyte‐specific growth inhibitor both in vitro and in vivo. The inhibitor released by CHL cells grown undisturbed in a suspension culture in vitro induces transcriptional activation and the subsequent retrotransposition of LINE, followed by CHL population death. The concentration of the inhibitor required to induce LINE sequence amplification is not known, but it may be relatively low based on the reasons discussed below. Synthetic pentapeptide inhibits granulocytic colony formation by approximately 40% at a concentration of 10⁻⁸ M in vitro,13, 14, 33 and conditioned medium obtained from approximately 10⁷ normal granulocytes after one hour of incubation in saline yields a preparation capable of inhibiting proliferating granulocytic precursor cells by 20‐40% in vitro.15 This suggests that the inhibitor concentration in normal granulocyte conditioned medium is approximately 10⁻⁸ M and, consequently, that the rate of inhibitor secretion is approximately 200 molecules/second/cell. If CHL cells secrete the inhibitor at a similar rate and if the inhibitor is short‐lived, the inhibitor concentration that activates LINE transcription in CHL suspension cell cultures might also be approximately 10⁻⁸ M. The cell count is doubled at the time of LINE retrotransposition and CHL cell death, and the inhibitor concentration is thus also presumably doubled. Nevertheless, the lethal concentration of the inhibitor in vitro appears surprisingly low. If these estimates are not fully erroneous, they hint that the CHL population might also be killed in vivo by the inhibitor added exogenously at reasonable concentrations. Support for this tempting viewpoint is provided by experiments in which the prolongation of life, and even a cure in some chloroleukaemic rats, was obtained by the exogenous administration of inhibitors,34, 35, 36 including the synthetic pentapeptide.36

The response of CHL cells to growth inhibitor‐induced LINE activation and subsequent cell death may simply be a curious biological phenomenon with no general meaning. However, recent reports indicate that this is not the case. For example, human retrotransposition was shown to be recognized by cells as a “genetic damaging event”, and massive retrotransposition was shown to trigger signalling pathways resulting in apoptosis.37 Consistent with this, the LINE‐1 retrotransposon reportedly induces apoptosis in human cancer cells,38, 39 and the cellular response to LINE‐induced DNA damage usually manifests as cell cycle arrest and cell death (apoptosis or necrosis).40 Finally, some de novo insertions of LINE in neuronal progenitor cells may induce apoptosis, and they may be responsible for the massive cell death that occurs during neurodevelopment.41

Considering these findings, the observed rapid regression of leukaemic cell populations in human patients treated with a semi‐purified inhibitor16, 42, 43 may be associated with LINE‐induced cell death, and the rate of leukaemic cell death increases after the onset of inhibitor treatment. For instance, the total leukaemic mass was reduced by an estimated 3 kg in 2 weeks in one patient. In general, the rate of leukaemic cell count reduction is similar to that observed after conventional drug therapy.42 At the time of that study,42 the immune system was suggested to play a leading role in the eradication of leukaemic cells, but it now appears that a more plausible mechanism is involved in the currently elucidated LINE retrotransposon‐associated death pathway.

CONFLICTS OF INTEREST

Potential conflicts of interest relevant to this article do not exist.

Rytömaa T, Grip‐Rytömaa K. Spontaneous death of rat chloroleukaemia cells induced by an endogenous growth inhibitor. Cell Prolif. 2018;51:e12421 10.1111/cpr.12421